Decentralization for improving the provision of public services in developing countries: A critical review

Abstract Decentralisation, which sends power and resources from the Central to local governments, comes with several theoretical promises. One such promise is the improvement in local service provision. However, achieving these theoretical promises of decentralization has been a big challenge in most developing countries. This study aims to address the question “does decentralization improve service delivery levels at the local level, as promised?” This study reviews work on decentralization and service provision in developing countries and bring out some lessons from these works. To do this, the study falls on a set of articles from world-class scholars on decentralisation and service provision and key findings and themes from their works. As has been stressed in the conclusion, decentralisation has made some gains in meeting the real needs of local people in terms of service provision. In some instances, increases in the expenditure of local governments have increased service provision levels. However, the level of quality of these service providers through decentralisation has remained questionable. The author argues that an increase in service provision is good; however, it is the quality of these service provisions at the local level that can reduce poverty at the local level. The study, therefore, recommends for a revisit of the design of decentralisation to embrace the type which seeks to empower the local people to demand transparency and accountability from local government officials. With this, the effectiveness of administrative, fiscal and political decentralisation along with clear channels for local accountability and transparency are necessary for improved and quality service provision.


Introduction
Decentralization has received increased global attention in recent times. Decentralisation, which involves the transfer of power from the central government to the local government has three parts. These are fiscal, administrative and political decentralization. Administrative part considers the organizational structure of the local governments, whereas the political dimension considers the politically appointed executives to head the local government. Fiscal dimension focuses on the financial and expenditure arrangements between the local governments and the Central government.
Many developing countries such as Ghana, Lesotho and Nigeria have embarked on donor-funded national reforms aimed at introducing decentralization for the transfer of power from the central government to the local governments. A large body of literature has explained the reasons for such reforms. For example, the classical theorist such as Tiebout and Musgrave provides the following theoretical justifications to support the reason why the national government should decentralize to local governments. These justifications are the proximity of local government to the local people compared to the central government. To them, local governments stand the best position to access first-hand information on the unique needs of the local people. Other writers also justify that local governments stand the best position to identify the economically viable areas of their region for taxation purposes, through proper fiscal decentralisation. By these advantages, local governments will invest their resources in the services which are directly in line with local needs, which gives the people the best utility, technically known as allocative efficiency.
From an economics perspective, these justifications serve to make fiscal decentralisation more efficient under the following points. These are: fiscal decentralisation allows local governments to determine the appropriate levels of service delivery. The provision of public services by different local governments creates competition and such competition created enables the citizens, local people, to select, by moving from one jurisdiction to another to obtain the optimal preference for service delivery and taxes. Furthering the works of Tiebout and Musgrave, Oates in 1995 consider decentralisation as a mechanism that ensures efficiency in the production of services leading to maximum satisfaction.
Empirical findings from many case studies prove that these theoretical gains of decentralization are not achieved on easily, and in some cases when achieved, faces quality challenges (Dick-Sagoe, 2017;Ibeanusi, 2011;Jutting et al., 2005;Doh, 2017;Glewwe & Muralidharan, 2016;Fox et al., 2011;Rodden, 2016). This means that certain conditions must be met at the local level and the decentralized bodies before these gains can be realized. In other words, decentralization is just a vehicle through which improved service delivery can be achieved. This vehicle needs a very good driver, which refers to the design of the decentralization policy itself and the mechanisms to enhance maximum participation, transparency and accountability measures. For example, Ibeanusi (2011) reports of failure of decentralization to improve power balance, accountability to local people and make government more efficient. These failures, which reduces the quality of service provided by local governments, are caused by factors such as political, incentives and the implementation of institutional design of decentralization (Ibeanusi, 2011). Doh in 2017 argues for low staff motivation and competences as the cause of failure in quality service provision for decentralized governments.
As objectives, this paper reviews the decentralization concept, its theoretical underpinnings and the linkage between decentralization and the provision of poverty-reducing public services like education and health. This is important in the sense that this paper will inform policymakers of the linkage between decentralization and service provision at the local level. Again, this paper will seek to improve the limited understanding between decentralization and service provision within the developing countries' perspective.
The organization of the rest of the paper is given below. It begins with reviews on the practical and theoretical arguments on decentralization and service provision and explains the missing link between decentralization and service provision in developing countries. The link between decentralisation and local service provision using empirical studies was reviewed. It further provides an explanation from empirical studies on decentralization and service provision using health and education as a case and finally concludes the study.

Linking decentralisation and improved service delivery: theoretical and practical arguments
This section explains how decentralisation creates efficiency gains in service provision. Decentralisation has an anticipated benefit; thus, it brings decision making closer to the local people and decision-makers. The classical theorists (Tiebout, 1956;Musgrave, 1959), on the benefits of decentralisation, argue that local decision-makers, compared to central authorities, have access to better information on local conditions. This advantage makes them capable of tailoring services and public spending patterns better to local needs and preferences. All things being equal, this point is expected to improve the efficiency and quality of services for local constituents. Oates (1972) introduces an economic argument which was derived from the heterogeneity of taste and spillovers from the public good. Oates used models to explain that local government can adapt outputs to local tastes. On the other hand, the central government produces a common level of goods for all local areas. More so, sub-national governments being closer to the local people can tune their budgets to meet local preferences, which best reflects their communities' preferences.
Commonly, economists assume decentralisation to lead to a better match between local preferences and outputs of local government. With this in mind, they (economists) rate local provision of services as more efficient, unless this situation is outweighed by spillovers or other efficiencies (such as economies of scale) in central government's provision (Oates, 1972). Tiebout (1956) argues that decentralisation is a vehicle to fulfil highly heterogeneous demand that may arise from different local governments.
Scholars also examine the efficiency argument supporting decentralisation from the perspective of consumers' gains due to allocative efficiency and producers' (e.g. government) gains in technical efficiency in delivering goods and services. Allocative efficiency may arise due to a more fitted bundle (i.e. set and composition) of services provided by the local government to their citizens. In other words, through the adjustment that may take place in the proportions of public spending geared to services such as education, health, water provision or others based on local government's response to local claims in a decentralised context. Higher technical efficiency is achieved when larger quantities and quality of goods and services are provided with the same amount of resources (Dick-Sagoe, 2016;Martinez-Vazquez & Mc Nab, 2003). Overall, devolving some of the centralized responsibilities to local levels has been envisaged, in most decentralisation agenda, as a way to improve both allocative and technical efficiency across different public services.
Several arguments have followed the efficiency gains that can be derived from decentralisation could also be outweighed by other efficiency gains arising from central provisions such as economies of scale and the ability to attract better personnel (De Mello, 2004). This indeed is a valid argument, but other scholars have also argued that those gains arising from central provision may also be overestimated (Oates, 1972;Prud'homme, 1995;Sarmistha & Jaideep, 2010). The possibility to achieve allocative and efficiency gains have important implications for improving public service delivery.
Efficiency gains in service delivery can also be achieved through accountability (Prud'homme, 1995;Treisman, 2002). Central government ministries are not incentivized to perceive beneficiaries of services as their clients (Rondinelli, 1981). Dellinger later in 1994 supported the idea and warned the system which concentrates large proportions of discretion on expenditure by the central government through its ministries to face challenges responding to the demands of their national constituencies. Decentralisation, therefore, provides more channels of communication with the central government.
The paragraphs below will provide explanations on how the various dimensions of decentralisation (viz., fiscal, administrative and political) shape service provision at the sub-national level. McLure in 2002 opines that sufficient fiscal resources and discretion over expenditure responsibilities are essential for any efficient service provision at the local level. McLure further notes that the fiscal resources are of two types. External sources and own sources (locally raised revenues), which can bring greater accountability on sub-national governments. Additionally, the sub-national government should have an adequate level of decision discretion on how to use those revenues to execute their public service functions expected from them (R. M. Bird, 1986;Sacchi & Salotti, 2014a).
The intergovernmental fiscal framework, through which sub-national governments finances local service provision has a four-pillar structure. These are revenue assignments, sub-national borrowing, transfers, and expenditure responsibilities. From this framework, sub-national governments finance their expenditure responsibilities (goods and services provided) through the following channels: first, self-financing using local tax revenues, user charges, or shared revenues with the central government; second, intergovernmental fiscal transfers, either through general-purpose block transfers or earmarked-specific purpose transfers; and third through sub-national borrowing. Within the confines of local service delivery, financing options include public-private partnerships, co-financing or co-production arrangements through which the users participate in providing services and infrastructure through monetary or labour contributions, and other co-participative schemes; all these avenues have also been encouraged by decentralisation processes (Litrack & Seddo, 1998;Saavedra, 2010).
Concerning the use of fiscal resources, there are different discretion levels that central governments establish. They are geared to assure a certain level of spending on specific goods and services provided by sub-national budgets. They depend on a variety of factors such as the local capacity to administer resources, fiscal considerations, national goals, political issues, and institutional constraints. From the fiscal dimension, the central government typically may be able to control spending allocations through strings attached in shared revenues and transfers to local governments (i.e. earmarked transfers or conditional transfers), through sub-national borrowing controls, or through other fiscal means (Arze & Martinez-Vazquez, 2003). The government can also place borrowing controls or even tighten local borrowing to solely raise resources for specific categories of goods and services provided at the local level (World Bank, 2004b). Because of all these (and other) considerations, measuring fiscal decentralisation presents several complexities and limitations when examining it empirically (R. Bird, 2010).
The powers of local officials are outlined in the administrative decentralisation. These local officials are responsible for delivering services in issues such as general management, personnel, service facilities and other administrative discretion in day-to-day operations.
Rondinelli in 1981 classified, administrative decentralisation into three types. These are devolution, de-concentration and delegation. With de-concentration, sub-national governments are given some sectoral responsibilities. However, ministerial branches make all relevant decisions. A typical model is for the central line ministries and agencies to have local representatives that manage services within the sub-national governments but respond hierarchically to their own central office (Rondinelli, 1981). Under this type of administrative decentralisation, local governments typically cannot hire or fire personnel, do not set salary levels, and cannot change the structure of the network of service facilities in place (i.e. number, size, and type of facilities). Local branches and representatives in charge of services simply manage day-to-day operations on behalf of the central ministry and under its watchful eye. Delegation involves the transfer of implementation functions to sub-national entities that deliver services. Through delegation, the central government transfers responsibility for implementation and administration to local governments, including service facilities not entirely controlled by central ministries, but ultimately accountable to it (Rondinelli, 1981). Delegation scheme is a mix of devolution and de-concentration. The levels of decision-making power vary significantly within that range across countries.
With devolution, administrative decision-making powers are wholly transferred to sub-national authorities. It empowers them with legal decision-making power and the ability to generate and control resources, including the sub-national public sector employees hiring and firing, career management and payment. Moreover, typically it provides local government with the ability to reallocate resources (including staff) across service facilities within their jurisdiction adapting to local circumstances (World Bank, 2004b).
The power given to citizens, through leaders elected by them, is called political decentralisation. Political decentralisation gives citizens political decision-making power. The local level decision reflects the diverse interests in society than those taken at the national level. Thus, it is expected that service delivery policies will be reflective of the diverse interests in society. Political decentralisation strengthens accountability, which is necessary for improved service delivery (World Bank, 2004b). The scenario is that if citizens are empowered to elect their executives or officials, they can re-elect or remove those officials based on their policy decision, through accountability, on services that affect the citizens.

Decentralisation and improving service quality in developing countries: why the missing link?
Quality service delivery needs decentralization. However, decentralization, in itself, does not lead to quality service provision. Available evidence indicates a failure of decentralised local governments in public service provision in developing countries (Lewis & Smoke, 2015;Martinez-Vazquez, Lagos-Panas et al., 2015b). Smoke and Lewis (2015) assert the difficulty in achieving the theoretical benefits of decentralization in the areas of socio-economic development improved service provision and accountability in developing countries.
Decentralisation and improved service delivery quality link has been hampered by several factors (World bank, 2013). These factors are keeping proper accounting procedures and capacity to manage public finances are lacking in most sub-national governments, thus negatively affecting service delivery levels. Partial decentralisation and political factors result in misalignment of responsibilities. A typical example of such can be seen from Ghana, for example, where subnational governments are responsible for education; the central government pays salaries of teachers. Local elites capture has resulted in low public participation in local governance. , Doh (2017) and Glewwe and Muralidharan (2016) argue that quality service provision cannot be achieved if service providers themselves are ignored in the process, through strategies to improve their performance. Aside from motivation for public staff, Doh (2017) considers staff competences (education, skills, and experiences) as another factor contributing to staff performance and thus improved service quality delivery. In the year 2008, Collier's study highlights the scarcity of qualified administrative staff, poor financial and managerial capacity of local governments' staff to account for poor service provision.
Assessing the quality of service provision for local governments in Ghana, Fox et al. (2011), attributes poor service provision to staff capacity, just like Doh (2017), and poor accountability mechanisms as a result of weak institutional structure among the local institutions in charge of local service provision, as the source of poor service provision. Smoke (92015b) identifies accountability to be the major challenge for service provision. On the structure of local government, Fox et al. (2011) observe that the existing structure has loopholes that facilitate improper accountability with the various institutions at the local level, thus affecting quality service provision at the local level. Rodden (2016) introduced the fiscal incentive model to explain the causes of poor service provision for decentralized governments. Rodden, just like Fox et al. (2011) also attacks the structure of local government as the cause of poor service provision. However, Rodden was more emphatic about the structure of the intergovernmental fiscal relations between the local governments and the central government.
In an attempt to measure if decentralization favours the poor, Jutting et al. (2005) develop a comprehensive model to determine quality service provision. To them, the model determines quality service delivery which meets the needs of the poor at the local. Their model considers political factors, administrative factors and fiscal factors. By political factors, they mean factors like political commitment, policy coherence, transparency and commitment to participation. Administrative factors consider central versus local powers, capacity building, and fighting elite capture and corruption. Finally, fiscal powers consider the stability and type of resources. Likewise, the insightful work of Donabedian (1980), though old, provides an easily assimilable starting point for understanding quality in health service provision in the health service. Donabedian outlines three dimensions for assessing quality in service provision in the public sector to include technical dimension (explained as what the service is intended to do?), non-technical dimension (socio-psychological interaction between the service provider and receiver), and the environment in which the service operates. All these provision models (by Jutting et al., 2005;Donabedian, 1980) provide the basis for classifying the available and disorganized factors identified by other researchers, which have been discussed above.
This section has proved that the mere provision of decentralized service provision does not guarantee the provision of quality service at the local level. Decentralized service provision is as good as making service available and accessible to local people. However, improving access does not mean that the service provided will be quality until the factors outlined here are met.

Decentralisation and service provision nexus: evidence from the literature
Emphatically, Akpan (2011) observes that decentralisation and service provision has received little research attention. This section of the study provides literature from relevant studies linking decentralisation and service provision from old and current works (Table 1).
People have questioned higher public spending on the specific sector following citizens' demand. They further claim that such higher expenditure which emanates from the desires of the citizens does not necessarily lead to improved outcomes (Ablo & Reinikka, 1998;Pritchett, 1996). For instance, in Cote d'Ivoire and Haiti, per capita health spending fell to below five dollars from the 1980s through the 1990s but with a different result in each of these countries: infant mortality rates worsened severely in the African country and improved in Haiti (World Bank, 2004b).
From an African perspective, this study has presented the works of Elhiraika (2007) in South Africa, Oriakhi (2006) in Nigeria, Akpan (2011) in Nigeria, Dick-Sagoe (2017) in Ghana and Aramov and Asante (2009) in Ghana. Details and research findings have been presented in Table 1. on the provision of these services by the local governments, over a period of 10 years, were also used. Correlation and multiple regression analyses were employed to analyse the panel data to test the public choice theories on service provision and its impact on poverty reduction, at an alpha of 5 percent.
With the main objective of measuring the impact of service provision (in terms of expenditure) on service outcomes (with particular reference to education and health) and assessing the impact of service provision on poverty levels in the region *Services provision has improved levels of poverty and more expenditure on service provision improves service outcomes at the local level, however, the corresponding increase in service delivery improvement was low.
Aramov and Asante (2009) The authors were interested in the access to public service disparities among decentralised districts in Ghana Using district and household level data To identify the disparities in access to local public service among the District Assemblies in Ghana *They conclude the causes of the inequalities in access to public service provision to respondents' characteristics (such as education and ethnicity) and location of the district (geography). Thus determinants of service delivery are ethnicity and geography.
(Continued) To observe how fiscal decentralisation contributes to efficient service delivery **He attributes his observation of poor service provision to mismatch between expenditure and revenue assignment and uneven vertical allocation formula, favouring federal government. The rest according to him were ineffective monitoring of public expenditures. Kaufmann et al. (2002) Survey of local agencies To assess the effects of decentralisation on access to public services **Their study found that both local and central service providers in Bolivia fell short in delivering an adequate quantity and quality of services, but local agencies were more successful in accessibility to citizens, particularly for the poorer brackets of population segments.
Arze and Martinez-Vazquez ( To assess the impact of fiscal decentralization on the efficiency of public service delivery **Fiscal decentralization can improve the efficiency of public service delivery but only under specific conditions (adequate political and institutional environments, sufficient degree of expenditure decentralization seems necessary to obtain favorable outcomes and decentralization of expenditure needs to be accompanied by sufficient decentralization of revenue) Source: Author's construct based on literature, 2019. *Decentralisation has improved service delivery. **Decentralisation has not improved service delivery.

Understanding the nexus between decentralisation and education
The theoretical basis of decentralisation is allocative and productive efficiency. These are critical for the efficiency and sustainability of service provision. It also creates equity in economic resource management. Decentralisation is expected to improve accountability, responsiveness, transparency and efficiency in local service provision. Thus with allocative efficiency, decentralizing education will fuse the needs and preferences of the local people within the educational system. This will eventually encourage participation; improve coverage and quality of education. Potentials of increased efficiency through decentralisation have attracted several questions, mainly due to a lack of consensus on literature. The reason is that community financing schemes (beneficiary cost recovery) provide the opportunity for such governments (mostly those facing several fiscal constraints) to off-load some of the fiscal burdens of education provision to the citizens (World Bank, 2001b).
Decentralisation of education by level (tertiary, secondary and primary) and non-formal education has always resulted in deep debate. Devolving decision making and by how much by educational level and to whom continues to be debated.
To Akpan (2011), several experiments are globally ongoing, ranging from devolution of limited functions to sub-national governments (local governments to community-based management, regional/state governments and schools' financing). Recent consensus reached puts the central government in charge of standards-setting and curriculum design, together with tertiary education. Devolution then should be for secondary and primary education. Local participation in school management improves responsiveness, accountability and fosters resource mobilization (Akpan, 2011).

Understanding the nexus between decentralisation and health care
The argument here is about improving service quality and service coverage. However, how these benefits can be realized is not known. Another unknown is the exact impact of different health systems (Akpan, 2011). Decentralising health care comes with the following benefits. Theoretically, these are improved implementation of health programmes, reduction in duplication of health services at the target communities and the greater community financing. The rest is greater community involvement, inequality between rural and urban areas is reduced, and local preferences are factored in rationalizing and unifying health services. Lastly, health programmes implementation is improved, cost containment from moving to streamlined, targeted programs, greater integration of activities from private and public agencies, and finally improved inter-sectoral coordination, particularly in rural development and local government activities (Faguet & Sanchez, 2009).
There is a need to intensify research to confirm the achievements of these theoretical health benefits. Little studies are available on the impact of health decentralisation on the services it delivers. Schwartz et al. (2002) state that the presence of anecdotal evidence which points poor and hastily implemented and designed decentralisation affects health delivery negatively. To Jiménez and Smith (2005), there is no analytical framework to isolate or generalize the factors behind (un)successful decentralisation.

Conclusion and policy recommendations
Improved service provision, which is responsive to local needs, is one of the prime concerns of decentralisation. Decentralised provision of services, through local governments, have been accepted globally due to its theoretical promises of allocative and productive efficiencies. Decentralisation is to enhance the participation of local people in governance. It is also to promote more accountability and transparency in service provision at the local level. A recent trend in research proves decentralised service provision in developing countries have achieved mixed result, as some areas have achieved success and others fail to benefit the citizens. This study has delved into decentralization and its effects on service provision, focusing on empirical reviews. The study reveals that the linkage between decentralisation and service provision is not a straightforward thing. Though studies have generally agreed on allocative efficiency and increases in expenditure on local services increasing service provision levels. Two main questions remain unanswered. These are: does the increase in expenditure on service provision results in a corresponding increase in service provision levels? The second question is: has decentralisation improved service delivery quality at the local level? Sow and Razafimahefa (2015), in their study, observe that increases in expenditure decentralisation should be supported by increases in revenue decentralisation, without which worse results will follow service provision at the local level. Dick-Sagoe (2017) observes that increases in revenues allocated for the provision of services did not result in a corresponding increase in service delivery outcomes for basic education services in the Central Region of Ghana.
The two findings here (Sow & Razafimahefa, 2015;Dick-Sagoe, 2017) agree that increasing expenditures devoted to service delivery is important for attaining positive result in service delivery; however, that does not necessarily result in a corresponding increase in service quality, measured in terms of service delivery outcomes. This means it takes certain institutional environment, in which Sow and Razafimahefa (2015), Smoke (2015b), and Dick-Sagoe (2017) both identify a strong capacity at the local level, strong accountability at various levels of institutions, good governance and effective the autonomy of local governments to make decentralisation more efficient in achieving quality service delivery, though increasing revenue decentralisation plays a major role. Improving the output of local service delivery is not just enough; however, the quality of the local service delivery counts a lot. A critical look at the literature reviewed in this paper hardly comments on the improvement in the quality of service provision as a result of decentralization. The question now remains, how decentralisation can improve the quality of local service provision at the local level?
This implies several conditions should be met first before decentralisation can fully result in improved quality of service provision at the local level. In the argument of Rodden (2016) and Oriakhi (2006), the design of local government contributes a lot towards the success of decentralisation, being its service provision at the local level. Jutting et al. (2005) provide other important supporting factors to ensure quality service provision at the local level. A design that seeks to limit the active participation and empowerment of the local people, who are supposed to be the beneficiaries of local service provision, results in the provision of poor service provision at the local level. Further, any design which seeks to position the local people to be unable to grip local government officials accountable for policy outcomes and budgetary allocations results in poor public service provision.
As a policy recommendation, the paper presents the following points. For the proper design of decentralisation, the following errors should be corrected. The first is a mismatch between expenditure and revenue assignment and uneven vertical allocation formula, favouring the central government. The second is ineffective monitoring and transparency of public expenditures by the local people. It is, therefore, recommended to align expenditure to community-based projects and community empowerment to hold the sub-national government to account for expenditure and revenue.