Who is responsible? Local government and accountability for service delivery in Kenya’s devolved health sector

One of the arguments in favour of local governments is their ability to deliver public services better because of their proximity to demand. This is typically achieved through decentralisation – the transfer of some level of formal responsibility, authority and/or resources to smaller, lower tiers of government. This paper examines service delivery in the health sector in Kenya within the context of the country’s newly introduced devolved system of government, which created 47 county governments under the 2010 Constitution. It examines three key responsibilities: i) resource mobilisation, distribution and administration, ii) decision-making, and iii) political accountability, and their bearing on outcomes for delivery of health services. The paper argues that devolution of the health function in Kenya has been only partial, leading to challenges of coordination between the national and county governments and ambiguity over responsibility for service delivery; both factors which risk undermining the opportunities that devolution creates or promises. Introduction One of the arguments in favour of local-level government is that it can better deliver public ‘goods’ because of its proximity to demand (Duncan 2004; Rao et al. 2014). Local governments can achieve quicker service delivery than centralised systems and gain more trust from citizens because they operate closer to the public and are more visible, accessible and accountable (Devas et al. 2008; Rao et al. 2014). Local-level governance is based on the concept of decentralisation: the transfer of some level of formal responsibility, authority and/or resources to smaller, lower units of government – usually counties, Ochieng Local gov accountability in Kenya’s health sector CJLG December 2017 159 districts, regions or states (Duncan 2004; Saltman et al. 2007). Such decentralisation may take different forms, including deconcentration, devolution, delegation and privatisation (Duncan 2004; Vrangbaek 2007; Devas et al. 2008). The existing literature on decentralisation and service delivery indicates that decentralisation can offer considerable advantages, although there is limited empirical evidence of positive outcomes (Smoke 2015). Advantages identified include greater effectiveness in mobilisation and efficiency in allocating resources (Devas 2005; Rao et al. 2014), deeper citizen participation (Blair 2000; Wong and Guggenheim 2006; Devas et al. 2008), and improved accountability mechanisms (Devas 2005; Lessmann and Markwardt 2009; Yilmaz et al. 2010). However, the literature also suggests that decentralisation has not been an easy route to effective service delivery, as it involves complex processes that influence quality and coverage of services (Wunsch 2001; Olowu 2003; Duncan 2004). For example, there are occasions when roll-outs of decentralisation programmes have faced rejection or resistance by sectoral/line ministries of national/central governments (Olowu 2003; Devas et al. 2008; Boex 2009). In other circumstances, the time-lag between in-principle decentralisation and its implementation has contributed to poor performance and sparked resentment from the public. Elsewhere, lack of experience amongst local elected leaders and the effects of electoral politics (for example a culture of patronage, or the capture of elected local governments by different interest groups) have also hampered the effectiveness of decentralised service delivery (Acosta et al. 2013). Notable successes, however, indicate that decentralisation can be very beneficial when actors address both the supply and demand sides of service delivery (UNCDF 2010): for example, when capacity development takes place within local and national governments in parallel, allowing local governments to maximise their autonomy while receiving support to balance local responsibilities with local discretion; or when local service delivery competencies are matched with adequate funding (Wunsch 2001; Duncan 2004; UNCDF 2005). This paper examines the framework for service delivery in the health sector in Kenya, within the context of the country’s newly introduced devolved system of government – county governments. The paper is based on a literature review of academic papers, reports by civil society organisations working on devolution, media accounts and analyses of constitutional devolution in Kenya. It examines three mandates devolved to county governments and their bearing on outcomes for delivery of services in the health sector. These mandates are i) resource mobilisation and distribution ii) decision-making and iii) political accountability. Ochieng Local gov accountability in Kenya’s health sector CJLG December 2017 160 Analytical framework Achieving smooth working relations between decentralised and national/central governments, with effective accountability and coordination mechanisms, is crucial for service delivery. Nonetheless, experience across the globe often finds such relations characterised by mistrust and wrangling, with the main source of tension being the tendency of central or national governments to exert excessive control, yet still expect high standards of accountability from decentralised units of government (Wunsch 2001; Crook 2003; Olowu 2003; Panday 2006; Rao et al. 2014). Some scholars maintain, understandably, that central governments are justified in their desire to remain involved in the affairs of decentralised units, in order to ensure accountability and prudent use of public resources (Olowu 2003; Crook 2003; Panday 2006; Boex 2009). However, others argue that national/central government involvement is only useful if it does not undermine local authority (Wunsch 2001; Delay 2008). From this tension, two key questions arise: 1) Do decentralised units of government need more autonomy in order to deliver services better, innovate boldly and be more accountable? and 2) What accountability mechanisms, both downwards to citizens and upwards to national/central governments, can allow them that autonomy while still engaging effectively with other tiers of government? The following analysis discusses the relationship between decentralised units of government and the national government (and independent national institutions) within the context of healthcare service delivery in Kenya. It explores issues of coordination and perceived responsibility for service delivery between national and county governments and considers their implications for achieving good outcomes at county government level. To do this, it examines relations between the two levels of government in three areas of responsibility: i) resource mobilisation, distribution and administration ii) decisionmaking and iii) overall political accountability. As illustrated in Figure 1, Kenya’s devolved system of government is a synthesis of the previous centralised system of government, collapsing eight provincial administrations to create 47 county governments, which now oversee areas previously covered by 175 local authorities, 280 deconcentrated local administrative units and 280 district administrations (Mwenda 2010; World Bank, 2011; Murkomen 2012). The health function is one whose responsibility has been transferred to county governments, but with some level of responsibility retained by the national Ministry of Health (Murkomen 2012). 1 Simon Delay points out that there is a fine line between excessive regulation and autonomy of local governments (Delay 2008): https://www.researchgate.net/profile/Munawwar_Alam2/publication/280317403_Financing_Local_Government /links/5764410508aedbc345ecc054/Financing-Local-Government.pdf?origin=publication_detail Ochieng Local gov accountability in Kenya’s health sector CJLG December 2017 161 Figure 1: How the county governments were formed Source: World Bank (2011) Analysis and discussion Resource mobilisation, distribution and administration Decentralised units of government are argued to be able to develop plans and programmes based on local needs and resources, and implement such programmes taking into account issues of political and social and economic marginalisation (Fjeldstad et al. 2006; Boex 2009; Rao et al. 2014). Some researchers have also argued that fiscal decentralisation could drive inclusive economic growth and promote sustainable development in localities traditionally bypassed by centralised governments (Shah 2007; Devas et al. 2008). However, in order for decentralised units of government to effectively deliver services and accrue the benefits of decentralisation, adequate financing is necessary (Shah 2007; Devas et al. 2008; Rao et al. 2014). This is actualised through fiscal decentralisation – the process of assigning not only financial resources to sub-national or local governments, but also the authority to make decisions on the management and application of, and accountability for, such resources (Duncan 2004; Devas et al. 2008; Boex 2009; Rao et al. 2014). The revenues that decentralised units of government spend on delivering services come from three main sources. The first is intergovernmental fiscal transfers (IFTs) – monies sent from national/central governments to local authorities to fund their programmes and activities (Bird and Smart 2001; Duncan 2004; Devas et al. 2008; Boadway and Shah 2007; Boex 2009; Alam 2014). The second is local collections – taxes, charges and other receipts which local governments are allowed to levy (Duncan 8 Provincial Administrations 47 devolved County Governments 175 Local Authorities 280 (+) deconcentrated local administrative units Overseeing liquor licensing, disaster management, and control of drugs and pornography Overseeing health, agriculture, livestock, fisheries, planning, housing, lands, transport, rural electricity, sports and culture, aquaculture, environment and conservation. 280 (+) District Overseeing solid waste management, public health, parking and street lighting, markets, abattoirs, water sewerage, dr


Introduction
One of the arguments in favour of local-level government is that it can better deliver public 'goods' because of its proximity to demand (Duncan 2004;Rao et al. 2014). Local governments can achieve quicker service delivery than centralised systems and gain more trust from citizens because they operate closer to the public and are more visible, accessible and accountable Rao et al. 2014).
Local-level governance is based on the concept of decentralisation: the transfer of some level of formal responsibility, authority and/or resources to smaller, lower units of governmentusually counties, CJLG December 2017 159 districts, regions or states (Duncan 2004;Saltman et al. 2007). Such decentralisation may take different forms, including deconcentration, devolution, delegation and privatisation (Duncan 2004;Vrangbaek 2007;Devas et al. 2008).
The existing literature on decentralisation and service delivery indicates that decentralisation can offer considerable advantages, although there is limited empirical evidence of positive outcomes (Smoke 2015). Advantages identified include greater effectiveness in mobilisation and efficiency in allocating resources (Devas 2005;Rao et al. 2014), deeper citizen participation (Blair 2000;Wong and Guggenheim 2006;Devas et al. 2008), and improved accountability mechanisms (Devas 2005;Lessmann and Markwardt 2009;Yilmaz et al. 2010).
However, the literature also suggests that decentralisation has not been an easy route to effective service delivery, as it involves complex processes that influence quality and coverage of services (Wunsch 2001;Olowu 2003;Duncan 2004). For example, there are occasions when roll-outs of decentralisation programmes have faced rejection or resistance by sectoral/line ministries of national/central governments (Olowu 2003;Devas et al. 2008;Boex 2009). In other circumstances, the time-lag between in-principle decentralisation and its implementation has contributed to poor performance and sparked resentment from the public. Elsewhere, lack of experience amongst local elected leaders and the effects of electoral politics (for example a culture of patronage, or the capture of elected local governments by different interest groups) have also hampered the effectiveness of decentralised service delivery (Acosta et al. 2013).
Notable successes, however, indicate that decentralisation can be very beneficial when actors address both the supply and demand sides of service delivery (UNCDF 2010): for example, when capacity development takes place within local and national governments in parallel, allowing local governments to maximise their autonomy while receiving support to balance local responsibilities with local discretion; or when local service delivery competencies are matched with adequate funding (Wunsch 2001;Duncan 2004;UNCDF 2005). This paper examines the framework for service delivery in the health sector in Kenya, within the context of the country's newly introduced devolved system of governmentcounty governments. The paper is based on a literature review of academic papers, reports by civil society organisations working on devolution, media accounts and analyses of constitutional devolution in Kenya. It examines three mandates devolved to county governments and their bearing on outcomes for delivery of services in the health sector. These mandates are i) resource mobilisation and distribution ii) decision-making and iii) political accountability.

Analytical framework
Achieving smooth working relations between decentralised and national/central governments, with effective accountability and coordination mechanisms, is crucial for service delivery. Nonetheless, experience across the globe often finds such relations characterised by mistrust and wrangling, with the main source of tension being the tendency of central or national governments to exert excessive control, yet still expect high standards of accountability from decentralised units of government (Wunsch 2001;Crook 2003;Olowu 2003;Panday 2006;Rao et al. 2014). Some scholars maintain, understandably, that central governments are justified in their desire to remain involved in the affairs of decentralised units, in order to ensure accountability and prudent use of public resources (Olowu 2003;Crook 2003;Panday 2006;Boex 2009). However, others argue that national/central government involvement is only useful if it does not undermine local authority 1 (Wunsch 2001;. From this tension, two key questions arise: 1) Do decentralised units of government need more autonomy in order to deliver services better, innovate boldly and be more accountable? and 2) What accountability mechanisms, both downwards to citizens and upwards to national/central governments, can allow them that autonomy while still engaging effectively with other tiers of government?
The following analysis discusses the relationship between decentralised units of government and the national government (and independent national institutions) within the context of healthcare service delivery in Kenya. It explores issues of coordination and perceived responsibility for service delivery between national and county governments and considers their implications for achieving good outcomes at county government level. To do this, it examines relations between the two levels of government in three areas of responsibility: i) resource mobilisation, distribution and administration ii) decisionmaking and iii) overall political accountability.
As illustrated in Figure 1, Kenya's devolved system of government is a synthesis of the previous centralised system of government, collapsing eight provincial administrations to create 47 county governments, which now oversee areas previously covered by 175 local authorities, 280 deconcentrated local administrative units and 280 district administrations (Mwenda 2010;World Bank, 2011;Murkomen 2012). The health function is one whose responsibility has been transferred to county governments, but with some level of responsibility retained by the national Ministry of Health

Resource mobilisation, distribution and administration
Decentralised units of government are argued to be able to develop plans and programmes based on local needs and resources, and implement such programmes taking into account issues of political and social and economic marginalisation (Fjeldstad et al. 2006;Boex 2009;Rao et al. 2014). Some researchers have also argued that fiscal decentralisation could drive inclusive economic growth and promote sustainable development in localities traditionally bypassed by centralised governments (Shah 2007;Devas et al. 2008).
However, in order for decentralised units of government to effectively deliver services and accrue the benefits of decentralisation, adequate financing is necessary (Shah 2007;Devas et al. 2008;Rao et al. 2014). This is actualised through fiscal decentralisationthe process of assigning not only financial resources to sub-national or local governments, but also the authority to make decisions on the management and application of, and accountability for, such resources (Duncan 2004;Devas et al. 2008;Boex 2009;Rao et al. 2014).
The revenues that decentralised units of government spend on delivering services come from three main sources. The first is intergovernmental fiscal transfers (IFTs)monies sent from national/central governments to local authorities to fund their programmes and activities (Bird and Smart 2001;Duncan 2004;Devas et al. 2008;Boadway and Shah 2007;Boex 2009;Alam 2014). The second is local collectionstaxes, charges and other receipts which local governments are allowed to levy (Duncan 2004; Blore et al. 2004;Devas et al. 2008). The third is external revenuesmainly loans 2 and grants obtained from external governments and/or financial institutions (Duncan 2004;Plekhanov and Singh 2007).
For decentralised units of government to benefit from any of these sources, deliberate and substantial concessions, in terms of both fiscal legislation and economic policy management, have to be made by national/central governments (Boex 2009). These are prerequisites for fiscal decentralisation to work and to enable local governments to deliver services. Such concessions must allow decentralised units of government a proportion of resources available at national level, commensurate with the amount of service delivery decentralised (Wunsch 2001;Devas et al. 2008). The concessions must also allow sufficient autonomy for local governments to design and execute local tax regimes adapted to the circumstances of the local population but also capable of attracting the revenues they require (Tanzi 2001;Wunsch 2001;Fjeldstad et al. 2014). Furthermore, national governments must allow them some legal space to effectively negotiate with international partners regarding grants or loans (Plekhanov and Singh 2007;Devas et al. 2008). Lastly, national/central governments must institutionalise effective mechanisms for developing the capacity of decentralised units of governmentfor example in public financial managementto efficiently administer and account for resources (Wunsch 2001;UNCDF 2005;Fjeldstad et al. 2006;Rao et al. 2014).
In Kenya, the framework for fiscal decentralisation has been characterised by wrangling between the national and county governments over their roles in mobilisation, distribution and accountability for health financing. There have been multiple reports in local media of tussles between county governors, through the Council of Governors, 3 and the national Treasury and Controller of Budgets over the freedom of county governments to borrow, attract grants and design tax regimes to generate more of their own financial resources (Were 2017). This wrangling has made it difficult to assign responsibility for the inadequacy of resources for health financing notable in many parts of the country. Although Kenya is a signatory to the Abuja Declaration of 2001, which commits the country to allocating at least 15% of national resources to the health sector, the national average expenditure on health has remained at about 6% over the past decade (although some counties have reported significant improvements) (Maina et al. 2016;Ochieng et al. 2015;Nesoba 2014 (Anami 2015). Further, in 2016 there were reports of financial crises faced by county governments due to delays in disbursement of funds by the national government (Wakhisi 2016). The outcome has been inefficiencies in procurement that have had devastating implications for availability of drugs at hospitals, remuneration of personnel, and the volume and standards of health facilities in many counties. 6 The national government attributes the delays and inefficiencies to a failure of county health management teams to comply with Treasury's fund management and administrative requirements.
County governments on the other hand argue that it is the responsibility of national government to put the requisite systems and structures in place to facilitate fiscal decentralisation (Anami 2015). They also argue that the inadequacy of systems must not be used by the national government as an excuse to limit the fiscal powers of county governments and restrict access to resources earmarked for service delivery at county level (Anami 2015 citizens seeking medical attention, the responsibility seems to lie squarely with the county governments, mandated by the constitution to provide healthcare. Yet the intricacies of the process of decentralising resources mean the situation is not that straightforward (Were 2017). County governments cannot deliver services without resources; but the public is told by national government that resources have been provided to county governmentsyet without explaining the shortcomings of the system of fiscal decentralisation.
Over the first two years of devolution in Kenya, county governments performed worse at local revenue collection than the former local authorities 7 (Okulo 2014;Were 2017 (Langat 2016). The national government however argues that without effective controls in several areas, namely i) public finance management ii) the scope of engagement with international development partners and iii) the tax regime selected, county governments could cause macroeconomic turmoil that must be strenuously avoided 8 (Tanzi 2001;Devas et al. 2008).
The result is a ham-strung system of fiscal decentralisation that makes it difficult for county governments to effectively deliver healthcare services. This is partly due to national government 7 "County governments perform dismally in revenue collection" (Okulo 2014). 8 Decentralisation may reduce room for maneuver for a national government to manage the economy, increase fiduciary risks, maybe even threaten macroeconomic stability  inefficiency, but nevertheless accountability remains squarely with county governmentsat least in the eyes of the voters. The questions that arise are therefore: Who is responsible when financing for health is inadequate? and Who is accountable to citizens at the county level when they are unable to access healthcare services?

Decision-making
One argument for decentralisation has been that allowing decisions to be taken closer to beneficiaries of public 'goods' improves allocative efficiency, responsiveness of policy, and downward accountability (Rao et al. 2014). Nonetheless, in multiple cases of decentralisation in Africa, progress has been slowed or thwarted by the reluctance of central or national actors to relinquish decision-making power (Wunsch 2001;Olowu 2003).
The devolution statutes in Kenya's 2010 Constitution clearly provide for transfer of decision-making prerogatives and authority to lower levels of governmentin the case of the health sector, county health management teams. This brings health system managers closer to service users and enables them to make decisions and act on them faster and more efficiently than through a centralised health system. Ochieng et al. (2015) studied the early impacts of devolution on Kenya's health sector. They found that whilst health sector stakeholders acknowledged that they had more decision-making autonomy since devolution, they still felt this autonomy was insufficient. Misconceptions and ambiguity over what decisions can be taken at what level of government are hampering decision-making at county level.
County government officers argued that for decision-making to be effective, and a 'game-changer', it must include authority over the most important health sector issues: namely resources, policy and human resources management (Ochieng et al. 2015).
In fact, however, most key decision issues remain subject to regulations and policy set by national government (Ochieng et al. 2015). For instance, Kenya's Controller of Budgets (an independent national-level post) retains power to cap how much a county government decides to spend on health (Menya 2014;Lakin 2015;Omondi 2016). This is a significant limitation to the power of elected county governors, who are nevertheless accountable to citizens for the availabilityor lackof resources for health. There have been attempts by county governments to bypass this, through resolutions passed by county assemblies, but without success (Menya 2014;Lakin 2015;Omondi 2016). . 9 This has been argued to be necessary for standardisation and strategic direction, amongst other reasons (Kibui et al. 2015;Ochieng et al. 2015). However, an unintended effect of this approach is that it limits the capacity of county health management teams to design and implement innovative, contextualised health systems suitable for their particular circumstances (Ochieng et al. 2015). Whether in procurement of drugs and other health commodities, or human resources management, national standards and regulations profoundly affect decisions made at the county level (Kibui et al. 2015;Ochieng et al. 2015).

Political accountability
A rather simplistic definition of political accountabilitybut one which is useful heredescribes it as a "specific form of relationship in which elected government officials are directly responsible to voters for their public actions including […] the provision of public goods" (Acosta et al. 2013, p. 6).
Essentially, political accountability depends on the number of agents involved in the relationship between the governed and the governingwith the governed holding those who govern to account on promises or policy, including service delivery (Acosta et al. 2013). In simple scenarios, this only involves elected officials and voters. In more complex variations, which are more common, there are more than two agents. For example, the relationship could include competing principalsas in the case of a legislator juggling the interests of an electorate and his party position or attempting to satisfy a powerful political figure (Acosta et al. 2013). Alternatively, the relationship could include non-elected state officers, with no or few obligations to the electorate (though in theory they should feel such obligations) who wield significant power capable of influencing the endeavours of the legislator (Acosta et al. 2013).
Unsurprisingly, such variations in the number and characteristics of the agents involved in an accountability framework have significant implications for decentralised service delivery (Panday 2006;Hoffman 2006;Acosta et al. 2013). They may negatively affect the ability and/or willingness of decentralised systems to effectively deliver services (Wunsch 2001;Crook 2003;Panday 2006 As a result, elected county governors and members of county assemblies seek to blame the national government for poor delivery of servicessometimes even for issues which are squarely within the county mandate. The national government then seeks to exonerate itself by pointing to the constitution and supplementary legislation transferring power to local governments. The electorate does not know who to believe and meanwhile service delivery continues to suffer. And ultimately, when pharmacies are empty or nurses at the local health facility go on strike, people do not go to camp at the doorstep of the Ministry of Finance in Nairobi, the Controller of Budgets or the Commission on Revenue Allocation. They go to the governor's office demanding answers.

Conclusion
Devolution as currently practised in the health sector in Kenya involves only a partial transfer of responsibility for service delivery to county governments, which unfortunately has created unbalanced accountability relations between different levels of government. Potentially, this imbalance could undermine the many positive opportunities that devolution creates or promises. Developments over the past two years in Kenya illustrate how devolution takes a long time to deliver sustainable change and how much effort, commitment and leadership is still required to succeed.

CJLG December 2017 168
This paper further finds that devolution of the health function in Kenya has introduced ambiguities of responsibility and accountability for service delivery which lead to challenges of coordination between the national and county governments. A great deal of decision-making capacity and resource management authority remain at national level, making it difficult for county governments to effectively deliver healthcare services, despite shouldering the burden of accountability. Finding the right accountability and coordination mechanisms to balance relations between central/national and decentralised governments is a daunting endeavour which will require patience and continuous action but is a vital one if Kenyan citizens are to secure the responsive and effective healthcare system they deserve.

Declaration of conflicting interest
The author declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding
The author received no financial support for the research, authorship, and/or publication of this article.