Bureaucratic Politics: Blind Spots and Opportunities in Political Science

Bureaucracy is everywhere. Unelected bureaucrats are a key link between government and citizens, between policy and implementation. Bureaucratic politics constitutes a growing share of research in political science. But the way bureaucracy is studied varies widely, permitting theoretical and empirical blind spots as well as opportunities for innovation. Scholars of American politics tend to focus on bureaucratic policy making at the national level, while comparativists often home in on local implementation by street-level bureaucrats. Data availability and professional incentives have reinforced these subfield-specific blind spots over time. We highlight these divides in three prominent research areas: the selection and retention of bureaucratic personnel, oversight of bureaucratic activities, and opportunities for influence by actors external to the bureaucracy. Our survey reveals how


INTRODUCTION
Across the globe, publicly funded bureaucrats administer government programs and provide public services.Bureaucrats in so-called street-level positions interact with citizens on a daily basis.Less visible, but equally important, are bureaucrats who engage in administrative work, designing and implementing policies.Just as service delivery and policy making are tasks universal to all polities, questions related to the functioning of public bureaucracies are relevant across geographic, political, and economic contexts.How are bureaucrats selected, retained, and deployed-and with what consequences?How are bureaucrats overseen, and by whom?When (and how) do external actors influence bureaucratic behavior?
In this review, we focus on political science research on bureaucracy published over the last two decades.We start by cataloging articles published in the American Political Science Review (APSR), the American Journal of Political Science (AJPS), and the Journal of Politics ( JOP).We document a significant growth in the absolute and relative number of bureaucracy articles published in the discipline's flagship journals.This growth reflects advances in data collection and availability, as well as a growing recognition within the discipline that the behavior of nonelected state agents is at the core of politics.
However, significant divides across political science subfields-in focus, methods, and evidence-hinder the speed at which knowledge can accumulate.We focus on the divide between the subfields of American and comparative politics-although similar or even larger gaps can be found between political science and other fields such as public administration (Bertelli et al. 2020), economics (Besley et al. 2022), and anthropology (Lea 2021).Building on our systematic review of articles in top journals, we document differences across subfields.We argue that increased dialogue across subfields will advance knowledge of bureaucratic politics more rapidly than retaining subfield divisions.Given the growing interest in bureaucracy studies, and especially the increased share of journal space dedicated to bureaucracies outside of the American context, now is an opportune moment to encourage greater inter-subfield learning and collaboration.
Our review uncovers blind spots in the study of bureaucracy that we argue are sustained by (at least) two factors.First, researchers' ability to access data varies.Governments may constrain access, or the practical costs of data compilation may discourage access.As simple a fact as this is, it needs to be accounted for as researchers attempt to evaluate general theories.We highlight the types of data that researchers in various settings employ with the aim of inspiring future data collection.The second factor is that some blind spots result from within-subfield path dependency.Professional incentives that prioritize certain questions and methods within each subfield reinforce this dependency.For instance, journal editors tend to rely on peer reviewers from within the subfield, which encourages research that meets established subfield parameters and may (unintentionally) discourage innovation.We highlight areas in which research opportunities for either Americanist or comparative scholars have been overlooked largely as a result of subfield tradition.
This review is designed to narrow the gap between American and comparative studies of bureaucratic politics.We do so by articulating the differences in focus, method, and data across subfields, which reveals opportunities for innovation.We focus on three promising areas of bureaucratic politics: personnel, oversight, and external influence.There are other important areas of research on bureaucratic politics, including historical development (Vogler 2023), macro-level models of bureaucratic governance (Dahlström & Lapuente 2022), citizen-bureaucrat interactions (Pepinsky et al. 2017, Grossman & Slough 2022), and corruption (Gans-Morse et al. 2018)-but we believe personnel, oversight, and influence are three areas that hold considerable promise for cross-subfield learning.By juxtaposing and contrasting recent research from American and comparative politics in these three arenas, we seek to foster a more productive research agenda on bureaucracies everywhere.

BUREAUCRACY TRENDS, 2000-2022
We assembled data on journal articles published from 2000 to 2022 in the top three generalinterest political science journals: APSR, AJPS, and JOP. 1 Focusing on these journals yields insights into the types of bureaucracy research that have been undertaken and rewarded in the discipline.Importantly, scholars of both American and comparative politics regularly publish in all three venues. 2Articles were classified as relevant if bureaucratic inputs or outputs were the primary independent or dependent variable.This exercise yielded 133 articles.
We coded articles based on their geographic coverage (i.e., countries studied), the level of bureaucracy (i.e., central, regional, or local agencies), and methods used (i.e., formal theory, regression, interviews, experiment).Classifications for the latter two variables were not mutually exclusive.An article could investigate multiple levels of bureaucracy and employ several types of analysis.Purely theoretical articles and those covering international bureaucracies (e.g., the World Bank or the European Commission) were coded as "Other." Figure 1 charts the relative and absolute increase in the number of all articles that focus on the bureaucracy during the last two decades.Table 1 presents our inventory in detail.Just under half of the articles focus on the United States (48%).However, the right panel of Figure 1 shows Rulemaking: the administrative process by which bureaucrats create legally binding policies, often by issuing a proposed policy and offering an opportunity for public comment The "Other" category comprises articles that do not focus on a particular country (e.g., those with formal models or a focus on international bureaucracies).Articles that include the US case among others were coded as "Comparative." there has been a significant decline in the share of articles that focus exclusively on the United States within the bureaucracy literature, from about 92% to 29% over the last 20 years.Beyond the United States (n = 64), the remaining literature is fragmented geographically.India (n = 6) is the next most studied case, followed by China, Russia, and Indonesia (n = 3 each).Perhaps surprising is the low number of European cases within these journals, given the ready availability of administrative data in these contexts.
The American and comparative politics literatures differ in the level of government they typically study.The former tends to focus on bureaucrats in the central government (61%), while the latter typically uses data from local agencies (64%).This may be due to the availability of data in different settings.In developing countries, it may be easier for scholars to collect data from regional or local agencies than from the central bureaucracy.This is partly because governments often prohibit access to data on central agencies and scholars cannot rely on Freedom of Information (FOI) requests because relevant legislation either does not exist or does not function in practice.In comparison, Americanist bureaucracy scholars often use data collected through FOI requests (for example, Carpenter & Ting 2007, Lowande 2018, Ritchie 2018).In addition, Americanists make use of regularly published data on central agencies (for example, the "Plum Book"), which are often absent in other contexts.
These differences affect the kinds of bureaucratic behavior studied.American and comparative studies generally use different units of analysis.American politics scholars tend to leverage variation across agencies, whereas comparative politics studies rely on variation across individual bureaucrats.Accordingly, American politics articles often focus on national-level agencies performing central, policy-setting tasks like rulemaking, while those in comparative politics typically consider local-level administrators or frontline service providers.Figure 2 places some of the reviewed studies along these three continua and reveals unexplored areas like services delivered by national governments and policy administration across regions or states.
Methodologically, the vast majority of articles in both subfields use quantitative data and some form of regression analysis.As discussed below, the richness and sheer quantity of data scholars work with have become increasingly impressive.A feature of the study of American bureaucracy is the use of formal models: Articles that solely present a formal model are confined to the US case.Within the US-focused literature, about 19% of the articles include a formal model.Comparativists are much more likely than Americanists to base their analysis at least in part on interviews (28% of articles versus 8%).

Policy/ administration
Frontline service delivery

Bureaucratic task
Figure 2 A three-dimensional typology of bureaucratic politics research.The figure displays selected bureaucratic politics studies from the comparative (blue) and American politics (red) subfields.
in the United States almost exclusively reports results from audit experiments involving a range of bureaucrats, including election officials (White et al. 2015, Porter & Rogowski 2018), housing officers (Einstein & Glick 2017), and marriage-license officials (Lowande & Proctor 2020).
Below, we review the recent bureaucratic politics literature on three key topics for which learning across subfields seems especially promising-selection, oversight, and influence.

PERSONNEL POLITICS
Who gets hired to do what jobs in public bureaucracies?How do formal and informal institutions shape patterns of hiring and firing?What impacts do hiring practices have on bureaucratic performance?Scholarship in personnel politics examines the selection and retention of bureaucrats, and the implications of both for government performance.Political patronage, in particular, is central to both the comparative and American politics literatures, although differences in word usage make the connections unclear.It is often of greatest interest to scholars when individuals are selected for political reasons, but the variety of such reasons creates disjunctures across research programs.In the US context, these political reasons mostly involve selecting bureaucrats with policy preferences similar to the principal, or what is sometimes called the official's loyalty to the person who hired them.In comparative studies, the political reasoning is usually about the distribution of government jobs to partisans-often party workers-or what is sometimes known as a spoils system.This terminology partly reflects institutional differences and history.Even after civil service reforms in the United States that began in the nineteenth century, several thousand national-level positions remain subject to patronage.These appointees mainly consist of high-level bureaucrats, Procurement: government purchase of goods and services from the private sector or state-owned enterprises a small subset of whom must be approved by the upper chamber of the legislature.Studies on the US case typically focus on these high-level positions-particularly the appointees' policy preferences and the selection process itself-all of which are animated by the possibility of divided party control of government.For example, Krause & O'Connell (2016) develop a latent measure of loyalty based on factors such as party identification, campaign contributions, and work history for the party.They show that, over time, US presidents come to rely less on partisan heuristics to select high-level officials.Bonica et al. (2015) argue that legislative checks on patronage selection lead to the appointment of extremists to the highest-level positions, which do not require legislative approval.Relatedly, Kinane (2021) shows that since high-level US appointments must be approved by a majority of the Senate, US presidents may have incentives to leave positions vacant or acting, rather than accede to a less loyal appointee.Though many countries have appointees in top policy-making roles, high-level positions remain relatively understudied outside of the United States.Most comparative work tends to focus on lower-level bureaucratic positions, and the subsequent performance-related effects of leaving hiring up to principals.
The collection of employment records at scale has allowed scholars to evaluate standard predictions about political patronage and its effects.Plainly put, it exists, and it is widespread.From federal civil servants in Germany (Bach & Veit 2018) to municipal employees in Brazil (Colonnelli et al. 2020) to teachers in Indonesia (Pierskalla & Sacks 2020), jobs seem to be subject to politically motivated selection, designed mostly to benefit the ruling party.
Recent work suggests that patronage exists on a continuum as opposed to being dichotomous.The important question is then not whether principals have discretion to hire partisans or loyalists, but for which jobs patronage occurs.While Colonnelli et al. (2020) find evidence of patronage hiring across the full spectrum of municipal jobs in Brazil, the influence of political connections is higher for managerial positions than for front-line positions and is concentrated in jobs for which civil service exams are not required.Conversely, Brierley (2021) finds evidence that patronage considerations influence appointments to entry-level, but not high-level, positions in local bureaucracies in Ghana.Recent studies have also highlighted variation that corresponds to the electoral calendar.For example, Pierskalla & Sacks (2020) find that elections lead to increases in the hiring of contract (but not permanent civil service) teachers in Indonesia.In turn, Toral (2022b) identifies electoral cycles in the hiring of both temporary and civil service employees in Brazil, in a pattern consistent with circumventing anticorruption laws that prevent spending and hiring around elections.The most similar research on the US context is, not surprisingly, historical.Calais-Haase & Rogowski (2021), for example, show differences in turnover rates between salaried and nonsalaried executive branch officials.They demonstrate that, after civil service reform, turnover was higher for individuals in lower-ranked positions.
Subfield differences also permeate studies of how personnel impact performance.In the US context, questions about the performance implications of personnel often go unasked.Most studies of the subsequent behavior of high-level officials are, in effect, studies of political control over policy making-which we review in a subsequent section.For research focused on personnel selection based on preference similarity, the appropriate performance benchmark is whether the official achieves the principal's preferred outcomes.However, there are exceptions.For example, Gallo & Lewis (2012) find that program managers selected via patronage had lower performance ratings than their nonpatronage counterparts.Additionally, Spenkuch et al. (2022) demonstrate that politically misaligned procurement officers and superiors are associated with cost overruns in federal contracts.
The rarity of this kind of study is partly a consequence of the level of analysis.When studies of selection focus on high-level administrators, their performance is defined by their jobs.But these jobs are diverse: They include collecting taxes, managing public lands, maintaining nuclear Principal-agent framework: theoretical model often applied to studies of bureaucratic politics, in which politicians (principals) delegate tasks to bureaucrats (agents), and agency problems arise because of hidden actions and hidden information arsenals, approving vaccines, running hospitals, and responding to tornadoes.That generates a difficult measurement problem.
Comparative politics research often identifies a particular type of bureaucrat and links selection to an outcome most scholars would agree impacts or is a measure of performance.One such outcome is corruption.Oliveros & Schuster (2018) show that bureaucrats perceive colleagues hired via exams rather than direct appointment as less corrupt.Patronage hiring can fuel corruption that aims to fund politicians' election campaigns, although such practices may vary by type of electoral system (Gingerich 2013) and level of party institutionalization (Sigman 2022).Service delivery is another area of focus.Akhtari et al. (2022) show that teacher turnover as a result of political forces decreases school test scores.Likewise, Barbosa & Ferreira (2019) argue that the hiring of politically connected workers crowds out the employment of unaffiliated teachers and doctors, which has downstream effects on schooling and mortality rates.Relatedly, Toral (2022c) finds evidence that health worker turnover increases and health service delivery worsens after incumbent electoral defeats in Brazil.Toral (2022a) also shows that local schools with appointed directors who lose their connections to the local government experience declines in service delivery.Xu (2018) argues that colonial governors in the British empire appointed via patronage (rather than open recruitment) tended to collect less tax revenue.Along similar lines, the sale of provincial governorships by the Spanish Crown in Peru may have reduced the health and welfare of residents for decades (Guardado 2018).
Focusing on lower-level officials and service delivery involves its own trade-offs.With some notable exceptions, the links between measures of politically motivated hiring and performance outcomes are often circumstantial.The hiring implications of patronage are measured at the individual level, but performance outcomes are often not.Sometimes, this even means studies examine the system-level impact of patronage versus merit rules, rather than the behavior of individuals who work in those systems.The greater the conceptual distance between the individual and the outcome, the more urgent the questions about what, precisely, drives the performance penalty or reward.Not surprisingly, there are lively, ongoing debates about the mechanisms that create performance bonuses for patronage-based hiring.In a principal-agent framework, they may be attributed to mutual trust between the principal and the agent (e.g., Jiang 2018), the ease of applying sanctions and rewards (Toral 2022a), distributive favoritism ( Jiang & Zhang 2020), or career concerns (Rivera 2020).In each case, however, indirect links between performance metrics and bureaucrat selection pose a challenge for adjudicating between these mechanisms.
In contexts where bureaucrats are selected for political reasons, patronage can be thought of as another solution to the problems of bureaucratic oversight because it reduces the distance between bureaucrats and politicians.Patronage may reduce the need for monitoring because political appointees have a direct stake in the incumbent's political survival.This dynamic has long been thought to govern the selection of high-level appointees in the United States (Lewis 2008).Leveraging survey experiments with local government employees in Argentina, Oliveros (2021) finds that government supporters in the bureaucracy provide political services because they expect their jobs to be tied to the electoral fate of the incumbent.If governments know that they can rely on some bureaucrats' self-interest to get what they want, they can strategically assign aligned bureaucrats to key posts in the administration.This is precisely what Hassan (2017) finds in her study of the security apparatus in Kenya.Of course, the gains of patronage for bureaucratic oversight may reduce citizens' welfare if politicians use patronage to extract rents.
Scholarship on personnel also considers career trajectories; many papers investigate how political turnover affects job retention or promotion.Research patterns in these studies are often reversed: The outcomes are individual-level retention or dismissal questions, whereas the independent variables are broader political forces like changes in party control of government.
Here, the collection of records has shown that career incentives often reproduce de facto patterns of political dismissal, even though they are most often de jure prohibited.In Sweden, Dahlström & Holmgren (2019) find that executives with fixed terms were still more likely to depart following changes in party control of government.Kim & Hong (2019) identify a similar pattern in South Korea for chief executives of state-owned enterprises.In the United States, Bolton et al. (2021) show that career senior executives with views that likely diverge from those of an incoming president are more likely to leave government.Again, though disparate in their institutional contexts, these studies point to similar theoretical arguments made about the determinants of turnover.Gailmard & Patty (2013) argue that the contraction of policy-making discretion reduces the attractiveness of public service.Similarly, Cameron & de Figueiredo (2020) show that, with the exception of extreme zealots who are willing to wait for the next election, policy-making interventions by superiors reduce subordinates' incentives to work hard or remain in their post.This reasoning is supported by surveys of public bureaucrats, who are more likely to report an intention to leave government if their influence or job security diminishes (e.g., Bertelli & Lewis 2013, Oliveros & Schuster 2018, Richardson 2019).Even absent formal mechanisms for political interference in staffing, politics informs the composition of bureaucracies via voluntary, individual-level incentives.
In any area of research that is active and innovative, answers to central questions are likely to develop unevenly.But in research on personnel politics over the previous decade, the availability of data and researchers' professional incentives are compounding obvious disjunctures.In the United States, the preoccupation with national-level bureaucrats generally comes at the expense of research on state and local governments.The professional returns to studying a handful of states-or even a single state-are known to be lower, all else equal.A similar point could be made about evaluating a single type of bureaucrat using a politically neutral performance metric.It is no accident that most exceptions to these tendencies are found in the booming study of local law enforcement (e.g., Ba et al. 2021), where the public salience of policing has driven up the professional rewards for studying it.These anomalies aside, the pressures of data availability and professional incentives mean that researchers essentially ignore a state-level public sector workforce that both outnumbers the federal workforce more than two to one and contains state governments with as many or more employees than some of the single-country studies we have mentioned.In our view, shifting attention to the state level would represent an important step toward addressing many of the blind spots discussed above.Performance is difficult to compare across high-level officials.There is also little institutional variation in selection mechanisms.Progress could be made on both fronts with the kind of lower-level, service-delivery-oriented study that is common outside the field of American politics.
Its parochial focus on high-level federal officials aside, research on the United States also offers lessons for comparativists.Its narrower focus on the upper echelons of the bureaucracy is justified by these officials' relative importance for designing and implementing policies.Comparative research lies at the other extreme, often focusing on street-level officials in service delivery roles.The measurement of patronage in comparative studies is typically coarse and limited by data availability.Politically motivated selection is often (and problematically) proxied with partisan alignment.More importantly, the central focus on service delivery seems to have generated puzzling findings about the circumstances under which patronage is helpful or hurtful.One of the leading explanations for a performance bonus to patronage is that aligned preferences imply reduced monitoring costs and that the officials for whom this is the case are typically at a higher level and in policy-making roles.As research on the United States has argued, these are exactly the kind of officials for whom we would expect principals to benefit from aligned preferences.In examining the selection of upper-level officials, it is common for US research to treat ideology and valence (or effectiveness) as distinct, separable dimensions of policy.Future studies should identify which of these dimensions matters for performance by bureaucrat type.

OVERSIGHT
How are bureaucrats monitored, and by whom?How much autonomy do bureaucrats enjoy?What are the impacts of monitoring and autonomy on what bureaucrats do and what governments deliver?Once bureaucrats are selected and deployed, these dynamics of bureaucratic oversight are a key determinant of public sector performance.Here we focus on top-down oversight, although other forms of monitoring and accountability (e.g., with respect to citizens, peers, or managers) are also key for bureaucratic performance.
Research questions on oversight contrast the actions of bureaucrats with the interests of politicians, a relationship typically modeled as a principal-agent problem.But the subfield pathologies identified in the previous section also persist in this body of research.In the US context, oversight research focuses on politicians' attempts to contain drift in bureaucratic policy decisions.By contrast, comparative studies typically emphasize attempts to minimize undesirable behavior by bureaucrats (e.g., corruption, absenteeism) and improve service delivery.Across subfields, the main policy tools studied are investments in monitoring and constraints on bureaucratic autonomy.
The US focus on national-level agencies means that, in addition to the president, the relevant oversight actors are often collective bodies such as Congress or congressional committees.American politics researchers have thus considered how the number of oversight actors affects outcomes.Scholars studying political control have found that oversight operates more effectively when there is a clearer oversight structure.For example, congressional influence over federal agencies decreases (relative to White House influence) as more committees are involved (Clinton et al. 2014), and agencies are more likely to comply with statutory deadlines under unified government (Bertelli & Doherty 2019).Using a formal model, Bils (2020) shows that unless agencies have divergent policy preferences, Congress achieves better outcomes by consolidating authority into a single agency.
Comparative research also investigates the structure of oversight institutions, but with the now-familiar focus on local agencies.For example, Gulzar & Pasquale (2017) study oversight by state-level politicians in India and find that policy implementation improves when bureaucrats are overseen by one as opposed to multiple politicians.Single-principal oversight may improve bureaucratic effectiveness by facilitating control and credit claiming, thereby incentivizing politicians to invest in monitoring.Similarly, Dasgupta & Kapur (2020) find that geographical congruence between politicians' and bureaucrats' districts leads politicians to increase bureaucratic resources and capacity.
In addition to institutional structure, the ideological alignment of bureaucrats and their overseers is a central concern in Americanists' research on the bureaucracy.The foundational prediction is that increasing the ideological distance between a principal and its agents invites greater monitoring and oversight.With respect to Congress, this principle is most often evaluated in terms of the design of statutes (Huber & Shipan 2002)-which can grant more or less leeway to agencies-and oversight hearings (McGrath 2013).Scholars have built on this principle to demonstrate many other-often informal-ways in which Congress engages in bureaucratic oversight.For instance, Bolton (2022) shows that congressional committee reports serve as additional, nonstatutory instructions to agencies, and that these reports become more stringent as ideological conflict increases.Likewise, ideology is thought to motivate a considerable portion of presidential oversight.
Scholars of American politics increasingly question the dominance of ideology as the primary driver of oversight.For instance, Lowande (2018) evaluates private letters from members of Congress to government agencies and finds that ideological (dis)agreement has a negligible impact on politicians' behavior.Bolton et al. (2016) demonstrate that although ideology is important to presidential review of regulations, the president's ability to control the centralized review process decreases if the reviewing office faces capacity constraints.Still other research suggests that ideological misalignment may motivate agents as well as principals.Potter (2019) argues that ideological conflict with principals may motivate bureaucrats to engage in "procedural politicking," or strategic maneuvers that leverage bureaucrats' procedural knowledge and make it more difficult for principals to engage in oversight.Another tactic that agencies can adopt involves implementing statutes in a way that assuages contemporaneous political coalitions (Acs 2016).
Investigating how ideology affects oversight relationships presents an opportunity for comparativists, as few studies in the comparative literature consider alignment.The lack of research in this area partly stems from the absence of certain institutional features outside the US context, such as divided government, as well as the limited role of ideology in some low-and middle-income contexts.However, comparativists can also expand the concept of preference divergence beyond ideology to consider, for example, divergences in policy implementation that stem from bureaucrats' ethnic, religious, or gender identities (Bozcaga 2020, Pierskalla et al. 2021, Purohit 2022).
Beyond formal institutions and alignment, principals' ability to monitor bureaucrats is another key driver of oversight.Scholars have started to challenge the assumption (common in both American and comparative politics research) that principals have the required expertise and incentives to monitor-which is itself a costly activity.In US states, low capacity among legislators (who often occupy this position as a part-time, low-paid job) has led to expanded participation by bureaucrats in the legislative process (Boushey & McGrath 2017), even in the creation of statutes that delineate bureaucratic authority (Kroeger 2022).Bureaucratic accountability in the administration of the US Social Security program has been shown to depend on the capacity of oversight agents at the state and federal levels (Drolc & Keiser 2021).A promising avenue to improve bureaucratic accountability is thus to train politicians.In a field experiment in Uganda, Raffler (2022) found that local politicians who received training on their mandate and rights, as well as financial information about the bureaucracy they oversaw, increased their monitoring and efforts to improve public service delivery-but only in local governments unaligned with the national ruling party.
Expanding or contracting bureaucratic autonomy is another key lever of oversight.Recent research has shown how constraining bureaucratic autonomy may contract bureaucratic performance, since it can depress gains from bureaucratic expertise and dampen workers' creativity and morale.For instance, Patty & Turner (2021) draw on a formal model to demonstrate that the existence of ex post review can lead some bureaucrats to make "pathological" (i.e., intentionally uninformed) policy choices.Experiments in the education sector in high-income countries have found gains from increases in school principal autonomy.In New York City, more school principal autonomy led to better student test scores (Wang & Yeung 2019), and in Denmark, greater autonomy led to more investments in bureaucratic expertise (Andersen & Moynihan 2016).Even in low-income settings, where Fukuyama (2013) theorized that performance would benefit from lower levels of autonomy, some studies have found gains from increasing it.For example, in the national civil services of both Ghana (Rasul et al. 2021) and Nigeria (Rasul & Rogger 2018), project completion rates are positively associated with increases in staff autonomy and discretion.In Pakistan, procurement officers who were given more autonomy improved their performance (Bandiera et al. 2021).
Future research should build on Fukuyama's (2013) argument to further theorize and test the conditions under which bureaucratic autonomy can boost performance.One reason bureaucratic autonomy may have been found to improve outcomes in developing countries is that bureaucrat capacity is not as homogeneously low as is often assumed.Instead, bureaucracies are often characterized by pockets of high capacity, either across agencies or across different types of jobs (Bersch et al. 2017, McDonnell 2017).In such contexts, granting bureaucrats greater autonomy can protect them from potential abuses of oversight from politicians that stem from weak vertical and horizontal accountability (Brierley 2020).Motivation is likely another important mediator in interventions affecting bureaucratic autonomy.Honig (2022) argues that increasing control over bureaucrats may crowd out their mission orientation, and that empowering bureaucrats to pursue the agency's mission with more, rather than less, autonomy may be a more effective means to improve performance, especially when their work is hard to monitor (Honig 2021).All in all, both Americanists and comparativists should consider how political institutions, agency-level characteristics (e.g., policy area, type of work, or bureaucratic norms) and individual bureaucrats' features (such as selection mode, level of human capital, or motivation) mediate the relationship between bureaucratic autonomy and performance.
The focus on local agencies within the comparative literature generates an important blind spot-oversight by national-level politicians.Scholars of US politics have made significant progress in opening the black box of oversight and uncovering a wide range of formal and informal means through which politicians hold the bureaucracy accountable beyond passing legislation or holding public hearings.This literature suggests that even informal and private forms of oversight can sometimes effectively control the bureaucracy (Ritchie & You 2019).Comparative scholars should consider how different avenues of formal and informal top-down oversight influences bureaucratic behavior.
Both American and comparative politics scholars should seriously consider the ways in which oversight by politicians interacts with accountability vis-à-vis other actors.For instance, how do judges shape politicians' ability to oversee bureaucrats?Wiseman & Wright (2020) show that US Supreme Court decisions constrain the ability of Congress to control the bureaucracy.Potter (2019) shows that in about half of rulemaking cases reviewed by courts, US federal agencies lose.The electorate is another key actor that can shape top-down oversight of the bureaucracy.Martin & Raffler (2021) show that Ugandan voters are less likely to hold politicians accountable when they believe politicians have limited power over bureaucrats.Slough (2022) suggests that politicians' incentives may bias bureaucratic accountability toward more politically active citizens.Finally, the media can also meaningfully influence citizens' understanding of oversight and bureaucratic behavior.In the United States, Ruder (2014) shows that media framing shapes voter perceptions of the extent of political control over federal agencies.In Mexico, Erlich et al. (2021) demonstrate that negative media coverage can increase bureaucratic responsiveness, while positive reports can reduce it.Taken together, these studies suggest that we should consider bureaucrats and politicians not as embedded in a linear principal-agent relationship but as involved in complex networks of accountability, which also involve citizens, judges, journalists, and other actors.

EXTERNAL INFLUENCE
Beyond oversight by politicians, bureaucrats are also influenced by external actors-especially businesses, nongovernmental organizations, and international donors.Who gets to influence bureaucrats' behavior, and through what means and strategies?How do such pressures shape bureaucratic governance and performance?Scholars often search for evidence of external influence in two areas: bureaucratic lobbying and public procurement.Lobbying the bureaucracy can take many forms, but-as we explain below-the vast majority of research on bureaucratic lobbying focuses on rulemaking, a process wherein bureaucrats have considerable power to set policy.Control over procurement also places significant power in the hands of bureaucrats because of the scale of public money involved; procurement constitutes an estimated 15% of global gross domestic product (World Bank 2021).
As in other areas of bureaucratic politics, American and comparative politics research on external influences diverges in its focus.Again, this divergence results partly from data availability and subfield path dependencies.The considerable attention devoted to biases that arise in the US rulemaking process is partly due to administrative data that are "astonishingly complete" (Carpenter et al. 2020, p. 425).By contrast, the study of procurement is similar across subfields, partly due to the widespread adoption of electronic procurement auctions across countries.
While interest groups can lobby the legislature before a law is passed, research on bureaucratic lobbying often centers on rulemaking, where bureaucrats set key parameters that guide implementation after a law is passed.Since rules can have tremendous social and financial consequences, civil society organizations and private companies vie to influence these bureaucratic decisions.In the United States, a rich literature explores the groups that participate in rulemaking and the conditions under which this form of lobbying is effective.Early research designs associated rule changes with public comments submitted by groups, revealing a pattern in which industry groups participate more frequently than less well-heeled interests and agencies are biased in favor of making changes requested by these powerful interests (e.g., Yackee & Yackee 2006, Kerwin & Furlong 2018).Yet, under the right conditions, more marginalized groups can also secure policy returns in rulemaking (Dwidar 2022).Recent scholarship has built on these earlier foundations to address the selection associated with studying comments received.Using a formal model, Libgober (2020b) shows that regulators anticipate groups' concerns, and that this strategic behavior suggests that existing empirical results are observationally equivalent with several potential mechanisms of regulatory bias.Meanwhile, Gordon & Rashin (2021) identify all potential stakeholders associated with a Medicare payment rule and demonstrate that anticipated losses-rather than potential gains-motivate participation decisions.
Studies on bureaucratic lobbying in the United States are in conversation with one another, a development made possible by the ready availability of rulemaking texts and their associated public comments from the Federal Register and from government websites such as Regulations.gov.This conversation has spurred research into related questions, such as how interest groups divvy up their bureaucratic lobbying resources.For example, using a formal model and data on lobbying associated with financial rules, You (2017) shows that larger firms are more likely to lobby Congress ex ante, whereas smaller firms are more likely to directly lobby agencies in the ex post stage.Also focusing on financial rules, Libgober (2020a) evaluates firm stock prices and shows that meetings with the relevant agency are more valuable to firms than submitting public comments.Finally, relying on a survey of interest groups, Yackee (2020) demonstrates that firms believe their lobbying has less influence over agency rulemaking than over agency guidance documents, which are a less transparent form of agency policy making.
The comparative literature has produced significantly less work on bureaucratic lobbying than its American politics counterpart.One exception is studies of lobbying in the European Union, where research has delved deep into lobbying trends (e.g., Klüver et al. 2015).In other contexts, the lack of a cohesive and deep body of work presents an opportunity.In some cases, the research gap can be attributed to the relative lack of structured venues for interaction (e.g., no established rulemaking process); in others, the gap is due to the difficulty in systematically observing exchanges between interest groups and bureaucrats.In developing countries, it may be more common for interest groups to influence the bureaucracy through informal avenues (e.g., through personal contacts or revolving doors) and at points besides rulemaking or other centralized bureaucratic tasks (e.g., at "last-mile" implementation stages).These may be important arenas of influence in the United States and the European Union as well, where they likely coexist with more formalized lobbying processes.Thus, scholars in both subfields may need to develop entrepreneurial ways to observe influence taking place in more informal and decentralized ways.
One area in which research is relatively comparable across country settings is public procurement, an activity that is universal to bureaucracies, and over which private firms have strong incentives to influence outcomes.Electronic procurement data are widely available in many countries, such as the United States and Italy, which has encouraged a notable research focus on these two countries.When available, these data can typically be accessed at the individual contract level and can be aggregated to higher units, including municipalities, agencies, and even countries.Electronic procurement records are also available in a handful of developing countries (e.g., Colombia).
The availability of fine-grained procurement data has allowed scholars to study both the mechanisms by which firms and interest groups attempt to influence procurement decisions and the political conditions that give rise to less competitive and more corrupt procurement practices.For example, one prominent mechanism is campaign donations; studies show that campaign donors experience better procurement outcomes than nondonors in numerous countries, including Brazil (Boas et al. 2014), Colombia (Gulzar et al. 2022, Ruiz 2021), the Czech Republic (Titl & Geys 2019), and the United States (Fazekas et al. 2023, Witko 2011).Turning to political conditions, scholars have uncovered a relationship between noncompetitive procurement practices and a host of largely undesirable structural conditions.Examples include low political competition in Sweden (Broms et al. 2019) and Italy (Coviello & Gagliarducci 2017), limited ex ante transparency in the procurement system in Italy (Coviello & Mariniello 2014) and in EU countries (Bauhr et al. 2020), less insulated and more politicized agencies in the United States (Gordon 2011, Dahlström et al. 2021, Krause & Zarit 2022), and reduced merit protections in European regions (Charron et al. 2017).
Procurement can also be used to gauge bureaucratic performance.Focusing on cost overruns and contract delays, scholars emphasize how poor contract performance is precipitated by ideologically misaligned bureaucrats (Spenkuch et al. 2022) and reduced agency competence (Decarolis et al. 2020) in the United States and inefficiencies in supervising courts in Italian regions (Coviello et al. 2018).
In the developing world, the general lack of systematic procurement data has pushed scholars to find other ways to study this topic.Their efforts have led to insights into ways in which procurement politics may differ in particular country contexts.For instance, Brierley (2020) combines interviews and survey experiments to study one form of procurement corruption, kickbacks, in the context of Ghanaian local governments.She finds that in districts where political overseers can punish bureaucrats by transferring them-a feature that is not present in all countries-corruption is more likely to occur.International organization spending related to development aid provides another opportunity to study procurement politics.Dávid-Barrett et al. (2020) use World Bankfunded aid projects to study procurement-related corruption in more than 100 countries that receive development aid; they find that procurement reforms are more effective in lower-capacity countries.Heinzel (2022) shows that international organizations that employ home country staff experience an improvement in the effectiveness of their aid-related procurement projects.
Path dependencies in the American and comparative politics subfields have shaped the direction of research on external influences, but these dependencies should not be considered determinative for future work.Further research on US lobbying, for example, could move beyond rulemaking and consider other venues for influence peddling, such as the revolving door in and out of the bureaucracy (Lee & You 2022) and the federal guidance process (Yackee 2020), as well as other ways lobbying may occur-for instance, when legislators lobby bureaucracies (Ritchie & You 2019).There are ample opportunities to explore bureaucratic lobbying beyond the US context, including research on the revolving door and its effects (e.g., Peci et al. 2022).Many countries also have public participation systems akin to the United States's notice-and-comment rulemaking process that could be mined.For instance, Kornreich (2019) finds that central bureaucrats in China are surprisingly responsive to public comments from street-level implementers.
Although research on public procurement is more unified across subfields, considerable opportunities for advancement remain.The transparency and stringency of procurement rules are central features of extant research, and scholars would do well to focus on the political influences shaping the formation of these rules in the first place (Dávid-Barrett & Fazekas 2020) and the downstream consequences of changing them (Bosio et al. 2022).Additionally, the fine-grained nature of procurement data allows scholars to screen for potential corruption risks (Gallego et al. 2021) and to answer questions about how purchasing shifts in response to distributive politics concerns.Future research could also examine how procurement outcomes are shaped by the selection, capacity, and careers of procurement officers themselves (Best et al. 2022).Finally, scholars have noted that, in the United States at least, the turn toward using procurement to acquire services from vendors-rather than products-has crowded out bureaucrats (e.g., DiIulio 2014, Verkuil 2017), although the political ramifications of this development are not yet well understood.

CONCLUSION
Within political science, the study of bureaucracy is fragmented.Scholars of American and comparative politics ask different questions, study different levels and types of bureaucracy, and apply divergent research approaches.Americanists generally tend to focus on policy making by national-level bureaucrats, relying on large-n quantitative studies-and often formal models too.Meanwhile, comparative work centers on implementation decisions by street-level bureaucrats, and comparativists frequently apply survey and experimental methods.
Though they are sometimes driven by institutional differences across country contexts, we believe that these differences are just as much a function of data availability, research incentives, and path dependency within subfields.Over time, the research divides we have identified have widened due to professional incentives that encourage conversations and research presentations within subfields.Professional networks are organized by subfield, rather than thematically.Convention leads journal editors to select peer reviewers from within, rather than across, subfields.Accordingly, researchers often have few incentives to read-let alone cite-research from the other subfield, regardless of its relevance.This is evident in the inventory of articles we collected for this review, which were published in general-interest political science journals.
Of course, the past need not dictate the future.Our review has demonstrated the discipline's increasing desire to understand bureaucratic politics.This is an opportunity.Within both research areas reviewed here, there are clear openings to apply the approaches of one subfield to the other.Yet, the field of bureaucratic politics faces a classic collective action problem: Each scholar's research would be improved by engaging with research from the "other side."In our view, studies of bureaucratic politics should be informed at every stage by perspectives across subfields.Ideally, researchers would circulate research designs and collaborate, and journal editors would solicit reviewers, with a focus on expertise in bureaucratic politics, rather than any specific country context.
Realizing this vision requires incremental change in the underlying professional networks in the study of bureaucratic politics.Suppose, for example, a journal mandated representation of each subfield during peer review.At this point, the most likely outcome is that the subfield blind spots we have identified in this review-the different terms of reference and subfield-specific yardsticks for what is deemed important or significant-would lengthen the process or lead to worse outcomes.This also means no individual researcher has an incentive to encourage cross-subfield review their own citation practices.
However, there are at least three individual steps that are incentive compatible.First, we view the contents of this review as an outline of subfield cleavages that can aid scholars if they are invited to review studies of bureaucracy outside of their subfield.We hope this information will be useful to reviewers and lead to higher-quality evaluations of manuscripts.Second, we recommend that scholars who host workshops and conferences and organize panels do so in terms of topics, rather than regions.If our analysis is right, a panel focused on the topic of bureaucratic personnel selection, rather than bureaucratic personnel in the United States or the Global South, will both foster more interesting dialogue and do more to promote conceptual and methodological innovations.Third, as educators, we recommend the creation of bureaucracy courses that expose students to a broader spectrum of the literature, regardless of subfield divides.In time, such action can propel a new generation of bureaucracy scholars who have a deeper understanding of bureaucracies across a variety of contexts.The problems that vex bureaucratic governance remain far too common around the world to be treated as separate and unique.

DISCLOSURE STATEMENT
The authors are not aware of any affiliations, memberships, funding, or financial holdings that might be perceived as affecting the objectivity of this review.
Figure 1 Trends in the study of bureaucratic politics, 2000-2022.(a) Share of total articles focusing on the bureaucracy in the American Political Science Review, the American Journal of Political Science, and the Journal of Politics.Dotted line indicates a linear regression line.(b) Total numbers of articles in the American and comparative politics subfields published by these journals.