Analysis

Targeting anticorruption interventions at the front line: developmental governance in health systems

Abstract

In 2008, Vian reported an increasing interest in understanding how corruption affects healthcare outcomes and asked what could be done to combat corruption in the health sector. Eleven years later, corruption is seen as a heterogeneous mix of activity, extensive and expensive in terms of loss of productivity, increasing inequity and costs, but with few examples of programmes that have successfully tackled corruption in low-income or middle-income countries. The commitment, by multilateral organisations and many governments to the Sustainable Development Goals and Universal Health Coverage has renewed an interest to find ways to tackle corruption within health systems. These efforts must, however, begin with a critical assessment of the existing theoretical models and approaches that have underpinned action in the health sector in the past and an assessment of the potential of innovations from anticorruption work developed in sectors other than health. To that end, this paper maps the key debates and theoretical frameworks that have dominated research on corruption in health. It examines their limitations, the blind spots that they create in terms of the questions asked, and the capacity for research to take account of contextual factors that drive practice. It draws on new work from heterodox economics which seeks to target anticorruption interventions at practices that have high impact and which are politically and economically feasible to address. We consider how such approaches can be adopted into health systems and what new questions need to be addressed by researchers to support the development of sustainable solutions to corruption. We present a short case study from Bangladesh to show how such an approach reveals new perspectives on actors and drivers of corruption practice. We conclude by considering the most important areas for research and policy.

Summary box

  • Corruption undermines good quality care.

  • Anticorruption interventions have failed to yield effective results.

  • New approaches to research and policy are called for.

  • The social, political and economic drivers of corruption must be taken account of in new research and policy.

  • Anticorruption efforts must be targeted at specific practices that undermine the health system and can be feasibly addressed.

  • ‘Developmental governance’ and targeted interventions offer a new way to manage informality and corruption within health systems.

Introduction

The critical role of governance in health systems strengthening,1 2 achieving Universal Health Coverage3 and equitable financing4 has long been recognised.5 Among health systems researchers, this has come with significant intellectual investment in the development of theory and methods,3 6 reflected in findings from a recent review which identified 17 governance frameworks used in health systems research.7 Theoretical development has been followed by considerable empirical work, with analysts moving beyond examinations of governance as a normative goal to descriptions of how people make and apply rules across the health system.8–10 This shift in focus has provided intellectual space for a more nuanced understanding of the formation and application of rules in different settings. Under the rubric of ‘everyday governance’ an analysis of ‘practical norms’11 has explored informal rules and norms that emerge as health workers and patients respond to different aspects of the contexts in which services are delivered.3 9 These studies have sought to persuade the global health community to recognise these informal practices,4 although agreement on how to manage informality when developing formal governance arrangements has proven elusive.

Corruption and anticorruption fall within the field of research on governance but our reading indicates that there has been less interest in these topics among health systems researchers than in other governance concerns (eg, participation, institution building and accountability). Often defined as practices by which individuals and groups exploit public resources for individual benefit, corruption in health systems may be ignored for various practical, social and political reasons.12–15 Not all forms of informality are corrupt, and some practices are forms of ‘survival corruption’ offering practical solutions to the difficulties of delivering care, especially in severely underfunded health systems.14 16–18 Work on corruption can also be hindered by the fact that it is difficult to define.14 19 Early descriptions, ‘the abuse of public resources for private gain’ ignored corruption within the private sector and failed to provide an explicit acknowledgement of the central role of power in its manifestation.20 More recent definitions, ‘the abuse of power for private gain’ enables a focus on both public and private sectors but continues to rely on a clear division between public and private spheres—a distinction that is not always clear or universally recognised.21 This definition continues to focus excessively on the individual. Moreover as Gaitonde argues, the system in which they are embedded and the enactment of corruption for the benefit of a group, organisation, party or others close to the person who is abusing power, is obscured in these definitions.19 It also makes it difficult to take account of the fact that corruption is socially construct and sits on an axis between moral/immoral and legal/illegal activity (with some forms recognised as legal but unethical and others as illegal but morally defensible).22

Policy-makers have sidestepped these debates by identifying actions or offences that can commonly be agreed on as corrupt (United Nations Convention Against Corruption 2003). In the health sector, these include theft (of money, medicines and consumables); demands for informal payments or bribes; absenteeism among staff; inappropriate referral and diversion of patients from public to private facilities; and inappropriate prescribing (often under pressure from pharmaceutical firms) and provision of misinformation19 23 Each of these is likely to significantly challenge the central goals of health systems—the realisation of the right to the highest attainable standard of health and financial protection.24 Recognition of the threat that corruption poses to the realisation of Sustainable Development Goal (SDG) 3 (Ensure healthy lives and promote well-being for all at all ages) and SDG16 (Promote peaceful and inclusive societies) has driven recent efforts by multilateral and international organisations and some governments.25 The World Health Organization (WHO), the Global Fund and United Nations Development Programme (UNDP) have proposed a Global Network on Anti-Corruption, Transparency and Accountability.26 Their work is, however, significantly undermined by a lack of data on effective strategies and solutions (even on a pilot basis) that have been tested in high, middle or low-income settings.19 27–29 Gaitonde’ striking conclusion in a 2016 Cochrane review states that there is ‘a paucity of evidence regarding how best to reduce corruption in any (ie, in high, middle or low-income) setting’.19 A Transparency International report suggests that many of the current anticorruption approaches within health are either ineffectual or ad hoc or both.30

This paper examines the potential for a new direction in anticorruption research and practice within health systems. It draws on the governance literature and in particular the idea of ‘developmental governance’. This approach is distinct from the dominant liberal approach to governance known as ‘good governance’ that has been advocated primarily by the World Bank and the International Monetary Fund. The good governance framework is heavily influenced by free-market economics, pays scant attention to historical change and is deeply problematic in its reliance on norms and benchmarks drawn from the features of (mostly) western, capitalist economies. Developmental governance, on the other hand, comes from heterodox political economy and uses a historical institutionalist approach to understand conflictual political processes of economic change. It examines the structural nature of informality and corruption as a feature of the economy and social relations, rather than something that has emerges from social norms and culture. Unlike the good governance framework, its concern is not with transforming whole societies but rather it focuses on anticorruption measures as central to tackling specific informal practices that are detrimental to development (the efficiency of the economy and the equitable distribution of high-quality services). It is known as ‘developmental governance’ because the framework demands that any intervention explicitly couple governance goals with identified development goals. It is particularly salient for the present discourse on decolonising public health because it is rooted in a rereading of history that explicitly recognises how colonialism has affected governance.

Anticorruption research has only recently started to draw on this framework, it has been used by researchers in Bangladesh to identify a strategy to tackle fraud in the skills training sector by improving organisational capability the demand for skilled workers31 and in Nigeria to make changes in the electricity sector.32 Anticorruption researchers from Nigeria, Bangladesh, Pakistan and Tanzania have also started to use this framework for anticorruption work in the health sector,23 33–36 but its theoretical implications for health systems have yet to be presented in detail. This paper seeks to add to the literature by settings out the theoretical and practice changes that a developmental governance approach to anticorruption work in health. Our case study focuses on front-line staff, in part responding to research that shows that clinics and health centres in many countries are dominated by informal practice3 and that policy-makers need frameworks and guidance on how to manage and when to act on this informality.

We are cognisant and supportive of the call for decoloniality within global health. This paper is the result of a long partnership between South-North institutions and is one of several papers to be published collectively. It has explicitly sought to present a theoretical framework that challenges ideas of improvements in governance as a northern-focused benchmarking process. The paper has three sections. First, it traces the history of corruption research in international/global health. It shows how dominant theoretical frameworks have largely ignored key contextual factors, namely political power, social networks and everyday forms of informal distribution of resources. Following this, it examines the potential of innovative approaches in anticorruption research from other sectors for health systems research. Finally, the paper presents research on absenteeism among doctors in rural Bangladesh to show how a nuanced account of the formal and informal structures in which health workers operate draws out hitherto unexplored elements of informal practice. This detailed knowledge provides the canvas on which effective and feasible anticorruption strategies can be formulated. We conclude by considering the potential impact of these theoretical approaches to the field and discuss possible intersections and collaborations between researchers and policy makers around these new approaches.

Research on corruption in health

Governance research and policy (of which anticorruption work is a subset) has developed through three more or less distinct phases.37 38 The first, emerged in the late 1980s and 1990s as neoliberal political orthodoxies challenged the position of the state as the driving force of economic development and provider of services in both high-income and low-income settings.38 39 The World Bank, with a particular focus on Africa argued that the problems of economic development lay with a crisis in governance at the national level.40 Although, as Mkandawire argues, the term ‘good governance’ had originally been devised by African economists wedded to developmental, democratic and socially inclusive policy-making,34 it was reworked at the World Bank. Good governance became a set of narrow and technical interventions devised with the explicit aim of protecting markets and private sector actors from what was seen as a predatory, neopatrimonial state.37 38 40 41 The law, bureaucracy and judiciary were critical areas for intervention: new legal frameworks and organisational practices that would support market economies, uphold the rule of law, and tackle rent seeking were established.37 38 40 41

First phase of corruption research in health: principals, agents and good governance

Whereas research on corruption initially focused on protecting private sector development, interest was then extended into governance in the social sectors with a new focus on creating more effective and accountable service delivery in education and health.38 42 43 Although ideas of social accountability were rare within health at the time, corruption research across the social sectors instrumentalised ideas of social accountability combining them with technocratic approaches to formulate anticorruption tools. An extended set of actors (citizens, civil society organisations as well as government authorities) were to be mobilised to curb the behaviour of corrupt officials.44 Many of these interventions were based on assumptions that increasing transparency would lead to more accountability, which coincided with an interest in promoting community participation in critical public services.

It was in this phase that the debates on corruption within public health began in earnest and three papers were written that came to dominate the field: Savedoff and Husemann (2006), Lewis (2006) and Vian (2008). In all three, the health sector is identified as having particular vulnerabilities—connected to unusually high government spending coupled with entrusting private providers (individuals and organisations) with important public roles. Each drew on ideas of good governance and neoclassical economic theory, to which Vian added issues of fraud and risk.

Savedoff and Husemann’s chapter in Transparency International’s Global Corruption Report drew on agency theory to identify critical relationships that, they argued, provide space and incentives for corruption in the sector.45 This was a significant innovation in health economics—where agency theory had been used to analyse inefficiencies for the first time. Two critical elements of the principal agent relationship were considered important in facilitating corruption.

‘asymmetric information leads to a series of problems that are usefully analysed within the framework of ‘principal—agent’ relationships. In such a framework, the ‘principal’ hires an ‘agent’ to perform some function. When the agent has interests that differ from those of the principal and when the principal cannot get complete information about the agent’s output, it is difficult to find contracts that are optimal.’ (Savedoff and Husseman 2006)

The authors recognised that forms of corruption differ between different types of health system, but their analytical model identified five main actors (regulators, payers, healthcare providers, suppliers and consumers/patients) present in all systems and two characteristics of principal–agent relations (diverging interests and informational asymmetry) that shape corruption across the world. According to them, all systems are vulnerable due to the uncertainties intrinsic in ill health and healthcare (in terms of when and who becomes ill and what treatments are likely to work) and the large numbers of dispersed actors working in the sector. For them, the only possibly response to corruption is a mix of accountability and transparency measures that would enable corruption to be ‘seen’ but their framework lacks a means to identify who would be able and motivated act on corruption. They are clear, however, that the amelioration of these agency problems can only come about with measures introduced across the whole sector and policy measures that will strengthen legal and regulatory frameworks.

Lewis’s paper also draws on agency theory but she broadens the analytical scope by considering the applicability of the entire World Bank’s good governance framework to service delivery within the health sector.46 It provides a classic analysis, in which the control of corruption is one of six areas that shape governance at national level (voice and accountability; political stability and lack of violence; government effectiveness; regulatory quality; rule of law; and, control of corruption). Three are identified as being particularly pertinent to service delivery: voice and accountability, government effectiveness and control of corruption. For Lewis, the focus on corruption comes in part from the lack of data on governance and the desire to find a proxy for good governance, the logic being that poor governance leads to high levels of corruption. Of interest, is the evidence that Lewis provides on voice and accountability, which at this early stage of anticorruption work in health showed a tenuous link to health outcomes—suggesting that citizens were not necessarily very effective in calling bureaucrats, politicians, and front-line providers to account in countries where democratic processes were not firmly embedded. Lewis is also circumspect about the potential effect of the involvement of communities in overseeing service delivery in health:

‘While voice matters in many service delivery instances, its role in health seems to be less straightforward. The specialized nature of medicine, the heterogeneous products of the sector and the status of physicians in most societies complicates oversight by communities.’ (Lewis 2006:43)

Considering anticorruption, like Savedoff and Hausmann, Lewis’ advocates a whole of government approach to policy making–an emphasis on what she terms ‘clean government’, with the expectation that improvements in transparency will curb decision-making space of officials and changes in incentive structures will address diverging interests and information asymmetry between principals and agents. Despite the lack of evidence on its effectiveness, she maintains that citizens and communities need to be involved directly in improving governance.

Vian’s paper, written 2 years later, drew these and on other approaches that had emerged in health, together with antifraud approaches, to set out the first and still dominant health sector-specific framework for understanding corruption within health. It combines an interest in relationships and individual decision-making processes (taken from principal agent theory), with good governance frameworks and risk-based models. Again, the focus is predominantly at an individual level, as she argues ‘corruption is driven by three main forces: government agents who abuse public power and position for private gain do so because they feel pressured to abuse (financially or by clients), because they are able to rationalise their behaviour or feel justified (attitudes and social norms support their decision) and because they have the opportunity to abuse power.’ Her explanation of these opportunities draws extensively on good governance models, arguing that the space in which corruption may be enacted is influenced by monopoly, discretion, accountability, transparency, citizen voice and enforcement.47 As with the other two models, transparency and accountability measures dominate alongside whole of sector approaches to anticorruption.

Second phase of corruption in health: from principals and agents to politics, social networks and practical norms

By the mid-late 2000s, across many sectors beyond health, there was growing disquiet that the basket of interventions that were being pursued by the World Bank, national governments, and bilateral development agencies were yielding very poor results.48–50

Researchers in other fields revisited the theory underpinning practice and several critiques of blind spots in principal agent theory and associated good governance models emerged (see reference 51 for a discussion of these), of which three are important for health. The first is that principal agent theory fails to consider the complexity and fluidity of relationships. In the health system, at different moments, the principal may be the state, the citizen or even health workers who may wish to call their managers or ministers to account (when, eg, medicines are scarce and commodities do not arrive in facilities). Second, while good governance interventions were logical and normatively appealing, very little work was given over to understanding how power and its distribution between individuals and among groups shaped governance: social, political and/or economic contexts in which corruption unfolded was not considered.37 48 52 53 Instead, the idea was that once better information was provided (through mechanisms promoting transparency) then asymmetry would disappear but in the context of developing countries where corruption is widespread, information asymmetry is often nested in deeper, structural asymmetries of power. The third point, closely related to the second, is that in places where corruption is endemic, it seemed likely that corruption (and anticorruption) was a form of collective rather than individual practice—associated with networks and groups, informal rules and organisations.50 52

Within health, two sets of new studies that focused on the relational nature of corruption demonstrated the need for theory that could make sense of the social nature of corruption and its complexity. The first set came from ethnographic accounts that focused more generally on African bureaucracies.11 In these, rather than being created through the actions of unregulated, individual agents acting in their own self-interest, rule breaking and corruption among African bureaucrats and front-line service providers emerges as part of a system of generalised informality. According to these accounts, corruption does not emerge from a ‘lack’ (of accountability, transparency, voice, etc) and ungoverned spaces. These ethnographic studies show how corrupt practice emerges in places where social and economic logics combine with local, sectoral and individual microdynamics to create informal rules and ‘tacit codes’ which dominate everyday practice.54 Pluralist systems of regulation dominate55–57 with the dialectic between formal rules and the realities of daily practice creating a moral and normative ambivalence about corruption among providers and recipients of care.58 59 Within this body of literature, studies of health systems show how political patronage determines, for example, payment for staff in Uganda.60 Practical norms shape who provides and gains access to the material resources that the health system offers (medicines, consumables and larger equipment) and on what terms, who provides and receives treatment, where, when, and how quickly; guiding decision making about how to distribute resources in settings where their availability never matches the demands made on the system.58 61 These accounts also provide an important warning for anticorruption researchers, that although corruption creates inequalities it is also a form of problem solving in adverse settings.

The second, interconnected way in which context has been brought into anticorruption research in health has been through research that focused on corruption as a collective action problem and so looked at the ways in which networks of actors were involved in corruption.11 62–64 This work is often less detailed than the ethnographic studies but nonetheless it usefully shows how networks of actors rather than isolated individuals are implicated in corruption within health systems.65 These complex, interconnected social and professional networks effectively constitute a second, informal system of exchange through which scarce resources (including access to services and medicines) are distributed.65 66 These are structured by informal but systematic patterns of rules which have different forms and qualities in different settings. They may draw on political or kin-based ties, or be formed of looser social relationships that can stretch across sectors and different types of street-level bureaucrats (teachers, social workers, health workers) so that resources from the education section (access to schools) may be traded with those from the health sector (more timely access to health workers).63 In places in which petty corruption is commonplace, these broader networks are often found, with more highly concealed and closed networks often emerging in countries where petty corruption is not tolerated.66

These bodies of work provide thick descriptions of corruption and demonstrate how any model or framework must account for the ways in which historical, social and economic factors do much more than create rationalisations for corruption (as suggested in Vian’s model). Instead these factors are embedded in the form that corruption takes and must be reflected in the design of anticorruption practice. They challenge the models for action as set out by Savedoff, Lewis and Vian, but beyond demonstrating that informality must be taken seriously, they provide little guidance on how this can be done. The question that many health systems researchers ask remains pertinent—how do we transform anticorruption practice so that it can factor these insights about informality, socioeconomic context and power into anticorruption theory and practice?

Developmental governance: new political economy approaches to anticorruption interventions

Although they rarely focus on health policy and practice, debates within development studies on the long-term political drivers of social and economic change38 67–69 offer an important starting point from which anticorruption practice can shift away from principal agent theory and ideas of good governance. Of these, Khan’s idea of ‘developmental governance’ offers the most theoretically developed approach to anticorruption.70 Like others, Khan is highly critical of good governance and the way in which it benchmarks progress across the world against Euro-American countries. Khan questions the fundamental assumption within the good governance framework: that it was enhancements in the rule of law, the protection of property rights and contract law that created the conditions through which highly productive capitalist economies emerged (see ref. 71 for a description of key Good Governance arguments). Reminding readers that the early stages of colonisation were devastatingly damaging for indigenous people, he argues that the evidence for Acemoglu et al’s influential thesis that settler (as opposed to extractive) colonisation lead to well-protected property rights and therefore economic development is scant72 ,73 . Instead, drawing on the example of the Enclosures of the 16th century, Khan argues that the capitalist transitions that underpinned high economic growth have never relied on ‘good governance’ mechanisms and have been anything but rule following. Similar examples can be made out of the history of oligopolistic chartered companies in 18th century in England and America’s infamous Gilded Age in the late 19th century. In countries that were colonised, the social re-engineering further complicated the already conflictual process of economic and social development.74 75 Policies which view high-income country institutions as a standard against which others should be judged ignores both the history of economic transitions but also colonial histories. While he argues, therefore, that good governance is essentially based on a fallacy, Khan does agree that governance matters for economic development. Presenting data from South Korea and Taiwan, he argues that the transformation of these economies has come about because they had governance capabilities that could address critical and specific growth-constraining characteristics (rather than first try and enforce rule of law and ensure impartial enforcement).70 74

These historical insights are important to understand because they create the basis for substantial shifts in the types of anticorruption reforms that are expected to yield results. Where good governance sees anticorruption measures as one part of wholesale governance improvements that needs to occur across sectors (as in Lewis 2006), developmental governance sees targeted anticorruption measures as the starting point for governance improvements. Finding pockets of change where action is feasible is complex and cannot rely on blue-prints or standardised solutions. Because they are targeted, anticorruption should only focus on forms of corruption that are highly damaging and should leave those that have less of an impact or serve as forms of problem solving to be resolved at a later date. It also recognises that some forms of corruption are highly damaging and inequitable but cannot be overcome without major political transformation. Such systemic political transformations cannot for the most part be made through policy change and intervention; and a policy change that destabilises a political settlement may have profound unintended consequences that could be worse than the corruption itself, increasing contestation among powerful parties and potentially leading to political violence and conflict.

The research that underpins developmental governance approaches to anticorruption has to be detailed and allow for a nuanced approach. Here, a health systems approach to developmental governance has to expand Khan et al’ model. Where they focus primarily on political economy factors that shape corruption and the potential for anticorruption intervention,76 health systems and policy research requires a more holistic approach that draws in a range of social, cultural and health systems factors that drive and shape practice. Baez Camargo and colleagues work alongside that of de Sardan and colleagues (see above) becomes extremely useful, making sense of the local drivers of practice, the forms of corruption that are most detrimental to the system and those which will be feasible to address through targeted policy change. Knowledge of the everyday realities of working in an underfunded or fragmented health system; an in-depth analysis of the informal rules that emerge; the networks in which resources are exchanged; the influence of powerful groups and organisations who distribute resources and opportunities informally within health systems.

Khan et al propose four strategies for action: changing individual and group incentives; creating policy that recognises and acts on differences between actors usually seen as a homogenous; creating forms of collective action among groups who are powerful enough to effect anticorruption; and rendering the rights of different actors transparent.76 For health systems, which are often underfunded we have added a fifth, targeted anticorruption investments.

Case study: absenteeism among doctors in rural Bangladesh

Absenteeism among doctors in Bangladesh fulfils the first criterion in Khan et al’s approach (corruption that is highly damaging) as it critically undermines access to effective, equitable and high-quality care.76 Widespread across the country, absenteeism is a particular problem in rural areas. Research from 2003 reported 41% of doctors to be absent77 and more recent comparisons between urban and rural areas shows a doctor, population ratio of 18:10 000 in urban and 1:10 000 in rural areas.78 It is further compounded by failures to recruit doctors into these positions,79–81 and has recently become an important priority for the Bangladesh government.82

To understand how policy, health systems resources and constraints combine with political and social networks to shape absenteeism among doctors in rural areas, thirty in-depth interviews were conducted with doctors about their experiences of absenteeism. The majority were junior doctors in their rural placements (n=18); five were mid-level doctors who had been assigned to a rural position within the past 5 years (n=5). We also interviewed seven senior doctors, most of whom were in management positions to provide us with a view of absenteeism as seen from the top to down.

Patient and public involvement

This research was done without patient involvement. Patients were not invited to comment on the study design or interpret the results. Patients were not invited to contribute to the writing or editing of this document for readability or accuracy.

Findings

From these interviews, we were able to group the doctors according to three courses of action. In the first group, which was relatively small, were those who were posted to a rural facility but either never arrived at the facility or arrived and then left shortly afterwards (often described as being there for a number of days) never to return. We did not meet any doctors who fell into this category. Instead, the actions of this group were based on cases described to us by colleagues so the processes involved in this form of absenteeism was mostly opaque. We often did not know whether the rural posting had been cancelled and the absent doctor had managed to get an ‘attachment’ or transfer to another health centre. In some instances, however, respondents described other doctors as being registered as absent for a few months after which time their names would disappear from the register and the positions would be relisted as unfilled. The second group of doctors were (or had been) those who were present at a rural facility some of the time but would take quite long periods of time (up to several weeks) away from their posts. Finally, a third group of doctors, mainly junior doctors who very rarely took time away from their positions. They often complained bitterly of the difficulty of ever taking leave (eg, to attend family occasions) and the difficult and extremely hard work that they had to do (in part to cover the work of those who were not there but also to cover posts that had not been filled).

Explanations of the differences in practice

Incentives and pressures within the health system

The study found few social, professional or economic incentives for doctors to stay at rural health facilities. Rural postings were described as difficult, stressful and dangerous. Male doctors complained of the dangers associated with being called out of the health facility and into villages to attend cases, especially if two families or factions were in dispute about the cause and consequences of an accident. Female doctors were also concerned about the lack of security in the health facility compounds, the dangers of undertaking night shifts because of a lack of security at night and, a complete lack of suitable housing either within the compound or beyond it (in one instance, in a very remote area). Both male and female doctors also described the ways in which local residents would put them under pressure to issue false or fabricated medical certificates and described verbal abuse and theft of their property.

Although policies present rural work as part of junior doctors’ career progression, all those we interviewed saw their time in the rural health centres as a threat to their careers and the achievement of a specialist training placement, seen as critical in a health system that affords little space for medical generalists. The workload made it difficult to find time to study for postgraduation exams necessary for entry to a specialist field. The highly competitive exams require considerable revision; training courses take place in urban settings and demand periods of absence that are not always sanctioned.

The influence of social and political networks

While these pressures from within the health system help to understand what drives doctors to be absent, it does not explain the differences among them: and why it was despite widespread dissatisfaction that some left the health facilities while others stayed. The explanation of the patterning of this absence lies in structures beyond the health system and the way in which they influenced the options of rurally posted doctors in Bangladesh.

In the first group (the majority being junior doctors), those who either arrived and left very quickly or those who never arrived at all, were those who were described as closely connected to politically powerful and/or, those with very strong professional networks, or those whose work could be covered by a family member (we encountered two cases where husbands, who were also doctors, covered their wives’ shifts). The following quotation was from a junior doctor in a rural area who was describing a case of absenteeism in her current rural health centre. Her discourse shows how it was logical to seek to move out of the health centre, if she/he has the financial resources.

‘They transferred 2 doctors (husband-wife) here but they left after joining and still now they are in the count. I have heard that their financial status was good, so they didn’t stay here. So, while here the doctor count is 5, ultimately [that appears that there is] no shortage of doctors. But, actually we are 2 in number to provide patients service.’

Whereas the quotation above points to wealth as a means of moving out of the rural placement, another junior doctor who described how she had had to pay to get a placement closer to her family and also made clear the ways in which those with powerful political and social connections did not have to make an unofficial payment to move from one health facility to another.

F:…as no one wants to go in places like X so it was not a preferred place to get the transfer that’s why the amount of unofficial payment [I paid] was also low. I offered only 2000 BDT. Everyone has to pay the money.

R: What was the cost of posting in X [a better area]?

F: It depends on the networking and the situation. Some people don’t need money even. The son of X [a senior politician] applied to change his posting from [rural health centre] to the nearest Institute of Health Technology. Immediately, after his application, staff were instructed to search for a vacant position for him and issue the order. He was transferred to the nearest location from his house from where he can continue [his training].

This case suggests that these forms of corruption are quite systematic and embedded within social and political structures of Bangladesh. More desirable postings cost more in terms of unofficial payments and more powerful networks enable doctors to move quickly and more cheaply to good quality positions.

Among the second group of doctors, (those who stayed at the facilities for a longer period of time but who were often absent as they sought new placements) were those who had access to political and social networks who could help them secure an urban position and who were able to pay the necessary unofficial payments as they lobbied different members of the civil service to find them. The difference between this group and the first is that these doctors took longer to move because they had poorer access to networks and/or were relatively less well-off so had to work much harder to secure a placement beyond the health facility and the effectiveness of networks in securing these places was not guaranteed. Below, three different junior doctors reflect on their own attempts to secure an urban placement and how this impacted on the time that they spent away from the rural health centre.

‘I tried to talk with different political leaders [to secure] my transfer using my social and personal network. I took 1-month’s leave and again, extended [it] for 21 days to manage my posting.’

‘I collected the information of vacant position and mentioned the area with position name in my application to the authorities at centre level . If you just submit the application and sit, your application will not move and you’ll not get your expected posting at your selected area. Doctors have to go repeatedly to take the news of the file and sometimes have to take appointments and meet them. Giving money to the peons, clerks and other staffs is also an important step of having transfer to a good area.’

‘I have taken forwarding from the head of the respecting health facility and submitted it to the divisional director. Then, I went to Dhaka for further communication. It took total 2 months to complete the process. For those two months, I had to go to every table to moving my file from one table to another. During this process many people have to spend their money [i.e. make an unofficial payment] on a regular basis. If anyone spent money, their transfer process will be fast compared to others.”

Whereas the powerful politician’s son did not have to look for a position himself or lobby for a change in posting, in each of these cases, the junior doctors had to have direct contact with those either in the bureaucracy or with powerful individuals who could influence those in the bureaucracy to move away from the rural health centre. Moreover, success was not guaranteed even if unofficial payments were made and networks were drawn on. In one case, a junior doctor who had been placed in a rural area with no safe house for her to stay in worked hard to get support from her political networks and those of her husband to lobby on her behalf for a new posting. She failed to secure enough support get a new posting and was left in her rural post in which there was no prospect that she would find secure housing or be able to study for a the post graduate examination. Eventually she resigned her position and left the profession.

The third group of doctors comprised those who very rarely took time away from their positions and who often had to work extremely hard to cover the work of those who were not there and those positions that had not been filled. Among this group, there was a minority of doctors who found their positions fulfilling but there was one interviewee who, despite the hard work, felt protected and valued by his family members who lived nearby. The majority, however, wanted to leave the rural area in which they worked and find a placement in a town or city where they could train for the postgraduation examination. Many appeared exhausted and with little enthusiasm for their work. With no time to study for the post-graduation exam, many also recognised that as they lacked powerful political connections or wealthy family members, it would be unlikely that they could find a way of doing so. One male junior doctor explained that he now regretted his failure to become involved in student politics as that would have provided him with the networks to support a shift to a new position.

‘Due to not having any political linkage or, student life political involvement, I’m unable to have my posting in Dhaka or, around Dhaka. If I [had] known earlier, I would be engaged in politics somehow. Some doctors have to stay at Rangamati or, Coxs Bazar for 10–15 years, as, they haven’t done politics or, haven’t been engaged in the politics.’

Linking the findings back to the frameworks

It is possible to examine these findings from Bangladesh through the models based on good governance, principal agent theory and risk. The desire for career progression and fears of safety could be classified in Vian’s model as a rationalisation of corrupt practice; and the opportunities to engage in corruption can be related to the fact that the Bangladesh health system lacks systems of accountability and transparency and limited enforcement of the formal rules. Such an interpretation is not wrong, but it leaves critical blind spots in the analysis and undermines the generation of a thick description of the context, health worker dissatisfaction and the political economy of power relations that underpins absenteeism, posting and transfer processes in many settings.83–85 It lacks the capacity to understand what we might consider a parallel system of accountability86 and the reasons why, when almost all doctors do not like working in rural areas, some stay while others are frequently or permanently absent.

Instead, absenteeism appears a form of problem solving—a means of managing unsafe worksites and threats to successful career progression and a commonplace and practical response to systemic problems within the health system. Camargo’s hypothesis (drawn from collective action theory) that it is networks of actors rather than individuals that are involved in corruption also resonates with our findings. In Bangladesh, these appeared to be more hierarchically organised networks than those in Tanzania and Rwanda that Baez Camargo and Koechlin examined63; but as in these African settings, it would be very difficult for individuals to step out of the informal system through which more desirable placements are distributed.

Once political and social structures are attended to, it is possible to take account of the differences between these doctors and, as Khan et al suggest, differentiate between actors who benefit and those who lose out from this informal system of distribution. Doctors are neither individuals seeking to take advantage of a lack of accountability nor are they a homogeneous group caught in a web of actors dominated by transactional social relationships. Instead, they appear as unequally positioned social agents who draw on social and political networks as they compete and cooperate for limited material, social and professional resources and rewards. Absenteeism is not an option for everyone; even for those who can use networks and afford to make unofficial payments, success is not guaranteed. The process is both difficult and costly to navigate. It seems likely that broader health systems investments, making improvements in terms of safety for health workers in rural areas, interventions to reshape incentives around career progression and create new forms of collective action the basis through which absenteeism could be tackled in Bangladesh.

Conclusion

The health sector faces particular challenges from corruption.45 Forms of corruption—informal payments, absenteeism, medicine theft, fraud and purchasing positions—can be found in health systems across the world.25 87–89 The effects of these activities are often greatest at the facility level during patient encounters, as providers shape what care is available to the population.

In health, anticorruption research continues to be dominated by frameworks that rely on neoclassical economic formulations of behaviour change, a focus on formal, technical fixes (government policy; and the need for institutional and organisational reform) that ignore the social and political context. Corruption is seen to emerge from ungoverned or poorly governed spaces in which profit maximising opportunists who seek to exploit weaknesses in the system are able to do so and those who do not are framed as more moral actors. New frameworks take account of the broader context of the health system and realities of the social, political and economic structures. They must take note of the considerable and growing critique of the techno-managerial approaches to public health and the evidence that demonstrates that informal practice, politics and political economy shape and often dominate practice. Incorporating developmental governance into health systems research and policy research provides new possibilities for research-driven intervention through which we can develop Gaitonde’s proposal for ‘locally tailored interventions born from a deeper understanding of local dynamics rather than focus on solutions that are universal or up-scalable’.14