Bureaucracy and Service Delivery∗

              Thomas B. Pepinsky†                Jan H. Pierskalla‡            Audrey Sacks§


                                     FIRST DRAFT: June 9, 2016
                                     THIS DRAFT: July 22, 2016


                                                 Abstract

              This essay reviews the literature on the politics of bureaucracy in the developing
         world, with a focus on service delivery and bureaucratic performance. We survey classic
         topics and themes such as the developmental state, principal-agent relations, and the
         efficient grease hypothesis, and link them to new research findings in political science,
         sociology, and economics. We identify the concept of embeddedness as an important yet
         still underexplored framework that cuts across disciplines and which may be used to
         understand bureaucratic performance and service delivery. Looking forward, we out-
         line a framework for conceptualizing bureaucratic action by exploiting variation across
         time, space, task, and client, and identify promising areas for further research on the
         bureaucrat-citizen encounter in developing countries.

             KEYWORDS: embeddedness, civil servants, street-level bureaucrats, developmen-
         tal state, principal-agent relations




  ∗
     This paper has benefited from the valuable feedback of Ron Herring, Daniel Ho, Phil Keefer, Margaret
Levi, Daniel Rogger, Nic van de Walle, and Michael Woolcock. We gratefully acknowledge the research
assistance of Jennifer Noveck and Darin Self.

   †
    Corresponding author. Department of Government, 322 White Hall, Cornell University, Ithaca, NY,
14853. pepinsky@cornell.edu.

   ‡
       Department of Political Science, the Ohio State University. pierskalla.4@osu.edu.

   §
       GPSURR, World Bank. asacks@worldbank.org.
1     Introduction

Scholars of comparative politics have made impressive strides in understanding political
institutions, voting, and political participation in developing countries. But most of politics
is not electoral politics, and the vast majority of experiences that citizens have with the state
are not electoral in nature. The face of politics for most citizens, instead, is a bureaucrat.
The interactions between citizens and bureaucrats—both the nature of the interaction itself,
and the results of that interactions—have consequences for mass political behavior, trust in
government, state legitimacy, and also for the functioning of bureaucracies themselves.

    Despite the importance of bureaucratic politics and the citizen-bureaucrat encounter in
developing countries, most work studying these issues comes from the disciplines of public
administration, sociology, and economics. The political science literature that would examine
these themes—the political economy of development—has tended to ignore the bureaucracy
itself, and instead studied conflict and elections, or focused on macrostructural variables like
the quality of property rights, geographic endowments, and the developmental state. The
theoretical literature on bureaucracies from within political science focuses on principal-agent
problems and is pitched at the types of problems and institutional environments that are
characteristic of advanced industrial countries. A rich literature on citizen-bureaucrat inter-
actions spells out their consequences for legitimacy, behavior, and citizenship, but focuses
overwhelmingly on the American and European experiences (Goodsell 1981, Mettler 1998,
Kumlin & Rothstein 2005).

    This review outlines a set of problems, questions, and puzzles in the study of the bu-
reaucracy in developing countries, focusing in particular on the politics of service delivery
as the key instance of citizen-state interactions. We pay special attention to “street-level
bureaucrats” (Lipsky 1980) in order to motivate our broader conclusion that conceptualizing
bureaucracy and service delivery requires attention both to what is being administered and
at what level of administration. Moreover, frontline service delivery has become a central


                                               1
plank in the global development agenda in the past thirty years, as can be seen through
the proliferation of social protection, decentralized service delivery, and related programs.
This means that understanding the bureaucracy tasked with frontline service delivery is in-
creasingly important for understanding the bureaucracy as a whole. A renewed focus on
bureaucracy and service delivery can show how political science is relevant for development
policymaking by conceiving of bureaucrats as political agents whose actions have political
effects. The study of the bureaucracy in developing countries can also provide new insights
into important questions that have long occupied comparative politics, such as the determi-
nants of democratic performance, development, statebuilding, institutional design, and the
possibility of programmatic politics in new democracies.

   We begin by reviewing existing literature on bureaucratic performance, focusing on the
two most prominent areas of research on the bureaucracy: the developmental state and
principal-agent models. While both important and useful, neither approach has sufficiently
explored bureaucratic politics and the determinants of effective service delivery in develop-
ing countries. Developmental state arguments largely ignore the delivery of health care and
education and focus on national-level policymaking and coordination. Principal-agent ap-
proaches in turn, while often studying civil servants in the health or education sector, often
narrowly cast bureaucratic politics as an internal management problem, while ignoring the
wider context of citizen-bureaucrat interactions. To further interrogate these shortcomings,
we discuss in Section 3 the role of street-level bureaucrats and the concept of embeddedness
as central to understanding most accounts of bureaucratic efficacy in developing country
contexts. This discussion highlights that, in order to understand bureaucratic performance,
we need additional research that carefully considers the specific administrative and political
context in which bureaucrats operate. To further this goal, we conclude by offering a frame-
work for conceptualizing bureaucratic action that explores variation by time, space, task,
and client. We also discuss a variety of methodological approaches that could be deployed
fruitfully to improve our understanding of bureaucratic politics in the developing world.


                                              2
2     State of the Art on the Politics of the Bureaucracy

2.1    Weber and Wilson

Two figures who occupy a central position in modern work on bureaucracies are Max We-
ber and James Q. Wilson. Weber’s Economy and Society (1978) characterizes many of the
quintessential features of the ideal-typical bureaucracy: esteem, professionalization, hierar-
chical organization, task routinization, and the impersonalization of bureaucratic authority.
Although Weber considered professional bureaucracies to be essential for modern states, he
also recognized that state bureaucracies were political creations with internal politics. Wil-
son’s Bureaucracy: What Government Agencies Do and Why They Do It (1991) embeds
bureaucracies in their political context, explaining why bureaucracies are so often character-
ized by red tape and inefficiencies that markets and firms would never tolerate. In his words,
“government bureaucracies are more bureaucratic than industrial ones in large part because
we—the people and our political representatives—insist that they be” (133). Relevant con-
temporary literatures that expand upon Weber’s and Wilson’s insights about bureaucracies
and their internal and external politics include the developmental state, good governance,
and principal-agent relationships.



2.2    The Developmental State and Quality of Government

The introduction of a relatively professional, meritocratic and efficient administrative struc-
ture was a critical factor in the rapid development of East Asia, including Japan, Korea,
Singapore, and Taiwan (Amsden 1989, Johnson 1982, Wade 1990). Although the exact role
the state played in guiding the economy varied across these states, the two distinguishing
features of East Asia’s developmental bureaucracies were meritocratic recruitment of knowl-
edgeable bureaucrats with network ties to the business community, and the overall coherence
of the bureaucracy (Evans 1995).


                                              3
   Networks, often informal, are crucial to bureaucratic coherence and identity and go be-
yond what meritocracy alone could provide. Although state institutions rely on networks
to hire civil servants, meritocratic hiring criteria break patronage links that would other-
wise produce incompetent or captured bureaucrats. By maintaining autonomous and close
network ties with industrial elites (Evans 1995), bureaucratic elites could effectively select
industries in which to invest, subsidize, and/or protect from competition.

   Chalmers Johnson’s classic MITI and the Japanese Miracle (1982) explicates the de-
velopmental state argument using the example of Japan’s successful economic model. The
presence of a meritocratic and capable bureaucracy enables a “plan-rational” development
model that can support economic growth over the long run because the bureaucracy enables
coordination among policymakers and the private sector, lumpy investments with uncertain
payoffs, and course changes when initial results prove unsatisfying. State institutions helped
to obtain the needed capital investment for industry. Bureaucrats rewarded firms that per-
formed well and sanctioned firms that did not meet strict performance standards (see also
Amsden 1989). Although Johnson does not invoke Weber as the foundation for his concep-
tion of the bureaucracy, his understanding of the bureaucracy is essentially Weberian (see
Weber 1978, especially Ch. 11).

   One body of research emerging from Johnson’s analysis of Japan has focused on the
politics of bureaucratic development to explain how effective and capacious developmental
states emerge and how Weberian bureaucracies facilitate development. Evans (1995) finds
that effective developmentalist bureaucracies balance autonomy and “embeddedness,” en-
abling them to learn from and collaborate with public and private partners without being
captured by their administrative subjects.

   Cross-national statistical work has found a that a simple measure of “Weberian-ness”
predicts economic performance in a sample of developing countries (Evans & Rauch 1999).
Conversely, the absence of a rational legal bureaucracy hinders development (Evans 1989,
Roth 1968). The origins of capable, developmental states are to be found in a political

                                             4
environment of vulnerability, paired with a lack of economic resources that forces political
leaders to invest in state structures that facilitate the delivery of essential public goods and
services and economic growth (Doner et al. 2005).1

    The contributors to Woo-Cummings (1999) examine the politics and economics of devel-
opmental states, continually returning to the role of bureaucrats and state structures. Kohli
(2004) considers structure of political competition to drive bureaucracy and performance,
and ties this ultimately to the varieties of colonial experience. These works share several
common refrains: (1) all bureaucracies are political, (2) efficacious bureaucracies are never
completely insulated from society, but (3) such bureaucracies do not become “mere tools”
either of politicians or of private interests. Ineffective bureaucracies, on the other hand,
come in many forms—unprofessional bureaucracies cannot carry out their tasks because bu-
reaucrats do not know how, weak bureaucracies are subject to capture by private interests,
and predatory bureaucracies consume the seeds of development before they germinate. Sub-
sequent work has measured different bureaucratic characteristics including career stability,
professionalization, salary scales, esprit de corps, and has used these variables to explain
                   ohm et al. 2011) and development (Cingolani et al. 2015).
corruption (Dahlstr¨

    More recent research in this tradition has focused less on bureaucratic characteristics
and more generally on the quality of government– understood as the quality of institutions
regulating government selection and replacement, the capacity to formulate and implement
sound policy, and the protection of institutions that govern economic and social interactions
(Kaufman 2005). Measurement is a central challenge in such research because of the general
difficulty of separating indicators of, for example, bureaucratic efficiency from the effects it
is supposed to predict (see Kurtz & Schrank 2007). A deeper problem is that bureaucratic
structures are both endogenous and highly correlated with other potential political sources

    1
      This represents a natural link to the related literature on European state-building (Tilly 1985), which
as been fruitfully employed to understand the presence (or lack) of strong states in Latin America or
Africa (Thies 2005, Herbst 2000).




                                                     5
of economic performance. Identifying the causal effects of bureaucracies—even assuming
away the problem of measuring their characteristics—is a hard problem.

   Both the developmental state literature and the quality of governance literatures are
characterized by a focus on macro-level measures and outcomes. The unit of analysis, typi-
cally, is the state. A focus on national bureaucracies is sensible for understanding complex
state structures and their effects, specifically because bureaucracies are features of national
politics. Case analyses such as Evans (1995) allow for a closer investigation of particular
policy choices, and highlight the complexities of bureaucracies in action, but still tend to
target national policies and organized public and private interests, with only indirect effects
on most citizens. The typical encounter of citizens with a bureaucracy, and the wider context
of this interaction, therefore goes unanalyzed. Moreover, while work on the developmental
state is useful for understanding the development on national capacity in macroeconomic
and industrial policymaking, it rarely engages challenges of service delivery in, e.g., health
care or education in the developing world. As such, it only captures one, albeit important,
facet of bureaucratic politics and performance.



2.3    Principal-Agent Problems

Quite distinct from the developmental state literature, another large literature addresses
the internal operation of bureaucracies and their interaction with elected political officials.
In particular, Weber’s view of the bureaucracy as a hierarchical organization with top-
down delegation (and monitoring) of tasks to subject matter experts has proven amenable
to theoretical investigations within a principal-agent framework. In political science and
economics this has become the dominant theoretical lens to understand bureaucracies (for a
review, see Gailmard & Patty 2012).

   Classic principal-agent accounts capture the hierarchical interaction between a principal
and an agent, e.g., between an elected official and a whole bureaucracy or between a su-


                                              6
pervisor and a frontline civil servant. In this interaction the principal delegates a task to
the agent, who in turn decides how much actual effort to exert. This type of interaction
can give rise to agency problems—divergence between the principal’s intended goals and the
agent’s actions—via moral hazard and adverse selection (Dixit 2002). The basis of agency
problems is the assumption that the agent has a preference schedule that differs from the
principal (if they were identical, then the agent would act exactly as the principal desires).
This assumption underlies most of the literature that we review here, although later in this
section, we entertain the possibility that the agent may intrinsically wish to implement the
principal’s policy.

   Research on American politics has used the principal agent framework to fruitfully in-
terrogate the interplay between elected principals and bureaucratic agencies (e.g. Whitford
2005, Snyder & Weingast 2000, Moe 1984), illuminating, e.g., the determinants of delegation
to independent, discretionary agencies (Huber & Shipan 2002). This work on delegation
of crucial policymaking tasks to politically independent agencies has also played a major
role in our understanding of central-banking in the developed and developing world (e.g.,
Adolph 2013, Keefer & Stasavage 2003). But what does this principal-agent perspective
tell us about frontline service delivery in developing countries? Principal-agent theories too
have become the theoretical bedrock for understanding the poor performance and high ab-
senteeism rates of civil servants tasked with service delivery in developing countries (Duflo
et al. 2012, Chaudhury et al. 2006). A large literature on corruption in the public sector
also relies heavily on the theoretical scaffolding of the principal-agent framework (Svensson
2005). The principal-agent framework has also been used to offer potential solutions: the
design, monitoring, and sanctioning of incentive-compatible contracts that allows principals
to induce high effort in agents (Dixit 2002). For example, performance pay contracts can
allow principals to induce increased effort by frontline civil servants, like teachers and health
care workers. Such incentive schemes, as well as other screening tools, can also be used to
affect the recruitment of individuals into the civil service to limit adverse selection problems.


                                               7
Efforts to limit corruption by civil servants also focus on civil service wages and monitoring
technologies employed by principals (e.g., Olken & Pande 2012).

    Increasingly, randomized controlled field trials have been employed to test the effective-
ness of performance pay and monitoring schemes in developing countries’ public sectors (for
a review of performance pay studies, see Hasnain et al. 2014). Explicit monetary reward and
penalty schemes for civil servants can, under very specific conditions, reduce absenteeism,
increase effort, and limit corruption (e.g. Duflo et al. 2012, Olken 2006; 2007). For example,
a field experiment by Khan et al. (2016), studying tax inspectors in Pakistan, finds that
high-powered financial incentives clearly increased revenue, without decreasing tax payer
satisfaction.2

    An alternative to explicit performance pay and increased monitoring, rotations and re-
assignments of civil servants can also be used to manage principal-agent problems. Rotation
places different agents in the same environment, which facilitates monitoring by helping
principals distinguish between outcomes due to the nature of the environment and those
due to the actions of particular agents. For example, in a private firms setting Hertzberg
et al. (2010) find that loan officers were induced to provide more accurate reports due to
rotations. States also use rotation to improve efficiency in challenging environments. In
premodern Europe, the central government rotated the most effective tax collectors to areas
in which collection was inefficient (Kiser & Kane 2001). Rotation can also be used to limit
informal ties between bureaucrats and their clients, decreasing corruption, increasing civil
servants’ dependence on the bureaucracy, and aligning the interests of civil servants with
those of their principals. As Stovel & Savage (2006) note, though, rotation is only effective
for tasks that do not require strong relationships between employees and clients and employee
knowledge of the local context. In fact, while rotation might be a useful tool for managing
principal-agent problems, little is known empirically about the effects of rotation on the

    2
      Interestingly, large parts of the revenue gains come from a small set of tax payers. They also find some
evidence of an increase in bribes to compensate bureaucrats for foregone incentive rewards.


                                                     8
quality of service delivery in the civil service more generally.

   Designing successful incentive schemes and management practices is challenging due to
the multidimensional nature of many public service tasks. Dixit (2002) argues that pub-
lic agencies face a multiplicity of principal-agent problems and challenges of multi-tasking,
which complicates simple insights from standard principal-agent models. For example, multi-
tasking—the requirement of civil servants to perform a number of distinct tasks—generates
a host of additional incentive problems and complicates moral hazard and adverse selec-
tion still further. Principals might only be able to effectively measure one of the many
tasks, inducing agents to neglect the other dimensions of their job (e.g., teaching to the test;
Holmstrom & Milgrom 1991). Similarly, Kiewiet & McCubbins (1991) argue that ‘collec-
tive principals’ generate additional agency losses because of disagreements among principals
about the tasks of the bureaucracy. Hammond & Knott (1996) demonstrate that multiple
principals can lead to agency losses even with perfect information about the bureaucracy’s
objectives and actions. The general result from this literature is that multiple principals
inhibit monitoring and increase agency slack.

   These theoretical concerns have been corroborated by empirical studies. For example,
in a detailed study of Nigeria’s civil service, Rasul & Rogger (2015) provide evidence that
increased autonomy correlates with higher productivity of bureaucrats, while increased mon-
itoring lowers output by bureaucrats. They show this lack of success in active management
by supervisors is due to poor targeting of incentives and multi-tasking problems. Similarly,
Gulzar & Pasquale (forthcoming) study a setting with multiple principals in India. Using
data on India’s National Rural Employment Guarantee Scheme, they show that locations
where local bureaucrats are accountable to a single politician perform better.

   Research in behavioral economics, sociology, and public administration also suggests
problems with explicit monetary reward and monitoring schemes to induce performance in
frontline civil servants. In contrast to spare principal-agent models that assume different
preference schedules between principal and agent, this type of work often emphasizes the

                                               9
esprit de corps or the intrinsic motivation of public servants as crucial bureaucratic char-
acteristics that affect overall productivity (Perry & Hondegheim 2008). Intrinsic motivation
or a strong organizational culture of efficiency and hard work operate as safeguards against
shirking (Grindle 1997). Such organizational cultures were held, in the literature cited above,
to be characteristic of public bureaucracies in developmental states.3 In general, there is little
work that traces the institutions and organizational structures that help to generate esprit
de corps within bureaucracies.4

   This perspective suggests that public-minded bureaucratic agents want to deliver services
and do not have to be made accountable via explicit, formalized monitoring and incentive
schemes. In fact, introducing pecuniary rewards that violate existing organizational norms
might demoralize current civil servants and deter intrinsically motivated people from enter-
ing the civil service in the first place (Keefer & Banuri 2013). Some studies though suggest
that high public sector wages and attractive career opportunities attract high quality appli-
            o et al. 2013) and that intrinsic motivation of civil servants amplifies the positive
cants (Dal B´
effects of increased monitoring (Callen et al. 2015). To complicate matters further, work on
corruption has also suggested that corrupt bureaucracies might attract bureaucrats who are
primarily interested in rent-seeking opportunities (Cowley & Smith 2013). Laboratory ex-
periments suggest that job seekers with aspirations for the civil service exhibit a greater
likelihood to cheat in experimental games and reveal a preference for corruption (Banerjee
et al. 2015).

   Theoretical work by Gailmard & Patty (2007) suggests that in order to recruit ‘zealots’
(i.e., publicly minded and intrinsically motivated individuals) into the civil service, civil

   3
       See, however, Ramseyer & Rosenbluth (1993) for a contrary view rooted in principal-agent theory.

   4
    One recent exception from the U.S. is Carpenter (2001), which explores how Progressive-era American
bureaucracies created “reputations for innovation, service, and moral protection that were embedded in
multiple networks” (32). Carpenter’s argument begins with the efficacy of the bureaucracy and explores
how this supports bureaucratic autonomy, rather than illuminating the origins of the norms within the
bureaucracy.



                                                     10
service wages must be below market wages to ensure that only zealots remain in the civil
service. This condition is quite unlikely to hold in most developing country contexts (Finan
et al. 2015), and a common refrain is that state agencies are populated by incompetent and
uncaring bureaucrats for whom the civil service is a safe and relatively lucrative career. Thus
merit and promotion procedures—even if they replicated those of Western bureaucracies—
are unlikely to produce a civil service populated by those intrinsically motivated to serve
the public. Career backgrounds and career aspirations can also affect civil servants’ policy
preferences and decisions. Using data from central banks from across the developed and
developing world, Adolph (2013) finds that central bankers with different career trajectories
choose predictably different monetary policies.

   A more recent set of studies has also investigated the effect of electoral competition on
monitoring, delegation, bureaucratic fragmentation, and bureaucratic performance. Elec-
toral competition should improve bureaucrats’ performance. Voters hold their representa-
tives accountable via elections, which in turn encourages politicians to monitor bureaucrats
to ensure that they provide public goods and services. In a study of the Nigerian civil
service, Rogger (2014) finds that electoral competition leads politicians to delegate public
goods provision to more capable bureaucrats, improving overall services provision. Evidence
from Pakistan also indicates that electoral competition is crucial for making politicians hold
civil servants accountable (Callen et al. 2014). Doctor absenteeism is substantially lower in
electorally competitive districts, higher for doctors that are personally connected to politi-
cians. They also find that interference with the sanctioning of health workers is higher in less
competitive areas, and that the use of smartphones for monitoring health workers’ assigned
visits to health care centers is only effective in competitive areas.

   Although competition may increase politicians’ incentives to monitor bureaucrats, politi-
cians who face less electoral competition and thus enjoy longer tenures might be able to
incentivize bureaucrats by offering future rewards. Looking at bureaucrats in India, Nath
(2014) finds that re-election concerns do not have a meaningful effect on the monitoring ef-


                                              11
forts of politicians. Instead, politicians with long tenures are able to provide better dynamic
contracts that incentivize effort by local bureaucrats. In Pakistan, Gulzar (2015) documents
the perverse incentives between ruling party incumbents and civil servants. Ruling party
politicians that win close elections are willing to tolerate shirking of civil servants, in ex-
change for their political support. This suggests that if politicians can leverage their control
over patronage opportunities and protect civil servants, bureaucratic service delivery will
suffer despite the presence of elections. Relatedly, Iyer & Mani (2011) show, using data from
India, that politicians prefer control over productivity, leading to assignments that mismatch
skilled bureaucrats with less important tasks. This incentivizes junior bureaucrats to invest
in loyalty instead of skills. In the case of Russia, the relative success of its public health
initiatives to reduce alcohol consumption, tobacco use and traffic fatalities is due, in part,
to the strong and clear signals Putin and his inner circle sent to the police about which
rules to enforce on the ground, such as writing fines for smoking in illegal places (Brooke &
Gans-Morse 2016).

   Monitoring does not have to be done by elected officials or high-level bureaucrats, but
can be outsourced to relevant constituents (i.e., citizens or businesses). Banks & Weingast
(1992) note that constituents perform an important monitoring function for politicians, but
that “agencies designed to serve unorganized groups allow bureaucrats to extract too much
of the gains from policymaking at the expense of both politicians and their constituents,
and hence politicians fail to create them because they are not worth the costs” (512). This
is particularly likely to characterize frontline service delivery in developing countries, for
the poor and marginalized are classic examples of unorganized interest groups. Although
Banks and Weingast argue that bureaucracies that serve unorganized interests should not
exist, their argument implies that when they do exist, they are unlikely to be effectively
sanctioned, and subject to high agency loss.

   In fact, donor organizations have invested significant resources in interventions designed
to mitigate the information constraints which they assume are hampering monitoring and


                                               12
service provision. The conventional wisdom is that information about the quality of govern-
ment services and politician performance will help the poor to demand greater accountability
from service providers and/or politicians and will lead to welfare improvements. The focus
of these interventions has been on information about public spending, especially in health
and education, corruption, and legislator performance. The results of these interventions,
however, have been mixed (This literature has been the subject of several literature reviews
including Kosack & Fung (2014).).

   Moreover and related to these mixed empirical findings, Banks and Weingast’s argument
presumes that (1) politicians actually consider their responsibilities to ensure efficient bu-
reaucratic function, and (2) that politicians are accountable to voters. If politicians can
respond to citizen monitoring with direct constituency service, then this would bypass the
bureaucratic function—the solution to a bureaucratic failure is not reform of the bureau-
cracy, but rather sidelining it. But if voters can imperfectly monitor politicians, then there
is a nested, two-level principal-agent relationship: bureaucrats are the agents of politicians,
who are in turn the agents of the mass public.

   Viewed through this lens, the requirements for a “McNollgast-style” (McCubbins et al.
1987) solution to the problem of bureaucratic performance—where administrative proce-
dures can preserve the preferences of the legislature—are particularly unlikely to generate
incentives for efficient and efficacious bureaucratic performance in most developing country
contexts. Frontline service delivery problems will be particularly acute: unorganized con-
stituents must monitor bureaucrats on behalf of politicians who may not have incentives to
respond to citizen demands, and who may not look to bureaucratic reform as the solution
to bureaucratic underperformance anyway. We would not be surprised to find that frontline
service delivery is poor under such conditions.

   While principal-agent models offer powerful insights, they also narrowly cast our un-
derstanding of bureaucratic processes as a form of internal management. The Procrustean
dominance of the principal-agent model may limit the types of questions we ask when it

                                              13
comes to the internal dynamics of bureaucracies, and more importantly, the interaction of
bureaucratic agents with politicians, citizens, and businesses. For example, while studies
of corruption have benefited from principal-agent models for understanding the behavior of
corrupt agents, Callen et al. (2014) also point out that corruption in the public sector is
often embedded in larger clientelistic machine politics, whose dynamics are not captured
solely with narrow principal-agent models. Cruz & Keefer (2015) demonstrate that good
bureaucratic performance emerges not from accountability to voters or citizens, but rather
from programmatic parties that allow politicians to resist the sorts of clientelist practices
that undermine bureaucratic effectiveness. Persson et al. (2013) argue that corruption in
general is not a principal-agent problem, but rather a collective action problem.

    As discussed in the preceding two sections, developmental state and principal-agent mod-
els offer useful insights for the study of bureaucracies, but they also come with inherent dis-
advantages when it comes to understanding the on-the-ground delivery of public services in
the developing world. Developmental state arguments focus too narrowly on national-level
macro policymaking. Principal-agent models have a lot to offer for understanding the in-
centives and management of frontline civil servants, but often focus too narrowly on explicit
monitoring and reward schemes. More generally, both principal-agent and developmental
state approaches have not sufficiently explored variations in the forms of social embedded-
ness in which frontline bureaucratic agents operate. We explore the issue of bureaucratic
embeddedness in depth with the example of street-level bureaucrats.



3     Embeddedness and Street-Level Bureaucracies

Street-level bureaucrats are the public service workers—including teachers, police officers,
social workers, public lawyers and nurses—who interact directly with citizens and who imple-
ment policy on the ground. Given the nature of street-level bureaucracies, they represent a
useful case to interrogate the bureaucratic politics of service delivery in the developing world.


                                               14
Like all civil servants, street-level bureaucrats enjoy discretion in policy implementation, but
street-level bureaucrats are distinct from others because they are the final step in the chain of
policy implementation. Weber recognized that by nature bureaucracies have a large amount
of unregulated social and political power. Due to the nature of their job, bureaucrats must
make decisions on the spot and exercise discretion in identifying the beneficiaries for ser-
vices and resources, the method by which services are distributed and by choosing how and
when to enforce government policies (Lipsky 1980). The discretion of frontline providers
in decision-making allows them to become de facto policy-makers through the implementa-
tion process (Lipsky 1980). In controlling the implementation of policies, frontline providers
directly influence the quality of life for many citizens. They determine which children are
educated and how, who receives life saving medication or an identity card, who is fined or
arrested for violating a law or regulation, and how much taxes are paid and by whom.

   Street level bureaucrats face a different type of workplace environment than other types
of officials: they have to deal with clients’ personal reactions to their decisions (Lipsky 1980,
9). Frontline service providers face unique pressures and dilemmas including inadequate
resources and often unclear policy goals and performance measures. Further, there is a
continual tug of war between the street-level bureaucrat’s desire for autonomy and his or
her supervisors’ desire for surveillance and control. Commonly, there is an uneasy if not
hostile relationship with clients on whom the street-level bureaucrats can impose substantial
punishment. Above all, there is the bedrock problem of whether to seek uniformity of
treatment in dealing with clients or whether, as best as possible, to tailor services and
punishments to the particular needs and problems of citizens or clients.

   While training for frontline service providers is important in equipping them with the
skills to confront these and pother pressures and to fulfill their tasks, very few studies look
at its effects. A randomized experiment on police in Rajasthan found that the provision of
training on professional (i.e., investigation and methods) and soft skills (i.e., communica-
tion and mediation) improved victim satisfaction with the police (Banerjee et al. 2014). A


                                              15
literature review on community health workers who are selected, trained and work in the
communities from which they come and who are not necessarily members of the civil service,
finds that improved training, continuous support and supervision tends to improve perfor-
mance and increase community members’ use of the health workers (Lehmann & Sanders
2007). Yet, the ambiguous role of community health workers within the broader health care
system and civil service can limit the training they receive and the scope of the services they
are able to provide, which weakens communities’ confidence in the health workers’ efficacy.



3.1     Embeddedness on the Front Lines

The focus of Evans (1995) and much of the developmental state literature is on a small
number of civil servants in core ministries involved in planning and policy and monitoring
the rest of the state and their linkages to powerful business interests. Until relatively recently,
the East Asian developmental states did not deliver substantial social services or transfers to
its populations.5 . Given the growth of public service delivery across developing states, our
focus is on the frontline bureaucrats who ultimately deliver these services, their links with
community members and the factors that enable for effective service delivery ranging from
health, education, public utilities, agricultural and social services to land administration,
identity cards and marriage certificates (Tsai 2007, Tendler & Freeheim 1994, Ricks 2016).

   Granovetter (1985) used the concept of embeddedness to refer to the social relationships
that influence an individual’s economic decisions. We use the concept to refer to the social
relationships that influence civil servants’ actions. Building on Granovetter’s insights into
the benefits of embeddedness for economic behavior, this literature suggests that street level
bureaucrats who are embedded in the communities in which they serve produce better policy
outcomes. The concept is particularly relevant to frontline bureaucrats by nature of their

   5
     One indicator of the relative smallness of the developmental state is Government Final Consumption
Expenditure, which between 1970 and 1995 was notably lower as a percent of GDP in Japan, South Korea,
and Singapore than in the United States, United Kingdom, or Australia, according to the World Development
Indicators.


                                                   16
frequent interaction with clients. Embeddedness reduces bureaucrats’ costs of information
gathering and information sharing. Providers who are embedded in the communities they
serve are likely to have a better sense of their needs and can target their efforts at fulfilling
those needs. In a study of water user associations in Indonesia, Ricks (2016) finds that
the frequency of contact between state officials and the beneficiaries, or farmer groups, is
much more important for the success of the group than training programs. Regular inter-
action between irrigation officials and farmers reduced the transaction costs for information
sharing. Through the frequent exchange of knowledge, the irrigation officials were able to
convince farmers of the usefulness of the water user groups. Training programs that were not
accompanied by other interactions between the officials and farmers were counterproductive
as farmers saw these one-off training events as only serving the officials’ rather than their
needs.

   A novel study by Bhavnani & Lee (2015) exploits the quasi-random assignment of bu-
reaucrats in India to document the value of local connections for the performance of the
civil service. They find that a civil servant’s education level, a proxy for their skills, does
not have an effect on their productivity, but bureaucrats who have established ties to their
assigned communities are able to provide a higher level of public goods.

   A different dimension of embeddedness pertains to the connections and environment civil
servants operate within the bureaucracy itself. This includes the importance of workplace
diversity for creative problem solving capacity (Lazear 1999, Shore et al. 2009). Rasul &
Rogger (2015) consider the effects of ethnic diversity in a bureaucracy for public service pro-
ductivity. Extracting information on the success of 4,700 distinct projects in Nigeria’s civil
service, they document a positive relationship between team diversity and project comple-
tion rates. Others document the conflict that emerges when different ethnic groups compete
for scarce public sector jobs (Brown 1999).

   Embeddedness also enables providers to frame their messages and perform their tasks
in culturally appropriate ways which will increase the likelihood that recipients adopt their

                                              17
providers’ recommendations and decisions. This is certainly true for health care workers who
are trying to persuade community members to adopt certain behaviors, such as giving birth
in health facilities rather than with traditional birth attendants, immunizing their children,
and feeding children more nutritious meals. Much of the literature addressing this topic
comes from field as diverse as public health (e.g., Kelly et al. 1991), network analysis (e.g.,
Burt 1999), and marketing (e.g., Chan & Misra 1990). More research is needed in political
science on the role opinion leaders play in helping frontline providers to accomplish culturally
sensitive tasks.

   Relatedly, repeated and positive interactions between providers and beneficiaries are
likely to generate views of the provider as trustworthy. Community members are more likely
to adopt the messages of those who they view as trustworthy. For example, frontline health
                    a, Brazil earned the trust of their clients by maintaining strong social
care workers in Caer´
ties to their communities, repeatedly visiting households and helping mothers with household
tasks, which went beyond their formal responsibilities. After repeated visits to households,
the health care workers were able to talk to families about sensitive health issues, such
as breastfeeding and better hygiene or nutrition, and the clients were responsive to their
messages. Knowing that community members viewed them as trustworthy motivated the
health agents to expend effort beyond what was required in their assignments (Tendler &
Freeheim 1994).

   Related to challenges of monitoring, embedded frontline workers may face direct com-
munity pressure that will prevent them from shirking and help to ensure that they are
accountable for public goods and services. In parts of China, local solidarity groups with
high moral standing in the community established and enforced unofficial norms and rules
by which bureaucrats and state officials were expected to abide. Through these groups,
communities are able to hold local officials directly accountable even when in the absence
of formal accountability institutions (Tsai 2007). In most cases, community members will
only exert social pressure on agents if they are aware of agents’ obligations and their rights


                                              18
as citizens, which is consistent with the literature on the role of asymmetric information in
principal-agent relationships (Besley 2006). In anticipating this problem, the state govern-
            a, Brazil empowered the public—including rejected civil service applicants—to
ment in Caer´
monitor civil servants. The state invested in public information campaigns that educated
citizens about the need to respect the civil service but also their obligation to monitor civil
servants, the obligations of service providers, and the steps they should take in response to
infractions (Tendler & Freeheim 1994).



3.2    Embeddedness, Corruption, and Clientelism

The same social networks that can put pressure on civil servants and elected officials to
provide the goods and services to which they are entitled can also demand favors from civil
servants (see Kiser & Sacks 2009). In many contexts, civil servants participate in what
outsiders perceive as corruption but might be viewed locally as reciprocal obligations of
kinship and community solidarity (Granovetter 2007). In some cases, when civil servants
receive salary raises, as part of overall civil service reform, they receive an increase in demands
from their kin networks, which may lead the civil servants to take bribes to compensate for
the higher expenses (Fjeldstad 2005).

   Related to corruption and embeddedness in community networks, in many developing
countries clientelism and patronage are also often pervasive (Kitschelt & Wilkinson 2007)
and used for political, social, and economic control. Clientelism is an inherent feature of
most civil service hiring in the developing world (Grindle 2012). Politicians in clientelistic
systems need access to state resources at all levels to sustain long-term clientelistic exchange
and finance vote and turnout buying during election times (Stokes 2005, Nichter 2008).
Control of the bureaucracy is also needed to sustain high-level multi-group coalitions via,
e.g, cabinet appointments (Arriola 2013). This becomes particularly relevant in, e.g., post-
conflict settings, in which the need to rebuild state capacity and provide essential services is



                                                19
competing with the need to sustain fragile political bargains via extensive patronage hiring
into the civil service.

   Public sector jobs are one of the most common and important currencies within clientelist
networks (Penfold-Becerra 2007). Using their access to state resources, politicians exploit
their control over the civil service to offer patronage jobs and promotions to political support-
ers in order to win elections (Golden 2003) and deliver benefits to their co-ethnics (Franck
& Rainer 2012).

   As a result, embeddedness in clientelistic networks represents a form of social ties that
makes civil servants essential cogs in an, albeit flawed, accountability mechanism between
voters and politicians. The question, though, is whether this type of embeddedness is ben-
eficial or detrimental for service delivery? The classic literature on clientelism argues that
patronage politics diverts important resources away from public goods and services delivery
(Hicken 2011, Keefer 2007). Service delivery that is conditioned by a political rationale and
controlled by opaque clientelistic machines runs counter to a modern understanding of a
neutral and effective civil service. Moreover, the practice of allocating civil service appoint-
ments to maintain clientelistic machines and ethnic alliances often leads to the unsustainable
increase in the state’s wage bill. Following Zambia’s independence, President Kuanda main-
tained support for a fragile UNIP and state by allocating appointments to state offices to
political leaders representing Zambia’s ethnic factions (Bratton 1992). From 1963 to 1968,
the number of Zambian central government employees increased from 1,357 to 7,509 (Szeftel
1982, 6). In Senegal, the use of the bureaucracy for rent-seeking and particularistic interests
paralyzed the Senegalese state in terms of its ability to implement programmatic economic
policies (Boone 1990, 352). While increasing the representative nature of a state’s bureau-
cracies increases the state’s wage bill, it can also mitigate conflict among groups who feel
excluded from the bureaucracy and the many benefits public sector employment offers, in-
cluding access to rents, patronage, stable employment and discretion in enforcing laws like
taxation and applying force through the police (Esman 1999, Wimmer 1997).


                                              20
   At the same time, clientelistic embeddedness might offer advantages that have gone
unrecognized in the literature thus far. For example, clientelistic exchange works best with
poorer voters (Weitz-Shapiro 2012). This implies that actual benefits are being delivered to
very poor members of society, even if with a political bias.

   “Effective” clientelism means effective delivery of targeted benefits, and likely requires a
modicum of state capacity. It could be that standard welfare services in democracies are cap-
tured by the middle class and do not cater to poorer citizens (Ross 2006). Similarly, Grindle
(2012) argues that many civil service reforms start in particular ministries or departments
and require the personal control of reform-minded officials in order to create new and effective
units. Influential clientelistic networks can generate opportunities for reform that eventually
transforms the whole bureaucracy. Grindle (2012) also points out that modern ideals of
a neutral, rule-based, and face-less bureaucracy often fails to deliver essential services and
cater to the needs of vulnerable populations, precisely because civil servants can hide behind
regulations and procedures, are insulated from societal pressures, and have little leeway to
adapt to local conditions. Especially in the context of developing countries it remains un-
clear whether formal, de jure reforms of the civil service, that institute meritocratic hiring
and are meant to limit the role of clientelism, can work. Such reforms potentially sever a
functioning accountability mechanism—albeit one that violates Western norm of democratic
accountability. Work by Andrews, Pritchett, and Woolcock (Andrews et al. 2013, Pritchett
et al. 2013) similarly argues that many reforms meant to improve state capacity are mere
‘isomorphic mimicry’ that do not improve functional capacity. By mindlessly trying to create
a bureaucracy that looks like Weber’s image of the Prussian state, many developing country
governments (and international donors) merely hide dysfunctional structures. Instead, they
argue, room should be given for civil servants to engage in experimentation, leveraging of
local information, and adaption of practices—all of which are tied to some form of social
embeddedness of frontline bureaucrats.

   The tensions reviewed in this section recall the debate about the “efficient grease” hy-


                                             21
pothesis (Kaufmann & Wei 2000), part of a broader debate about the role of corruption and
development usefully reviewed by Bardhan (1997). Although this debate probably can never
be resolved, our review of frontline service delivery suggests that rather than prescribing a
single institutional form, political scientists might instead concentrate on exploring which
factors explain civil servant effectiveness.



3.3    Consequences of the Interaction of Civil Servants and Citizens

Many citizens encounter government primarily through their interactions with frontline ser-
vice providers. Street level bureaucrats provide individuals with information about govern-
ment’s goals and how it works, its responsiveness, and how it values their particular social
group (Soss 2005). Beyond corruption, a fairly large literature documents the frustrating,
arbitrary nature and often, pernicious, interactions between citizens and frontline service
providers (e.g., Gupta 2012, Auyero 2011). Gupta (2012), for example, documents how
Indian bureaucrats tasked with identifying beneficiaries of old age pensions guess potential
beneficiaries’ ages rather than rely on objective criterion like birth certificates which are often
non-existent. In many contexts, police are the source of citizens’ fear rather than protection
(Diaz-Cayeros et al. 2015). Studies from China document labor dispute plantiff’s negative
evaluations of the legal systems in terms of its fairness and effectiveness (Gallagher 2006)
and the resistance of villagers to excessive and often illegal fees and levies, as well as brutal
tax collection methods (Bernstein & L`
                                     ‘u” 2003, O’Brien & Li 1995).

   Whether positive or negative, citizens evaluate government based in part on their in-
teraction with frontline providers (Lipsky 1980). The interaction between citizens and civil
servants shape citizens’ willingness to voice grievances about the particular program, their
perceptions of political efficacy, their willingness to participate in civic engagement and po-
litical action, and ultimately, and their compliance with rules and regulations (Soss 2005,
Tyler 1990, Levi et al. 2009). Based on an analysis of survey data from Chicago, Tyler (1990)



                                               22
finds that citizens’ decisions about whether to obey legal rules is based more on the per-
ceived fairness of the way police officers and judges make decisions than on the favorability
or fairness of the outcome. For many, to be respected and heard is actually more important
than winning a case (Tyler 1990, 149-150).

    The literature on the political consequences of citizen-bureaucrat interactions outside of
the U.S. and Europe is still in its infancy, although recent work has begun to explore such
phenomena. For example, using Afrobarometer data, Levi et al. (2009) find citizens’ judg-
ments of the fairness in government procedures and decisions appear to correlate with their
beliefs regarding taxation. Other work asks whether non-state actor provision of services
affects citizens’ beliefs about their government’s performance and legitimacy. When citizens
think that foreign actors such as donors and international NGOs are involved in service de-
livery, they are more likely to credit their government for helping to leverage the aid and, in
turn, more likely to view their government more positively (Sacks 2012, Dietrich & Winters
2015). Future work will draw on the concepts and theories of existing scholarship (Goodsell
1981, Mettler 1998, Kumlin & Rothstein 2005) while adding the unique insights from the
developing country context.



4     Conclusion: New Directions for Old Questions

The performance of national bureaucracies in sustaining national economic performance and
implementing national policy has long been central to understanding political development.
But as our review has illustrated, bureaucracies are not monolithic institutions. The ways
that citizens, firms, and constituents interact with bureaucracies vary, with implications
for understanding both how bureaucracies function and their political and economic effects.
We conclude this essay, then, by outlining a framework through which to capture these
effects by disaggregating bureaucracies across time, space, task, and client. Understanding
different forms of embeddedness across these different dimensions is critical for understanding


                                              23
variation and change in bureaucratic performance and citizens’ evaluation thereof.

   Variation in bureaucracies across time can reflect both changes in external environment
or changes in internal organization. Examples of changes in the external environment may
include global technological developments that create new regulatory needs (i.e. internet
regulation), international economic integration that reduces bureaucrats’ scope for effective
intervention, or the very process of economic development itself. Changes in internal orga-
nization may include reforms that target bureaucratic organizational structures or political
realignments that merge, split, or redirect existing bureaucracies. Internal and external
drivers of bureaucratic change may interact, with unexpected results.

   Spatial variation in bureaucracies emerges from the observation that all national bu-
reaucracies are hierarchical organizations whose regional or local structures may vary. For
example, local offices of a bureaucracy may vary in staff numbers, or staff qualifications.
Alternatively, the same bureaucratic function may be more or less difficult to carry out de-
pending on the local context. For example, regional offices of a bureaucracy tasked with
frontline service delivery may have larger case loads in communities with greater service
needs. In the former case, bureaucracies vary spatially because their own structures vary;
in the latter cases, bureaucracies vary spatially because the local context in which the same
bureaucratic structure operates varies. In either case, this variation may lead to spatial
differences in bureaucratic performance, or how citizens interact with bureaucracies.

   Bureaucracies also vary according to the tasks that they carry out. Among bureau-
cracies that have face-to-face interactions with citizens, we can distinguish, for example,
among those that award licenses and permissions, those that provide services, and those
that create infrastructure. License- and permission-granting bureaucracies should generate
relatively more opportunities for bribery, corruption, and holdup than will service-providing
bureaucracies, although informal fee-for-service arrangements may exist in the latter as well.
Infrastructure-creating bureaucracies create opportunities for corruption during procurement
and contracting that are invisible to the beneficiaries of that infrastructure. Embeddedness

                                             24
for frontline service workers in local communities may be more beneficial for task effectiveness
than would be embeddedness of licensing officials in the firms that they regulate.

   Finally, bureaucratic politics varies in the kinds of clients with whom bureaucrats interact.
Some bureaucracies interact directly with individual citizens: granting licenses or delivering
health services, for example. Among those, some bureaucracies interact primarily with poor,
disadvantaged, or marginalized population. Other bureaucracies’ main interactions may be
with firms, trade groups, and other bureaucracies. Within the same bureaucratic institution,
there are a variety of kinds of clients: trade ministries interact horizontally with ministries
(finance, labor, planning), and vertically with both trade associations and individual firms.

   A final consideration is that national factors may condition each of these kinds of varia-
tion. Frontline service delivery in a geographically compact developing nation may exhibit far
less spatial variation than does frontline service delivery in an archipelagic nation such as the
Philippines or in a geographically diverse country such as Cameroon or Brazil. Bureaucratic
tasks may vary more in certain contexts than in others: a ministry of finance or a taxation
authority in a unitary state is more likely to have uniform regulatory codes than in a federal
system. Accountability rules and bureaucrat-politician linkages also vary across countries,
as do the types of tasks delegated to particular bureaucracies and institutions. As a result,
drawing inference about bureaucratic policies from cross-national comparisons—the preoccu-
pation of much of the classic literature on bureaucracies and the developmental state—may
be less productive than drawing inferences that exploit variation within countries.

   Conceptualizing variation in bureaucratic politics across time, space, task, and client
helps to illuminate what aspects of bureaucracies are being captured in existing work. It
also reveals that a single reform or policy change may result in changes across various dimen-
sions at the same time, with attendant consequences for how to interpret the consequences.
For example, a retasking of a health ministry to provide mosquito nets to underserved com-
munities involves a change in both task and client. Deconcentrating planning from national
to regional offices induces spatial variation while potentially changing bureaucratic clients as

                                               25
well. Some central questions that political scientists are well positioned to answer include


  1. Does bureaucrats’ embeddedness have different consequences across tasks and clients?

  2. If monitoring is not possible, then what (if anything) sustains good bureaucratic per-
     formance?

  3. For what kinds of tasks, and at what level of administration, is bureaucratic discretion
     desirable?

  4. How does bureaucratic performance shape citizenship, and does this vary across tasks,
     clients, or bureaucracies? Do personal or impersonal citizen interactions have different
     effects?

  5. What tasks, conditions, or organizational structures facilitate cooperation with citi-
     zens?

  6. How do conditions for bureaucratic effectiveness vary across national or regional con-
     text?


   Disaggregating bureaucracies and exploring variation will require political scientists to
leverage research designs that have been underutilized in the study of bureaucracies to date.
Classic work on bureaucracies relied on historical case studies, cross-case comparisons, and
cross-country regression analysis. Today, there are opportunities for creative research designs
that include, but also go beyond, current trends towards causal identification. Naturally,
large-scale field experiments will become increasingly viable for the study of bureaucratic
performance, as governments and donors become more willing to systematically evaluate
civil service reform programs (for some current examples see Finan et al. 2015). Likely of
larger importance is the wealth of administrative data, collected by bureaucracies themselves,
that can shed light on the politics of bureaucratic performance. Bureaucracies, by their very
nature, are data producing institutions. They often collect systematic data on bureaucrats

                                              26
themselves, documenting their work history, their assignments, performance evaluations,
salaries, and promotions. Bureaucracies also keep track of other inputs in the bureaucratic
production functions: agency budgets and procurement are increasingly tracked digitally.
Moreover, some bureaucracies also document core outputs—e.g., number and type of health
services provided. Last, bureaucracies’ main output—text—has been the focus of several
qualitative, including ethnographic, studies (Hetherington 2011, Gupta 2012, Hull 2012a;b).
Modern text processing methods may allow for the quantitative analyses of bureaucratic
documents and produce novel insights about bureaucratic politics.

   Such data can be used for rich, micro-level studies of bureaucratic politics, although
negotiating access to this internal information will remain a crucial challenge for social
scientists. Increasingly, data produced by bureaucracies themselves can be validated and
complemented by external data. Low-cost cell phone surveys can be used to elicit civil
servants and citizens attitudes about bureaucratic performance, while modern information
technology has also allowed citizens to better document corruption and failures in service
delivery.

   We believe accessing this potential wealth of data, combined with increased attention to
design-based inference, will produce a number of useful insights for the study of bureaucra-
cies. At the same time, we also advocate the return to classic qualitative methods in the
study of bureaucracies. Serious ethnographies of bureaucrats, life histories, and qualitative
explorations of social networks can shed light on the role of embeddedness, the values and
self-understandings of bureaucrats, and the micro-logic of bureaucratic action (Oldenburg
1987, Marcesse 2016). Such work can provide the necessary grounding for broader, quanti-
tative investigations; it can also reveal what is missing from quantitative analyses, including
helping to understand the production of data itself.




                                              27
References

Adolph C. 2013. Bankers, bureaucrats, and central bank politics: The myth of neutrality.
  Cambridge University Press

Amsden A. 1989. Asia’s Next Giant: South Korea and Late Industrialization. New York,
  NY: Oxford University Press

Andrews M, Pritchett L, Woolcock M. 2013. Escaping capability traps through problem
  driven iterative adaptation (PDIA). World Development 51:234–244

Arriola LR. 2013. Capital and Opposition in Africa: Coalition Building in Multiethnic Soci-
  eties. World Politics 65:233–272

Auyero J. 2011. Patients of the state: An ethnographic account of poor people’s waiting.
  Latin American Research Review 46:5–29

Banerjee A, Chattopadhyay R, Duflo E, Keniston D, Singh N. 2014. Can Institutions be
  Reformed from Within? Evidence from a Randomized Experiment with the Rajasthan
  Police. Working Paper

Banerjee R, Baul T, Rosenblat T. 2015. On self selection of the corrupt into the public sector.
  Economics Letters 127:43–46

Banks JS, Weingast BR. 1992. The Political Control of Bureaucracies under Asymmetric
  Information. American Journal of Political Science 36:509–524

Bardhan P. 1997. Corruption and development: A review of issues. Journal of Economic
  Literature 35:1320–1346

Bernstein TP, L`
               ‘u” X. 2003. Taxation without representation in contemporary rural china.
  Cambridge ; New York: Cambridge University Press




                                              28
Besley T. 2006. Principled Agents? The Political Economy of Good Government. Oxford
  University Press

Bhavnani RR, Lee A. 2015. Leadership and Development: Education, Embeddedness and
  the Performance of Indian Bureaucrats

Boone C. 1990. State Power and Economic Crisis in Senegal. Comparative Politics 22:341–
  357

Bratton M. 1992. Zambia starts over. Journal of Democracy 3:81–94

Brooke HSG, Gans-Morse J. 2016. Putin’s crackdown on mortality: Rethinking legal nihilism
  and state capacity in light of Russia’s surprising public health campaigns. Problems of
  Post-Communism 63:1–15

Brown DR. 1999. Ethnic politics and public sector management in Trinidad and Guyana.
  Public Administration and Development 19:367–379

Burt RS. 1999. The social capital of opinion leaders. The Annals of the American Academy
  of Political and Social Science 566:37–54

Callen M, Gulzar S, Hasanain A, Khan Y. 2014. The Political Economy of Public Employee
  Absence: Experimental Evidence from Pakistan. Working Paper

Callen M, Gulzar S, Hasanain A, Khan Y, Rezaee A. 2015. Personalities and Public Sector
  Performance: Evidence from a Health Experiment in Pakistan. Working Paper 21180,
  National Bureau of Economic Research

Carpenter DP. 2001. The Forging of Bureaucratic Autonomy: Reputations, Networks, and
  Policy Innovation in the Executive Agencies, 1862-1928. Princeton, NJ: Princeton Uni-
  versity Press

Chan KK, Misra S. 1990. Characteristics of the opinion leader: A new dimension. Journal
  of advertising 19:53–60

                                              29
Chaudhury N, Hammer J, Kremer M, Muralidharan K, Rogers FH. 2006. Missing in Ac-
  tion: Teacher and Health Worker Absence in Developing Countries. Journal of Economic
  Perspectives 20:91–116

Cingolani L, Thomsson K, de Crombrugghe D. 2015. Minding Weber More Than Ever? The
  Impacts of State Capacity and Bureaucratic Autonomy on Development Goals. World
  Development 72:191–207

Cowley E, Smith S. 2013. Motivation and mission in the public sector: evidence from the
  World Values Survey. Theory and Decision 76:241–263

Cruz C, Keefer P. 2015. Political Parties, Clientelism, and Bureaucratic Reform. Comparative
  Political Studies 48:1942–1973

       ohm C, Lapuente V, Teorell J. 2011. The Merit of Meritocratization: Politics, Bu-
Dahlstr¨
  reaucracy, and the Institutional Deterrents of Corruption. Political Research Quarterly
  65:656–668

     o E, Finan F, Rossi MA. 2013. Strengthening State Capabilities: The Role of Financial
Dal B´
  Incentives in the Call to Public Service. Quarterly Journal of Economics 128:1169–1218

Diaz-Cayeros A, Magaloni B, Romero V. 2015. 10. caught in the crossfire: the geography of
  extortion and police corruption in mexico23. Greed, Corruption, and the Modern State:
  Essays in Political Economy :252

Dietrich S, Winters MS. 2015. Foreign aid and government legitimacy. Journal of Experi-
  mental Political Science 2:164–171

Dixit A. 2002. Incentives and Organizations in the Public Sector: An Interpretative Review.
  Journal of Human Resources 37:696–727




                                            30
Doner RF, Ritchie BK, Slater D. 2005. Systemic Vulnerability and the Origins of Develop-
  mental States: Northeast and Southeast Asia in Comparative Perspective. International
  Organization 59:327–361

Duflo E, Hanna R, Ryan SP. 2012. Incentives Work: Getting Teachers to Come to School.
  American Economic Review 102:1241–78

Esman MJ. 1999. Public administration and conflict management in plural societies: The
  case for representative bureaucracy. Public Administration & Development 19:353

Evans P. 1995. Embedded Autonomy. Princeton, NJ: Princeton University Press

Evans P, Rauch JE. 1999. Bureaucracy and Growth: A Cross-National Analysis of the Ef-
  fects of “Weberian” State Structures on Economic Growth. American Sociological Review
  64:748–765

Evans PB. 1989. Predatory, developmental, and other apparatuses: A comparative political
  economy perspective on the Third World state. Sociological Forum 4:561–587

Finan F, Olken BA, Pande R. 2015. The Personnel Economics of the State. Working Paper
  21825, National Bureau of Economic Research

Fjeldstad OH. 2005. Corruption in Tax Administration: Lessons from Institutional reforms
  in Uganda. Tech. rep., Chr. Michelsen Institute (CMI), Bergen, Norway

Franck R, Rainer I. 2012. Does the Leader’s Ethnicity Matter? Ethnic Favoritism, Education,
  and Health in Sub-Saharan Africa. American Political Science Review 106:294–325

Gailmard S, Patty JW. 2007. Slackers and Zealots: Civil Service, Policy Discretion, and
  Bureaucratic Expertise. American Journal of Political Science 51:873–889

Gailmard S, Patty JW. 2012. Formal Models of Bureaucracy. Annual Review of Political
  Science 15:353:77


                                            31
Gallagher ME. 2006. Mobilizing the law in china:informed disenchantment and the develop-
  ment of legal consciousness. Law & Society Review 40:783–816

Golden MA. 2003. Electoral Connections: The Effects of the Personal Vote on Political Pa-
  tronage, Bureaucracy and Legislation in Postwar Italy. British Journal of Political Science
  33:189–212

Goodsell CT. 1981. The Public Encounter: Where State and Citizen Meet. Bloomington:
  Indiana University Press

Granovetter M. 1985. Economic action and social structure: The problem of embeddedness.
  American Journal of Sociology 91:481–510

Granovetter M. 2007. The Social Construction of Corruption, In On Capitalism, eds. V Nee,
  R Swedberg. Stanford: Stanford University Press

Grindle MS. 1997. Divergent cultures? When public organizations perform well in developing
  countries. World Development 25:481–495

Grindle MS. 2012. Jobs for the Boys: Patronage and the State in Comparative Perspective.
  Cambridge: Harvard University Press

Gulzar S. 2015. Can Political Alignment be Costly?

Gulzar S, Pasquale B. forthcoming. Politicians, Bureaucrats, and Development: Evidence
  from India. American Political Science Review

Gupta A. 2012. Red tape: Bureaucracy, structural violence, and poverty in India. Durham:
  Duke University Press

Hammond TH, Knott JH. 1996. Who Controls the Bureaucracy?: Presidential Power, Con-
  gressional Dominance, Legal Constraints, and Bureaucratic Autonomy in a Model of Multi-
  Institutional Policy-Making. Journal of Law, Economics, and Organization 12:119–166


                                             32
Hasnain Z, Manning N, Pierskalla JH. 2014. The Promise of Performance Pay? Reasons
  for Caution in Policy Prescriptions in the Core Civil Service. The World Bank Research
  Observer 29:235–264

Herbst J. 2000. States and Power in Africa: Comparative Lessons in Authority and Control.
  Princeton, NJ: Princeton University Press

Hertzberg A, Liberti JM, Paravisini D. 2010. Information and incentives inside the firm:
  Evidence from loan officer rotation. Journal of Finance 65:795–828

Hetherington K. 2011. Guerrilla auditors:          The politics of transparency in neoliberal
  Paraguay. Durham: Duke University Press

Hicken A. 2011. Clientelism. Annual Review of Political Science 14:289–310

Holmstrom B, Milgrom P. 1991. Multitask Principal-Agent Analyses: Incentive Contracts,
  Asset Ownership, and Job Design. Journal of Law, Economics & Organization 7:24–52

Huber JD, Shipan C. 2002. Deliberate Discretion? The Institutional Foundations of Bureau-
  cratic Autonomy. New York: Cambridge University Press

Hull MS. 2012a. Documents and bureaucracy. Annual Review of Anthropology 41:251–267

Hull MS. 2012b. Government of paper: The materiality of bureaucracy in urban Pakistan.
  Berkeley: University of California Press

Iyer L, Mani A. 2011. Traveling Agents: Political Change and Bureaucratic Turnover in
  India. Review of Economics and Statistics 94:723–739

Johnson C. 1982. MITI and the Japanese Miracle. The Growth of Industrial Policy, 1925-
  1975. Stanford: Stanford University Press

Kaufman D. 2005. 10 Myths About Governance and Corruption. Finance and Development
  42

                                              33
Kaufmann D, Wei SJ. 2000. Does ‘grease money’ speed up the wheels of commerce?

Keefer P. 2007. Clientelism, Credibility, and the Policy Choices of Young Democracies.
  American Journal of Political Science 51:804–821

Keefer P, Banuri S. 2013. Intrinsic motivation, effort and the call to public service. Tech.
  Rep. WPS6729, The World Bank

Keefer P, Stasavage D. 2003. The Limits of Delegation: Veto Players, Central Bank In-
  dependence, and the Credibility of Monetary Policy. American Political Science Review
  97:407–423

Kelly JA, St Lawrence JS, Diaz YE, Stevenson LY, Hauth AC, et al. 1991. Hiv risk behavior
  reduction following intervention with key opinion leaders of population: an experimental
  analysis. American Journal of Public Health 81:168–171

Khan AQ, Khwaja AI, Olken BA. 2016. Tax Farming Redux: Experimental Evidence on
  Performance Pay for Tax Collectors. Quarterly Journal of Economics 131:219–271

Kiewiet DR, McCubbins MD. 1991. The Logic of Delegation. Chicago: University Of Chicago
  Press

Kiser E, Kane J. 2001. Revolution and state structure: The bureaucratization of tax admin-
  istration in early modern England and France. American Journal of Sociology 107:183–223

Kiser E, Sacks A. 2009. 11 improving tax administration in contemporary african states:
  Lessons from history1. The new fiscal sociology: Taxation in comparative and historical
  perspective :183

Kitschelt H, Wilkinson SI. 2007. Citizen-politician linkages: an introduction. In Patrons,
  Clients, and Policies. Patterns of Democratic Accountability and Political Competition,
  eds. H Kitschelt, SI Wilkinson. New York: Cambridge University Press, 1–49



                                            34
Kohli A. 2004. State-Directed Development. Political Power and Industrialization in the
  Global Periphery. New York: Cambridge University Press

Kosack S, Fung A. 2014. Does Transparency Improve Governance? Annual Review of Polit-
  ical Science 17:65–87

Kumlin S, Rothstein B. 2005. Making and Breaking Social Capital The Impact of Welfare-
  State Institutions. Comparative Political Studies 38:339–365

Kurtz MJ, Schrank A. 2007. Growth and Governance: Models, Measures, and Mechanisms.
  Journal of Politics 69:538–554

Lazear EP. 1999. Globalisation and the Market for Team-Mates. Economic Journal 109:C15–
  40

Lehmann U, Sanders D. 2007. Community health workers: what do we know about them.
  The state of the evidence on programmes, activities, costs and impact on health outcomes
  of using community health workers. Geneva: World Health Organization :1–42

Levi M, Sacks A, Tyler T. 2009. Conceptualizing Legitimacy, Measuring Legitimating Beliefs.
  American Behavioral Scientist 53:354–375

Lipsky M. 1980. Street-level Bureaucracy. Dilemmas of the Individual in Public Service. New
  York: The Russell Sage Foundation

Marcesse T. 2016. Public policy and informal institutions: The new moral economy of cor-
  ruption in rural India

McCubbins MD, Noll RG, Weingast BR. 1987. Administrative Procedures as Instruments of
  Political Control. Journal of Law, Economics, & Organization 3:243–277

Mettler S. 1998. Dividing Citizens: Gender and Federalism in the New Deal Public Policy.
  Ithaca, NY: Cornell University Press


                                             35
Moe TM. 1984. The New Economics of Organization. American Journal of Political Science
  28:739–777

Nath A. 2014. Bureaucrats and Politicians: How Does Electoral Competition Affect Bureau-
  cratic Performance?

Nichter S. 2008. Vote Buying or Turnout Buying? Machine Politics and the Secret Ballot.
  American Political Science Review 102:19–31

O’Brien KJ, Li L. 1995. The politics of lodging complaints in rural china. The China Quar-
  terly 143:756–783

Oldenburg P. 1987. Middlemen in Third-World corruption: Implications of an Indian case.
  World Politics 39:508–535

Olken B. 2006. Corruption and the Cost of Redistribution. Micro Evidence from Indonesia.
  Journal of Public Economics 90:853–870

Olken B. 2007. Monitoring Corruption: Evidence from a Field Experiment in Indonesia.
  Journal of Political Economy 115:200–249

Olken BA, Pande R. 2012. Corruption in Developing Countries. Annual Review of Economics
  4:479–509

                                                                       avez’s Misiones.
Penfold-Becerra M. 2007. Clientelism and Social Funds: Evidence from Ch´
  Latin American Politics and Society 49:63–84

Perry JL, Hondegheim A. 2008. Building Theory and Empirical Evidence about Public
  Service Motivation. International Public Management Journal 11:3–12

Persson A, Rothstein B, Teorell J. 2013. Why Anticorruption Reforms Fail—Systemic Cor-
  ruption as a Collective Action Problem. Governance 26:449–471




                                             36
Pritchett L, Woolcock M, Andrews M. 2013. Looking Like a State: Techniques of Persistent
  Failure in State Capability for Implementation. Journal of Development Studies 49:1–18

Ramseyer JM, Rosenbluth FM. 1993. Japan’s Political Marketplace. Cambridge: Harvard
  University Press

Rasul I, Rogger D. 2015. The Impact of Ethnic Diversity in Bureaucracies: Evidence from
  the Nigerian Civil Service. American Economic Review 105:457–461

Ricks JI. 2016. Building Participatory Organizations for Common Pool Resource Manage-
  ment: Water User Group Promotion in Indonesia. World Development 77:34–47

Rogger D. 2014. The Causes and Consequences of Political Interference in Bureaucratic
  Decision Making: Evidence from Nigeria

Ross M. 2006. Is Democracy Good for the Poor? American Journal of Political Science
  50:860–874

Roth G. 1968. Personal Rulership, Patrimonialism, and Empire-Building in the New States.
  World Politics 20:194–206

Sacks A. 2012. Can donors and non-state actors undermine citizens’ legitimating beliefs?
  World Bank Policy Research Working Paper

Shore LM, Chung-Herrera BG, Dean MA, Ehrhart KH, Jung DI, et al. 2009. Diversity in
  organizations: Where are we now and where are we going? Human Resource Management
  Review 19:117–133

Snyder SK, Weingast BR. 2000. The American system of shared powers: the President,
  Congress, and the NLRB. Journal of Law, Economics, and Organization 16:269–305

Soss J. 2005. Making Clients and Citizens: Welfare Policy as a Source of Status, Belief, and
  Action. In Deserving and Entitled: Social Constructions and Public Policy, eds. A Schnei-
  der, H Ingram. Albany, NY: State University of New York Press

                                            37
Stokes SC. 2005. Perverse Accountability: A Formal Model of Machine Politics with Evidence
  from Argentina. American Political Science Review 99:315–325

Stovel K, Savage M. 2006. Mergers and mobility: Organizational growth and the origins of
  career migration at Lloyds Bank. American Journal of Sociology 111:1080–1121

Svensson J. 2005. Eight Questions about Corruption. Journal of Economic Perspectives
  19:19–42

Szeftel M. 1982. Political graft and the spoils system in Zambia: The state as a resource in
  itself. Review of African Political Economy :4–21

Tendler J, Freeheim S. 1994. Trust in a rent-seeking world: Health and government trans-
  formed in Northeast Brazil. World Development 22:1771–1791

Thies CG. 2005. War, Rivalry, and State Building in Latin America. American Journal of
  Political Science 49:451–465

Tilly C. 1985. War Making and State Making as Organized Crime. In Bringing the State
  Back In, eds. P Evans, D Rueschemeyer, T Skocpol. New York: Cambridge University
  Press, 169–191

Tsai L. 2007. Accountability without Democracy. New York: Cambridge University Press

Tyler TR. 1990. Why People Obey the Law. Princeton, NJ: Princeton University Press

Wade R. 1990. Governing the Market: Economic Theory and the Role of Government in
  East Asian Industrialization. Princeton: Princeton University Press

Weber M. 1978. Economy and Society: An Outline of Interpretive Sociology. Berkeley: Uni-
  versity of California Press

Weitz-Shapiro R. 2012. What Wins Votes: Why Some Politicians Opt Out of Clientelism.
  American Journal of Political Science 56:568–583

                                            38
Whitford AB. 2005. The Pursuit of Political Control by Multiple Principals. Journal of
  Politics 67:28–49

Wilson JQ. 1991. Bureaucracy: What Government Agencies Do and Why They Do It. New
  York: Basic Books

Wimmer A. 1997. Who owns the state? understanding ethnic conflict in post-colonial soci-
  eties. Nations and nationalism 3:631–666

Woo-Cummings M, ed. 1999. The Developmental State. Ithaca, NY: Cornell University Press




                                             39