How do external donors influence national health policy processes? Experiences of domestic policy actors in Cambodia and Pakistan.

Although concerns have historically been raised about the influence of external donors on health policy process in recipient countries, remarkably few studies have investigated perspectives and experiences of domestic policymakers and advisers. This study examines donor influence at different stages of the health policy process (priority setting, policy formulation, policy implementation and monitoring and evaluation) in two aid-dependent LMICs, Cambodia and Pakistan. It identifies mechanisms through which asymmetries in influence between donors and domestic policy actors emerge. We conducted 24 key informant interviews-14 in Pakistan and 10 in Cambodia-with high-level decision-makers who inform or authorize health priority setting, allocate resources and/or are responsible for policy implementation, identifying three routes of influence: financial resources, technical expertise and indirect financial and political incentives. We used both inductive and deductive approaches to analyse the data. Our findings indicate that different routes of influence emerged depending on the stage of the policy process. Control of financial resources was the most commonly identified route by which donors influenced priority setting and policy implementation. Greater (perceived) technical expertise played an important role in donor influence at the policy formulation stage. Donors' power in influencing decisions, particularly during the final (monitoring and evaluation) stage of the policy process, was mediated by their ability to control indirect financial and political incentives as well as direct control of financial resources. This study thus helps unpack the nuances of donor influence over health policymaking in these settings, and can potentially indicate areas that require attention to increase the ownership of domestic actors of their countries' health policy processes.

Key Messages

Better understanding of low- and middle-income country domestic policy actors’ experiences of donor influence during priority setting, policy formulation, policy implementation and monitoring and evaluation of policies, and what mechanisms drive this, can be critical for strengthening national ownership of health policies.

This study involving high-level policy actors in Pakistan and Cambodia indicated that control of financial resources was the most commonly identified route by which donors influenced priority setting and policy implementation.

However, unequal power relations may be perpetuated in subtle ways beyond control of financial resources, including exclusionary practices in knowledge production, dissemination, and utilization for policy and planning. Thus, a truly ‘new aid approach’ should reconsider not only financing and lending modalities, but also important issues in the daily practice of donor–recipient relations, including the extent to which local expertise is supported, valued and involved at all stages in the policy process.

Introduction

The influence of governments, multilateral agencies and private agencies that provide funds or conduct activities with the stated aim of improving health in low- and middle-income countries (LMICs) (collectively termed ‘donors’) remains prominent in the health policy process of recipient countries (Ollila 2005; Fraser and Whitfield 2009; Ravishankar ). It is well documented that donor influence, or their ability to direct the decisions or priorities of national health policymakers, occurs when there is substantial reliance of recipient countries on external funding. For example, this can occur through the use of conditionality in policy-based lending or through the funding of vertical programmes, informed by particular policy approaches (Okuonzi and Macrae 1995; Rutkowski 2007; Groves and Hinton 2013). However, in some countries, donor influence on health priority setting has been prominent even in the absence of substantial funding flows (Sridhar and Gomez 2011). Even when donors have reduced conditionality on funding, or provided direct budgetary assistance, questions have been raised about the use of alternative mechanisms by donors to continue to influence national policy processes (Mosley ; Koeberle 2003; Swedlund 2013). Indeed, Harrison (2001) has used the term ‘post-conditionality’ to reflect new modalities by which donors continue to influence recipient countries through more routine and centralized practices, such as national plans, surveys or budgeting and monitoring exercises; although these mechanisms have not yet been fully investigated, Harrison’s findings resonate with what Molenaers and Renard (2008) refer to as a ‘new aid approach,’ in which donors have started to support changes to planning and decision-making structures and processes (in addition to previous modalities of vertical project funding or horizontal budget support).

Numerous papers have highlighted the problems that can arise from donor dominance in the health policy processes of LMICs. These include overshadowing of recipient countries’ existing programs and priorities, overlooking strengths and absorptive capacities of national health systems, and their ability to sustain gains once donor funding ends (Travis ; Ollila 2005; Khan and Coker 2014). There are also more fundamental governance challenges that external influence can raise in terms of accountability to local populations and country ownership over policy—key principles that many global health actors at least purport to endorse (Okuonzi and Macrae 1995). In recent decades, the global community has made concerted efforts to rethink the way that development assistance for health is utilized in response to a number of challenges around donor fragmentation, effectiveness and influence (cf Paris Declaration 2005; Accra Agenda for Action 2008); however, studies indicate that progress is uneven and slow (Woods ), with some arguing that the increased emphasis on global programs and priority setting initiatives—such as the Global Polio Eradication Initiative and the Global School Health Initiative—is undermining national health policy process in LMICs (Yamey 2002).

Although concerns have been raised about the influence of donors on health systems and national sovereignty in LMICs (Yamey 2002; Shiffman, 2008; Biesma ; World Health Organization Maximizing Positive Synergies Collaborative Group 2009; Hafner and Shiffman 2013; Khan and Coker 2014), relatively few studies have actually investigated in depth the perspectives and experiences of domestic policymakers and advisers with regard to donor influence over health policy (Okuonzi and Macrae 1995; Hanefeld 2010; Spicer ; Chima and Homedes 2015; Parkhurst ). Understanding LMIC policy actors’ experiences of donor influence, and what mechanisms underpin the relationships of LMIC institutions with donors, however, is crucial to strengthening national ownership of health policies and promoting good governance more broadly. Considering the dearth of information on this topic from Asian contexts, our study examines donor influence and dynamics between donors and domestic policy actors during different stages of the health policy process in two Asian LMICs, Cambodia and Pakistan.

Study setting

We analysed the roles of donors in two Asian countries—Cambodia and Pakistan—to capture diversity in health systems, as well as commonalities in terms of aid dependence, while drawing on the researchers’ strong working relationships and links with stakeholders in these countries. As eliciting the ‘true’ views of policy actors can be challenging, and can impact on the quality of policy research findings (Walt ), our country selection ensured that data collection was conducted by experienced policy researchers who had lived in the study country for significant periods of time, with strong local links and contextual knowledge.

Although Pakistan and Cambodia represent diverse contexts in terms of population size, history and health system structure, commonalities include dependence on external aid and low government investment in health, which has meant that external funding has been a major contributor to resources for health in both countries (Table 1); this is known to affect the level of autonomy a country has in setting and implementing its health policies (Goldsmith 2001). Government expenditure on health is similar (just under $15 per capita) in both countries, although as a percentage of growth domestic product (GDP) Cambodia spends 6% on healthcare, whereas Pakistan spends half of that, lower than a commonly cited 5% WHO benchmark (Savedoff 2007). Pakistan, which is a more populous country, also roughly receives four times the amount of development aid than Cambodia does. However, when analysed per capita, Pakistan received less than half as much as Cambodia, and has also experienced more fluctuations in annual aid levels than Cambodia.

Table 1.

Key indicators from Cambodia and Pakistan (World Health Organization 2014; The World Bank 2016)

Indicators	Cambodia					Pakistan
	2010	2011	2012	2013	2014	2010	2011	2012	2013	2014
Population (millions)	14.4	14.6	14.8	15.1	15.3	170.0	173.7	177.4	181.2	188.9
Poverty headcount ratio at national poverty lines (% of population)	22.1	20.5	17.7	NA	NA	36.8	36.3	NA	29.5	NA
Net official development assistance (current US$ million)	733	795	807	805	799	3020	3498	2016	2191	3612
Net official development assistance per capita (current US$)	51	55	55	53	52	18	20	11	12	19
Total health expenditure per capita (current US$)	47	50	59	60	61	31	37	34	34	36
Government health expenditure per capita (current US$)	10	11	12	12	13	11	12	14	14	14
Health expenditure, total (% of GDP)	6	6	6	6	6	3	3	3	3	3

NA, not available

Both Cambodia and Pakistan have faced challenges in recent decades which have weakened or slowed development, and there are signs that domestic policymakers are trying to improve the situation. In the early 1990s, Cambodia emerged from two decades of civil conflict, embarking in a process of democratic transition and state reconstruction. Cambodia’s political transition ended a long period of isolation, opening the country to greater engagement with the international community and large flows of foreign aid assistance. In the process, international actors and institutions played a central role in setting the policy agenda, defining priorities and approaches through the politics of funding. In recent years, however, further changes have occurred. While the government is still dependent on foreign aid in many sectors (Ear 2013), economic growth and the strengthening of institutional structures, such as the Department of Planning and Health Information, have increased local ownership and management of decision-making processes.

Similarly, Pakistan’s health and human development has been affected by conflicts along the country’s northern borders, problems with internal security, governance challenges, high population growth and natural disasters in the form of major floods and earthquakes between 2000 and 2015 (Nishtar ). Steps are being taken to improve governance and institutional capacity. For example, the devolution of health as part of the 18th Constitutional Amendment in 2010 was implemented to increase accountability and policymaker capacity at the provincial level (Government of Pakistan 2010). Building on the principles outlined in the National Health Vision—bringing together provincial and national level policy makers as well as academics and private sector representatives—a major reform of the health sector was launched in 2016, after a 15-year gap without any significant policy change (Government of Pakistan 2016).

Methods

Data collection

We defined policy actors as high-level decision-makers who inform or authorize health priority setting, allocate resources and/or are responsible for policy implementation. We focused on domestic policy actors, including those working at the national and provincial levels. Consultants, in-country donor representatives or non-governmental organization actors that liaise closely with policymakers were also included to collect multiple perspectives on the domestic policymaking environment.

Initially, five policy actors in each country were identified purposively based on professional connections of the research team through ongoing or previous public health research in the countries. During the initial interviews, the other participants were identified through snowball searching in which each policy actor introduced researchers to one or two potential informants in their network. In total, we conducted 24 in-depth, semi-structured interviews (14 in Pakistan and 10 in Cambodia). To ensure broad representation, we interviewed high-level policy actors involved in planning or implementation in a range of health areas and contexts, including maternal and child health, infectious diseases, and in primary and tertiary healthcare delivery, to explore common and contrasting experiences. These policy actors worked in government agencies and non-governmental organizations agencies in senior positions as national policy advisors, civil servants, program directors/managers and providers of technical expertise. Our topic guide aimed to: (1) elicit policy actors’ perceptions about the influence of donors at four key stages of the policy process (priority setting, policy formulation, policy implementation and monitoring and evaluation) (Parsons, 1995) and (2) identify mechanisms through which asymmetries in influence between donors and domestic policy actors emerge, and how they can be addressed.

Interviews in each country were led by one of the researchers in English or the local language depending on the interviewee’s preference. If consent was obtained, interviews were audio-recorded. Interviews lasted between 45 and 60 min and hand-written notes were taken during and after each interview. Recorded interviews were transcribed verbatim in full. Interviews that were conducted in languages other than English were translated into English and the interviewer reviewed the translated transcripts to verify accuracy of translation. No contacts refused to participate, although three participants preferred not to be recorded.

Analysis

We conducted a thematic analysis starting with deductive coding and grouping of transcripts in relation to three means of influence: financial resources, technical expertise and indirect financial and political incentives. We decided to focus on these routes of influence based on former studies that documented their relevance to policymaking (Dalglish ) and on theories of power including Foucault’s (Bunton et al. 2002), in which power and knowledge are interconnected, and Lukes’ conceptualization (Lukes 2005), who identified three ‘dimensions’ of power (or means by which power is exercised in policymaking): power of decision-making, power of setting the agenda and ideological influence. Specifically, we focused on financial resources, technical expertise and intersectoral leverage. In our study, influence mediated through the direct control of financial resources relates to decisions about resource allocation, including the time frame of resource availability; therefore, it can be considered as one of the most ‘visible’ means of exercising influence across all dimensions (Lukes 2005; Dalglish ). Technical expertise indicates influence gained through the ability to produce, interpret and disseminate knowledge and information to policy actors (akin to what Shiffman has termed ‘epistemic power’ in global health agenda setting) (Shiffman 2014). Finally, intersectoral leverage refer to means of influence operating outside of the health sector, such as effects on the international image or standing of countries which can impact on areas including trade and tourism (Lin and Gibson 2003; Harris and Siplon 2007). Deductive coding was followed by an inductive coding phase to identify emerging sub-themes within the three routes of influence, applying techniques from the constant comparative method (Boeije 2002), including line by line analysis of initial interviews (performed independently by two researchers), the use of subsequent interviews to test preliminary assumptions, and the comparison of codes across countries and health areas (Strauss 1987; Parsons 1995).

Results

Our interviews with health policy actors in both countries indicated that, as expected, donors were perceived to exert strong influence across the four stages of the policy process. The three routes of influence studied, and specific mechanisms that were important in establishing donor influence, varied at each stage, as described in the following sections (and summarized in Table 2).

Table 2.

Means of power exercised across the different stages of the policy process in Pakistan and Cambodia, as perceived by domestic policy actors

	Routes of influence
Stages	Intersectoral leverage [influence from impact outside of health sector such as international tourism or trade restrictions]	Financial Resources [control of resource allocation, including time frame of resource availability]	Technical expertise [advantage through ability to produce, interpret and disseminate knowledge]
Agenda/Priority Setting	Impact on international reputation and tourism from failure to address donor priorities (Cambodia) Potential trade or travel restrictions (Pakistan)	Donors select which health areas are provided funding for, thereby setting agenda Donors prioritize which research or surveys they fund to provide the evidence base to inform agenda setting (Cambodia and Pakistan)
Policy Formulation			Donors have greater proficiency in using data from surveys/studies to develop policies Donors can commission surveys/studies to fill knowledge gaps Donors have better coordination to collaborate on policy formulation (Cambodia and Pakistan)
Policy Implementation		Financial resources from donors shape the areas of work of non-governmental organizations (Cambodia and Pakistan) Control timing of availability of resources for programme implementation; sudden stops and starts (Pakistan)
Monitoring & Evaluation (M&E)	Donors set (M&E) targets which must be met to maintain international standing (Cambodia and Pakistan)	Donors set (M&E) targets which must be met to receive funding (Pakistan) Donors influence which health areas receive funding to strengthen M&E systems (Cambodia and Pakistan)

Agenda or priority setting

Overall, policy actors in Pakistan and Cambodia felt that the level and availability of external funding often dictated which issues were placed high on national health agendas as well as the types of interventions that were selected to address the health issues. Financial resources were identified as the main mechanism through which donors either directly shaped national health priorities or indirectly exerted influence by determining which research or surveys they fund to provide the evidence base to inform agenda setting and advocacy. In relation to donors directly shaping which health areas are prioritized for action through funding availability (or lack thereof), one international NGO representative and policy advisor (C1) used mental health in Cambodia as an example to illustrate how much dependence on donor funding impacts the health policy agenda. She explained that even though domestic stakeholders were aware of the urgent need to address mental health issues—owing to the genocide perpetuated in the country—this was not a priority health area until 2016 (when the first strategic plan was initiated) because of a lack of donor funding, on which the government is reliant.

In addition to directly influencing policy setting through funding availability, some interviews (P3, C1, C9) indicated a form of indirect influence donors could have by shaping the areas for which health information or evidence is available to policymakers during the agenda setting process. One interviewee in Cambodia explained:

Many times, however, research is driven by funding, not demand. And this type of research is less relevant to the country. (C9)

Related to this, an international NGO country director and another international NGO manager who had previously worked in the public sector (P3, C1) expressed frustration at the lack of funding made available by their governments for research, which they identified as a reason for the limited power of policy actors in influencing what evidence is available, and through this, overall health priority setting.

A majority of policy actors who experienced such a power dynamic felt they had a limited voice in their country’s health and health-related research priority setting, despite being in high level positions (P2, P6, P10, P11, C3, C5). A smaller group of policy actors (P3, P5, C1) conveyed stronger negative sentiments about donor influence, for instance stating that external aid could be harmful when it is not aligned with national policymakers’ priorities. Policy actors (P3, P12, C1) were clear in acknowledging that the imbalance in power was related to the relatively small amount of funding from national sources.

[Donors] do play a bit of a negative role, because they’re pushing for their own issues they see as their priority. There’s funding for that… a big proportion of the health budget is still funded by external partners and they mostly decide what they want to fund. (C1)

Indeed, two specific negative consequences of the (perceived) limited influence of policy actors in their country’s health and health-related research agenda were expressed in the two countries. Firstly, as described above, specific health issues or approaches become prominent in countries—even if they do not fit with the overall national strategy—because these were better resourced by donors. The second negative consequence of the perceived lack of influence of domestic policy actors on priority setting or research was that important areas considered can be neglected, especially when donors focus on narrow, pre-determined policy goals. This view was common in both countries (P5, P7, P9, P10, C1, C2). In particular, for both research and health programs, there was a common feeling among interviewees that aspects related to health system strengthening, such as prevention and primary care, received less attention because donor funding targeted disease specific programs.

Apart from financial resources, which was found to be the most salient of the three routes of influence analysed in relation to agenda or priority setting—bilateral and multilateral donor countries’ power to influence the recipient country’s standing in sectors beyond health (intersectoral leverage) also played a role in some instances. Two policy actors in Cambodia (C4, C9) explained that the way that donors and UN agencies portray Cambodia globally matters because it has a direct impact on international reputation and tourism; therefore, high-level national policymakers feel the need to particularly pay attention to priorities of donors and international agencies such as the World Health Organization. Similarly, two interviewees from Pakistan (P2, P3) who had held managerial roles in both public and private organizations believed that the threat of travel and trade restrictions being introduced by international organizations if polio was not controlled was important in placing polio high on the national agenda.

Policy formulation

In contrast to agenda and priority setting, in which control of financial resources played a major role in mediating donor influence, we found that technical expertise of donors appeared to be a key route of influence at the policy formulation stage. Donors were perceived to have greater proficiency in using data from surveys and research studies to develop strong policies and strategic plans. Policy actors also felt that donors were better at filling gaps in evidence whether by commissioning specific research or relying on their data and knowledge base to extrapolate findings to inform policies. Several interviewees in both Cambodia and Pakistan (P2, P4, P5, C1, C5, C6) felt that local capacity for analysing data to inform policies was lacking, and that policy formulation was therefore either slow or not based on sufficient data analysis, as illustrated by one policy actor in Pakistan working at the provincial level:

We don’t look at statistical information. We don’t run regressions. We don’t look at correlations or causations. We just decide. (P2)

Two advisers to national policymakers (P12, C2) specifically identified the language through which technical information and policy-relevant research was presented as working to disadvantage policy actors or reinforce the influence of donors. Technical reports used to inform policy formulation were described as lengthy, written in English and utilizing complex terms that served as barriers to the accessibility of the information serving to inform policy choices. As one interviewee explained: Start first of all with the English language, it has already created a barrier for those at the grassroots level to really connect with the technical expertise.(C2)

When talking about the imbalance between themselves and donors in terms of capacity to analyse data and formulate policies, some interviewees, particularly in Pakistan, questioned donors’ motives around truly wanting to build local capacity (P3, P5), while the manager of a disease control program in Cambodia (C5) provided examples of donors that had demonstrated higher and lower levels of commitment. Further, one informant in Pakistan reported:

I think they (donors) also want to “burn” their money. They just want to spend the money. Their aim is not to make Pakistan independent. They also do not take exactly evidence based decisions. (P3)

We also found that the level of coordination and collaboration among donors was perceived by some interviewees to give them collective power in forming health policies. Examples of specific donor coordination platforms described include the Health Partners Meeting in Cambodia and the Technical Resource Facility in Pakistan (Mott MacDonald 2017; TRF Pakistan 2017). These platforms are set up specifically with the aim of enhancing coordination between donors and, in the opinion of interviewees that described them (Ca8, Ca6, P12), allow donors to present a coherent and powerful position to influence policy development. For example, a policy advisor in Cambodia (C6) shared details about the Health Partners Meeting, which involves the participation of bilateral and multilateral donors, and international agencies, such as the World Health Organization. He believed that external donor and technical support agencies holding a closed meeting one week before the monthly Technical Working Group for Health meeting (in which health policies are discussed with the Cambodian government) helps them to prepare a unified and well composed plan to present to high level policymakers. In contrast, it appeared from account of several interviewees (P9, P12, C5, C6, C8) that domestic policy actors had no similar mechanisms to organize themselves in the same way to be able to effectively influence policy design.

Policy implementation

The main mechanism identified through which donors could influence policies being implemented on the ground was by using financial resources to shape the areas of work of non-governmental organizations (NGOs) in the country. An advisor to policymakers in Cambodia (C2) explained that he had seen sudden growths in NGOs focusing on specific topics, often unlinked to national priorities or even the NGOs own mandate, as they were dependent on winning grants to continue their operations. In Pakistan, a similar concern over donor influence emerged from multiple interviewees (P6, P10, P11) who noted that funding for tuberculosis control activities is now being controlled by a single NGO, which is the primary recipient of a US$39 million grant from a multilateral donor (Results International 2016). Since financial flows to this NGO dwarf independent budgets of the national and provincial tuberculosis control programme, a major shift in decision-making authority of the government policy actors was felt by those interviewees involved in tuberculosis control in Pakistan.

Interviews further revealed that there was also a strong donor influence on the timing of implementation of various health initiatives. There were instances reported in which donors’ could derail progress towards the national strategy because ongoing external funding was often linked with the achievement of time-sensitive goals or political commitments (P2).

Monitoring and evaluation

Finally, in the evaluation stage of the policy process, several policy actors interviewed (P2, P3, P5, P7, P9, P10, C1, C2) felt that donors dictated targets that needed to be achieved for certain national health programs. Targets included numbers of patients to be diagnosed or started on treatment, proportions of patients receiving a selected intervention and numbers of diagnostic devices introduced into health facilities. Donors often exerted influence through conditionality of financial resources on achieving targets, and through intersectoral leverage; the latter was related to donor influence on the country’s international standing and reputation when targets they push countries to adopt are aligned with global initiatives. For example, an interviewee in Cambodia (C1) explained that targets based on the global 90–90–90 HIV strategy have been powerful in influencing programme implementation in the country because of international support for this strategy (UNAIDS 2017). Similarly, in Pakistan an interviewee (P12) felt that the National AIDS program mobilized quickly because HIV was high on the global agenda even though domestic policy actors did not see it as an urgent priority in the Pakistan context owing to very low HIV prevalence.

While a potential positive effect of donors’ power to push for achievement of specific targets was that this could improve the speed and efficiency of national and provincial health programs, we found that targets linked to global health programs hold considerable weight and there is political pressure to adopt these ‘uniform’ targets even when domestic actors know it is not appropriate for the context. For example, three public sector programme managers involved in tuberculosis control in Pakistan (P6, P10, P11) independently explained that the global strategy calling for a rapid scale-up multidrug resistant (MDR) tuberculosis treatment is not what they would recommend based on their knowledge of health systems constraints in monitoring adherence to treatment and managing serious side-effects. One interviewee argued that donors should evaluate success of a policy based on strengthening of broader capabilities rather than on narrow targets:

If you want to achieve 2000 MDR cases, then we should train our people on them, we should have our expert machines in places, in proper places, we should have the right linkages, right communications, right capacity, those processes should be strengthened instead of looking at the target –target chasing only. It should be the process that should be strengthened all the time. And, my discussion with the [donor name] that I keep on saying is that the target should not be the patients, the target should be systems instead. (P10)

Just as donors were found to have influence in putting certain disease specific, vertical programs high on the national health agenda by making resources available for them, we also found that they were able to influence the strength of monitoring and evaluation of selected health areas in both countries. We identified two main mechanisms by which this occurred. Firstly, donors influenced which health areas information systems were enhanced for by channelling financial resources towards infrastructure development. This included investments in standardized record keeping, moving from paper-based to electronic information storage and capacity building of healthcare providers to use the information systems effectively. Secondly, donors could be instrumental in ensuring the targets for monitoring were clearly defined, and made resources available for regular monitoring by independent organizations.

Finally, this study indicated that the perceived lack of influence of policy actors when negotiating health targets may have been exacerbated by limitations in their power to decide which health areas are covered by strong health monitoring and information systems. For example, one Cambodian policy adviser explained that without credible independent data, national policy actors were unable to resist unrealistic targets set forth by donors or advocate for alternative health priorities, even if they disagreed with the evidence presented (C2).

Discussion

Donors are known to exert influence over policy and practice in low resource settings, but to date only limited work has explored the implications of power imbalances at different stages of the policy process in aid-recipient nations. We recognize that the four stages we analyse separately—priority setting, policy formulation, policy implementation and monitoring and evaluation—do in fact overlap in reality and are not discrete or linear (Walt ). Nonetheless, by considering them one-by-one we were able to draw useful insights and organize the research material in a logical manner. Another key contribution of this study is its direct focus on perceptions and experiences of domestic health policy actors, many of whom appeared to be struggling to gain or maintain power in one way or another, and the variety of mechanisms through which donors may shape policy making and interventions. Given the qualitative nature of the study, and the focus on two particular countries, our findings may not be generalizable or relevant beyond them.

With this limitation in mind, however, we must note that a striking point emerging from the comparative analysis of the interviews is the essential agreement of participants in Cambodia and Pakistan on fundamental issues concerning their relations with international donors. Despite significant differences in health systems, history and engagement with the international community, policy actors in both countries raised similar concerns over the ways that donors may influence the policy process, leading to policies which they felt were often misaligned with local needs and capacities. In both countries, we found that control of financial resources was the most commonly identified lever by which donors influenced policy, particularly at the priority setting and implementation stages. Many policy actors in Cambodia and Pakistan revealed a mismatch between what health activities they believe are important for their countries and what happens in practice. While others have documented that control of financial resources directly influences health policy (Buse ), in this study settings we additionally found that control of technical expertise through the management and strategic presentation of knowledge can play an important role, with those laying claim to expertise exercising influence and gaining authority based on a privileged relationship to knowledge and stronger capacities to use evidence for policy and planning.

This study also found that the influence conferred by greater technical capacity was not only related to skills and expertise of donors, but also to better organizational mechanisms for coordination and collaboration among donors and international technical agencies and platforms they have set up to maximize interaction with policy elites. In contrast, policy actors in Cambodia and Pakistan acknowledged that domestic structures to support priority-setting were weak and collaboration with local research bodies and institutions was lacking. Donors’ control of financial resources can also allow them to indirectly influence health policy agendas by making research funding available to generate evidence in donor priority areas and not in others. In these ways, the creation of evidence could be seen as part of a political process, reflecting what has been described an ‘issue bias’ (Parkhurst 2017). The indirect influence of power can also be seen to reflect broader Foucaludian ideas of a ‘power-knowledge nexus’ existing in society—by which power works to construct knowledge and knowledge constructions further work to establish power relations (Nola and Irzik 2005). In such conceptualizations, power is not just identified at discrete decision points, but rather seen to also be more diffuse: built into systems of interactions (and discourses), which end up shaping what is considered relevant knowledge in the first place (often to the advancement of particular interests). For instance, the discourses and long-term interactions which work to establish particular ideas of ‘technical expertise’ as a legitimizing source of authority can thus be privileging donor positions. Although it is worth noting that in a pure Foucaldian sense, power is often seen as ‘subject-less,’ rather than wielded by any particular actor, as well as being a constructive force in society through its production of relationships and ideas (cf Gaventa 2003). In this study, when it came to the setting of policy goals and targets for monitoring and evaluation, donors’ influence was strengthened by linking goals they are pushing for to global initiatives and norms, as well as providing financial resources to support selected assessment and reporting systems.

Our findings also illustrate that power relationships between donors and aid recipients are more complex and multifaceted than simply donors having direct influence over decisions by controlling resource allocation. Indeed, all three of Lukes (2005) ‘dimensions’ of power are in operation within aid relationships for health policymaking. Donor influence over the agenda, for instance, reflect long-standing concerns over donor influence which reflect Luke’s first face of power—power as ‘decision making’. In addition, we found many instances of how power was exercised outside discrete decision-making points. Political influence exercised through concerns about impacts outside the health sector impacts (such as on international reputation), for instance, illustrates how power of donors can be structurally established in ways that end up influencing which issues get on decision agendas in the first place—thus reflecting the second ‘dimension’ of power, at times called ‘non-decision making’—or what Lukes explains as the power ‘to decide what is decided’ (Lukes 2005, p. 111). Lukes’ third dimension of power, however, refers to the construction of ideas itself—and of dominant hegemonies or ideologies (which are said to potentially shape the ‘very wants’ of a particular group) (Lukes 2005, p. 27). In our case, this is captured in the definition and creation of policy relevant pieces of information by those controlling research and evaluation processes, or by establishing international ‘consensus’ about priorities and needs as explained by Shiffman (2014).

In addition, our findings about specific mechanisms that can result in donors having greater power—beyond direct control of financial resources for health—may have implications for addressing the power imbalance. Although increased funding for health from national and provincial governments in lower-income countries would be one way to alter power dynamics, this is not straightforward to achieve; the competing demands for budget allocation and development assistance for health from donors has been shown to reduce government spending on health in LMICs (Lu ). However, policy actors in lower income countries could address the perceived power imbalance in technical expertise even with limited resources, for example, through better coordination of domestic stakeholders and organization of platforms for agenda setting and policy formulation.

Concerns about power imbalances must also be kept in mind considering the global health and development community’s continued use of language of ‘evidence based policymaking’ to justify particular decision-making strategies, systems and norms in the health sector. Appeals to technical evidence—typically of intervention effect or cost effectiveness measured over a small set of outcomes—is common; yet our findings illustrate just how many other concerns may be at stake in health decision making, and further point to important governance concerns around the process by which evidence is brought to bear and used to prioritize, legitimize, or justify particular policy actions. Concerns over national autonomy, local accountability, local capacity building, and competing social values rarely are directly addressed in health policy development processes, yet all were touched on as important in these settings. We further saw some examples of activities by international actors to not only use evidence to inform specific decisions, but to build structures and institutions within countries as well that may shape how evidence is created and utilized to inform decision making. It may be that ‘new aid approaches’ (or ‘post-conditionality’ approaches)—involving supporting national data systems, establishment of technical expert bodies or funding of research agendas—can have important governance implications (Harrison 2001).

Our findings on existing power imbalances must also be considered in light of the countries’ historical background and ongoing changes in decision-making dynamics. Though Cambodia and Pakistan faced many challenges in the past four decades, which have slowed development and weakened state and health system infrastructure and institutions, recent institutional reform and economic growth in both countries have bolstered local capacities for decision-making and programme implementation. Therefore, a shift in power balance may occur going forward. In 2007, the Cambodian government introduced a Midwifery Incentive Scheme, which aimed to reduce maternal mortality rates by paying midwives with cash incentives based on the number of public health facility-based deliveries they attended. This policy, which is entirely implemented and financed by the national government has been successful (Ir ) and illustrates a shift to local leadership in policy formulation and management. The Cambodian government has also taken greater financial responsibility for health policies which were originally introduced and supported only by international actors—such as the Health Equity Funds (Annear )— another key development which is likely to improve sustainability of interventions and their alignment with local structures and capacities. Although domestic policy actors’ role in funding, designing and implementing of health policies in Cambodia has increased, capacity for monitoring and evaluation of health programmes is still developing (University Health Sciences of Cambodia 2015).

As with Cambodia, in Pakistan there are indications of an increasing role of domestic policy actors in agenda setting, policy formulation and policy implementation. For example, Pakistan has recently been recognized internationally for developing locally appropriate public health innovations—such as the community-based Lady Health Worker programme—and policymakers from Pakistan have impacted on health policy on a national and global arena (Horton 2013). Unlike Cambodia, increased government spending on health is a critical area which has not yet seen substantial improvements (Khan and Van den Heuvel 2007; Nishtar ), although hospital budgets for medicines and renovations have increased following devolution (Zaidi ). The volatility of international aid commitments has also been a challenge for Pakistan; for some policy actors in Pakistan this appears to have resulted in a mistrust of donors and NGOs supported by international donors (Bano 2012).

Lastly, a number of limitations in our study must be noted. First, as acknowledged by others, the perspectives of policy actors can often be difficult to investigate, particularly when researchers require insider access to domestic policy elites, and skills in building rapport with interviewees to discuss potentially sensitive topics (Walt ). Second, we recognize that—despite identifying a number of recurring themes—we may not have achieved saturation. Since this study focused on two countries and did not capture donor perspectives, further research should be conducted to refine our conclusions and enhance their theoretical value. Third, we found that our analysis touched upon deeper and broader structural elements than initially expected, but we could not explore these issues in-depth. Indeed, the ways that donors exert power by shaping policy-relevant evidence, ideas, and discourse, or the structural changes donors may be making in the name of informing policymaking, could each be subjects of their own further investigations. For example, we are unaware of substantial research in lower-income settings exploring in depth the effects of donor efforts to shape structures and mechanisms for the use of evidence in the policy process. As such, this may serve as a useful area for future work to understand the exercise of power by donors within health policymaking in aid-recipient nations, not just over individual decisions or programmes, but at a systemic level as well.

Conclusion

National structures for decision-making have improved in Cambodia and Pakistan. Nonetheless, many participants in this study expressed some frustration with international donors, and their ability to influence the policy process through financial means, unequal distribution of expertise and imbalances in technical and organizational resources for strategic planning. There was a recognition among domestic policy actors that low investment in health by their own governments was partly responsible for the power imbalance. Awareness of these perceptions is important in the current debate on international development. Despite changes in the aid architecture, grievances about donor approaches remain deep seated in some LMICs, reflecting wider imbalances in the context of global political economy and international relations.

As we have seen and other studies documented, unequal power relations may still be perpetuated in subtle ways, including exclusionary practices in knowledge production, dissemination, and utilization for policy and planning. Thus, a truly ‘new aid approach’ should reconsider not only macroeconomic aspects, such as financing and lending modalities, but also important issues in the daily practice of donor–recipient relations, including the extent to which local expertise is supported, valued and involved at all stages in the policy process.