Background
Local health governance, and particularly accountability towards citizens has been an object of study and debate since the global health community shifted attention to primary healthcare, and engaged with the good governance and decentralisation paradigms. Already the 1978 Alma-Ata declaration espoused people’s “right and duty to participate individually and collectively in the planning and implementation of their health care” [
1]. The Bamako initiative operationalised the declaration by promoting decentralisation of health service financing and management and universal access to health [
2]. Around the same time the view that decentralisation of public services, and good governance of these services, including health was essential for their enhanced performance, gained wider traction [
3,
4].
Despite increasing attention and the development of multiple analytical frameworks for health governance a need remains to empirically test and validate existing frameworks [
5]. Relationships of power, responsibility and accountability between health systems actors and the institutions that shape this are considered central to health governance. Yet, the functioning of local governance is often studied from the perspective of decentralisation or provider accountability towards communities [
6‐
8], while a gap remains in understanding how accountability relations between the multitude of governance actors shape local health systems, particularly in the former Soviet Union and Central Asia. Furthermore, a need for a stronger consideration of the political and contextual factors influencing accountability relations remains [
9].
This study aims to address this gap by offering an analysis of principal-agent linkages in the local health system of two districts in Tajikistan. Health governance in Tajikistan, or Central Asia in general, has received little attention, and has mainly been focused at central level reforms, such as the introduction of a Basic Benefit Package of health services (BBP) and the role of international aid in this [
10‐
15]. Fragmentation in health financing regulations and stewardship functions coupled with insufficient donor coordination, leading to a proliferation of vertical and pilot programmes suggests the existence of a patchwork of different health governance arrangements across the country [
14]. Closer insight into the way local stakeholders in the health system relate to each other, and how the local context shapes these patterns of power is therefore warranted. To analyse governance relations at the local level in Tajikistan a triadic principal-agent framework is used. This framework allows for an exploration of principal-agent linkages between three sets of actors that are considered to form the heart of health governance: government, providers and citizens. Particular attention is paid to the core components that have been identified to contribute to accountability as a principal-agent relationship.
This paper proceeds as follows. The next section outlines the methods used for this qualitative study, and an elaboration on the choice and assumptions underlying the principal-agent analytical framework of this paper. The following section presents the main health system actors in two distinct districts in Tajikistan and an exploration of their accountability relationships. The implications of the key findings are next discussed in light of the wider literature on accountability and governance, both in terms of the lessons from the Tajik experience and the use of agency theory for health governance analysis. Lastly, the paper offers some concluding thoughts on future entry points for policy, practice and research.
Discussion
This study has provided an understanding of the nature of principal-agent relationships in the local health sector of two districts in Tajikistan and their underlying power dynamics. Beyond that the findings from the two settings can serve to yield insight into the complexity of accountability relations and the way the different components in the process of accountability can relate to each other. The application of agency theory to two cases in this study has also served to highlight its use and limitations. These insights will be discussed and elucidated below in the context of study findings from other settings and relevant conceptual tools and theory.
Despite their different socio-economic, historical and political contexts the qualitative data suggests that in both districts relationships between key governance actors are fraught with generally similar constraints on accountability for equitable and quality service provision as proposed in the triadic accountability model. This however does not exclude the existence of accountability relationships with a different nature. In the face of weak or absent formal accountability mechanisms it appeared that informal interpersonal and inter-organisational behaviours play an important role in establishing an accountability relationship, which confirms theoretical reflections in the field of (health) governance and accountability [
23,
34‐
36]. These formal and informal accountability relationships form a complex institutional web in which agents sometimes also act as principals and vice versa. Particularly health providers, as street-level bureaucrats [
37], find themselves to be in this role, being held accountable “from bottom-up, top-down as well as sideways” [
38] as they face (sometimes conflicting) demands from state actors, citizens and development agencies.
In the relationship between health providers and state actors the findings suggest accountability for the delivery of the BBP is limited by insufficient resources to carry out this mandate, a rigid resource allocation rationale that is de-coupled from population needs or provider performance and monitoring activities that appear more aimed at finding faults in record-keeping and opportunities for resource-extraction through fines and (informal) co-payment revenue than at support for service delivery. This rent-seeking behaviour, which was reported in both districts irrespective of the co-payments associated with the BBP pilot, is in line with patterns in the wider bureaucracy as documented in a related study [
14]. It is important to recognise that the negative, punitive character of this supervision style was found to be an important factor in health staff demotivation and attrition elsewhere and stands in contrast to the more supportive or coaching supervision approach by managers, which has been identified as a strong motivator for health workers in a broad variety of low and middle income settings [
39‐
43]. The lack of decision space, limited resources and capacity to exercise effective accountability has also been found to be critical in other rural low resource settings [
44,
45]. Its combination with a bureaucratically-engrained rent-seeking rationale, which turns monitoring and supervision into a power tool to incentivise the agents (health providers) to serve in the principal’s interest, particularly skews internal accountability away from provider performance, as has been documented extensively in India as well [
46].
According to Hirschman [
27] voice is one of the two important ways, together with
exit, in which people respond to inadequate services. By extension Paul [
47] considered them the two main factors that influence accountability. The findings of this study suggest that accountability between state authorities and citizens in the two districts is hampered by a disaffection among citizens with the severely limited opportunities for them to express their voice and the lack of effective formal enforceability mechanisms accessible to a wider public, i.e. a strong local legislative power that is chosen through free and fair elections. The lack of voice towards government actors resulting from a lack of belief in the possibility of answerability in Tajikistan corroborates findings from elsewhere in GBAO [
48] and also echoes findings from other settings with recent experiences of authoritarian government [
49]. This could also be a factor in explaining the lack of ‘rude accountability’ [
50], found in this study. The instrumental use of threats through shaming or violence as a mechanism for frontline negotiations by citizens towards service providers, constituting this ‘rude accountability’, appears to be going hand in hand with a greater awareness of rights and rising expectations on social service provision in Bangladesh, where this was found [
50]. Tajikistan markedly contrasts with this setting, as expectations of what the state is able or willing to deliver have massively reduced [
51] and bureaucratic and democratic legitimacy have significantly eroded since the short-lived period of openness in the Soviet Union’s last years [
52].
Based on the recognition that the ‘long route of accountability’ is often insufficient or ineffective in incentivising services to be more responsive (international) non-governmental organisations in comparable contexts have over the past decade initiated social accountability mechanism aimed at strengthening the ‘short route of accountability’ between citizens and health providers [
6,
8]. In the two districts of this study however, the primary focus of the community-based organisations that have been formed, particularly in the GBAO district, is not on promoting active decision-making with state authorities. The degree to which they can enforce providers to be answerable appears to depend largely on ‘weak’ or ‘bridging’ social capital ties [
53] that some community representatives manage to establish or nurture to voice their expectations and concerns [
32], whilst information asymmetry between providers and citizens hampers the ability of the latter to do so. The in-kind ‘payments’ or support provided to local rural health facilities, in the GBAO district particularly, can be interpreted as a token in the creation of a social bond, or debt, with an obligation to reciprocate, as elaborated by Mauss and others [
54], or as a limited form of co-production [
55]. As Abimbola noted, this type of collective action by non-state (community) actors to keep primary healthcare services afloat can be found across LMICs, and can be seen as an informal example of collective governance [
56].
The lack of strong or formal channels for citizens to voice their expectations or concerns around health services is particularly significant given the severely limited ‘exit options’ for people, particularly the poor [
57]. Widespread poverty, geographical isolation, bad road infrastructure and a lack of private healthcare provision in addition to a limited network of public facilities contribute to this. The observation that both exit and voice options are severely limited can perhaps explain citizens’ efforts to contribute to the functioning of their local health centres. In other words, the lack of exit and voice options lock them in a type of continuing loyalty of ‘making do’ with the limited services that are available. This is a hypothetical inversion of Hirschman’s theory that high loyalty to a company, organisation or state works to limit people’s voice and exit options [
27].
A number of limitations to the application of the triadic principal-agent model to the study of health systems of low-resource settings have surfaced in this study. First of all, the triadic model does not take account of the influence of the main external development agencies on citizens, providers and state authorities and the relationships between them in such settings. The findings from this study suggest the AKDN and project SINO have mainly dealt directly with health providers, establishing a principal-agent relationship parallel to that between state actors and health providers. This fits a pattern that donors in Tajikistan have mainly worked directly with beneficiaries instead of trying this through the government [
58]. This pattern, although highly time-bound and likely to evolve, is unlike other fragile settings or areas of precarious statehood where external development actors and local health authorities engage in an unbalanced mutual dependency relationship [
45] or networked governance of the health sector [
59] in an imperfect attempt to foster local ownership and systems strengthening.
The cases explored in this study give insight into the internal divisions, power asymmetries and varying or sometimes competing interests, partly stemming from the inherited Soviet Semashko health system. It shows how complex practices of power and contestation over resources within the bureaucracy are influential in shaping policy implementation, mirroring the contestation over resources among local governance actors in South Africa [
60]. The triadic model with its theoretically homogenous actors categories does not serve to understand these complex relations of power. Heterogeneity and competition within the actors categories was particularly evident in the ‘state’ and ‘providers’ groups. The competing interests of the Hospital Director, PHC manager, district health team and the district financial department revolved largely around access to and autonomy in decision-making over the allocation of scarce funds. The side-lining of the district health team in both districts, which was facilitated by the elusive formal mandate of this body and the double hat worn by the Hospital Director, as head of the most important health provider in the district, and key local government health official are the most striking examples of this. These examples confirm that the idea of holding a single agent category, such as ‘the state’, to account can be problematic in practice, summed up as ‘the problem of many hands’ that have contributed to any policy (outcome) [
61].
Altogether these limitations throw up some fundamental question on the use of principal-agent theory for analysing health governance
. Not only are the relations between actors more complex than suggested in the triadic model, the uncovering of rent-seeking rationales and co-production initiatives at community level suggest a need for other concepts and tools to explore practices of power, contestation and collaboration in local health governance. The application of principal-agent theory to governance has been criticised for ‘theoretically mischaracterising’ governance problems assuming the existence of ‘principled principals’, who are willing to hold agents to account whilst embodying the public interest, and the emphasis on individual incentive calculations [
62]. This fails to recognise the implication of principals themselves in abuse of power for private gain, which is suggested in this study, and the expectation of others to be implicated in that. This influence of the (expected) behaviour of others on individual behaviour highlights the collective, rather than individual nature of rent-seeking and the importance of collective norms in perpetuating behaviour that is irrational from the viewpoint of the public good. Approaching governance as a collective action phenomenon, centred around the question what factors can help to overcome harmful equilibria of particularistic interests dominating in governance has therefore been seen as a more useful approach [
63,
64]. Ultimately, however, the two approaches can also be considered complementary, as Marquette and Peiffer argue [
65]. As this study shows, the application of principal-agent theory in the study of local health governance helps to unpack the incentives under which key stakeholders in the system relate to each other. It has provided insight into the challenges to the different components making up an effective accountability relationship, such as an unclear mandate, the lack of effective channels for voice or insufficient resources to carry out a mandate. Future research could help to further explore the phenomena found in this study, with attention to the function and the role of norms in rent-seeking behaviour, as well as the use of collective action theory to understand the role of (mis)trust in governance relations.
Research limitations
This study has been subject to several limitations and its results are time-bound to the period of field research. Policy and organisational details have changed since data collection, and will continue to change as lessons are drawn, funding cycles change and new reforms are piloted or implemented. The limitations in data collection pertain to the relatively short period in which data collection took place, and the limited number of FGDs with citizens. Future studies could rely more on longer-term immersion and participant observation to better flesh out the complexities of agent’s motivations, strategies and practices of power. Respondent bias cannot be ruled out as AKDN, SDC’s project Sino and WHO facilitated entry to the study settings, although their representatives were excluded from interviews and FGDs with other stakeholders. Together with the closed political environment this may have biased respondents’ answers. Some limitations pertain to the topic of inquiry itself. Exploring power relations is sensitive in any setting, but particularly in an authoritarian environment with a legacy of large statist dominance of basic services, the economy and society [
66] that has been penetrated by a more patrimonial type of governance by central elites [
52,
67]. This requires provisions in the presentation of results to protect informants. Lastly, quantitative methods could have helped to gather more representative views on the daily practices of health workers and citizen’s voice.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.