The World Bank as an Enforcer of Human Rights∗
Kelebogile Zvobgo†Benjamin A.T. Graham‡
Journal of Human Rights pre-publication version
Since the 1990s, the World Bank’s Inspection Panel and Compliance Advisor/Ombudsman (CAO)
have responded to hundreds of human rights complaints ﬁled by or on behalf of project-aﬀected
communities. Yet, little is known about complaint outcomes and factors that enhance the likelihood
of complaint success. We theorize that complaints involving indigenous communities, or pertaining to
involuntary resettlement and severe environmental impacts, will be more likely to result in a favorable
outcome due to the Bank’s novel policies governing these issues. We also expect that NGO-supported
communities are more likely to succeed due to organizational capacity, expertise, and advocacy. We
evaluate these expectations using an original dataset of Inspection Panel and CAO complaints (1993-
2017). We ﬁnd strong support for our expectations about indigenous communities, weak support for
projects involving involuntary resettlement, and no support for environmentally-risky projects. NGO
support is strongly associated with complaint success, may be principally due to screening.
∗We wish to thank Claire Bond, Kaitlyn King, Sarah Lu, Amanda Scott, Markus Sherman, J. Benjamin Wu, and
Madeline Zheng for excellent research assistance. For their invaluable advice and comments on earlier drafts, we are
grateful to Miriam Barnum, Karen Brown, Mark Buntaine, Suparna Chaudhry, Feryal Cherif, Gabrielle Cheung, Andrew Coe,
Shareen Hertel, Ellen Kim, Suzie Mulesky, Rebecca Sanders, Anita Weiss, and two anonymous reviewers for The Journal of
Human Rights. Previous versions of this article were presented at Pomona College, the University of Southern California,
the 2017 Annual Meeting of the American Political Science Association, the 2017 International Studies Association-West
Annual Conference, the 2018 Annual Conference on the Political Economy of International Organizations, and the 2018
International Studies Association Annual Convention. This work is supported by the University of Southern California’s
Provost Fellowship in the Social Sciences. In addition, this material is based upon work supported by the National
Science Foundation Graduate Research Fellowship Program under Grant No. DGE-1418060. Any opinions, ﬁndings, and
conclusions or recommendations expressed in this material are those of the authors and do not necessarily reﬂect the views
of the National Science Foundation. Both authors contributed equally to this article.
†Dept. of Political Science and International Relations, University of Southern California; Global Research Institute,
William & Mary, firstname.lastname@example.org.
‡Dept. of Political Science and International Relations, University of Southern California, email@example.com.
The number, strength, and scope of international legal instruments, compliance monitors, and courts have
grown considerably since the end of the Second World War (Koremenos, 2016; von Stein, 2018; Hafner-
Burton, Mosley and Galantucci, 2019). Human rights courts, in particular, have become increasingly
powerful and accessible due to “new-style” design features, such as compulsory jurisdiction and private
litigant initiation rights (Alter, 2014; Haddad, 2012). Yet, accountability for human rights violations
and other harms has not been, and can not be, solely accomplished through judicial institutions (Lupu,
2013). Access to formal courts is limited by a range of hurdles, including standing requirements and
Over the past four decades, quasi-judicial bodies have proliferated to help ﬁll the accountability gap
left by courts, including in the area of human rights (Orentlicher, 2007; Villa-Vicencio, 2000; Zvobgo,
2019, 2020). These mechanisms have jointly received thousands of complaints and often deliver signiﬁ-
cant relief to complainants, yet we lack a ﬁrm understanding of their performance. Notably, scholarship
has yet to conduct a systematic quantitative evaluation of two of the most high-proﬁle and pioneering
of these quasi-judicial bodies, the World Bank’s Inspection Panel and Compliance Advisor/Ombudsman
(CAO), which investigate human rights violations and other harms associated with international de-
velopment projects. In this paper, we explore the extent to which the World Bank, through these
accountability mechanisms, is responsive to complaints about the adverse impacts of its projects at the
local level. Understanding responsiveness at the local level helps shed light on human rights gains at the
Since their creation, the Inspection Panel and CAO have heard nearly 250 complaints, extending
international administrative law into development ﬁnance. Through the Inspection Panel and CAO,
the World Bank has deﬁnitively established that “international organizations have a legally signiﬁcant
contractual relationship with private parties that is independent of [...] a member state” (Bradlow, 1993,
554). These bodies have been widely copied by other international ﬁnancial institutions (IFIs): similar
accountability mechanisms now operate in at least 15 other development banks, including the Asian
Development Bank (ADB), the European Investment Bank (EIB), and the Inter-American Development
Bank (IDB) (Accountability Counsel, 2015; Gutner, 1999; Nanwani, 2014). However, until now, no
holistic assessment exists regarding the types of complaints brought before them, the conditions under
which complainants prevail, or the types of remedies that complainants receive. In short, these quasi-
judicial bodies are being widely replicated with little knowledge of whether or not they function as
The Inspection Panel and CAO are sometimes cited as evidence of the World Bank’s increasingly
rights-based approach to development and poverty reduction (Okogbule, 2011; Sano, 2000). Scholars
have also suggested that the Inspection Panel and CAO enhance transparency, promote accountability,
and represent important developments in international administrative law (Benvenisti and Downs, 2009;
Bradlow, 1993; Buntaine, 2015; Tignino, 2016). Indeed, the World Bank was the ﬁrst international
organization (IO) to eﬀectively tie its hands, submitting to the evaluations of two independent panels of
experts, as well as to external monitoring by civil society actors who can ﬁle complaints with or on behalf
of project-aﬀected communities. And, as our research demonstrates, vulnerable populations can and do
ﬁnd relief and redress through both the Inspection Panel and CAO.
The World Bank has had a long and complicated history with projects involving indigenous peoples,
involuntary resettlement, and severe environmental impacts. Following intense criticism from the human
rights community in the 1960s and 1970s, the World Bank developed novel safeguard policies to govern
projects that (1) involve indigenous peoples (Operational Directive 4.20), (2) cause involuntary resettle-
ment (Operational Directive 4.30), and (3) pose the highest risk to the environment (Operational Policy
4.01). As elaborated below, these policies represent what we might understand as the Bank’s strongest
‘governing law’ and have helped to set the global standard for governance of development ﬁnance (Fox,
2000; Sarfaty, 2004/2005). Certainly, the Bank’s broader policy framework covers additional issues, for
example labor and working conditions in project sites, community access to information, corruption, etc.
However, the policies pertaining to indigenous communities, involuntary resettlement, and high environ-
mental risk are much more stringent and provide communities additional protections and entitlements
(Fox and Brown, 1998; Sarfaty, 2004/2005). Moreover, these policies were catalyzed by civil society
advocates and donor governments who demanded the Bank take these issues more seriously, and these
actors continue to monitor closely the Bank’s compliance with these policies (Accountability Counsel,
2015; Buntaine, 2015). Thus, we theorize that complaints covering indigenous communities, involuntary
resettlement, and severe environmental impacts will be more successful relative to all other complaints
because: (1) the stringency of policy in these areas makes violations that require redress more likely, (2)
there exists a more explicit framework in which communities and their representatives can couch their
complaints, and (3) the Bank has faced and continues to face intense scrutiny and pressure around these
issues. Other issues that lead to complaints but are not targeted by speciﬁc operational directives include
labor rights violations and economic and cultural damage (See Figure 3).
Similarly, the Bank has had both a conﬂictual and cooperative relationship with non-governmental
organizations (NGOs) (Clark, Fox and Treakle, 2003; Lavelle, 2011). NGO activism played an important
role in the creation of the Inspection Panel and CAO and has contributed to the Bank’s broadening
and strengthening of its safeguard policies (Mallaby, 2006; Shihata, 1994; Woods, 2000; Ziai, 2016).
As in international courts and other quasi-judicial bodies, NGOs monitor the World Bank by ﬁling
Inspection Panel and CAO complaints with or on behalf of project-aﬀected communities (Buntaine,
2015; Uçarer, 2014). And, we expect that NGO involvement is strongly associated with complaint
success. Project-aﬀected communities often lack the capacity to eﬀectively ﬁle a formal complaint with
an IO like the World Bank and, indeed, many communities may not even know that the Inspection
Panel and CAO exist (Hunter, 2003). In contrast, NGOs possess the knowledge, professional expertise,
and organizational capacity to be eﬀective co-complainants. We, thus, theorize that NGO involvement
increases the likelihood of complaint success. This is likely to happen via two mechanisms: complaint
quality and external advocacy. In addition to possessing technical expertise and administrative capacity
useful for ﬁling high-quality complaints, NGOs also enjoy the relationships and public proﬁle to advocate
eﬀectively outside of the complaint process.
We introduce a new comprehensive dataset covering the universe of Inspection Panel and CAO cases,
from their inception through the spring of 2017. The primary empirical contribution of this paper is
descriptive: we oﬀer the ﬁrst holistic portrait of these bodies’ performance. Our data cover: (1) the
types of actors that bring complaints, (2) the issues raised in complaints (and the types of projects that
give rise to them), and (3) the outcomes of complaints and the remedies provided to communities who
ﬁle successful complaints. Our data also allow us to our test theory regarding the factors that inﬂuence
Descriptively, we ﬁnd that environmental damage, economic damage, and involuntary resettlement
are the most-frequently raised issues in Inspection Panel and CAO cases, and formal acknowledgements
of harm, project change or termination, and compensation are the most-frequently provided remedies.
Roughly 50% of complaints garner at least one positive outcome. However, punishment of perpetrators,
e.g., host country governments, project partners, or contractors, is exceedingly rare.
Consistent with our theoretical expectations, we ﬁnd that indigenous project-aﬀected communities
are more likely to receive a formal acknowledgment of harm and secure rulings for project change, though
they are not more likely to obtain judgments for compensation than other complainants. We also ﬁnd
that complaints alleging involuntary resettlement are more likely to result in an order for compensation,
but not in harm acknowledgment or project change. Contrary to expectation, we ﬁnd no evidence that
complaints alleging environmental harm or complaints ﬁled with respect to projects deemed especiallyT
environmentally sensitive are more likely to result in a favorable outcome.
We also ﬁnd that NGO involvement is a strong predictor of rulings for project change and compen-
sation for project-aﬀected communities, though NGO involvement does not enhance the likelihood of
communities receiving a formal acknowledgment of harm. These correlations are consistent with our
expectation that NGOs capitalize on their expertise in information, leverage, and accountability politics
to gain remedies for communities. However, in our interviews, NGO representatives also conﬁrm our
expectation that NGOs screen complaints, limiting their support to complaints that they think credi-
bly assert violations of the Bank’s social and environmental safeguard policies. Thus, at least part of
the positive relationship we observe between NGO involvement and complaint success is driven by this
To help assess the extent to which the relationship we observe between NGO involvement and
complaint success is driven by NGO expertise and capacity, we conduct a mediation analysis of NGOs’
eﬀect on success via complaint quality. We expect that a core mechanism by which NGO expertise
and capacity lead to success is the ﬁling of higher quality complaints. We ﬁrst estimate the eﬀect of
NGO involvement on the fulﬁllment of “complaint fundamentals”—a latent variable that we construct to
capture the basic elements that we expect contribute to success. We then estimate the eﬀect of complaint
fundamentals on complaint success. As expected, we ﬁnd that NGO involvement is positively correlated
with complaint fundamentals. However, the mediated eﬀect is not statistically signiﬁcant. Thus, the
eﬀect of NGOs on complaint success occurs primarily through other channels—through assistance not
reﬂected in the observable quality of the written complaint. Our interviews suggest that screening and
external lobbying are inﬂuential for complaint success.
This paper contributes both theoretically and empirically to the scholarship on international courts
and quasi-judicial bodies, as well as the business and human rights literature on access to remedy.
The Inspection Panel and CAO were created contemporaneously with several “new-style” international
courts and possess many similar features, notably delegated compliance review and enforcement powers,
and case initiation by private persons and organizations. Since project-aﬀected communities do not
have standing in any court to sue the World Bank for the ﬁnancing and execution of its projects, the
Inspection Panel and CAO were and remain their only formal venue for recourse. Therefore, these bodies
deserve to be included in debates on enforcing human rights and increasing access to justice. In addition,
we provide the ﬁrst comprehensive quantitative analysis of quasi-judicial bodies at the intersection of
economic development and human rights, and bring them into ongoing debates around the “protect,
respect, remedy” principles for business and human rights. This framework has been applied almost
exclusively to states and corporations (Davitti, 2016). We extend the framework to encompass IOs,
which, as the World Bank has established, are also duty-bound to protect, respect, and remedy harms
experienced by private individuals and communities in and around operational sites.
2 The Inspection Panel and CAO: Novel Quasi-Judicial Bodies for Enforcing Hu-
For decades, the human rights community rebuked the World Bank’s ﬁnancing and execution of devel-
opment projects, namely infrastructure and extractive projects that damaged local environments and dis-
placed populations, including indigenous communities. (Fox, 2000; Gutner, 1999; Rich, 1994). Scrutiny
and pressure from transnational advocacy groups and donor governments came to a head in the 1980s,
culminating in the Bank developing policies to to prevent, mitigate, and provide redress for social and
environmental harms (Clark, 2002; Mallaby, 2006; Nielson and Tierney, 2003; Shihata, 1994; Woods,
2000; Ziai, 2016). Continuing censure induced the creation of the Inspection Panel and CAO in the
1990s—two novel internal accountability mechanisms designed to respond to complaints made by per-
sons harmed or likely to be harmed by Bank-funded projects (Carrasco and Guernsey, 2008; Mallaby,
2006; Ziai, 2016).
The Inspection Panel was established to respond to complaints regarding projects funded through
the International Bank for Reconstruction and Development (IBRD) and the International Development
Association (IDA); that is, public sector projects. And, the CAO was created to respond to similar
complaints regarding projects funded through the International Finance Corporation (IFC) and guaranteed
through the Multilateral Investment Guarantee Agency (MIGA); that is, private sector projects. Since
1993 and 1999, respectively, the Inspection Panel and CAO have provided a venue for project-aﬀected
communities to communicate the harms they experience, gain public acknowledgment of these harms,
and receive relief and compensation. There is also evidence that they shape the Bank’s lending behavior
(Buntaine, 2015). However, until now, the overall eﬀectiveness of these bodies in granting relief and
redress to project-aﬀected communities has been diﬃcult to assess in the absence of comprehensive data
on their performance.
2.1 Quasi-Judicial Bodies and Development Finance
The Inspection Panel and CAO are a novel addition to a long list of quasi-judicial bodies with “a mandate
to monitor compliance with a body of norms, settle disputes regarding those norms, or make determina-
tions on the basis of investigations” (Tignino, 2016, 242). Other such bodies include truth commissions
which inquire into historical political violence, the Aarhus Compliance Committee which assesses inter-
national environmental compliance, and the Committee on Economic, Social and Cultural Rights which
oversees implementation of the 1966 Covenant (Bissell, 1997; Hayner, 2000; Minow, 1998; Tignino,
Each of the major regional development banks1has attempted to replicate the Inspection Panel and
CAO, and single-nation based development banks2have followed suit. The proliferation of institutional
“copies” raises the importance of evaluating the “originals”. Our ﬁndings can serve as a baseline against
which scholars can ascertain gains and losses associated with institutional replication. And, the strengths
and weaknesses we identify can inform the reﬁnement of analogous bodies.
2.2 Quasi-Judicial Bodies in an Era of “New-Style” International Courts
Though quasi-judicial in nature, the Inspection Panel and CAO resemble in many ways “new-style”
international courts (ICs). Indeed, the Inspection Panel and CAO were created contemporaneously
with a wave of new-style ICs, including the Andean Tribunal of Justice, the European Free Trade Area
Court, and the International Criminal Court. Alter argues that, “New-style ICs reﬂect the reality that
states have tasked ICs with helping to enforce international law, and with reviewing the application of
international law by state and international administrative and legislative actors” (2014, 7). The World
Bank has similarly tasked the Inspection Panel and CAO with monitoring compliance with its social and
environmental safeguard policies and enjoining appropriate reparative acts in cases of noncompliance.
Among the most important characteristics of new-style ICs is the right of private persons and organi-
zations to initiate litigation. This is in contrast to older ICs where the right to initiate litigation is enjoyed
exclusively by states. Scholars argue that private actor access increases courts’ independence and activity
(Alter, 2014; Haddad, 2012, 2018; Tallberg et al., 2014). With the creation of the Inspection Panel and
CAO, the World Bank became the ﬁrst IO to create a forum for private actors to hold IOs accountable for
harms created by their operations—a signiﬁcant development in international administrative law (Hunter,
2003; Macdonald and Miller-Dawkins, 2015; Nanwani, 2014; Keck and Sikkink, 1998). The Bank has
empowered these two bodies to make decisions that host country governments, and even the Bank itself,
may dislike but that project-aﬀected communities and their advocates may prefer.
2.3 Limits of the Inspection Panel and CAO
Despite newly-developed formal access to IOs via the Inspection Panel and CAO, real access is limited
(Buntaine, 2015). Worldwide knowledge of the bodies is low and the bodies themselves have failed
to seek out project-aﬀected communities and advertise their dispute resolution and compliance review
functions (Accountability Counsel, 2015; Hunter, 2003). In addition, scholarship problematizes the In-
spection Panel’s inability to grant and oversee relief to project-aﬀected communities independent of Bank
Management and the Board of Executive Directors (Carrasco and Guernsey, 2008). This contrasts the
CAO which, in addition to investigating noncompliance and issuing recommendations, may independently
facilitate dispute resolution between the relevant parties and supervise the implementation of compliance.
Nevertheless, as elaborated in the analysis, the work of both the Inspection Panel and CAO has led to
formal acknowledgments of harm, orders for project change, and compensation.
3 Drivers of Complaint Success
This paper provides a systematic empirical evaluation of the Inspection Panel and CAO’s performance,
including an examination of the relief and compensation mandated, and the conditions under which
complainants are most likely to be successful. We begin our investigation by developing expectations
about the types of communities and complaints that are most likely to prevail. Speciﬁcally, we expect
that complaints covered by the Bank’s most stringent safeguard policies—that is, complaints related to
indigenous communities, involuntary resettlement, and severe environmental impacts—will be more likely
to gain favorable outcomes from the Inspection Panel and CAO.
3.1 Stringent Safeguards
The World Bank was the ﬁrst development ﬁnance organization to establish a set of policies to prevent,
mitigate, and provide redress for social and environmental impacts created by its projects. Chief among
these are the Bank’s policies on indigenous peoples, involuntary resettlement, and projects that pose se-
vere environmental impacts (Mallaby, 2006; Nanwani, 2014; Sarfaty, 2004/2005; Shihata, 1994; Woods,
2000; Ziai, 2016). This innovation was prompted by adverse impacts of large-scale infrastructure and
extractive projects, and proceeded to catalyze the development of similar policies across several develop-
ment banks, including the ADB, EIB, and IDB. These rules expand the scope of possible violations, and,
by extension, increase the likelihood of meritorious complaints and favorable outcomes for complainants.
3.1.1 Indigenous Communities
Following criticism regarding projects’ adverse impacts on indigenous communities in the 1960s and 1970s,
the Bank developed operational policies on the rights and entitlements of indigenous project-aﬀected
communities. In 1982, the Bank developed Operational Manual Statement 2.34, a novel framework to:
(1) recognize and protect indigenous people’s land rights and (2) provide essential services to indigenous
peoples, including health and education. In the following years, the Bank hired a team of specialists
in indigenous rights and, coordinating with the International Labor Organization, revised its policy on
indigenous peoples. The revised policy, Operational Directive 4.20, broadened the Bank’s deﬁnition of
indigenous peoples, expanding protections to communities not formally recognized as indigenous by their
governments. According to OD 4.20, any person or community that claims to be part of an indigenous
group is so considered by the Bank. Unlike in the past, these individuals and communities do not have to
be formally recognized by their states. This has alleviated the problem of powerful actors not recognizing
these communities in an eﬀort to deny their rights and entitlements (World Bank, 2011).
Most recently adopted is Operational Policy (OP) 4.10, the World Bank’s hallmark policy on in-
digenous peoples, which holds the Bank to a higher standard when operating in lands and territories
traditionally owned, customarily used, and/or occupied by indigenous peoples. Bank-funded projects
operating in these contexts require, among other things: (1) a culturally appropriate community consul-
tation and consensus evaluation, (2) equitable sharing of project-related proﬁts, and (3) identiﬁcation
of alternative project designs which avoid or mitigate harm, including physical relocation of communi-
ties (World Bank, 2013b). For clarity, each set of policies described represents an additional layer of
protection for indigenous communities relative to other communities.
3.1.2 Involuntary Resettlement
Historically, involuntary resettlement has been among the Bank’s best-documented failures (Clark, 2002;
Webber and McDonald, 2004). Displacement causes economic harm, namely the loss of shelter, assets,
and income (Shihata, 1994). It can also create physical and psychic harms, such as “loss of identity,
culture, and customary livelihoods” (World Bank, 2013b). Thus, it has been among the priorities for
global civil society actors and donor governments seeking to regulate governance of development ﬁnance.
Similar to the Bank’s indigenous peoples policies, OD 4.30 holds the Bank to a very high standard for
projects that may involve involuntary resettlement.
In general, involuntary resettlement is to be avoided. However, when it cannot be avoided, Bank
policy requires that resettlement is minimized and communities indemniﬁed. More concretely, OD 4.30
requires that communities receive relocation assistance and residential housing, or be granted sites to set
up their own houses and agricultural systems. In addition, displaced communities’ prior standard of living
must be restored, if not improved, in their new location. Moreover, OD 4.30 requires that displaced
persons share in project-generated beneﬁts(World Bank, 1990). Of note, project-displaced communities
are the only non-indigenous communities who are entitled to such beneﬁts under Bank policy.
3.1.3 Severe Environmental Impacts
Finally, perhaps the most-publicized Bank failure is environmental damage generated by certain projects
(Gutner, 1999; Rich, 1994; Shihata, 1994). For decades, transnational advocacy groups documented
severe environmental impacts of infrastructure projects—for example, the Sardar Sarovar projects on the
Narmada River in India and the Bíobío hydroelectric project in Chile. These project failures reinforced the
Bank’s poor reputation for human rights (Clark, Fox and Treakle, 2003; Lavelle, 2011). In response, the
Bank developed environmental safeguards, culminating in its environmental assessment policy, OP 4.01
in 1989. At the heart of OP 4.01 is a system of environmental screening that the Bank undertakes to:
(1) evaluate potential risks to the environment, (2) identify possible improvements to project selection,
design, and implementation such that adverse environmental impacts are either prevented, minimized, or
mitigated, and (3) map a plan to compensate communities for adverse impacts.
At the design stage, World Bank projects are placed into one of four environmental categories—A,
B, C, or FI—depending on the nature and scale of the project, its location, the nature of the risks posed,
and the predicted severity of environmental impacts. Category A captures those projects that are “likely
to have signiﬁcant adverse impacts that are sensitive, diverse, or unprecedented” (World Bank, 2013a).
Relative to Category B, C, or FI projects, Category A projects require a full, comprehensive environmental
impact assessment (EIA). This involves consulting independent experts to: (1) collect detailed data on
baseline site conditions, (2) analyze the costs and beneﬁts of the project as currently designed, and (3)
recommend feasible alternatives. Category A projects additionally require timely, prior consultations with
communities and local NGOs, as well as throughout the project life-cycle. There are many ways OP 4.01
can be violated and violations addressed by the Inspection Panel and CAO. We illustrate just two such
ways. First, if a full EIA is not executed for a project, complainants may secure a judgment for project
change. Second, if environmental impacts are not adequately prevented or mitigated for an especially
risky project, complainants may secure an order for compensation.
3.1.4 Stringent Safeguards Hypotheses
Collectively, these policies place stringent demands on the Bank when interfacing with indigenous com-
munities, executing projects that force resettlement, and implementing projects predicted to have severe
environmental impacts. All else equal, it should be more diﬃcult for projects to meet these elevated
standards. Thus, we expect complaints pertaining to indigenous communities, involuntary resettlement,
and severe environmental impacts to be more successful, ﬁrst because the stringency of policy in this area
makes violations requiring redress more likely and, second, because there exists a more explicit framework
in which communities and their representatives may couch complaints.
As we show in our descriptive statistics, 23% of complaints were submitted by indigenous communities,
25% allege involuntary resettlement, and 47% involve projects with a high predicted environmental
impact. Some complaints fall into two or all three of these categories, and 33% fall into none. We
expect that each of these three factors has an independent and positive eﬀect on the probability of
Hypothesis 1a Complaints submitted by or on behalf of indigenous communities are more likely to be
Hypothesis 1b Complaints alleging involuntary resettlement are more likely to be successful.
Hypothesis 1c Complaints about projects with a predicted high environmental impact are more likely
to be successful.
3.2 NGO Advocacy in the Complaint Process
In addition to developing expectations about the types of communities and complaints that are most
likely to prevail, we develop and test theory about the inﬂuence of NGOs in the complaint process.
We extend Alter’s (2014) altered politics framework, moving from the domain of international courts
to international quasi-judicial bodies, to generate expectations about the role of NGOs. Similar to the
courts that Alter (2014) evaluates, the Inspection Panel and CAO enforce compliance with Bank policy
through alliances with domestic and international actors. The Inspection Panel and CAO’s compliance
partners, among them NGOs, are motivated by substantive goals, such as preserving a clean and healthful
environment, compensating involuntarily resettled communities, or restoring damaged cultural heritage
sites. Compliance partners can also be motivated by self-interest and by normative commitments to
general human rights principles (Alter, 2014, 53). Whatever their distinct motivations, NGOs’ advocacy in
the Inspection Panel and CAO processes should enhance the likelihood of success, as these organizations
draw on their expertise in information, leverage, and accountability politics to help deliver successful
outcomes for project-aﬀected communities.
We expect that NGOs support communities in producing and ﬁling high-quality formal complaints,
while also leveraging their public proﬁle and long-term relationships with the Bank, donor governments,
and other stakeholders to advocate outside the formal complaint process. It is also likely that NGOs
screen complaints and are more likely to support cases that they believe are more meritorious and/or
more likely to prevail. This screening plays an important role in our empirical evaluation, as it inﬂuences
how we interpret the relationship between NGO involvement and complaint success.
3.2.1 Complaint Quality
The most direct way that we expect NGOs to increase the likelihood of complaints success is through the
resources and technical expertise they provide to help communities ﬁle high-quality complaints. Project-
aﬀected communities—people who have been dispossessed of their lands and their livelihood, whose
environments have been degraded, and who have suﬀered physical harm—are comparatively less likely
to possess the material, informational, language, and human resources to independently and successfully
lodge a quasi-legal complaint, whereas NGOs do possess these resources (Islam and Morgan, 2011). NGOs
can assist project-aﬀected communities in clearly connecting the harms experienced to discrete violations
of World Bank operational policies—arguably the greatest barrier to complaint success (Uçarer, 2014).
The president of a Latin American environmental NGO (ENGO) that has supported complainants at the
Inspection Panel and CAO conﬁrmed to us that NGOs oﬀer technical expertise otherwise unavailable to
project-aﬀected communities, devoting resources so that communities “understand the technical aspects
of the projects and their impacts.” (Names and organizations are included for interviewees who gave
their express permission. Names and organizations are omitted for interviewees who requested to remain
anonymous. Our interview protocol was declared exempt by the University of Southern California’s
Institutional Review Board under study number UP-17-00438.)
What’s more, NGOs have established networks from which they may draw to assist project-aﬀected
communities. Apolinar Tolentino, the Asia Paciﬁc regional representative of the labor rights organization,
Building and Wood Workers International (BWI) reports that NGOs draw on diverse contacts when
preparing a complaint. The ﬁrst time BWI prepared a CAO complaint, the organization sought advice
from the OECD’s Trade Union Advisory Council. Tolentino aﬃrms, “They have the experience. Since
this [was] our ﬁrst experience it help[ed] to have a colleague coach us.” In turn, Vladlena Martsynkevych,
former regional director of the National Ecological Centre of Ukraine (NECU), shared with us that Danish
colleagues, working on a human rights court case in the Ukrainian courts, alerted NECU of an agribusiness
project-aﬀected village that could beneﬁt from the organization’s assistance. NECU’s inquiry into the
community led to the identiﬁcation of communities in seven other villages aﬀected by the same project.
Martsynkevych contends that the complaint was successful because it involved so many communities
which, of course, was only possible with NGOs at the helm.
3.2.2 External Pressure
In addition to being experts in information politics, many NGOs are experts in leverage and accountability
politics, and have been and are catalysts for change, including at the World Bank (Keck and Sikkink,
1998; Uçarer, 2014; Woods, 2000). NGO pressure partly explains the Bank’s adoption of a strategy for
poverty reduction in the 1990s that includes “the promotion of safety nets and targeted social programs
to support those who fall through the cracks” (Fox and Brown, 1998, 493). And, as previously discussed,
NGO advocacy contributed to the creation of the Inspection Panel and CAO, themselves. Given that
NGOs have successfully pressured the World Bank to modify its strategy for development, NGOs are
likely to similarly pressure the Bank, via the Inspection Panel or CAO, for relief and redress for speciﬁc
Moreover, unlike most project-aﬀected communities, NGOs are “repeat-players” (Galanter, 1974).
While it may be possible to silence a vulnerable community, it is much more diﬃcult to marginalize a
high-proﬁle local, national, and/or international actor, such as an NGO. Our interviewees corroborate
our expectation that the Inspection Panel and CAO “pay attention” when complainants are supported by
NGOs. In addition, because NGOs often interface with the World Bank at the project, national, and/or
international level, the Bank is more responsive to NGO concerns. Indeed, the Bank encourages NGO
participation in the “project cycle,” including in project design and evaluation (Fox and Brown, 1998).
Hypothesis 2 Complaints submitted by or with the assistance of NGOs are more likely to be successful.
3.2.3 Selection into NGO Involvement
While NGO involvement may have a positive eﬀect on complaint success, we also recognize that NGOs
may be more likely to select cases that already have a high likelihood of success. Given limited time
and resources, NGOs must distinguish meritorious cases or causes from less- or non-meritorious ones
and thereby utilize resources eﬃciently (Ron, Ramos and Rodgers, 2005).3Since NGOs’ credibility and
eﬀectiveness rely on the accuracy and relevance of the information they furnish, it is essential to ﬁnd
and support project-aﬀected communities with “good facts” (Barnett and Finnemore, 2004; Keck and
Sikkink, 1998). This might involve, for example, NGOs supporting project-aﬀected communities who raise
issues for which there is a speciﬁc World Bank operational protocol. We interviewed the director of an
international workers’ rights organization, who described to us how the organization ﬁrst “determine[s]
whether there are grounds to claim that the actions violate the safeguards that the institutions have
adopted” before proceeding to support a complaint (emphasis added).
Some NGOs only support project-aﬀected communities’ complaints after conducting their own in-
vestigation. For example, NECU representative, Vladlena Martsynkevych, reports that NECU collects
information using open sources on persons harmed or likely to be harmed by Bank-funded development
projects. The organization then contacts the relevant individuals/community, requesting supporting doc-
uments and other evidence of harm. After receiving this information, the organization then visits the
site to conﬁrm the accuracy of the claims and supporting evidence. Thus, it is likely that some of the
cases that proceed to the Inspection Panel and CAO without NGO support do so because NGOs have
investigated project-aﬀected communities’ claims and have themselves estimated that they lack merit.
In the empirical sections that follow, this selection process shapes how we interpret the strong rela-
tionship we observe between NGO involvement and complaint success. It also motivates the additional
mediation analysis that we conduct.
4 Research Design and a New Dataset
The data we introduce cover the universe of cases brought before the Inspection Panel and CAO since
their inception. We capture all cases concluded by May 2017, a total of 239 cases. These data include
nearly 100 variables covering: (1) characteristics of the project under scrutiny, (2) characteristics of the
aﬀected community and complaint ﬁlers, (3) complaint fundamentals, (4) nature of the alleged harms
and identity of the perpetrators, (5) progress of the complaint through the Inspection Panel and CAO
processes, and (6) ﬁnal outcomes of the complaint. The data provide great descriptive value and allow
us to test our hypotheses regarding the drivers of complaint success.
To evaluate our hypotheses, we use information on the groups bringing complaints (i.e. data on
indigenous communities and NGOs) and the issues they raise to predict several diﬀerent types of complaint
success, notably harm acknowledgment, project change, and compensation, while controlling for a range
of potentially confounding factors, such as the nature of the project that gave rise to the complaint.
Because we expect that NGOs strategically select the cases they support, we then use a combination
of quantitative and qualitative evidence to assess the importance of complaint quality for our results.
Quantitatively, we evaluate whether NGO involvement leads to higher quality complaints, as measured by
our latent variable, Complaint fundamentals, and if complaint quality, in turn, leads to complaint success.
Qualitatively, we draw on elite interviews with representatives from NGOs that support Inspection Panel
and CAO complaints. These interviews help us ascertain whether or not these NGOs screen cases for
merit before assisting with complaint ﬁling. The interviews also help us understand in what manner and
to what degree they apply pressure on the World Bank outside of the formal proceedings of the Inspection
Panel and CAO. The following sections describe the main variables used in the analysis. More detailed
descriptions and summary statistics for all variables used in the analysis are provided in the supplementary
appendix at the JHR Harvard Dataverse site.
4.1 Complaint Success
Complaint success is captured by three variables: Harm acknowledged,Project change, and Compensa-
tion.Harm acknowledged refers to the Inspection Panel or CAO’s formal acknowledgment of project-
related harm suﬀered by complainants. While harm acknowledgment is a minimalist form of success, its
delivery by a formal authority can be an important part of the healing process for survivors of human
rights violations and other harms (e.g. Hayner, 2000, 135-140). Project change captures amendments
or alterations ordered by the Inspection Panel or CAO, for example the introduction of additional envi-
ronmental safeguards and monitoring. Project change also includes project termination. Compensation
includes monetary and other economic reparations (e.g., land parcels for involuntarily resettled commu-
nities) to be paid to project-aﬀected communities, either by the Bank or by the host country government
at the request or instruction of the Bank. Compensation takes a value of 1 only if compensation is
ordered in response to a complaint. Sometimes this occurs when compensation is added to a project
where it had not been previously planned, and sometimes this occurs when a complaint forces payment
of planned compensation that was, for some reason, withheld prior to the complaint.
4.2 Stringent Safeguards and NGO Involvement
Indigenous community is a dichotomous indicator that takes a value of 1 if a complaint is ﬁled by or
behalf of an indigenous community. Involuntary resettlement is a binary variable that captures whether
involuntary resettlement is alleged in a complaint. High environmental risk takes a value of 1 if the
project was originally classiﬁed under Environmental Category A—that is, determined to pose the most
severe environmental impacts. NGO involvement is a binary variable that captures whether at least one
NGO ﬁled or signed the initial complaint made to the Inspection Panel or CAO.
4.3 Complaint Fundamentals
We estimate a latent variable, Complaint fundamentals, to capture fulﬁllment of seven items that we
expect make complaint success more likely: (1) raw data, including maps, (2) personal testimony by
project-aﬀected persons, (3) petitions, (4) typed complaints, (5) complaints written in English, (6)
references to speciﬁc Bank operational protocols, procedures, and/or performance standards that com-
plainants claim has been violated, and (7) a description of the link between the alleged Bank policy
violation and the harm suﬀered by the community or individuals bringing the complaint. The latent
variable model allows us to estimate complaint fundamentals even for observations missing one or more
of our indicator variables, and allows us to estimate both the value of the latent trait and the uncertainty
of that estimate. Summary statistics for the seven indicators used to estimate Complaint fundamentals,
the latent variable model’s assumptions, and information on item diﬃculty and discrimination parameters
are provided in the supplementary appendix.
4.4 Descriptive Statistics: Assessing the Big Picture
The Inspection Panel and CAO were created to enforce policies for mitigating social and environmental
impacts of World Bank projects, and to provide relief and redress to communities when said policies are
violated. Figures 1 to 5 illuminate these central issues, speciﬁcally what types of complaints are lodged
and what types of outcomes complainants secure. As more and more development ﬁnance institutions
attempt to replicate the Inspection Panel and CAO, it is crucial that we gain a comprehensive and
accurate picture of the Inspection Panel and CAO’s performance, allowing these newer bodies to both
build on the strengths of the Inspection Panel and CAO, and, perhaps, make adjustments to address
The distribution of cases by geographic location and by project- and issue-type illuminate where the
Inspection Panel and CAO work and the type of cases they hear. Figure 1 shows that complaints are
concentrated in the global South, with India, Peru, Argentina, Kenya, and Uganda as the top ﬁve origin
sites for complaints. This broadly comports with the global distribution of World Bank projects.
Figure 1: Geographical Spread of Inspection Panel and CAO Cases
As shown in Figure 2, two project types dwarf all others in terms of complaints ﬁled with the Inspection
Panel and CAO: infrastructure projects trigger 56% of Inspection Panel complaints and 22% of CAO
complaints, while mining or resource extractive projects trigger 49.5% of CAO complaints and 8% of
Inspection Panel cases.
Figure 2: Projects Triggering Inspection Panel and CAO Complaints
As reported in Figure 3, environmental damage is the most-raised issue in both the Inspection Panel
and CAO. Economic damage takes second position in the Inspection Panel, while involuntary resettlement
takes second position in the CAO. Finally, labor rights violations and involuntary resettlement respectively
take third position in proceedings before the Inspection Panel and CAO. This generally comports with
the qualitative literature on the Inspection Panel and CAO, which has focused primarily, though not
exclusively, on environmental damage and involuntary resettlement (Clark, 2002; Fox, 2000; Rich, 1994;
Webber and McDonald, 2004).
Figure 3: Primary Issues Raised in Inspection Panel and CAO Complaints
Figure 4 describes the overall eﬀectiveness of the Inspection Panel and CAO in delivering speciﬁc
types of remedies. Notably, 42% of Inspection Panel cases and 61% of CAO cases result in some type of
positive outcome for complainants. With respect to diﬀerences between the Inspection Panel and CAO,
a greater percentage of CAO cases result in a formal acknowledgment of harm, a determination of Bank
noncompliance, and judgments to halt and change projects, while a greater percentage of Inspection
Panel cases result in an order for compensation to project-aﬀected communities. We believe that this
is primarily due to the Inspection Panel handling more involuntary resettlement cases than the CAO, as
this type of project-related harm requires compensation.4
The Inspection Panel and CAO oﬀer some form of relief to a high percentage of complainants. In
many cases this relief is not trivial, with projects ordered to be revised or halted and/or compensation
paid. In short, these institutions are not toothless and they do not appear reluctant to judge in favor of
communities that lodge complaints with the Bank.
Figure 4: Success at the Inspection Panel and CAO
Figure 5 shows a more granular distribution of outcomes across issue areas. In both Inspection Panel
and CAO cases, rulings for compensation are most likely in cases alleging involuntary resettlement. They
are also sometimes made in response to claims of economic or environmental damage. Physical integrity
rights violations represent the most severe allegations in our data; however, these cases are also few in
number, making it diﬃcult to generalize about them.5
Figure 5: Outcomes by Primary Issue Raised
5 Hypothesis Testing
We theorize that complaints are more likely to be successful when an indigenous community is involved
(H1a) or when communities ﬁle complaints about involuntary resettlement (H1b) or projects that posed an
especially high environmental risk (H1c). We also expect that complaints are more likely to be successful
when an NGO is involved (H2). We evaluate these hypotheses using Indigenous community,Involuntary
resettlement,High environmental risk, and NGO involvement to predict our three main measures of
complaint success. Models 1 and 2 in Table 1, below, present the results of logit regressions with Harm
acknowledged as the dependent variable. Models 3 and 4 predict Project change, while Models 5 and 6
predict Compensation. In each model we control for a range of factors that may be causally related to
complaint success and correlated with our variables of interest, including project characteristics and the
region in which the alleged violation took place.6
We ﬁrst analyze the relationship between complaints covered by stringent safeguards and the like-
lihood of complaint success. Consistent with Hypothesis 1a, we estimate a strong positive eﬀect for
the involvement of indigenous communities on complaint success. This eﬀect is statistically signiﬁcant
(p<0.05) with respect to Harm acknowledged and Project change, but not with respect to Compensation.
To estimate the substantive size of these eﬀects, we evaluate the eﬀect of indigenous involvement while
holding all other variables constant. Substantively, the predicted probability of gaining a formal acknowl-
edgment of harm for a typical indigenous community is roughly double that of a similar non-indigenous
community, 46% instead of 26%7The diﬀerence is similar with respect to judgments for project change:
the predicted probability for a typical indigenous community gaining an order for project change is 64%,
compared to 30% for a comparable non-indigenous community. Finally, the estimated probability of
compensation orders rises to 23% for indigenous communities compared to 13% for all others.
The estimated eﬀect of involuntary resettlement allegations is positive and statistically signiﬁcant
with respect to Compensation (p<0.05). It is also positive, though not statistically signiﬁcant, with
respect to Harm acknowledgement and Project change. Thus, we ﬁnd weak support or Hypothesis 1b.
If we hold all other variables constant, the predicted probability of securing a ruling for compensation
for complaints alleging involuntary resettlement is roughly triple that of complaints that do not allege
involuntary resettlement, 37% instead of 11%. The predicted probability of project change rises less,
from 39% to 45%. In contrast to these results, we estimate a negative relationship between involuntary
resettlement cases and harm acknowledgment, with the predicted probability falling from 34% to 28%.
Inconsistent with Hypothesis 1c, the environmental sensitivity of projects is not strongly correlated
with complaint success, nor are or allegations of environmental harm. Their respective estimated eﬀects
are substantively small and statistically insigniﬁcant. In the supplementary appendix we estimate the
eﬀect of environmental risk without controlling for issue area (including environmental damage), and
vice-versa, and still ﬁnd no eﬀect for either high environmental risk or environmental damage allegations.
Consistent with H2, we ﬁnd that NGO involvement is also a positive predictor of all three types
of complaint success, but this eﬀect is only statistically signiﬁcant (p<0.05) with respect to Project
change and Compensation, and not with respect to Harm acknowledged. NGO involvement increases the
probability of gaining rulings for project change from 32% to 47%, while more than tripling the probability
of orders for compensation from 7.5% to 25%. Finally, the estimated eﬀect on harm acknowledgment is
smaller but still non-trivial, rising from 28% without NGO involvement to 37% with NGO involvement.
Table 1: Stringent Safeguards, NGO Involvement, and Outcomes
DV = Harm Acknowledged DV = Project Change DV = Compensation
(1) (2) (3) (4) (5) (6)
Indigenous Community 1.189** 0.997* 1.442** 1.434** 0.161 0.758
(0.460) (0.478) (0.436) (0.460) (0.482) (0.612)
NGO Involvement 0.466 0.473 0.752* 0.771* 1.660* 2.012**
(0.343) (0.347) (0.366) (0.376) (0.658) (0.735)
High Environmental Risk 0.275 0.292 0.0840 0.107 0.188 0.276
(0.384) (0.376) (0.339) (0.340) (0.473) (0.531)
Infrastructure 0.0684 0.161 0.442 0.444 0.0508 -0.172
(0.398) (0.405) (0.419) (0.435) (0.593) (0.584)
Mining/Resource Extraction -0.279 0.0269 0.413 0.446 0.0228 -0.681
(0.441) (0.498) (0.479) (0.528) (0.664) (0.795)
Bank Commitment (log) -0.161 -0.161 -0.439** -0.427** 0.149 0.144
(0.147) (0.164) (0.135) (0.135) (0.171) (0.167)
CAO 0.918* 0.832+0.382 0.309 -0.0121 -0.173
(0.370) (0.429) (0.350) (0.395) (0.537) (0.596)
Alleged Harms and Perpetrators
Involuntary Resettlement -0.282 -0.251 0.310 0.321 1.858** 1.960**
(0.476) (0.472) (0.459) (0.456) (0.599) (0.658)
Environmental Damage 0.191 0.127 0.148 0.144 0.419 0.681
(0.372) (0.368) (0.414) (0.422) (0.490) (0.477)
Economic Damage -0.445 -0.381 0.0996 0.126 0.803 0.672
(0.340) (0.338) (0.351) (0.351) (0.493) (0.492)
Labor Rights Violations 0.0298 0.107 0.148 0.0815 1.477* 1.169
(0.697) (0.666) (0.691) (0.676) (0.752) (0.714)
Physical Integrity Rights Violations -0.276 -0.315 -0.912+ -0.927+ -1.398+ -1.406
(0.509) (0.529) (0.486) (0.489) (0.844) (0.859)
Bank is Alleged Perpetrator 0.00278 -0.181 -1.154*
(0.437) (0.409) (0.491)
Africa and Middle East -0.573 -0.0827 1.838*
(0.491) (0.475) (0.754)
Asia -0.123 -0.159 1.617*
(0.415) (0.420) (0.745)
Europe -0.674 -0.314 1.655*
(0.578) (0.460) (0.802)
Constant 1.293 1.527 6.139* 6.161* -6.989* -7.742*
(2.702) (3.040) (2.435) (2.534) (3.192) (3.481)
Observations 220 220 220 220 221 221
All models report standard errors clustered by country.
+p<0.10, * p<0.05, ** p<0.01
Overall, we estimate consistently positive eﬀects for indigenous community involvement, involuntary
resettlement, and NGO involvement. While the eﬀects are not statistically signiﬁcant for all outcomes,
the results show that complaints in these two categories are systematically more likely to receive at least
one positive outcome for complainants. The following section parses the NGO eﬀect more ﬁnely.
5.1 Drivers of NGO Success: Complaint Quality, External Advocacy, and Selection
We expect that NGOs increase complaint success via professional expertise and organizational capacity,
which yield higher-quality written complaints. We also anticipate that NGOs apply pressure on the Bank,
both internal and external to the formal complaint process. However, it is also possibly the case that
NGOs are more likely to support complaints they consider promising. Thus, we investigate whether NGO
eﬀorts actually increase the quality of ﬁled complaints, and whether increases in complaint quality allow
NGOs to move the needle on complaint success beyond simply picking winners.
We focus our quantitative analysis on complaint quality because this the most direct way in which
we expect NGOs to have an eﬀect. More pragmatically, complaint quality is also more readily observable
than external pressure; external pressure can take many forms, some of which are intentionally hidden,
such as back-channel lobbying.
5.1.1 Complaint Fundamentals
We evaluate the inﬂuence of NGO expertise and resources on complaint quality by estimating the eﬀect
of NGO involvement on “complaint fundamentals,” or the observable characteristics of complaints that
we expect inﬂuence the prospects of complaint success. While these fundamentals do not capture
every potential avenue through which NGOs may inﬂuence complaint quality, if NGOs make important
contributions to complaint quality, one observable implication is that NGO-supported complaints should
contain more of these fundamental elements.8
Figure 6 depicts fulﬁllment of four complaint fundamentals in cases with and without NGO involve-
ment. The relationship here is striking: across each indicator, NGO-supported complaints are more likely
to contain elements that we believe are fundamental to complaint quality. The diﬀerences are particu-
larly stark with respect to the inclusion of raw data and citations to speciﬁc World Bank policies. In the
absence of NGOs, complaint fundamentals are markedly lacking.
Figure 6: NGO Involvement and Complaint Fundamentals
Table 2 evaluates whether NGO involvement is a strong predictor of Complaint fundamentals when
we control for other factors. Consistent with the descriptive pattern shown in Figure 6, we estimate a
positive relationship between NGO involvement and Complaint fundamentals, though this relationship is
not statistically signiﬁcant at a conventional error level. In contrast, we ﬁnd that indigenous communities
and projects with a high environmental risk are positive and statistically signiﬁcant predictors of the
outcome (p<0.05). Indigenous-ﬁled complaints and complaints about projects with a high environmental
risk are both more likely than other types of complaints to include two particular elements of complaint
quality: citations to speciﬁc bank policy and a link between that violation and the harm experienced.
63% of indigenous complaints cite bank policy and link the alleged policy violation(s) with experienced
harms. Similarly, 55% of complaints about projects with a high environmental risk cite bank policy and
link the alleged violation(s) with harms. In so doing, communities strengthen their claims to relief and
redress. This is consistent with our theoretical argument that the speciﬁcity and stringency of Bank
policies—with respect to indigenous communities and environmentally-sensitive projects—make it easier
for them to obtain complaint success.
Table 2: NGO Involvement and Complaint Fundamentals
DV = Complaint Fundamentals
NGO Involvement 0.123 0.198+
Indigenous Community 0.243∗0.265∗
Infrastructure -0.119 -0.183
Mining/Resource Extraction 0.112 0.0363
High Environmental Risk 0.311∗∗ 0.341∗
CAO -0.435∗∗ -0.520∗∗
Involuntary Resettlement 0.0384 0.0196
Environmental Damage 0.184 0.244+
Bank Commitment (log) -0.0210
Economic Damage -0.0275
Labor Rights Violations 0.333∗
Physical Integrity Rights Violations 0.0567
Observations 150 134
Standard errors in parentheses
All models report standard errors clustered by country.
+p<0.10,∗p<0.05,∗∗ p< .01
The causal mediation results presented in Figure 7 allow us to assess whether NGO involvement
inﬂuences the likelihood of complaint success through its eﬀect on complaint fundamentals, or whether
NGO involvement operates independently of these complaint fundamentals.9Displayed in the ﬁgure are
the estimated average direct eﬀect (ADE) of NGO involvement on complaint success and the estimated
average mediated eﬀect (ACME) of NGO involvement. We do not ﬁnd support for a mediated eﬀect of
NGO involvement on success before the Inspection Panel and CAO. The ACME is positive for each success
outcome; however, the eﬀects are small and are not statistically signiﬁcant. In contrast, the estimated
direct eﬀects are much larger, albeit only statistically signiﬁcant with respect to compensation. The
direct eﬀect includes all channels other than complaint fundamentals, including selection and external
Taken together these results show that, while NGO involvement is positively associated with complaint
fundamentals, these fundamentals do not account for most of the observed relationship between NGO
involvement and complaint success.
Figure 7: Maximum eﬀects of NGO Involvement (Mediated through Complaint Fundamentals) on Harm Acknowl-
edgment, Project Change, and Compensation
To better understand the nature of the direct eﬀect of NGO involvement on complaint success,
we conducted a series of interviews with representatives from NGOs that supported recent cases in
our data. Consistent with merit screening, NGO representatives consistently report that they screen
prospective cases for merit before agreeing to support a complaint. Interviewees also describe a range
of technical abilities they bring to the table, consistent with our expectation that NGO involvement
increases complaint quality. In addition, several NGO representatives highlighted NGOs’ ability to apply
public pressure on the World Bank, which they believe compels the Inspection Panel and CAO to take
certain complaints more seriously, even though the mechanisms are technically independent of the rest
of the Bank. To give two examples, a representative from a workers’ rights organization reports,
“I think unfortunately maybe they do pay more attention when it has some international
attention and attention from an organization [...] which has some international clout and
The president of the Latin American ENGO highlighted the organization’s pursuit of high-proﬁle
cases in which it can lobby the World Bank directly and indirectly, via donor countries “where there was
interest to lobby against the project.”
Our results do not allow us to rule out the possibility that the relationship between NGO involvement
and complaint success is primarily driven by selection. However, there is some evidence that NGO
involvement improves complaint quality, even if we do not ﬁnd that complaint quality, in turn, drives
complaint success. Our interviews suggest that the work that NGOs undertake to enhance the likelihood
of complaint success goes well beyond simply supporting communities in drafting better complaints. As
future work seeks to understand the mechanisms driving NGO success (and dissecting what types of
NGOs make the most eﬀective community advocates), it will be necessary to look beyond the formal
Our results oﬀer three main implications. First, quasi-judicial bodies deliver important outcomes to
vulnerable populations, including redress for human rights violations and other harms. Given that private
persons do not have standing in formal courts to sue the World Bank for the ﬁnancing and execution of
its projects, and that the organizations in the World Bank Group may fail to provide relief and redress of
their own accord, the Inspection Panel and CAO may be the most eﬀective venues for project-aﬀected
communities to seek relief and redress.
Our research motivates inquiry into other quasi-judicial bodies that seek to protect human rights and
promote good governance and, in particular, dispute resolution and compliance review mechanisms in
regional and national development banks. While quasi-judicial bodies receive less attention than formal
courts, such as the Inter-American Court of Human Rights, these venues may be equally important, if
not more important, for providing substantive relief to some of the world’s most vulnerable populations.
Second, NGO support is strongly correlated with success at the Inspection Panel and CAO. While
at least some of the positive relationship we observe between NGO involvement and complaint success
can be attributed to NGOs screening cases on merit, we estimate that NGO involvement leads to higher
scores on complaint fundamentals (though this diﬀerence is not statistically signiﬁcant). We also oﬀer
qualitative evidence that NGOs provide advocacy support to complainants outside of the formal Inspection
Panel and CAO processes, which may also have an eﬀect. Reassuringly, we do not ﬁnd evidence that less-
professional complaints have lower success rates, suggesting that the Inspection Panel and CAO can and
do accommodate low-capacity complainants. However, the high correlation between NGO involvement
and complaint success suggests that NGOs may, nonetheless, be necessary for the Inspection Panel and
CAO to provide eﬀective relief, notably compensation, to communities.
Third, safeguard policies for speciﬁc groups and issue areas have mixed eﬀects—safeguards related
to indigenous communities and involuntary resettlement appear to enhance accountability, while envi-
ronmental safeguards seem less eﬀective. We speculate that environmental safeguards may fail to yield
higher complaint success rates for reasons unique to the nature of environmental complaints—particularly
that some degree of environmental change may be, from the Bank’s perspective, an acceptable conse-
quence of economic development. Thus, grey areas exist in which communities may reasonably perceive
negative environmental eﬀects from projects that nonetheless comply with Bank guidelines.
More broadly, we hope that the patterns in outcomes we observe in the Inspection Panel and CAO will
inform the development and reﬁnement of cognate bodies at other IFIs. While they often order project
change or enjoin compensation, the Inspection Panel and CAO are notably lacking in their ability to punish
actors responsible for human rights violations and other harms, be they host country governments, their
partners, or contractors.10 This is where a concerted campaign to apply the “protect, respect, remedy”
framework to IFIs, and the World Bank in particular, may be valuable for project-aﬀected communities.
In addition, IOs in general and IFIs in particular should contemplate whether the policies against which
compliance is assessed provide elevated protections for the groups they believe to be most vulnerable in
the contexts in which they operate. Lastly, these organizations should assess whether their compliance
mechanisms are designed in ways that make NGO representation necessary for complaint success in many
contexts and, if so, whether there exists the requisite NGO infrastructure to meet the needs of potential
Since their creation in the 1990s, the Inspection Panel and CAO have adjudicated nearly 250 human
rights complaints in more than 60 countries. During this period, other development ﬁnance institutions
have emulated the World Bank, creating similar dispute resolution and compliance review mechanisms.
Inspection Panel and CAO cases focus predominantly on three issue areas: environmental damage,
economic damage, and involuntary resettlement. Thus, they address issues of serious and substantive
concern to project-aﬀected communities, meeting global civil society and donor countries’ stated goals.
Our study demonstrates that these bodies are not toothless: More than a third of complaints result in
an order for the alteration or abandonment of the project under scrutiny, and 15% of complaints result
in orders for compensation for project-aﬀected communities.
This paper oﬀered the ﬁrst comprehensive quantitative analysis of the operation and eﬀectiveness
of development ﬁnance institutions’ accountability mechanisms, enlarging our understanding of quasi-
judicial bodies. We examine the ways in which factors such as the individual harms raised in complaints,
NGO advocacy in the complaint process, project type and size aﬀect outcomes of interest, notably
the Bank’s acknowledgment of harm, and orders to change or terminate projects, and to compensate
We theorized that complaints are more likely to succeed if they are covered by the Bank’s stringent
safeguards for indigenous communities, involuntary resettlement, and/or severe environmental impacts,
or when complainants are represented by an NGO. In our regression analysis, we ﬁnd strong support
for our expectations about indigenous communities, weak support for our expectations about projects
involving involuntary resettlement, and no support for our expectations about environmentally-sensitive
projects. We also ﬁnd that NGO representation is strongly associated with complaint success. In a
more detailed analysis, we then unpacked and clariﬁed the mechanisms through which NGO involvement
aﬀects complaint success. We ﬁnd evidence that NGO involvement enhances complaint quality, but also
that NGOs screen complaints, seeking to support those they estimate meritorious.
Much remains to be accomplished in this research area. In particular, our research provides a bench-
mark for evaluating gains and losses associated with institutional replication of the Inspection Panel and
CAO. Do institutional copies function similarly to the originals on which they are modeled? Additionally,
do successful cases brought before the Inspection Panel and CAO enhance the security of human rights
in the broader domestic realm, either by empowering individuals and communities, or sending signals
to governments and ﬁrms that accountability is possible? Finally, what are the eﬀects of accountability
mechanisms on the IFIs that created them? One interviewee representing a workers rights organization
has found partnerships between NGOs, within and without Inspection Panel and CAO complaints, useful
for “improv[ing] language in the safeguards and implementation procedures.” When we consider the uni-
verse of venues through which individuals and communities may seek justice for human rights violations
and other harms, and through which the landscape of development ﬁnance and international law can
be changed, the Inspection Panel and CAO deserve consideration as serious and eﬀective, if still ﬂawed,
1See the ADB, EIB, and IDB, as well as the African Development Bank (AfDB) and the European Bank for
Reconstruction and Development’s (EBRD).
2See, for example, the Brazilian Development Bank, the Canadian Oﬃce of the Extractive Sector Corporate
Social Responsibility, the Japan Bank for International Cooperation, and the U.S. Overseas Private Investment
Corporation Oﬃce of Accountability.
3NGOs may also concentrate their eﬀorts on communities whose experienced harms correspond to their professional
experience and organizational missions—communities for which they can more expertly advocate. Ron, Ramos
and Rogers (2005) argue and ﬁnd that NGOs have incentives to pursue causes and campaigns that promote their
“brand” and inspire public conﬁdence or trust. Thus, cases within NGOs’ issue portfolios may be more likely to gain
NGO support than complaints in other issue areas. In our empirical analysis of the eﬀects of NGO involvement
we control for issue area.
4Involuntary resettlement is raised in 25 complaints to the Inspection Panel compared to 7 complaints to the CAO.
5Physical integrity rights violations constitute the primary alleged harm in only 9 complaints.
6Dummy variables are included for each region with The Americas as the omitted region. The indicator variables
for diﬀerent alleged harms capture whether each category of harm was among those raised in a complaint. Many
complaints allege more than one type of harm and we code up to three per complaint.
7Substantive eﬀects are calculated based on Models 1, 3, and 5.
8In estimating complaint fundamentals and conducting the mediation analysis, we use only those cases with the
original ﬁling available. This limits us to 150 of our 239 total cases. Perhaps surprisingly, the availability of
original ﬁlings is only weakly correlated with the year in complaints were ﬁled (Pearson’s rho = 0.06).
9The full regression results for the mediation analysis are provided in the supplementary appendix.
10To our knowledge, the only case of perpetrator punishment is a Uganda Transport Sector Development Project
case in which the Government of Uganda provided compensation to survivors of sexual assault perpetrated by
road workers contracted by the Ugandan National Road Authority (UNRA) and prosecuted those responsible. In
turn, the World Bank suspended new lending to Uganda while conducting a review of their entire lending portfolio
in the country. This case is not in our data because it was not concluded by May 2017.
Accountability Counsel. 2015. “Accountability Resource Guide: Tools for Redressing Human Rights &
Environmental Abuses in International Finance and Development.”.
Alter, Karen J. 2014. The New Terrain of International Law: Courts, Politics, Rights. Princeton University
Barnett, Michael and Martha Finnemore. 2004. Rules for the World: International Organizations in
Global Politics. Ithaca: Cornell University Press.
Benvenisti, Eyal and George W. Downs. 2009. “National Courts, Domestic Democracy, and the Evolution
of International Law.” European Journal of International Law 20(1):59–72.
Bissell, Richard E. 1997. “ Recent Practice of the inspection panel of the World Bank.” American Journal
of International Law 91(4):741–744.
Bradlow, Daniel D. 1993. “International Organizations and Private Complaints: The Case of the World
Bank Inspection Panel.” Virginia Journal of International Law 34:553–613.
Buntaine, Mark T. 2015. “Accountability in Global Governance: Civil Society Claims for Environmental
Performance at the World Bank.” International Studies Quarterly 59(1):99–111.
Carrasco, Enrique R. and Alison K. Guernsey. 2008. “World Bank’s Inspection Panel: Promoting True
Accountability through Arbitration.” Cornell International Law Journal 41:577.
Clark, Dana, Jonathan A. Fox and Kay Treakle. 2003. Demanding Accountability: Civil Society Claims
and the World Bank Inspection Panel. Lanham: Rowman & Littleﬁeld Publishers.
Clark, Dana L. 2002. “The World Bank and Human Rights: The Need for Greater Accountability.”
Harvard Human Rights Journal 15:205–226.
Davitti, Daria. 2016. “Reﬁning the Protect, Respect and Remedy Framework for Business and Human
Rights and its Guiding Principles.” Human Rights Law Review 16(1):55–75.
Fox, Jonathan A. 2000. “ The World Bank Inspection Panel: Lessons from the First Five Years.” Global
Fox, Jonathan A. and L. David Brown. 1998. “Assessing the Impact of NGO Advocacy Campaigns on
World Bank Projects and Policies.” Center for Global, International and Regional Studies .
Galanter, Marc. 1974. “Why The “Haves" Come Out Ahead: Speculations on the Limits of Legal
Change.” Law & Society Review 9(1):95–160.
Gutner, Tamar L. 1999. Banking on the Environment: Multilateral Development Banks and their Envi-
ronmental Performance in Central and Eastern Europe. Cambridge: MIT Press.
Haddad, Heidi Nichols. 2012. “Judicial Institution Builders: NGOs and International Human Rights
Courts.” Journal of Human Rights 11(1):126–149.
Haddad, Heidi Nichols. 2018. The Hidden Hands of Justice: NGOs, Human Rights, and International
Courts. Cambridge University Press.
Hafner-Burton, Emilie M., Layna Mosley and Robert Galantucci. 2019. “Protecting Workers Abroad and
Industries at Home: Rights-Based Conditionality in Trade Preference Programs.” Journal of Conﬂict
Hayner, Priscilla B. 2000. Unspeakable Truths: Confronting State Terror and Atrocity. New York:
Hunter, David. 2003. “Using the World Bank Inspection Panel to Defend the Interests of Project-Aﬀected
People.” Chicago Journal of International Law 4:201.
Islam, M. Rezaul and William J. Morgan. 2011. “ Non-Governmental Organizations in Bangladesh: Their
Contribution to Social Capital Development and Community Empowerment.” Community Development
Keck, Margaret E. and Kathryn Sikkink. 1998. Activists Beyond Borders: Advocacy Networks in Inter-
national Politics. Ithaca: Cornell University Press.
Koremenos, Barbara. 2016. The Continent of International Law: Explaining Agreement Design. Cam-
bridge University Press.
Lavelle, Kathryn C. 2011. “Multilateral Cooperation and Congress: The Legislative Process of Securing
Funding for the World Bank.” International Studies Quarterly 55(1):199–222.
Lupu, Yonatan. 2013. “Best Evidence: The Role of Information in Domestic Judicial Enforcement of
International Human Rights Agreements.” International Organization 67(3):469–503.
Macdonald, Kate and May Miller-Dawkins. 2015. “Accountability in Public International Development
Finance.” Global Policy 6(4):429–434.
Mallaby, Sebastian. 2006. The World’s Banker: A Story of Failed States, Financial Crises, and the Wealth
and Poverty of Nations. New York: Penguin.
Minow, Martha. 1998. Between Vengeance and Forgiveness: Facing History after Genocide and Mass
Violence. Boston: Beacon Press.
Nanwani, Suresh. 2014. “ Directions in Reshaping Accountability Mechanisms in Multilateral Development
Banks and Other Organizations.” Global Policy 5(2):242–252.
Nielson, Daniel L. and Michael J Tierney. 2003. “Delegation to international organizations: Agency
theory and World Bank environmental reform.” International organization 57(2):241–276.
Okogbule, Nierum S. 2011. “Modest Harvests: Appraising the Impact of Human Rights Norms on
International Economic Institutions in Relation to Africa.” The International Journal of Human Rights
Orentlicher, Diane F. 2007. “ ‘Settling Accounts’ Revisited: Reconciling Global Norms with Local Agency.”
The International Journal of Transitional Justice 1(1):10–22.
Rich, Bruce. 1994. Mortgaging the Earth: The World Bank, Environmental Impoverishment and the
Crisis of Development. Boston: Beacon Press.
Ron, James, Howard Ramos and Kathleen Rodgers. 2005. “Transnational information politics: NGO
human rights reporting, 1986–2000.” International Studies Quarterly 49(3):557–587.
Sano, Hans-Otto. 2000. “Development and Human Rights: The Necessary, but Partial Integration of
Human Rights and Development.” Human Rights Quarterly 22:734.
Sarfaty, Galit A. 2004/2005. “The World Bank and the Internalization of Indigenous Rights Norms.”
Yale Law Journal 114(2004):1791–1818.
Shihata, Ibrahim F.I. 1994. The World Bank Inspection Panel. Oxford: Oxford University Press.
Tallberg, Jonas, Thomas Sommerer, Theresa Squatrito and Christer Jönsson. 2014. “Explaining the
Transnational Design of International Organizations.” International Organization 68(4):741–774.
Tignino, Mara. 2016. Quasi-Judicial Bodies. In Research Handbook on the Theory and practice of
International Lawmaking. Cheltenham: Edward Elgar Publishing Ltd. pp. 242–260.
Uçarer, Emek M. 2014. “Tempering the EU? NGO advocacy in the Area of Freedom, Security, and
Justice.” Cambridge Review of International Aﬀairs 27(1):127–146.
Villa-Vicencio, Charles. 2000. “Why Perpetrators Should Not Always Be Prosecuted: Where the Inter-
national Criminal Court and Truth Commissions Meet.” Emory Law Journal 49:205.
von Stein, Jana. 2018. “Exploring the Universe of UN Human Rights Agreements.” Journal of Conﬂict
Webber, Michael and Brooke McDonald. 2004. “Involuntary Resettlement, Production and Income:
Evidence from Xiaolangdi, PRC.” World Development 32(4):673–690.
Woods, Ngaire. 2000. “ The Challenge of Good Governance for the IMF and the World Bank Themselves.”
World Development 28(5):823–841.
World Bank. 1990. Operational Directive 4.30: Involuntary Resettlement.
World Bank. 2011. Implementation of the World Bank’s Indigenous Peoples Policy.
World Bank. 2013a.Operational Policy 4.01: Environmental Assessment.
World Bank. 2013b.Operational Policy 4.10: Indigenous Peoples.
Ziai, Aram. 2016. “ The World Bank Inspection Panel in Ghana: Civil Society Protest and the Glocalization
of Accountability.” Journal of Poverty Alleviation & International Development 7(1).
Zvobgo, Kelebogile. 2019. “Designing Truth: Facilitating Perpetrator Testimony at Truth Commissions.”
Journal of Human Rights 18(1):92–110.
Zvobgo, Kelebogile. 2020. “Demanding Truth: The Global Transitional Justice Network and the Creation
of Truth Commissions.” International Studies Quarterly. Online First.