70274 A WORLD BANK STUDY Decentralization and Governance in the Ghana Health Sector Bernard F. Couttolenc A W O R L D B A N K S T U D Y Decentralization and Governance in the Ghana Health Sector Bernard F. Cou olenc © 2012 International Bank for Reconstruction and Development / The World Bank 1818 H Street NW, Washington DC 20433 Telephone: 202-473-1000; Internet: www.worldbank.org Some rights reserved 1 2 3 4 15 14 13 12 World Bank Studies are published to communicate the results of the Bank’s work to the development community with the least possible delay. The manuscript of this paper therefore has not been prepared in accordance with the procedures appropriate to formally edited texts. This work is a product of the staff of The World Bank with external contributions. Note that The World Bank does not necessarily own each component of the content included in the work. The World Bank therefore does not warrant that the use of the content contained in the work will not infringe on the rights of third parties. The risk of claims resulting from such infringement rests solely with you. The �ndings, interpretations, and conclusions expressed in this work do not necessarily reflect the views of The World Bank, its Board of Executive Directors, or the governments they represent. The World Bank does not guarantee the accuracy of the data included in this work. The boundaries, colors, denomi- nations, and other information shown on any map in this work do not imply any judgment on the part of The World Bank concerning the legal status of any territory or the endorsement or acceptance of such boundaries. Nothing herein shall constitute or be considered to be a limitation upon or waiver of the privileges and immunities of The World Bank, all of which are speci�cally reserved. Rights and Permissions This work is available under the Creative Commons A ribution 3.0 Unported license (CC BY 3.0) h p:// creativecommons.org/licenses/by/3.0. Under the Creative Commons A ribution license, you are free to copy, distribute, transmit, and adapt this work, including for commercial purposes, under the following conditions: A ribution—Please cite the work as follows: Cou olenc, Bernard F. 2012. Decentralization and Governance in the Ghana Health Sector. Washington, D.C.: World Bank. DOI: 10.1596/978-0-8213-9589-9. License: Creative Commons A ribution CC BY 3.0 Translations—If you create a translation of this work, please add the following disclaimer along with the a ribution: This translation was not created by The World Bank and should not be considered an official World Bank translation. The World Bank shall not be liable for any content or error in this translation. All queries on rights and licenses should be addressed to the Office of the Publisher, The World Bank, 1818 H Street NW, Washington, DC 20433, USA; fax: 202-522-2625; e-mail: pubrights@worldbank.org. ISBN (paper): 978-0-8213-9589-9 ISBN (electronic): 978-0-8213-9590-5 DOI: 10.1596/978-0-8213-9589-9 Library of Congress Cataloging-in-Publication data has been requested. Contents Foreword .............................................................................................................................vi Acknowledgments..........................................................................................................viii About the Author ...............................................................................................................ix Acronyms and Abbreviations .......................................................................................... x Executive Summary ......................................................................................................... xii 1. Introduction ......................................................................................................................... 1 Objectives .............................................................................................................................. 1 The Ghana Health System .................................................................................................. 2 Main Issues in the Health System ..................................................................................... 6 2. Literature Review on Decentralization .......................................................................... 8 Conceptual Framework ...................................................................................................... 8 Experience in Developing Countries .............................................................................. 12 Lessons Learned ................................................................................................................ 20 3. Regulatory and Policy Framework on Decentralization in Ghana......................... 31 General Legal Regulatory Framework ........................................................................... 31 Current Legislation and Policies for Decentralization in Health ............................... 40 4. Rapid Assessment of Local Capacity and Challenges............................................... 46 Approach and Methods .................................................................................................... 46 Size and Organization of District Health Systems ....................................................... 46 Governance and Accountability of DHAs ..................................................................... 48 Structure and Staff ............................................................................................................. 51 Planning and Budgeting ................................................................................................... 52 Finance and Expenditure ................................................................................................. 54 Human Resources.............................................................................................................. 56 Procurement ....................................................................................................................... 57 Health Care Management ................................................................................................ 58 Knowledge and Perceptions of Decentralization.......................................................... 59 Conclusion .......................................................................................................................... 65 5. Analysis of Regional Resource Allocation .................................................................. 68 Data Sources and Issues ................................................................................................... 68 Some Evidence on Fiscal Decentralization in Health ................................................... 69 The Financing of District Health in Ghana .................................................................... 71 Resource Allocation Pa erns ........................................................................................... 75 Regional Variation in Health Expenditure per Capita ................................................. 79 Summary of Findings from the Resource Allocation Analysis ................................... 84 iii iv Contents 6. Main Challenges and Policy Options........................................................................... 87 Capacity Strengthening .................................................................................................... 89 Coordination Mechanism ................................................................................................. 89 Policy Framework ............................................................................................................. 90 Integrated Planning and Budgeting................................................................................ 90 Strengthening of DAs’ Structure and Capacity............................................................. 91 Management at the Facility Level ................................................................................... 92 Financing Framework for Decentralization................................................................... 93 Procurement Process ......................................................................................................... 93 Human Resources Management ..................................................................................... 94 Information and M&E Systems ....................................................................................... 94 Legal Framework............................................................................................................... 95 Appendix A: Main Legislation and Policy Documents Regarding Decentralization and Local Government in Ghana .................................. 96 References................................................................................................................................ 100 Tables Table 1.1: Distribution of facilities by type and ownership ..................................................3 Table 1.2: Facility number and density by region ..................................................................4 Table 1.3: Mean size of hospitals by type, 2008.......................................................................5 Table 1.4: Mean size of other facilities and administrative units by type, 2008 .................5 Table 2.1: Mapping the decision space: Ghana, Philippines, Uganda, and Zambia .......21 Table 2.2: Main advantages and disadvantages of the modalities of decentralization .........27 Table 2.3: Expected impacts and risks in decentralization ..................................................29 Table 3.1: Mapping out the decision space for Ghana districts regarding health ...........43 Table 4.1: Structure and number of questions in the instrument .......................................47 Table 4.2: Mean density of public health infrastructure per district..................................48 Table 4.3: Weaknesses in capacity and process by level of government and function .........66 Table 5.1: Mean expenditure* by type of facility or unit .....................................................78 Table 5.2: Mean features of hospitals by category ................................................................79 Table 6.1: Illustrative assessment of prospect for decentralization by function ..............91 Figures Figure 1.1: Levels and institutions in the Ghana health system ...........................................4 Figure 2.1: Relation between scope, depth, and types of decentralization.......................12 Figure 4.1: Participants in the workshop ...............................................................................47 Figure 4.2: Administrative lines of authority ........................................................................48 Figure 4.3: Lines of accountability ..........................................................................................49 Figure 4.4: Role of the commi ee............................................................................................50 Figure 4.5: Contracting arrangements with other public entities ......................................51 Figure 4.6: Administrative units present at RHAs and DHAs ...........................................52 Contents v Figure 4.7: Health Office involvement in integrated planning...........................................53 Figure 4.8: Responsibility for budget execution ...................................................................54 Figure 4.9: Contracting arrangements with other public entities ......................................55 Figure 4.10: Authority for determining staff numbers and composition ..........................56 Figure 4.11: Authority for hiring staff ....................................................................................57 Figure 4.12: Responsibility for drugs procurement .............................................................58 Figure 4.13: Supervision of and interaction with District Hospitals .................................59 Figure 4.14: Supervision and interaction with sub-district health services ......................60 Figure 4.15: Awareness of role and responsibilities under decentralization....................60 Figure 4.16: Expected main responsibilities under decentralization .................................61 Figure 4.17: Perceptions about transfer of DHAs to DA authority (percentage of agreement) ............................................................................................................ 61 Figure 4.18: Perceptions about transfer of facilities to DAs: Health Centers (percentage of agreement) .................................................................................. 62 Figure 4.19: Perceptions about transfer of facilities to DAs: Hospitals (percent of agreement) ............................................................................................................ 62 Figure 4.20: Perceptions on the feasibility of decentralization ...........................................63 Figure 4.21: Expectations on positive impact of decentralization on health system performance ......................................................................................................... 63 Figure 4.22: Rating of capacity of DHAs (top) and DAs (bo om) .....................................64 Figure 4.23: Mapping of DAs capacity for typical functions or dimensions ....................65 Figure 5.1: Fiscal decentralization in health across countries.............................................70 Figure 5.2: NHA expenditure data for 2002 ..........................................................................71 Figure 5.3: Health expenditure composition by WHO estimates ......................................72 Figure 5.4: Financing flows for district health services .......................................................74 Figure 5.5: Distribution of recurrent health expenditure by type and level, 2006–08...........75 Figure 5.6: National and district health spending by head item, 2008 ..............................76 Figure 5.7: Financing sources by head item in district expenditure ..................................77 Figure 5.8: Health expenditure by level and type of service, 2008 ....................................77 Figure 5.9: Expenditure allocation at regional and district levels ......................................78 Figure 5.10: Variation in sub-national health expenditure per capita across regions, 2008 ......................................................................................................... 79 Figure 5.11: Total health expenditure per capita by district, 2008 .....................................80 Figure 5.12: Expenditure per capita and welfare across regions........................................81 Figure 5.13: Health expenditure per capita and welfare across districts ..........................81 Figure 5.14: Health expenditure by level in regional capitals ............................................82 Figure 5.15: Primary care expenditure per capita (DHAs and SDGs) across districts, 2008 ........................................................................................................ 82 Figure 5.16: Hospital expenditure per capita across districts, 2008 ...................................83 Figure 5.17: Mean and range of size by type of hospital (in no. of beds) .........................84 Foreword D ecentralization of health care delivery, �nancing and management has been an im- portant trend of recent decades, in developed and developing countries alike; it has usually been undertaken under the belief that a decentralized system will be more efficient and effective. However, decentralization is no magic wand, and is a complex endeavor. In order to produce the expected bene�ts, it requires a correct balance be- tween different elements, including strong leadership, clear vision, policy and strate- gies, adequate local capacity, and strong information and monitoring systems. On the other hand, decentralization takes many different flavors and features depending on the political, social and economic characteristics of each country. There is no one single model or approach that would meet the needs and expectations of different countries, but international experience suggests approaches more likely to be successful and criti- cal elements of the process. The international literature on decentralization usually focuses on government de- centralization in general and its main modalities and features in particular. Not much has been said about how decentralization works out in the health sector, the speci�c challenges it faces in addition to those facing general decentralization, and how to ad- dress them. In particular, it is uneasy to strike a balance between true community partici- pation at the local level and local governments’ autonomy, on the one hand, and system coordination and integration across government levels, and consistency between local initiatives and national goals and objectives, on the other. Ghana has since its independence, promoted decentralization in successive waves of legislation and initiatives, with several steps back along the way. This process has been heavily influenced by the political instability that characterized the �rst decades of the post-independence period, which limited continuity and consistency over time. Nev- ertheless, several important building blocks have been put in place, and an abundant legislation has been passed. In the health sector, progress has been slower and incom- plete, and different models and approaches are competing, leading to fragmentation of structures and regulations. This book brings together the �ndings from different analyses of the Ghanaian health sector decentralization process and a review of the lessons learned from interna- tional experience. It assesses past initiatives and the current status of decentralization as it relates to health; along the way, it brings together prior studies on the subject in Ghana, and offers additional evidence and insights from a �eld survey conducted for the purpose of assessing local preparedness for and perceptions on decentralization. The evidence brought together in this book point to a signi�cant dissociation between policy and implementation, and over-reliance on general legislation at the expense of well-thought implementation strategies and the critical elements that make it work. This dissociation has also been found in other countries. The book makes a signi�cant contribution to the identi�cation of gaps and weak- nesses in the Ghanaian process and can thus provide policy makers in Ghana with use- ful insights on how to strengthen it and align its different elements into a consistent na- vi Foreword vii tional policy. But it is also relevant to other developing countries in Sub-Saharan Africa and elsewhere that are in the process of decentralizing their own health system. This book is also a welcome contribution to the area, by throwing some light at the intricacies and complexities of health system decentralization. Dr. Elias K. Sory Director General Ghana Health Service Acknowledgments T he author is grateful to Karima Saleh, Senior Economist (Health) and task team leader for this assignment, for her support and guidance; GNV Ramana and Finn Schleimann (World Bank) for their careful review of the draft document and suggestions; and Smile Kwawukume (World Bank, Accra) for his kind and active support. The author also extends a special thanks to Dr. Elias Sory, Director General of Ghana Health Service, for his insightful comments during our many conversations on this difficult subject. Of course, this work would not have been possible without the fruitful support and participation of the members of the CSR Technical Sub-Commi ee on Governance and Decentralization, including Daniel Degbotse and Emmanuel Owosu-Ansah (MOH); and the many participants who provided their input and comments in the two work- shops organized as part of the study. Special thanks are owed to the moderators and discussants in these workshops, including Nana Boachie-Danquah, Senior Lecturer, University of Ghana Business School; Dr. Irene Agyepong, Regional Director for Greater Accra, GHS; Dr. Daniel Kertesz (WHO representative, Accra); Dr. Yao Yeboah, Director of the Church of Pentecost Social Services; Dr. Moses Adibo, Lecturer, Health Policy and Planning; School of Public Health, University of Ghana; Dr. Sylvester Anemana, Ag. Chief Director Minister of Health; Kwame Owusu-Bonsu-MLGRD; Mr. Daniel Nyankamawu, Chief Director-MLGRD; Levina Owusu-MLGRD; Mr. Akwasi Oppong- Fosu, Head of Local Government Service Secretariat; Dr. George Amofah, Deputy Director General-GHS; Dr. Ebeneezer Appiah Denkyira, Director HRH, MOH; Dr. Frank Nyonator, Director PPME-GHS; Dr. Kyei Faried, Head, Disease Control, GHS; Dr. Koku Awonoor-Williams, Regional Director Upper East, GHS; George Dakpallah, Ag. Director PPME, MOH; Dr. Erasmus Agongo, Regional Director, Eastern; Mr. Kwamena Ahwoi, Principal Lecturer, Graduate School, Ghana Institute of Management and Public Administration. viii About the Author B ernard F. Cou olenc holds an MBA and a PhD in health economics. He has worked for many years in executive positions in public and private hospitals, health system planning, and �nancing. He has 20 years of experience in consulting in developing countries for international organizations, in projects focusing on health sector reform, health �nancing, hospital management and reform, health care �nanc- ing and evaluation, and public-private partnerships. He held a teaching position at the University of São Paulo in health economics and now heads a policy research center in São Paulo, Brazil. ix Acronyms and Abbreviations BMC Budget Management Center CAG Comptroller Accountant General’s Office CHAG Christin Health Association of Ghana CHPS Community Health Planning and Services CSR Country Status Report DA District Assembly DACF District Assemblies Common Fund DDHS District Director of Health Services (chief of DHA) DHA District Health Administration (GHS district offices) DHIMS District Health Information Management System DHO District Hospital DHS Demographic and Health Survey GDP Gross Domestic Product GES Ghana Education Service GHC Ghana Cedi GHS Ghana Health Service GOG Government of Ghana GPRS Ghana Poverty Reduction Strategy HIPC Heavily Indebted Poor Countries (a debt alleviation program) HQ Headquarters HRM Human Resource Management IGF Internally Generated Funds LG Local Government LGA Local Government Authority M&E Monitoring and Evaluation MMDA Metropolitan, Municipal and District Assemblies MOH Ministry of Health MLGRD Ministry of Local Government (and Rural Development) MOFEP Ministry of Finance and Economic Planning NDAP National Decentralization Action Plan NHA National Health Accounts NHIA National Health Insurance Authority NHIS National Health Insurance System NRC(D) National Redemption Council (Decree) OOP Out-of-pocket health expenditure x Acronyms and Abbreviations xi PARDIC Public Administration Restructuring and Decentralization Implementation Commi ee PER Public Expenditure Review PETS Public Expenditure Tracking Survey PNDC(L) Provisional National Defense Ruling Council (Law) PPME Policy, Planning, Monitoring and Evaluation (unit) RCC Regional Coordinating Council RDHS Regional Director of Health Services (chief of RHA) RHA Regional Health Administration (GHS regional offices) RHO Regional Hospital RPCU Regional Planning and Coordinating Unit THE Total Health Expenditure THO Teaching Hospital Executive Summary I n recent years, many countries, both developed and developing, have engaged in a pro- cess of decentralization of health service delivery and/or other functions of the health system. In most cases, decentralization has been adopted to improve accountability to local population, efficiency in service provision, equity in access and resource distribution, or to increase resource mobilization. In Ghana, the government has embarked in a decentralization policy since the 1980s, which was strengthened and ampli�ed by the Local Government Act of 1993. However, policies have so far been inconsistent and the process incomplete in imple- mentation. Effective decentralization still faces considerable challenges, especially in social sectors involving large structures. The public health sector is one that has not fully embraced the decentralization model adopted by the government of Ghana (GOG). Some functions and responsibilities have been decentralized, but others remain central- ized or simply deconcentrated. The process also faces signi�cant challenges that will be described below. The review of decentralization experiences of developing countries such as Uganda, Nigeria, Tanzania, the Philippines, Brazil, Colombia and others indicates that decentral- ization is a complex process, that reflects each country’s history and administrative, geo- graphic, demographic, and health sector characteristics. Its success and impact depends of correct design features and implementation strategy. Some of these key features for success are: (i) a comprehensive national policy on decentralization and a clear strat- egy for its implementation; (ii) a choice of decentralization to a government level that has sufficient capacity and sufficient scale to absorb the decentralized responsibilities; (iii) a model and approach to decentralization appropriate to the country’s history and characteristics, but that ensures sufficient real autonomy to the decentralized entities; (iv) a clear de�nition of responsibilities and roles across government levels and institu- tions; (v) strong transparency and accountability mechanisms to both local communities and the higher government level; (vi) adequate resources at the local level to ful�ll the decentralized functions and responsibilities; (vii) sufficient number of quali�ed staff at all levels, and clear policies regarding the status and hiring regime of health workers under local governments (LGs). Since the 1980s, Ghana has made signi�cant progress in decentralization, and several building blocks for a devolved health system have been put in place. However, important weaknesses remain regarding several of the key requirements mentioned above. Inter- national comparisons indicate that Ghana shows a narrow effective decision space with respect with most health functions and subfunctions when compared with several other developing countries (such as Uganda, Zambia, and the Philippines or Brazil and Colom- bia). This suggests that in spite of the progress accomplished since independence and the several waves of decentralization reforms, Ghana public sector remains less decentralized than comparable countries. Progress and effectiveness of decentralization so far has been xii Decentralization and Governance in the Ghana Health Sector xiii hampered by issues related to regulatory inconsistencies, unclear policies, incomplete implementation, resistance to change, weak managerial capacity, centralized authority over key resources, weak capacity at central and regional level to monitor and support implementation, and weak economic base of many districts. The critical issues identi�ed in the main report are: ■ Several of the basic elements of administrative decentralization are already in place, in the form of district political and administrative units; District Assem- blies (DAs) are endowed with a signi�cant management structure, though with varying size and capacity, and the Ghana Health Service (GHS) District Health Administration (DHA) offices already undertake a number of decentralized functions. Further, in many districts the District Assembly is supporting the DHA office in concrete ways: providing office space, hiring staff to complement DHA own staff, contracting maintenance or support services (security, cleaning and others). However, these activities are unsystematic and li le coordinated. If brought together under a single authority, local health structures would make for the basic structure of a decentralized health system. ■ A number of useful information systems and management tools have been developed and implemented, including planning and budgeting systems, reporting and information systems, performance measurement, and �nancial transfer mechanisms to LGs, among others; however, their effectiveness is lim- ited by many overlaps and duplications, the fragmentation among systems, and their inability so far to produce reliable critical information. ■ The existing regulations, assessments and policy documents have correctly identi�ed the main challenges in implementing effective government decen- tralization in the form of devolution to LGs. But only recently has the Minis- try of Health (MOH) produced consistent policy documents addressing the speci�c issues of decentralization in the health sector. A comprehensive and clear policy framework to guide implementation of devolution in health is still lacking. ■ The existing legal framework concerning health is confusing and contradictory; successive waves of laws and regulations offer changing and conflicting views of what decentralization should look like, and are quite vague as to which func- tions are to be devolved to whom. One of the main conflicts is the often high- lighted contradiction between GOG’s general policy of devolution and MOH/ GHS model of delegation cum deconcentration: while the Local Government Act and other general legal documents have de�ned the Ghanaian process as one of full devolution to districts, the MOH itself has delegated the responsibility of managing its facility network to an autonomous entity, the GHS, which has in turn established a deconcentrated structure with Regional and District Health Administrations (RHAs and DHAs).1 However, other regulatory conflicts have built up over the years, as indicated in this report. ■ As a result, governance and accountability of local health institutions is weak- ened; lines of authority and accountability are blurred, with important overlaps and duplications among DHAs, RHAs, MOH, GHS, and the DAs; actual respon- sibilities for resource allocation, personnel management, or procurement, are split between different levels of government and offices, without a clear policy xiv A World Bank Study or regulation de�ning who is responsible for what. For example, broader legisla- tion calls for “full devolution to local governments�, but management of local staff has been centralized in a parallel Local Government Service (LGS), which in effect withdraws from local governments the authority over the major resource they need for managing local services. Multiple procurement lines and staff hir- ing regimes also contribute to fragmentation of authority. ■ Mechanisms for local communities’ participation and voices have been established—in the form of local councils and commi ees—but these have mostly advisory or consultative roles; in many areas they have not been func- tioning as an effective channel of participation in decision making and planning. It is unclear from the information gathered for this report whether mechanisms to prevent local elites’ capture of devolved authority exist. ■ Financing mechanisms to local health facilities and programs have been estab- lished, including transfers to LGs through the DAs’ Common Fund. However, the �nancing pa ern for local-level health functions is fragmented and con- fusing: different funding sources2 specialize in �nancing speci�c line items or programs, and the DAs’ resource allocation to the health sector—though man- dated in current legislation—is not transparent or clearly recorded. In addi- tion, �scal decentralization in Ghana is more apparent than real: over 50 per- cent of public health expenditure is allocated to the district level, but the larger part of these resources are allocated and controlled by the central government; local authorities—whether DAs or GHS District Offices and facilities—have li le real decision power on resource allocation. ■ Effective local decision power is further reduced by substantial delays and unpredictability in the transfer and release of funds, both by the GOG and the National Health Information System (NHIS), these delays greatly reduce the real autonomy of local authorities and facilities alike, as well as �nancial accountability and budget transparency. ■ The pa ern of geographical allocation of resources—namely across districts—is not transparent and does not follow a clear pa ern, in spite of existing regula- tions and proposed allocation formulas; large variations in public health expen- diture per capita are observed (7 districts had a value of GHC 20 or more, while 31 districts showed a value lower than 1 Cedi, against a national average of GHC 5.52 in 2008). The observed variation does not seem related to population, poverty levels, but appear to be strongly influenced by existing health infra- structure. These wide variations are strongly suggestive of (i) inequalities in resource allocation, and (ii) lack of standards and transparent criteria for budget allocation, in spite of existing formulas; they are a critical issue in implementing effective decentralization. ■ Capacity for implementing and managing a truly decentralized health system needs strengthening, not only at the district level but across all levels of govern- ment, although the weaknesses may be different at each level; many districts are hampered by their small scale and/or their remoteness, and have li le leverage for a racting capable staff. Another bo leneck is the lack of reliable information for decision making, monitoring and evaluation; for instance, no information is currently available on how much funding is available to each district, including all different sources. Decentralization and Governance in the Ghana Health Sector xv ■ A major obstacle to effective implementation in the health sector has been that many stakeholders have limited awareness and understanding of the process objectives, prerequisites and implications, as a survey of regional and district officers clearly showed; this makes it difficult to build consensus and support for the process. 37 percent of DHAs indicate having li le or no knowledge of the government’s decentralization policy, and many local officers show skepti- cism and reluctance about devolution’s feasibility and impact. This skepticism constitutes a potentially important obstacle to the devolution policy of GOG in health. Insufficient information and discussion on the policy, and a perception of weak capacity of DAs and local offices generally, appears to contribute to this skepticism, but may also reflect GHS staff their reluctance and suspicion of the decentralization policy. Overall, Ghana has over the years established several of the building blocks needed for a successful decentralization, but these efforts lack cohesion and unity of purpose. The two key players in local health systems—the DAs and DHAs—have a low level of collab- oration and integration, even though formal mechanisms have been established, such as the formal participation of the Director of Health Services (in DHAs) in the DAs’ social or health commi ees, and composite planning. The GOG—and MOH when regarding the health sector—needs to bring together these many policies, instruments and systems and make them work for effecting decentralization. Several policy initiatives would greatly contribute to reduce the fragmentation and inconsistency in the current health system. These are: Capacity strengthening: The planning and implementation of decentralization would greatly bene�t from a systematic assessment and mapping of the DHAs’ and DAs’ capacity and conditions for taking responsibility for speci�c functions to be devolved. Coordination mechanisms: The discussion and de�nition of a decentralization policy framework for the health sector would have to mobilize all stakehold- ers, both for strengthening technically the �nal proposal that will emerge, and to build consensus and support around this proposal. Such an endeavor will require strong and commi ed leadership from the part of the overseeing insti- tution in the health sector, the Ministry of Health. A Coordinating Commi ee jointly led by the MOH and the Ministry of Local Governments (MLG) could be established to coordinate the formulation and implementation of decentraliza- tion in the health sector. Policy framework: A health system decentralization framework is greatly needed, and would further clarify and detail the responsibilities and functions of each government level and agency; the de�nition of particular functions to be decen- tralized should take into account factors such as: economies of scale (espe- cially in procurement of drugs and other strategic supplies and services) and the highly technical nature of some functions and services. The design of a strong policy framework should thus encompass a detailed technical discus- sion on which functions and responsibilities should be decentralized to local (or regional) level, and which would remain centralized. Integrated planning and budgeting: As most assessments and policy documents have pointed out, the current “composite budget� policy has not yet taken root, xvi A World Bank Study though it is a critical step in moving toward full devolution, and needs to be strengthened in the short term. This could be done by the development of prac- tical guidelines for effective joint planning and budgeting in health at the dis- trict level; this means revising existing guidelines for the “composite budget� and actual practices to promote effective participation and involvement of the District Assembly in the discussion and preparation of the district health plans and budget. Strengthening of the DAs’ structure and capacity: Several aspects are to be consid- ered in strengthening LG capacity. First, once effectively devolved, the differ- ent responsibilities will be carried out in different places by different groups of staff. It is thus necessary to break down the general capacity assessment at the “local level� and clearly identify the different types of capacity that will be needed within the LG and de�ne where exactly they should be invested. Some responsibilities will be carried out at the facility level, others by the (yet to be transferred) local health offices, and others at the DA level and its management commi ee and staff. Management at the facility level: As part of the discussion of capacity and auton- omy of LGs, health authorities at all levels need to discuss and de�ne what degree and form of managerial autonomy will be given to what types of health facilities. GHS has over the years deconcentrated some responsibilities to facili- ties, but not in a homogeneous way, and—as shown in the rapid assessment— activities actually performed at the facility level vary signi�cantly across facilities of similar type and size. The role and capacity, and thus the structure, of sub- district entities and communities, also need to be clari�ed. Finally, increasing autonomy at the facility level also requires that the bo lenecks to real autonomy be addressed appropriately. Financing framework for decentralization: A clearer �nancing framework for LGs on the health sector would be a great contribution to decentralization imple- mentation; this framework should seek to streamline the multiple existing flows and funds and take advantage of the District Financing Fund to consolidate these flows. As a �rst step, it would be important to estimate �nancial needs (expenditures) to upgrade LGs’ capacity, and to meet the devolved responsi- bilities. Secondly, the framework would de�ne �nancing sources and flows for decentralized levels, including the incentives structure needed to promote effective implementation and a ract staff to more remote areas. As part of this framework, the current policy and formula for budget allocation across regions and districts could be revised, so as to emphasize the objective of equitable redistribution of funding that is part of the decentralization process. Oppor- tunities for testing and implementing performance-based �nancing schemes should be seriously considered, as international experience has shown that such schemes can provide a proper incentive structure for improving performance. Human resources management: Revising and de�ning regulations and policies regarding human resource management in a decentralized system would help reduce duplications and fragmentation, and formulating a clearer regulatory framework that at the same time homogenizes processes and provides minimal standards, without limiting LGs’ autonomy to manage staff; this framework Decentralization and Governance in the Ghana Health Sector xvii should necessarily include provisions for transferring staff from the central to local level, and a structure of incentives to encourage staff to transfer. Monitoring and evaluation: To be er promote and support the decentralization process, and monitor and evaluate the decentralization process and its impact, the capacity of central and regional levels for monitoring and evaluation (M&E) would have to be strengthened—in terms of human resources, systems and instruments; this activity could take advantage of several important initiatives regarding M&E and performance assessment that have been adopted in recent years, and adapt them to a decentralized system. Legal framework: As a result and consolidation of these efforts, the �nancing and function of decentralization policy frameworks would be consolidated into one legal framework for health system decentralization; this legal framework could be prepared once policy documents have de�ned all main dimensions and aspects of decentralization, to avoid contradictions and regulatory revisions. In summary, Ghana has along the years put in place several important building blocks for a truly decentralized health system. But these efforts have been hampered, and their effectiveness diminished, by the absence of a strong regulatory and policy framework for health, regulatory conflicts and duplications accumulated by several waves of regu- lations, weak capacity to coordinate and manage a devolved health system, fragmented management systems regarding staff management, procurement and budgeting and �nancing. Political instability and the resulting lack of continuity and consistency in the process during most of the period have also prevented the process to take root, and have contributed to the current fragmentation in the health system. But taking advan- tage of the structures and features that have been put in place, strengthening the policy framework, ensuring effective coordination, and addressing the weaknesses highlighted above, would very likely make decentralization of the health system more effective and consistent with GOG general policy. Notes 1. Ghana Health Service Administration and Teaching Hospitals Act (2003). 2. District-level health services and programs are funded through �ve main sources and channels: (i) Central government budget allocated to local facilities, programs and administrative offices; (ii) Internally Generated Funds (IGF), which are the main source of �nance for recurrent costs; (iii) central government transfers to DAs (which are not clearly recorded and consolidated); and (iv) local government own revenues. Foreign funds, an important contributor to health �nance, are usually—but not always—channeled to local services and facilities through the government budget. CHAPTER 1 Introduction I n recent years, many countries, both developed and developing, have engaged in a process of decentralization of health service delivery and/or other functions of the health system. In most cases, decentralization has been adopted to improve account- ability to local population, efficiency in service provision, equity in access and resource distribution, or to increase resource mobilization. Ghana has a long history of local government, going back to pre-independence times of the nineteenth century. By 1859 Municipal Councils were established in the major coastal towns of the then Gold Coast. Native Authorities, Councils and Courts were also estab- lished to administer law and order under the indirect authority of the colonial govern- ment; the limitations of this system was repeatedly put forward in the 1930s and 1940s, and reforms were introduced in 1951 by the Local Government Ordinance (Ahwoi 2010). The government has embarked in a decentralization policy since independence, which was strengthened and ampli�ed by the Local Government Act of 1993 and other legislations. At the present the Government of Ghana (GOG) is commi ed to strengthen the implementa- tion of decentralization, and for that purpose revise and strengthen the policy and regula- tory framework governing decentralization. In spite of this long history and successive waves of decentralization reforms, effec- tive decentralization in the country still faces considerable challenges, especially in large social sectors involving large structures. The public health sector is one that has not fully embraced the decentralization model adopted by the GOG—decentralization by devolu- tion to the districts—for a number of reasons that will be discussed in this report. Some functions and responsibilities have been decentralized, but others remain centralized or simply deconcentrated. Objectives This study on Decentralization and Governance was commissioned by the World Bank as part of a Country Status Report (CSR) to assess the state of decentralization in the Ghana health sector, assess the readiness of the health system to implement decentraliza- tion by devolution, identify inconsistencies in current policies and regulations regarding decentralization, and build consensus for reform through a be er understanding of the policies, strategies and obstacles of health system decentralization. This volume therefore brings together the �ndings of �ve activities: ■ A review of the international literature on health sector decentralization and experiences in developing countries ■ A review of relevant documents and information regarding the status of decen- tralization in Ghana, including policy documents and regulations ■ An assessment of the capacity of the different government levels and health sec- tor structures to implement decentralization, based on a sample assessment of regional and district health administrations 1 2 A World Bank Study ■ An analysis of district-level distribution of health expenditures and its implica- tions for equity and �scal decentralization ■ A discussion of measures and strategies to strengthen the GOG’s proposed decentralization policy and its implementation. This study looks at decentralization policies and process across the different levels of gov- ernment and public institutions in the Ghana health sector, with a particular emphasis on the preparedness for full decentralization of the main public health system—in other words, the Ghana Health Service (GHS) district-level structure and facilities. The study also discusses the role in decentralization of other health structures and institutions. The rest of this section provides an overview of the Ghana health sector and espe- cially of the public system. Section 2 reports on a review of the international literature and experiences of decentralization experiences in other countries, and draws some les- sons from these experiences. Section 3 reviews the regulatory and policy framework on decentralization in Ghana, both from a general government perspective and especially as it applies to the health sector, and assesses the main weaknesses of the current frame- work for devolution in the health system. In Section 4, the report presents the �ndings from a rapid assessment of the capacity of local health structures and the challenges they face in the implementation of full decentralization under the devolution model; special a ention is given to the relationship between the District Health Administrations (DHAs) and the District Assemblies (DAs). Section 5 analyzes the geographic distribu- tion of health resources and expenditure and provides some evidence on equity issues within a decentralization context. The last section summarizes the main challenges fac- ing full implementation of decentralization in the health sector, and describes some pol- icy options to address them in key strategic policy areas. The Ghana Health System The Ghana health system is made of 1832 facilities, 49 percent of which (897) are government-owned (Service Availability Mapping 2006). 509 private facilities are accredited by the National Health Insurance Scheme, in addition to 626 providers of diagnostic and other services. The health sector in Ghana encompasses six sub- systems, which will be discussed further later in the report; ■ GHS. Under Act 525 (Ghana Health Service and Teaching Hospitals Act), the Ministry of Health has transferred to GHS the responsibility for managing and operating nearly all public facilities. The GHS is a semi-autonomous agency with the mandate to “ensure access to health services at the community, sub-district, district, and regional levels.� This subsystem manages most health facilities and the largest part of the public �nancial resources in the sector (see table 1.1). ■ MOH. The Ministry of Health (MOH) is responsible for carrying out central-level activities—such as policy making, regulation, and planning coordination—and several vertical programs (mostly focused on speci�c diseases and public health). It also manages the three teaching hospitals—the largest in the country. ■ CHAG. MOH directly funds a signi�cant number of private, non-pro�t (mission) facilities grouped under the Christian Health Association of Ghana (CHAG). This para-public subsystem is an important—and in some areas the sole—provider of health services. Decentralization and Governance in the Ghana Health Sector 3 Table 1.1: Distribution of facilities by type and ownership Facility type Government Quasi-Govt Mission** Private Others TOTAL Teaching 2 0 0 0 0 2 Hospitals Regional 8 0 0 0 0 8 Hospitals Specialized 3* 0 1 12 0 16 Hospitals District/local 61 7 34 72 2 176 Hospitals Polyclinics 8 0 0 2 0 10 Health Centers 416 2 42 10 2 472 Clinics 162 24 92 375 37 690 CHPS 172 0 0 0 0 172 Maternity 35 1 1 199 5 241 Homes Others 30 0 12 3 0 45 TOTAL 897 34 182 673 46 1832 Source: GHS (Summary of Health Facilities 2007). * Three psychiatric hospitals. ** Mostly facilities run by the Christian Health Association of Ghana (CHAG). ■ Local governments (LGs). MMDAs (Metropolitan, Municipal and District Assem- blies) provide certain health-related services, mostly involving support to MOH and GHS activities; they do not run or manage facilities, and the incipient role of these government structures is a direct consequence of the decentralization pro- cess as it has progressed in recent years. ■ Private sector. The private sector in Ghana includes about one-third of all health facilities in the country, especially clinics and maternity homes. In some areas private hospitals are the only hospital and play the role of district hospitals. ■ Quasi-public. A number of non-health public institutions—such as the Armed Forces, the Police Force and the prison system and a few universities—have a signi�cant number of health facilities, which usually however cater to their own target groups (members of the forces) rather than the general population. The regional distribution of health infrastructure shows signi�cant but reasonable varia- tion across regions. The density of public hospitals and beds varies within a range of 1:2 approximately, while the density of ambulatory facilities varies in the range of 1:11, but drops to 1:3 if the two extremes—The Greater Accra and Upper West regions—are excluded (see table 1.2). MOH and GHS current health structures are structured in �ve main levels or layers (as shown in �gure 1.1): i) The MOH, which is responsible for the general coordination and oversight of the system, and has delegated much of the operational responsibilities to the GHS, except for three teaching hospitals 4 A World Bank Study Table 1.2: Facility number and density by region Public Health Hospital Population Public Hospitals/ Health Facilities/ Hospital Beds/ REGION 2008 Hospitals 100,000 Facilities* 100,000 Beds 100,000 Ashanti 4,720.912 23 0.49 196 4.15 3,302 69.94 Br. Ahafo 2,211.90 8 0.36 154 6.96 1,769 79.98 Central 1,882.11 10 0.53 123 6.54 1,781 94.63 Eastern 2,354.534 13 0.55 234 9.94 2,675 113.61 Gr. Accra 4,100.701 11 0.27 63 1.54 4,006 97.69 Northern 2,270.95 9 0.40 159 7.00 1,310 57.68 Upper East 1,004.24 5 0.50 125 12.45 819 81.55 UpperWest 659.83 4 0.61 115 17.43 692 104.88 Volta 1,901.18 12 0.63 229 12.05 2,214 116.45 Western 2,476.13 12 0.48 176 7.11 1,977 79.84 TOTAL 23,582.50 107 0.45 1574 6.67 20,545 87.12 Source: GHS (Summary of Health Facilities 2007). *Includes CHAG facilities ii) At the regional level (10 regions overall), GHS’ Regional Health Administra- tion (RHA) offices provide secondary hospital care through Regional Hospitals (RHO), and coordinate the districts’ health activities and planning iii) District-level facilities include most health centers and other primary care facili- ties, managed by the DHA offices of GHS, and district hospitals; there are pres- ently 124 districts and 46 Metropolitan and Municipal administrations (up from the original 62 District Councils established in 1974)1 Figure 1.1: Levels and institutions in the Ghana health system Source: Author’s elaboration. Decentralization and Governance in the Ghana Health Sector 5 Table 1.3: Mean size of hospitals by type, 2008 Total Number Total Number Mean Number of Patient-days Mean Number Facility type N* of Beds of Beds (1,000) of Staff Teaching hospitals 3 2,805 935 752.2 1,995 Psychiatric hospitals 3 1,186 395 438.2 427 Regional hospitals 8 1,745 218 447.1 NA District hospitals 99 7,235 73 1,402.6 123 CHAG hospitals 59 6,039 102 1,178.8 NA Quasi-Government 24 1,689 70 96.8 NA hospitals Muslim hospitals 4 262 66 32.5 NA Private hospitals 71 1,377 19 NA NA Source: GHS Bedstate 2008. *N represents the number of hospitals for which data are available. iv) Sub-districts (some 1,500 overall), are supposed to be the main providers of pri- mary health care and report to DHAs, but have in fact limited responsibilities and capacity v) The lowest level rests with Community Based Health Planning and Services (CHPS), which are responsible for the provision of certain community health activities. The �rst three levels of the system—central, regional, and district—correspond to the political-administrative units in the country. The last two are not political entities, although they are supposed to take over some responsibilities under the decentralization model to be implemented (see Section 3 for a description of this model and the proposed role of each level of the system). By Act 525 of 1996, the provision of regional and local services was delegated to GHS, including regional, district-, and sub-district-level services and facilities. For plan- ning and budgeting purposes, these administrative units are structured into Budget Management Centers (BMCs): one for GHS headquarters, 10 RHAs, 8 Regional Hospi- tals, 110 DHAs, 95 District Hospitals, and 110 Sub-district BMCs.2 The different levels and types of facilities are associated to different sizes,3 as shown in tables 1.3 and 1.4 below. Teaching hospitals are, expectedly, the larger ones (935 beds Table 1.4: Mean size of other facilities and administrative units by type, 2008 Facility/Unit type Number Mean Number of Staff Maternity Homes 9 62 Polyclinics 10 147 Clinics 156 9 Health Centers 1,059 4 Sub-districts (CHPS) 285 9 RHA 10 165 DHA 110 64 Sources: GHS (Summary of Health Facilities 2007), WHO 2005. 6 A World Bank Study on average), and represent 17 percent of the total number of patient days in public hos- pitals. Psychiatric and regional hospitals follow in size, and account with 10 percent of patient days each. District hospitals are much smaller (73 beds, but with a wide varia- tion) are the largest group in volume, with 32 percent of the total. Mission hospitals (CHAG) are a bit larger, and constitute the second largest group with 27 percent; they are thus a critical player in the publicly funded system. Muslim and private hospitals are smaller players in the system. Main Issues in the Health System An independent review undertaken for the MOH acknowledged signi�cant advances in recent years but also highlighted a number of problems in the current system (Ministry of Health 2009—Independent Review). The main difficulties are: ■ Increasing fragmentation of the health sector, with weak coordination and fre- quent duplication of efforts among and between public agencies, private orga- nizations, and volunteers ■ Low effectiveness of health expenditure relative to other African countries ■ Disconnection between existing policy documents and plans and actual imple- mentation and practices; plans are short-sighted and do not reflect local priorities, and need to be redrawn according to �nancial availability; and budget preparation is incremental, with no strategic or policy approach; therefore, �nancial resources tend not to follow stated priorities ■ Mismatch between plans and budgets on one hand and disbursements and �nan- cial reporting on the other, due to frequent delays—of several weeks or months— in fund release or payment, hampering plan implementation and operations; resources are not transferred to regions and districts in a timely fashion ■ Slow reimbursement by the National Health Insurance Service (NHIS) due to internal bo lenecks at the different levels of the system; these delays amounted at the end of 2008 to three to four months of total IGF revenue ■ Cumbersome planning, budgeting, �nancial processes and practices, which are of li le help to health professionals and managers ■ Multiple parallel and disconnected budget, accounting and �nancial management information systems or sub-systems,4 many of them manual; they are designed for central level monitoring and control (Including the District Health Informa- tion Management System—DHIMS), and thus of li le use for local level manage- ment; systems used or �lled in at the district and facility level are often manual and on spreadsheets, but may include separate and speci�c software or systems ■ Substandard quality in many facilities for some services such as neonatal care; ■ Low coverage of certain key public health activities, including immunization (around 50 percent of children are fully immunized), and deliveries by skilled professionals (39 percent) ■ Weak role and participation of local communities in health planning and de�ni- tion of policy priorities ■ Low level of skills and competencies for most staff, due in part to inadequate training, especially at the local level; this is compounded by weak policies and capacity for staff a raction and retaining in disadvantaged and remote areas Decentralization and Governance in the Ghana Health Sector 7 ■ Signi�cant inequities in the distribution of health professionals among service agencies, within agencies and across regions, in spite of recent policies and efforts (e.g., posting enforcement) ■ Low staff productivity, and lack of standards for staffing norms and agreed measures of productivity ■ Excessive proportion of wages within the health budget (93 percent in 2008), also due in part to the absence of staffing norms. These problems reduce the effectiveness of the public health system and resources, and also tend to complicate the process of decentralization. Notes 1. Most of the data presented in this report cover 138 districts, municipalities, and metropolitan areas, as of 2008. 2. See the section titled “Planning and budgeting� for a short discussion of BMCs. 3. It is worth noting that the categories of health facilities, their numbers and corresponding beds and staff are not consistent across documents, producing discrepancies in the ratios shown. Some of these differences are likely to be due to changes over time (since different data were not available for the same years), but not all. 4. In 2008, these included: BPEMS, Activate, ACCPAC, and others based on Excel spreadsheets. These issues relate to the public sector as a whole, and are also found in the MOH/GHS system. CHAPTER 2 Literature Review on Decentralization D ecentralization has become over the last decades an increasingly popular reform for health systems in developing as well as developed countries. Most countries have been or are in the process of decentralizing responsibilities to local governments. Overall, a 1974 study showed that 84 percent of 75 developing and transitional countries with population greater than 5 million, had engaged in some type of decentralization (Dillinger 1994). According to Lister and Betley (1999), “There is usually a presumption that decentralization is a good thing which goes along with democracy, good governance, a market economy, poverty alleviation and efficiency in public expenditure.� However, depending on the objectives of decentralization in different countries, and the strategic choices made to design and implement it, decentralization takes on many forms, and produces a wide variety of institutional arrangements. These are shortly reviewed in the following section. In addition, the actual impact of decentralization is not necessarily consistent with the widespread assumption that it is “a good thing�; successes and failures to achieve the desired results are found in international experience. The sec- tion titled “Experience in Developing Countries� reviews some of these experiences, and the section titled “Lessons Learned� summarizes the main lessons learned from them. Conceptual Framework Broadly, decentralization has to do with the transfer of responsibilities from the central government to other government levels or institutions. However, there is no consensus on a precise de�nition of decentralization, which has been understood by researchers and policy-makers alike in different ways. Several conceptual approaches have been applied to the study of decentralization, including: the public administration approach (Rondi- nelli and Cheema 1983), the local �scal choice approach (Musgrave and Musgrave, 1989), the social capital approach (Putnam 1993), and the principal agent approach (Pra and Zeckhauser 1991; Griffith 1966). Every approach contributes different elements to the understanding of decentralization processes. For instance, the social capital approach suggests that localities with a long tradition of community organization, civic networks and solid local institutions will be more likely to be successful in a decentralization pro- cess. The principal agent approach allows examining the relationships between a central Ministry of Health and local governments and how the former can influence the behav- ior of the la er. The public administration approach produced a useful and commonly used typol- ogy of modalities of decentralization based on the level and type of institution responsi- bilities are transferred to. According to Bossert (1998): Deconcentration is de�ned as shifting power from the central offices to peripheral offices of the same administrative structure (e.g., Ministry 8 Decentralization and Governance in the Ghana Health Sector 9 of Health and its district offices). Delegation shifts responsibility and authority to semi-autonomous agencies (e.g., a separate regulatory commission or an accreditation commission). Devolution shifts responsi- bility and authority from the central offices of the Ministry of Health to separate administrative structures still within the public administration (e.g., local governments of provinces, states, municipalities). Privatiza- tion transfers operational responsibilities and in some cases ownership to private providers . . . Another approach looks at three dimensions of decentralization: political, admin- istrative, and �scal. Political decentralization involves establishing local government structures and community participation mechanisms. From this perspective, decentral- ization should provide increased political accountability, transparency and representa- tion (Pallai 2001). Administrative decentralization is usually de�ned in terms of the administrative structures and systems needed at the different levels of government, and where respon- sibilities should be vested. It implies the (re)organization and integration of administra- tive bodies at the local level to carry on the decentralized functions, and the responsibility over staff. It thus relates to the issues of administrative capacity and accountability. Authority over staff and its management is often a source of conflict and misalignment in decentralization processes, since execution responsibilities may be decentralized while staff at local level may still be appointed centrally. Fiscal decentralization is the assignment of responsibility for mobilizing, managing and allocating funds to and within subnational governments. It focuses on the main issues of who can raise revenues (�scal autonomy) and who can spend them (�nancial autonomy). It relates to the issues of intergovernmental transfers, revenue mobilization at the local level, the budgeting process across government levels and �scal/monitoring by the central government among others (Farrant and Clarke n.d.). The �scal aspects of decentralization are quite important and tend to affect the accountability mechanisms of local governments and other dimensions of the process. Decentralization—in any form or model—needs to balance and link responsibilities for function (service delivery), �nance (�nancial and taxing autonomy) and administra- tion (resource management, including staff), since international evidence indicates that imbalances or mismatch among these three dimensions are likely to lead to decentraliza- tion failure. An important and unresolved question in decentralization, closely related to �s- cal decentralization, is the “appropriate� distribution of responsibilities and revenues across levels of government. There is no ideal proportion of health spending that should go to each level of government. From a �scal federalism perspective, the prin- ciple of subsidiarity assumes that responsibilities are primarily vested in local govern- ments, and are transferred to higher levels only when local governments are unable to carry them out efficiently (Dafflon n.d.). In a centralized system, the opposite is true: the central level holds a priori all competencies, and can decide which to decentralize to local governments. The framework used in this report draws on several of the theoretical contributions outlined above—especially Bossert (1998)—but takes a more empirical perspective, based 10 A World Bank Study on the three key questions to be asked in examining decentralization processes across countries; these are: 1. Why was decentralization undertaken? What were the main objectives it sought to achieve? 2. How was it designed? What model was adopted? Particularly, what responsi- bilities were decentralized, to whom, and with what degree of autonomy? 3. What was the impact of decentralization? Were its stated objectives achieved? What were the main issues and challenges? In relation to the �rst question, governments move to decentralized systems for a variety of reasons: technical (to improve service delivery effectiveness), political (to increase local autonomy and participation—or, in a broader sense, “democracy�), or economic/�nancial (to increase efficiency and cost control or improve accountability).1 For instance, in the transition states of Eastern Europe and the former Soviet Union, and in China and Viet- nam, decentralization came along embedded in a broader set of political and economic reforms. In other countries, it was a consequence and feature of political democratiza- tion (as in Brazil and other Latin American countries). In some African and East Asian countries, it was linked to a reform process aimed at modernizing government. Improv- ing service quality and effectiveness was usually less of a priority in developing coun- tries, and more so in North America, Chile, Australia and New Zealand (United Cities and Local Governments 2009). However, objectives of decentralization are not always very clear, and most often mix political, economic and technical objectives in varying combinations. Decentralization has often been seen—and implemented—as an end to itself. How- ever, Bossert (1998) highlighted that “decentralization is not an end in itself but rather should be designed and evaluated for its ability to achieve broader objectives of health reform: equity, efficiency, quality and �nancial soundness�. The particular objectives of decentralization—and even their lack of clarity—tend to influence the design, imple- mentation and practice of the process, as well as its impact. To answer the second set of questions, we apply Bossert’s (1998) useful concept of decision space. Based on a modi�ed principal agent framework, this approach de�nes decentralization “in terms of the set of functions and degrees of choice that formally are transferred to local officials�. In other words, what functions are transferred, to whom, and with what degree of autonomy. Bossert identi�ed three key elements of decentralization, “the amount of choice that is transferred from central institutions to institutions at the periphery of health systems, what choices local officials make with their increased discretion, and what effects these choices have on the performance of the health system�. At one end, centralized systems have all or most of decision mak- ing authority concentrated at the central level (usually the Ministry of Health in health systems). At the other end, fully decentralized systems have most decision making authority transferred to the local levels of government (e.g., municipalities or counties). In relation to the third question, measuring the impact of decentralization is much more difficult than categorizing its model, for two main reasons. First, decentraliza- tion is a multi-faceted and complex process that interacts with many factors. Second, it rarely takes place in isolation; most often decentralization is undertaken along with other far reaching reforms, whether political or economical, which makes it quite diffi- cult to identify and measure the impact of decentralization alone. However, identifying Decentralization and Governance in the Ghana Health Sector 11 the challenges and weaknesses of the process in different countries can go a long way in avoiding the pitfalls of decentralization. From the preceding paragraphs, several key policy areas and issues can be identi- �ed in the process of decentralization, and require careful design and implementation. ■ Scope: What activities or responsibilities are to be decentralized (planning, human resources management, budget preparation and/or execution, procure- ment, and other functions)? This depends on a number of factors: economies of scale required for particular functions or activities, local capacity, the cost of building structures and capacities at the local level, and the model of decen- tralization chosen. In the health sector particularly, decentralization does not usually apply equally to all functions; the general rule is to “assign to the lowest level of government possible, those local public goods and services which can best be delivered at that level� (Oates 1972), but this rule—and others—offers li le practical guidance as to which functions exactly should be decentralized to which level. ■ Depth or degree of autonomy: what degree of autonomy should local gov- ernments enjoy in managing the transferred responsibilities? This involves bal- ancing the potential conflict between national policies and planning, and local needs and preferences. ■ Model: What model be er meets a given country’s characteristics and objec- tives (deconcentration, delegation, devolution)? ■ Level: To whom—which level of government or entities—are these functions decentralized? For instance, should responsibilities be decentralized to regions or states, or to municipalities or districts, or to still lower tier levels? ■ National policy and strategy: successful decentralization requires a clear and comprehensive national policy bringing together all aspects and issues of decentralization and guiding its implementation through a clear strategy. ■ Organization of LG units: how are LG units to be structured in order to ensure proper representation of communities and accountability? Organization of LG units needs to balance the quest for greater representation and participation (which assumedly increases as LGs get smaller) with the need for economies of scale; in addition, ethnic or geographic factors may influence the size and boundaries of LGs. ■ Institutional capacity: Are these decentralized entities prepared to take over their new responsibilities? Local capacity involves having sufficient numbers of quali�ed personnel to handle decentralized responsibilities, staff contract- ing and management systems, merit-based remuneration, and corruption control. ■ Financing: How are decentralized services to be funded? What mix of transfers from central government, local revenues and donors’ funds, is appropriate? In most countries, LGs cannot be �nancially self-sufficient, and need to be �nanced largely by transfers from central or regional governments. LGs have different capacities to raise revenues, due to existing legislation (can they establish their own taxes?), technical capacity (do they have the technical, human, and �nancial capacity to set rates, manage revenue collection?), and local population capacity to pay such taxes. A critical issue is whether LGs are given the ability to borrow, and if such ability has some sort of central supervision or control. 12 A World Bank Study ■ Accountability and participation mechanisms: Which accountability mecha- nisms are put in place and what incentives exist for local entities to perform and achieve their goals? To what extent local citizens actually participate in deci- sion making, planning and priority se ing, or management supervision, and through what channels or mechanisms? A key issue here is the mode of select- ing and nominating the LG chief executive, and to whom he is accountable to: in countries like Brazil, LG chief executives are directly elected, while in others, there are appointed by the local legislative or by the central government. ■ Impact: What is the (expected or effective) impact of decentralization on health system objectives (efficiency, cost-control, quality and effectiveness, equity, and access)? The �rst two areas de�ne Bossert’s concept of decision space, and are illustrated in Fig- ure 2.1. Usually, the different modalities of decentralization—deconcentration, delegation, and devolution—involve increasing scope of functions and degree of autonomy (depth). Most of the others areas have been highlighted by Campbell (in United Cities and Local Governments 2009). Another important issue to consider is whether health system decen- tralization takes place in isolation (separately from decentralization in other areas of gov- ernment) or as part of a greater movement of government decentralization, involving political, administrative and �scal decentralization. In Ghana, the second case is true. Experience in Developing Countries The reasons why countries have decentralized responsibilities from central to local governments have varied: it was mostly political in some cases (as part of the political Figure 2.1: Relation between scope, depth, and types of decentralization Depth of decentralization (more power) Devolution Delegation Deconcentration Scope of decentralization (more functions) Source: Author’s elaboration. Decentralization and Governance in the Ghana Health Sector 13 transition to democracy in Latin America), part of general economic and political transfor- mation (as in Eastern Europe, the former Soviet Union, China, and Vietnam), a response to ethnic and regional conflicts (as in South Africa, Indonesia, and Sri Lanka), or a strategy to improve the delivery of health and education services (as in Chile, Uganda and Cote d’Ivoire).2 However, whatever the main motivation for reform, improving social services delivery through strengthening incentives of government agents is often a strong com- ponent of decentralization. In most African countries, decentralization is slowly evolv- ing from a strategy of administrative organization within a basically centralized regime, toward real transfer of decision powers to local authorities (United Cities and Local Governments 2009). In this section we review the health sector decentralization experience of a sample of developing countries, based on the framework developed in the section titled “Con- ceptual framework.� In each case we identify the motivation and context of decentraliza- tion, its main features and, whenever possible, its impact. The main conclusions of this international review are presented in the “Lessons Learned� section below. Uganda As in many former British colonies, British rule in Uganda emphasized decentraliza- tion as a strategy for maintaining control (Dunlop n.d.). But following independence, in 1965 a new government initiated an era of centralized authority and internal con- flict, and it was only in 1993–94 that decentralization was reinstated, when basic health services provision and control over medical personnel were transferred by devolution to 56 local districts (with a mean population of 450,000); the �rst 13 districts began receiving unconditional block grants from the central government (Akin 2001). Central government transfers accounted for the largest part of local governments’ resources (81 percent of total revenue in the mid-1990s) and are transferred through the Ministry of Local Government, with 18.7 percent being unconditional grants and 62.6 percent coming from several conditional grants (e.g., for primary care). Donors contributed 12 percent, and LG revenue 6.5 percent. Overall, district-level spending increased greatly as a proportion of public health expenditure, to represent most of recurrent expenditure except personnel salaries. This system of block grants was fairly advanced among African countries, and dis- tricts could contract with non-public providers, as well as hire and �re staff. However, the actual level of decision power given to local governments was limited by �scal constraints. Funding for salaries and vertical programs were rigid and accounted for the majority of LG resources; only 25 percent of total funding was subject to LG discretion (Bossert and Beauvais 2002). Donors funding, most often linked to particular programs, also allowed limited allocative autonomy. Decentralization somewhat followed the British model, with primary care the respon- sibility of districts and hospitals remaining in MOH hands, to be converted into autono- mous organizations. MOH, the Ministry of Local Government and the Ministry of Finance are jointly responsible for monitoring the performance of districts, and districts officials are held accountable for their district’s performance (especially with respect to improve- ments in people’s health status). Ethnic rivalries and inequalities across regions and ethnic groups posed signi�cant challenges to decentralization. On the other hand, the general desire to move away from the chaos of the 1970s and 1980s, grassroots social cohesion 14 A World Bank Study with clear rights and responsibilities, and economic growth since the mid-1990s helped the process. However, survey evidence indicates a number of problems. Districts’ health systems are not faring well �nancially, as the central government puts a cap on payroll expenditure and locally raised user fees revenues were suspended in 2001 (Dunlop n.d.). Local pro- curement of drugs is difficult, so it remains centralized. Evidence shows that local govern- ments were not transferring budgeted resources to schools, and the central government started an initiative of disseminating budgetary information in the press. New surveys indicated that this strategy had a positive impact (Khemani 2005). Akin’s study of decentralization in Uganda (Akin 2001) has indicated that for the provision of public goods—such as public health services—decentralization of deci- sion power to local governments may in fact be inefficient (from a society’s perspec- tive). This is because when given the possibility to choose, citizens tend to prioritize the provision of goods and services which bene�t them directly (private goods) rather than those which bene�t larger groups or society as a whole (public goods). Sensitive to their citizenry, local governments also tend to prioritize spending on the former. In Uganda, Akin found evidence that local governments spent relatively more on curative, personal services rather than public health and preventive ones. Devolution also did not correct the allocative bias toward urban areas and hospitals. In addition, decentralized provi- sion of public (health) goods tend to be reduced as one district investing in it bene�ts the surrounding districts, which are then discouraged to do the same, in a “free rider� phenomenon. The Brazilian experience (see below) seems to con�rm the �rst �nding: when disease surveillance and other typical public health services were decentralized to municipalities, the system broke down (partly because of weak capacity and expertise at the local level), and had to be recentralized. An interesting feature of the Ugandan experience is that the country invested signif- icantly in capacity-building at the local level, through Local Government Development Program grants, linked to an annual evaluation of performance in good governance— based on a prede�ned set of criteria—that determines whether the districts are eligible for a reward or penalty (Farrant and Clarke n.d.). Nigeria Nigeria has been a federal state since 1954 (with shared responsibilities between the federal and state governments for service provision), but in 1976 Local Government Authorities (LGAs) were established with signi�cant devolution of decision powers. LGAs had the responsibility to provide most local public services, and were jointly �nanced by the central government and the states.3 In 1980, all primary health care provision was made the sole responsibility of LGAs, while states are responsible for training, planning and technical support to LGAs and the MOH is in charge of national policy making and monitoring and evaluation (M&E). The constitution of 1999 strengthened the process by clarifying the dis- tribution of power between the federal government, 36 states and 774 local governments. Government revenue was allocated 24 percent to state governments and 20 percent to LGs. However, the central government retains important authority and influence over subnational governments, and overlapping of responsibilities “creates policy conflicts, duplication of efforts and inefficient use of resources� (Egbenya 2010). Process planning and management have been inadequate, with li le consultation with local stakeholders. Resource constraints prevent LGs to meet their decentralized functions. Decentralization and Governance in the Ghana Health Sector 15 A survey of health facilities and LGAs showed that the vast majority of facility man- agers recognized the LGA as the main decision maker in most management areas— investment decisions, se ing fees, procurement, use of fees revenue, managing and transferring staff—followed by the community and facility management (83, 8.5 and 6.5 percent respectively—see Khemani 2005). Even payroll payment was appointed as the near-exclusive responsibility of LGs (95 percent). However, this apparently clear responsibility of local governments was not without problems: the study also indicated that salary payment was delayed 6 months or more for 42 percent of staff respondents in one state, and only 13 percent of respondents in both states reported no delay (Khe- mani 2005). This indicates a serious issue of �nancial sustainability and/or accountability (associated with possible misallocation of resources) for local governments in Nigeria. The study author concludes that these �ndings challenge the general conventional wis- dom that decentralization helps improving accountability by bringing decision making closer to users and the community. Tanzania Tanzania started its decentralization process in the 1970s, and followed the deconcentra- tion model (Mills et al. 1990). Regional and District Development Commi ees were estab- lished, with the responsibility for planning and implementing development programs. While �eld offices of the MOH had to report to the senior regional or district officer, the main line of authority and overall control remained with MOH headquarters. MOH and local health services were integrated. Later reforms actually devolved increased authority to local governments. They fostered a rede�nition of the roles of the different govern- ment levels: the local level (District Councils) was charged with implementation/execution of most social, security and economic functions, while the regional level was responsible for support and supervision of these councils. However, the weakness of capacity at the regional level (including in terms of human resources) prevented the regions to play a signi�cant role in the process. The management of teaching hospitals has been delegated to an autonomous entity. LGs were funded by a health block grant with a clear rule for distributing funds across districts: 70 percent is based on population size, 10 percent on the poverty count, 10 percent on district vehicle routes, and 10 percent on under-�ve mortality. This formula-based allocation eliminated the need for budget negotiation among govern- ment levels and improved �scal predictability at the local level. Overall, the MOH spends nearly 50 percent of recurrent health expenditure (and most of investment spending), against 40 percent for LGs and 6 percent for the regions. However, �nancial management and use of resources at the local level remains an issue: about 10 percent of LG expenditures during the period 1999–2002 have been questioned, most of which because expenditures were not vouchered or improperly vouchered. Another impor- tant problem is the inability of districts to implement the plans due to lack of resources and capacity at the local level (Farrant and Clarke n.d.). Zambia The main motivation for Zambia’s decentralization process was to improve “local par- ticipation and the extension of democratic values to health service development and management� (Lake et al. 2000: 39). The guiding policy document, National Health Policies and Strategies (NHPS) of 1992, de�ned as the main goal to provide “equity of 16 A World Bank Study access to cost-effective, quality health care as close to the family as possible.� In a pro- cess similar to Ghana, Zambia’s Ministry of Health delegated operational responsibili- ties to an autonomous entity, the Central Board of Health, and retained policy-making and regulatory functions; as a result, its staff was reduced from 400 to 67 (Fran et al. 2004). The Board, in turn deconcentrated responsibilities to regional, district and hos- pital boards through annual contracts. Districts were seen as the key government units for implementation and management of health service delivery. Autonomous District Health Boards were established for that purpose, with the responsibility to oversee Dis- trict Health Management Teams. Larger hospitals were managed by autonomous Hospi- tal Management Boards, but these boards were appointed by MOH, and hospitals plans and budgets had to be approved by the Central and District Health Boards, thus limiting the real autonomy of hospitals. District boards had limited managerial autonomy. They were responsible for developing and managing budgets and plans within centrally de�ned guidelines and allocation ranges, and with approval from the Central Board. Most procurement was centralized, though districts could purchase supplies occasionally. The reform was supported by Cooperating Partners (as donors are known in the country) through a Sector-Wide Approach program, with a special focus on the implementation of a dis- trict health basket fund to cover district health recurrent costs and essential program expenditures. As a result, the proportion of total public funding directed to the districts greatly increased, from 5.7 percent in 1991 to 52 percent in 1999 (Lake et al. 2000), and �nancing predictability and targeting to poor districts improved. Budget transfers were conditional on performance as audited by Regional Directorates. The impact of the reform was limited in the �rst years, and the reform was nearly aborted for that reason. While some indicators have improved, especially in the late 1990s and early 2000s (including infant mortality, malaria deaths, immunization cov- erage), others have remained stagnant or deteriorated (including maternal mortality, STI incidence, supervised deliveries). These less-than-expected outcomes are mostly due to bo lenecks such as low quality referral hospitals, inadequate drug supplies, human resources and capital expenditures. An additional issue is one of sustainability: the increase in the basket funding has come almost exclusively from donors increased participation, while government funding has stagnated. A shortage of human resources, and a disconnect between health reform and the general government decentralization process, constitute other signi�cant challenges. Philippines By the mid-1980s the Philippines had a centralized health system based on a network of some 2,000 Rural Health Units providing primary care, but very fragmented and weakly coordinated. It showed relatively poor results relative to its neighbors and important regional inequities. Aimed mostly at improving the effectiveness of the health system, the sweeping reform of 1991–93 devolved funding, all primary care facilities, 600 hos- pitals and 46,000 staff to some 1,600 Local Government Units (LGUs) (77 provinces, 60 autonomous cities and 1,548 municipalities). The Philippine Department of Health (DOH) was responsible for monitoring and supporting LGUs, and retained control of most larger hospitals, some of which were converted into autonomous units or priva- tized. Funding was devolved, mostly through an Internal Revenue Allotment from the central government. Fiscal decentralization was signi�cant: the proportion of LGU Decentralization and Governance in the Ghana Health Sector 17 spending increased from 20 percent of government revenue before decentralization to 40 percent within three years; health accounted for two-thirds of LGU expenditure (Bossert and Beauvais, 2002). In spite of this large scale devolution, a recent review pointed to less than expected impacts of decentralization. Compared to the African countries reviewed above, Philippine LGUs have much broader real autonomy, and much less constraints in planning, allocating resources, and managing operations. The major constraint was the requirement that LGUs maintained the devolved staff and applied the uniform national pay scale. Given the salary increase and bene�ts mandated by the Magna Carta for Health Care Workers of 1992, and the gradual transfer of staff funding responsibility to LGUs, this implied a signi�cant and increasing �nancial constraint. The decentralization process and the parallel reforms (such as the hospital reform) have been challenged all along by civil servants unions, and a strong medical estab- lishment. The reform seemed to further increase system fragmentation, as LGUs had ample autonomy but li le managerial and �nancial capacity (Philippines 2000). Further, they were mostly very small to build signi�cantly their capacity (the average population size is 38,000). An important part of LGUs’ responsibilities were in fact carried out by regional offices of DOH. In order to reduce fragmentation and improve coordination, health care agreements were signed between DOH and LGUs, and some small LGUs pooled together to form “Inter-Local Health Zones�. However, the issue of low mana- gerial capacity at the local level, and weak monitoring and evaluation capacity at the regional and central levels, are still pending issues. Transfer of staff was made possible by passing a generous set of bene�ts which later challenged economic sustainability. Finally, the budget transfers to LGUs did not keep up with the expansion of devolved functions. Successive policy documents, starting with the 2000 Health Sector Reform Agenda, have addressed these issues, with mixed results. Indonesia As in the Philippines, the main motivation for the Indonesian health decentralization reform was dissatisfaction with the system’s performance. Initial reforms, focused on improving incentives and control mechanisms, was not very successful. The Indonesian decentralization experience is noteworthy for its “big bang� approach, which involved transferring several million public servants to local authorities. Its basic laws were passed in 1999 and implemented by 2001. The process was coordinated by a Decentralization Unit (DU) within MOH, and other MOH departments (Directorates General—DGs) were turned into specialized supervisory and supporting agencies. National Health Grants were established to support poorer regions. Districts have ample budget authority, and do not need to refer to MOH or central government to set fees and de�ne their budget size and composition, even though a large part of their budget comes from intergovernmental transfers (Kruse et al. 2009). However, local civil servants employment rules are deter- mined by the central government. The process did not happen without difficulties. During early implementation the DU failed in leading the process, and the DGs provided li le technical support to prov- inces and districts. Overall, decentralized levels have been more proactive than MOH, which has been unclear about its new role and responsibilities (Lieberman n.d.). Deal- ing with staff-related issues—including overstaffing, low productivity, and incentives for transfer—has been a signi�cant issue. Finally, local community engagement and 18 A World Bank Study responsibility implications of decentralization have not been clearly acknowledged and followed up. Brazil Brazil is the most decentralized country in Latin America, and maybe a good exam- ple of health decentralization gone too far. It is a federal country with 27 states and 5,600 municipalities, most of them small (the average municipal population is 35,000). Prior to the 1980s, the system was very fragmented. The MOH was mostly responsible for public health activities (such as immunization campaigns and disease control), while the Ministry of Social Security provided curative services to employees of the formal sector and their dependents, through its own facilities and a large number of contracted private providers. Many states and some municipalities (mostly the larger ones) had their own facility network and provided health services mostly to the poor. Private providers provided care to groups covered by private insurance schemes, and public corporations and organizations had their own insurance scheme (such as the army and civil servants). There was no coordination between government levels or entities, or clear distribution of responsibilities among sector players. In the mid-1980s, following 15 years of military regime, the country initiated a process of re-democratization, which in the health sector translated into a signi�cant reform of the public health sector. This sanitary reform was characterized by (i) de�ni- tion of roles and responsibilities in the public sector; (ii) establishment of coordinating mechanisms across levels of government (vertical coordination) and across political enti- ties at the same level (horizontal coordination) as well as mechanisms to ensure social participation; (iii) decentralization of responsibilities for primary and secondary care to municipalities; and (iv) joint �nancing of the new “Uni�ed Health System� by the three levels of government, in a large part through matching grants. The decentralization model chosen was one of devolution to municipalities, which were turned into the main institutional level for public health care delivery. This transfer to municipalities was made possible by an important transfer of �nancial resources to local governments as part of the political decentralization process (Fleury et al. 1999). However, decentralization was not complete (some functions were not decentralized) or even (because too many Brazilian municipalities are too small and have limited capac- ity).4 Functions that were decentralized included primary and most secondary care, but the MOH—which in the process incorporated the health care arm of Social Security— retained many of its hospitals. Disease surveillance and other public health functions were partially decentralized, but the MOH retained control over several vertical, disease- centered programs. Under the new system, the MOH is responsible for de�ning national health policies, ensuring the overall coordination of the system, and is the main �nancer (with around 50 percent of the total), while states contribute 24 percent and municipali- ties 26 percent. Because of this co-�nancing arrangement, the MOH retained an important leverage over state and municipal secretariats of health, which by constitutional mandate are fully autonomous. To address the issue of small size and different capacities, two different speeds and statuses were de�ned in the process of decentralization. In addition, a number of munic- ipalities have joined efforts and established “municipal consortia� (groupings of several municipalities for a speci�c purpose) to gain economies of scale in the management of their local health systems. But these initiatives have not resolved the problem of small Decentralization and Governance in the Ghana Health Sector 19 scale and capacity. The pace and form of the decentralization process also produced important problems. “At the time of the handover, municipalities were neither adminis- tratively nor technically prepared to handle the responsibilities of managing health care services. States were, however, and this led to municipal bureaucratic inefficiencies and a renewed interest by governors in taking back these administrative functions and with- holding state-level support for much-needed administrative reforms� (Gómez 2008). Recent legislation mandates that the three levels of government spend at least a given proportion of their total revenue on health (15 percent for municipalities, 12 percent for the states, and for the federal government a rule ensuring that health expenditure will increase at the pace of the gross domestic product—GDP). However, many government entities have been shown to shirk on these requirements. The de�nition of allocation criteria for federal transfers was always a major issue, one that remains hotly discussed. Federal transfers were split in two broad grants. The �rst was for primary care, which was partly based on a �xed per capita amount per year, and partly (20 percent) on municipalities performance in terms of expanding coverage and implementing speci�c national health programs (such as the Family Health Program, Disease surveillance program, and others). The other part, aimed at �nancing specialized and hospital care, was mostly based on the volume of services produced. Overall, nearly 100 different payment mechanisms were de�ned, each with its parallel planning and account rendering requirements. With respect to accountability mechanisms, to accommodate the political autonomy inherent of a federal state, and the co-�nancing arrangement, a complex set of negotiating councils and instruments was put in place, including mechanisms for coordinated plan- ning, budgeting and account rendering. Health councils have been established at each level of government and a ached to each health secretariat (totaling nearly 6,000 over- all). In larger municipalities, local (submunicipal) councils have also been established. The reform and its main decentralizing feature has had some positive impacts: it expanded coverage to the whole population, it contributed to reducing inequalities in resource allocation, it de�ned roles and responsibilities (although these still need to be further clari�ed), it established channels for community participation at all levels, and improved coordination between the different types and levels of care. The main limita- tions of the process include: high transaction costs due to the complex and multi-layered system, li le impact on efficiency (which remains very low system-wise), incomplete decentralization in the sense that public facilities were not given the necessary autonomy to function adequately, and li le improvement in the area of public governance. In addi- tion, decentralization did not contribute to management autonomy, since all govern- ment levels are subject to the same public sector regulations regarding civil servants and procurement. Colombia In Colombia, decentralization toward state and municipal governments began in the late 1980s, and was passed into Law 10 of 1990. It was aimed at improving the efficiency and effectiveness of government programs, increase coverage, accessibility and quality of public services. The reform was deepened by the Law on Social Security for Health (Law 100) of 1993. The Colombian reform included systematic contracting with private pro- viders through management contracts, and privatization of some providers. Evidence gathered by Molina (2009) showed a number of difficulties faced by the decentralization process: lack or fluctuating political will in support of the process, quality of staff linked 20 A World Bank Study with political patronage, weak information systems and administrative processes, poor institutional capacity and real autonomy, imbalance in the distribution and allocation of health resources (with, in some cases, an expenditure shift from primary care to curative care), difficulties in effectively targeting the poor and improving their access to care, limited role of and support from national and provincial health authorities, weak com- munity participation. Political patronage at the local level increased signi�cantly. Mexico and Argentina In Mexico and Argentina, the decentralization process was similar in several respects. First, in both cases the main motivation was to improve �scal imbalances by raising rev- enues at the decentralized levels. Second, authority was devolved to the provincial/state level rather than the local level. Third, decentralization was partial and fragmented, since it focused on certain programs and activities, while coverage extension and other health programs were devolved separately under the umbrella of a centralized social security institution (in Brazil, the health care branch of Social Security was merged into the MOH and the new system). Fourth, the role of the central health authority was greatly reduced in the process, with few control, normative or regulative powers, and weak steering capacity, leaving it to each state to develop its own brand of decentralized health system (contrary to Brazil where decentralization gave rise to the Uni�ed Health System). In Mexico, the health system reform and decentralization began in 1983–88, and was followed by a second wave in 1994–98, which expanded decentralization to all states, with a focus on new �nancing mechanisms and increased social participation. The operation of social services was delegated to “decentralized organizations�, in fact structured as autonomous entities with their own legal status, governing boards, and funding sources (Mills 1990). Arredondo and Orozco (2006) found that the reform increased diversi�- cation of �nancing sources, including of locally raised funds. This however had some negative effects on equity given the increase in user fees. The authors also found limited opportunities for participation in decision making and the establishment of health pri- orities, poor information on the decentralization process among health care providers and other stakeholders, and unclear lines of authority. Control mechanisms set in place maintained a signi�cant level of authority in the hands of the central government. In Argentina, a parallel reform followed a more liberal orientation: the management of some hospitals was privatized (Obras Sociales), and several public health functions— regulating, monitoring and auditing health care delivery—was delegated to an autono- mous agency, the National Health Services Authority (Superintendencia Nacional de Servicios de Salud). Province �nancing comes from the Fiscal Pact arrangement, and they have ample autonomy for its allocation. In the mid-1990s, 71 percent of these funds went to provincial governments, 16 percent to municipal governments, and 13 percent to the MOH (Fleury et al. 1999). An important weakness of the reform was that it had no design for the future, allowing for fragmentation and differentiation of the process across provinces. Lessons Learned The international experience shows that decentralization is a complex process, which reflects each country’s administrative, political, geographic, demographic, and—in this case—health sector characteristics. Because of this wide variation, it is uneasy to compare the features and impacts of decentralization across countries, and most comparisons only take into account a few features or dimensions. A useful exercise was provided by Decentralization and Governance in the Ghana Health Sector 21 Bossert and Beauvais (2002). Applying the concept of decision space, the authors looked at the decentralization processes in four developing countries from Africa and Asia, to map out the level of decision space in several health functions. Table 2.1 summarizes the �ndings of the study. The conclusion is that compared to the other countries— Uganda, Zambia, and the Philippines—Ghana shows a narrow decision space with respect to all functions and sub-functions considered, except in �nancing. In this area, Ghana enjoys a moderate diversity of funding sources, relative freedom at the local level for charging and using income from user fees, and executing expenditures. For all other functions, local governments in Uganda, Zambia, and especially the Philippines, enjoy much greater decision space than in Ghana.5 This suggests that in spite of the progress accomplished since independence and the several waves of decentralization reforms, Ghana’s public sector remains less decentralized than comparable countries. However, the authors point out the lack of information and evidence to properly assess the impact of the different approaches. Table 2.1: Mapping the decision space: Ghana, Philippines, Uganda, and Zambia Degree of Decision Space Function Narrow Moderate Wide Financing • Sources of revenue Zambia Ghana, Uganda Philippines • Expenditures All four Philippines • Income from fees Ghana, Uganda, Zambia Service organization • Hospital autonomy Ghana, Zambia Uganda Philippines • Insurance plans Ghana, Uganda Philippines Philippines, Zambia • Payment mechanisms Ghana, Uganda Zambia Zambia • Contracts with private providers Uganda Human resources • Salaries All four Philippines • Contracts Ghana Philippines, Uganda, Zambia Uganda, Zambia • Civil service Ghana Access rules Ghana Philippines, Uganda, Zambia Governance • Local government Ghana, Zambia Philippines, Uganda • Facility boards All four • Health of�ces Ghana, Philippines Uganda, Zambia • Community participation Ghana, Uganda Philippines, Zambia Source: Bossert and Beauvais (2002). According to the review above, the impact of decentralization has been mixed, especially in developing countries (Ahmad et al. 2005). The main �ndings are summa- rized below, structured according to the main policy areas identi�ed in the conceptual framework. Model and Approach ■ Countries undergoing decentralization have adopted a wide variety of models and forms of decentralization, which often mixed the four theoretical models described above (deconcentration, delegation, devolution, and privatization); Bossert (2000) has shown that assessing the amount of choice given to speci�c 22 A World Bank Study functions—in other words, the decision space—is more important for character- izing decentralization processes than the traditional administrative approach of decentralization models. ■ Decentralization is not always a good thing, and can be counterproductive if structural problems relating to governance, management and �nance are not addressed properly. ■ A “one-size-�ts-all� approach to decentralization is unlikely to be successful, especially when local realities are very different (in terms of size, management capacity, and cultural preferences). Level and Organization of Local Governments ■ Countries have decentralized to different levels of government: states or provinces in some federal countries (Canada, Australia, Argentina, Mexico, and Ethiopia), municipalities or counties in others (Philippines, Uganda, Colombia, and Scandi- navian countries), or both (Spain and Brazil6); others yet have retained a central- ized system with some decentralized functions (United Kingdom). ■ The number of decentralized units has often been excessive and the size of such units too small; when local units are too small, or in large metropolitan areas, lower-tier LGs may partner or amalgamate in larger units to achieve economies of scale and improve efficiency. Policy and Strategy ■ Many countries undertaking decentralization lack a comprehensive national policy on decentralization and a clear plan or strategy for its implementation; in many countries, decentralization policies and regulations are fragmented and uncoordinated, sometimes conflicting. Many of the countries reviewed in United Cities and Local Governments, 2009 had not developed a decentralization plan, and among those who did, like Bolivia, South Africa, and Indonesia, this plan was not comprehensive and lacked a clear implementation strategy; the lack of a clear vision and strategic planning contributes to poor information and limited support from stakeholders. ■ In many cases, clear information on the process objectives and strategies is not passed on to key stakeholders, further leading to resistance to change; change management is a key element of the process but has often been overlooked. ■ Decentralization can be a highly disruptive process, which can easily disorganize already fragile health systems and institutions, and cause breakdowns in service delivery (Bossert and Beauvais 2002); these disruptions have often resulted in backlash against decentralization reforms, and limiting them requires carefully planning and a clear strategy for implementation. Roles and Decision Space ■ Clear de�nition of responsibilities and roles across government levels and insti- tutions is crucial to effective decentralization, but oftentimes these responsibili- ties are unclear or blurred, or may change over time; decentralization is often weakened by conflicting or unclear distribution of responsibilities, associated in many cases with an incomplete process (Pakistan is one example); this in turn results in either duplication of services, or absence of supply when different levels of government fail to deliver in face of unclear regulations. Decentralization and Governance in the Ghana Health Sector 23 ■ Formal (legal) decentralization can be quite different from the actual auton- omy local authorities enjoy, and this may contribute to less than expected impacts; in many cases, decentralization is undermined by centrally de�ned policies, constraints and rules, which restrict the effective degree of auton- omy of local governments; examples include payment for civil servants of local governments (which tend to account for the major part of local health expenditures), rules on procurement of critical inputs (such as drugs), rigid and standardized public sector regulations (e.g., for procurement or staff hir- ing), centrally de�ned budget ceilings, or limiting conditions on transferred funds; these constraints greatly reduce the actual autonomy of LGs and, thus, the effectiveness of the process. ■ The functions and responsibilities decentralized to LGs vary widely across countries, from judiciary, pension and economic development in large Chinese cities, to a more typical set of core local services, including water and sanita- tion, waste management, streets maintenance, urban planning, primary care and primary education, in many developing countries; the degree of autonomy enjoyed by LGs also varies a lot across countries. ■ Not all functions work well when decentralized, and decentralization may have different effects depending on the particular functions decentralized; broader decision space in planning and �nancing has been shown to be associated with be er performance at the local level, while broader decision space in procure- ment, supply management, information systems, training and client contact was associated with poorer performance in Ghana (Bossert and Amenyah 2004); in Brazil, devolution of disease surveillance and immunization had negative effects and these functions were later recentralized. ■ Decentralization involves difficult balancing acts; harmonizing and balanc- ing local government planning process with national planning priorities represents a signi�cant challenge; quite often, a so-called decentralized or bo om-up planning and budgeting process ends up being li le more than an endorsement by local governments of centrally de�ned objectives, policies and priorities. Local Capacity ■ Lack of management and �nancial capacity of subnational governments has hampered decentralization in many countries, and reduced its effec- tiveness and impact (Uganda, Tanzania, and Ethiopia are some examples); decentralization only makes sense when the decentralized entities have a sufficient scale and capacity; although capacity can to some extent be built during the process, signi�cant capacity is required before transfer of powers can be effected. ■ Managerial capacity is not required from local governments only; experience shows that signi�cant capacity has to be built at regional levels (when respon- sible for supervision and coordination of local governments), and at the cen- tral level (process design and strategy, policy making, and M&E); where these capacities were not addressed appropriately, the decentralization process suffered. ■ A crucial aspect of decentralization is the changing role of central ministries and—depending on the level of decentralization—regional authorities; in many 24 A World Bank Study countries, MOH and/or regional health authorities faced great difficulty in adapting to their new roles and functions, and “got lost� in the process; addi- tionally, there is a more conceptual dilemma on the role of central government (MOH): to induce local governments to adopt or follow national policies or pri- orities (through matching or conditional grants for instance, which in Brazil is seen as a rest of centralism) or allow full autonomy at local entities and risk a greater fragmentation of the system? Transparency and Accountability ■ Even though decentralization is expected to increase accountability and trans- parency, this is not a natural effect of the process; in fact, experience has shown that in many cases accountability—especially �scal/�nancial—has deteriorated in the process; improved accountability requires that speci�c mechanisms be imbedded in the design of decentralization. ■ Decentralization usually improves accountability to local communities, and can increase social participation in decision making and priority se ing if proper mechanisms are in place; in a number of countries, weak community participa- tion and transparency has been a frustrating result of decentralization, often because of the capture of the devolved autonomy by local leaders or interest groups (as the Indonesian experience has shown). ■ Weak accountability and control of local governments, coupled with �scal autonomy, have in some cases led to over-borrowing and threatened �nancial stability and sustainability (as in Argentina and Brazil, where the federal gov- ernment ends up “picking the tab�). ■ Limited local accountability and transparency may greatly reduce the advantages of decentralization (as Farrant and Clarke have shown in Tanzania, Ethiopia, Rwanda, and Ghana). ■ Information dissemination (on budget, expenditure, planning priorities) is key to improving accountability of local governments, both downward (to local communities) and upward (to central government); strong information systems are thus critical. Financing ■ Financial autonomy is key to ensure effective managerial autonomy, and thus the impact of decentralization; decentralizing executive responsibilities without the proper �nancial autonomy usually greatly reduces the bene�ts of decentral- ization; in other words, “money needs to follow functions�; this is especially true in the case of devolution. ■ In most developing countries (and many developed ones), local govern- ments do not have the capacity to be �nancially independent and �nance the decentralized functions through their own resources; intergovernmental transfers from the central government usually accounts for more than two- thirds of local revenues; locally generated taxes and user charges accounted around year 2000 for 23 percent in Uganda to 36 percent in Ghana (Farrant and Clarke n.d.). ■ The design of appropriate mechanisms for transferring central funds to local governments is an important issue in decentralization; the establishment of conditional or matching grants is often a means for MOH to retain leverage for Decentralization and Governance in the Ghana Health Sector 25 inducing implementation of national policies and priorities (such as pro-poor programs) and ensure allocative efficiency, but they may also end up reducing local governments’ autonomy. Evidence from Uganda shows that once transfers were made conditional on certain conditions, the pro-poor orientation of local expenditure increased (Farrant and Clarke n.d.); in Brazil, a complex system of conditional grants ensures consistency of health policies across the country— but at the expense of reduced focus on local needs and priorities. Also, multiple payment mechanisms (27 in Uganda, 100 in Brazil) impose a signi�cant mana- gerial burden on local governments, since each mechanism usually has its own reporting and accounting requirements. ■ Leakages of centrally transferred funds at the local level are common; in Brazil for instance, many municipalities use transfers earmarked for health services or particular health programs for other ends (such as sanitation, road maintenance and others); similar �ndings were encountered in Uganda (Akin et al., 2001) and other countries; these leakages demonstrate the inadequacy or insufficiency of existing accountability mechanisms. ■ In some countries decentralization was adopted as a strategy for mobilizing local revenues (because local communities tend to be more willing to pay taxes or levies if these are used for local activities and services); international experi- ence shows that decentralization has often resulted in increasing local revenues, whether through local taxes or user charges; but this was often balanced by a reduction in central government �nancing (as in Brazil), so that the overall health envelope was not changed. ■ In several countries, the increase in local revenues through user charges has had a negative impact on equity. Human Resources ■ Human resources are a critical factor in decentralization processes, for different reasons; �rstly, a key issue in decentralization is the status and hiring regime of health workers under LGs; different countries have adopted different approaches: uniform centralized national civil service (as in Papua New Guinea and the Phil- ippines); decentralized civil service (as in Uganda); mixed models, with central civil servants being transferred under their original regime but new employees being hired under LGs’ own regime (as in Brazil and Jamaica); separation from national civil service, where decentralized units are staffed under a new regime (as in Ghana) and old civil servants are transferred but “delinked� from the central civil service (as in Zambia). ■ Second, civil servants often feel threatened by decentralization, since it often means the transfer of large numbers of staff to local governments (62,000 in the Philippines, or 62 percent of the Department of Health’s original staff) and the risk of losing acquired bene�ts; they have often mobilized against it (as in the Philip- pines and Uganda); they are a key stakeholder in the process and can derail it if they don’t buy in. ■ Third, human resources are the key element on which to build capacity at each level of government; even though most countries have provided some training to local government or transferred staff, general information on the process objec- tives and strategies is often insufficiently or not disseminated to civil servants. 26 A World Bank Study Impact of Decentralization ■ Decentralization in many countries has been associated with unrealistic objec- tives and expectations, which tend to lead to frustrating results; the expected advantages of decentralization do not flow naturally or necessarily from the process, but actually depend on building into it speci�c elements and mecha- nisms (for accountability, �nancing, participation, and others as discussed above) to ensure these results. ■ Formal decentralization is no guarantee of improvements in efficiency, account- ability, quality or equity, due to intervening factors either not under the control of the decentralizing policy or authority, or not taken into account in the process design and implementation: some examples include varying local characteris- tics, resistance to change, poor community participation, weak implementation, low capacity at the local level, low M&E capacity at the central level, inadequate incentives structure; for instance, decentralization can contribute to increased efficiency only insofar as the decentralized entities have the incentives, auton- omy and capacity to perform more efficiently. ■ Decentralization may hinder the reduction of inequalities or even increase them, since be er equipped and capable local governments are likely to make be er use of resources (both transferred from central government and their own), have greater capacity to increase their revenues, and may also have eas- ier access to central funds because of their relative success; therefore, a clear and transparent set of criteria is needed to de�ne allocation of central gov- ernment funds and transfers to the different local governments, in order to promote greater equity. In some countries this has been achieved through the establishment of equalization funds. ■ Decentralization usually implies increased costs in three categories, which may or may not be balanced by gains in efficiency and cost control: (i) infrastruc- ture development (buildings, equipments and vehicles), as local governments tend—or need—to build infrastructure for the newly decentralized services; (ii) administrative, as local health offices need to be set up or strengthened to take care of the new responsibilities; this may imply replicating in each local government a unit similar—but smaller—than the central MOH; and (iii) staff, as local governments need to hire new staff to take care of decentralized responsibilities while personnel working at the central level may be reluctant to be transferred to the provinces. The additional cost in infrastructure was very clear in Brazil, where municipalities, which became the main public provider and spender of health services, started build- ing a large number of small hospitals to respond to their constituency’s expectations. During the 1980s and 1990s, over 1,500 hospitals were opened by municipalities, most of them very small (average size, 17 beds).7 Political interests (“me-too� new hospi- tals), policies aiming at improving access to hospital care were the main factors in this process. After more than twenty years of building hospitals, the government now realizes that most of them are too small to operate efficiently or offer good quality services, and is �guring out a strategy to close or transform them into ambulatory referral units. The main advantages and disadvantages of decentralization and its dif- ferent modalities are summarized in table 2.2. Decentralization and Governance in the Ghana Health Sector 27 Table 2.2: Main advantages and disadvantages of the modalities of decentralization Decentralization Model Advantages Disadvantages Decentralization (relative Effective way to meet local needs May increase regional disparities to centralization) Greater ef�ciency, lower costs Possible capture by local elites Better targeting of the poor Weaker central government More equitable spatial distribution of May weaken accountability resources Improved access to services High administrative costs (new bureaucracies, of�ce space, staff, equipment, posts) Reduced congestion at the center Local councils may not feel accountable to local communities if they are not elected Increased accountability to local Weaker control of �scal policy and �nancial community management Greater participation, civic consciousness Mobilization of local resources/ Willingness to pay for services More effective coordination, reduced duplications Reduced patronage from center Deconcentration MOH leadership in policy making and Decision making afar from users coordination Clear authority lines Low community engagement Often appointed, unelected of�cials Policies and services may not meet local realities and needs Lower accountability to community Control and constraints of local autonomy by central government Delegation Separation between policy/�nancing Limited community involvement and provision/operational functions Clear lines of authority and Decision power can be centralized at accountability headquarters Easier coordination Devolution Decision making closer to users Weak enforcement of national policies Potentially greater ef�ciency in Dif�cult for MOH to adapt to new role resource use (if local level has the capacity) Clearer rules for resource allocation Unclear role of regional levels More dif�cult to control and monitor, Higher risk of fragmentation Cost of replicating administrative units Sources: Author’s elaboration. Note: Both advantages and disadvantages are expected and potential, but do not necessarily materialize in any particular experience. 28 A World Bank Study In summary, if decentralization is to achieve certain objectives, speci�c elements and conditions need to be built into the process design and implementation. Bahl (n.d.) has indicated key requirements for �scal decentralization to be successful, but these requirements to some extent also condition the success and impact of decentralization as a whole; they are: “(i) enough skilled labor, access to materials, and capital to expand public service delivery when desired, (ii) an efficient tax administration, (iii) taxing power sufficient to capture signi�cant portions of community income increments, (iv) an income-elastic demand for public services, (v) popularly elected local officials, and (vi) some local discretion in shaping the budget and se ing the tax rate�. These condi- tions are not present in many developing and transition countries, and wherever they exist, they are mostly found in the wealthier provinces. As noted by Bossert and Beauvais (2002), “the issue is not whether or not to decen- tralize, but rather how to design and implement be er decentralization policies to achieve national health policy objectives�. In addition, it is important to note that, empir- ically, decentralization implies a complex balancing act regarding different trade-offs: central authority for policy making and planning vs. local autonomy, vertical programs vs. integrated care at local level, “equal� allocation of resources vs. equity promotion through “unequal� allocation. Different countries have addressed these issues in quite different ways. There is thus no such thing as an ideal model or arrangement for decentralization. The choice that a country makes depends on political, historical, cultural, and economic factors that are peculiar to that country. A model that works well for one country may not work as well for another. However, experience suggests that once the model is cho- sen, its success depends on the correct balance and arrangements between the key ele- ments of decentralization. In other words, the success of decentralization depends more on the consistency between objectives, design and implementation of the process, than on the theoretical model itself. According to Bossert (2000), there is “reason to believe that it is not so much decen- tralization itself but how decentralization is designed and implemented, that will make the difference in equity, efficiency, quality and �nancial soundness�. Once a decentraliza- tion policy is designed and passed, its implementation and impact depend on a number of factors often unrelated to the policy itself, and relating especially to the way local man- agers take advantage—or not—of the broader autonomy given to them, in other words, how they respond to the new reality; they may choose to conduct things as usual, or they may decide to innovate, that is to say, doing new things or doing things in a different way they used to prior to decentralization. An issue here is how to measure the effect of decentralization, as opposed to changes introduced by the central government inde- pendently of it. The behavior of local managers in face of their decentralized powers will in turn produce—or not—the expected effects on efficiency, quality, access, and overall performance. In that sense it is important to distinguish formal and actual authority. Local Gov- ernments’ autonomy is often limited by �scal, policy and budget constraints de�ned by the central government. Bach et al. (2009) devised a “scope of spending power� indicator to assess the real degree of �scal autonomy local governments enjoy. The key idea is to measure the latitude LGs have in changing public expenditure at their level. The indica- tor takes into account �ve main areas of autonomy: policy autonomy (the authority to de�ne local policies and such aspects as the set of services provided), budget autonomy Decentralization and Governance in the Ghana Health Sector 29 (related to authority for budget allocation against earmarked grants or spending limits), input autonomy (control over the different inputs used in service provision, including staff, contracting, etc), output autonomy (control over the type, quantity and quality of services), and autonomy over M&E (the ability to set goals and standards and evaluate performance). As a result, decentralization can help improve the performance of health systems in some cases, but in others it “has been linked with increasing bureaucratic costs, inef- �ciency and political manipulation. . . . It has also been associated with increasing orga- nizational fragmentation and constraints on the development of national health policy objectives and strategic planning� (Collins et al. 2000). Table 2.3 summarizes the expected effects of decentralization and its risks. Table 2.3: Expected impacts and risks in decentralization Area Expected Impact But . . . Financing (resource Increased because of �scal autonomy and Heavy dependence on central transfers mobilization) willingness to pay for local services Limited by the income level of local population Ef�ciency and costs Increased due to closer/better control by Possible weak participation and oversight communities by community Increased by reduced layers of Risk of capture by local elite bureaucracy Increased due to greater cost- Local units may replicate central consciousness bureaucracy and rigid rules May decrease due to weak controls Management Autonomy Increased “by de�nition� May be greatly limited by central controls, policies and requirements; Depends on model of decentralization and actual decision space Accountability and Increased because closer to users and Risk of capture by local elite or interest groups transparency community participation Possibly ineffective local oversight Weak accountability mechanisms toward the center Needs strong information systems Equity in resource allocation Within LGs: Increased because closer to Possibility of resource capture by local and access population elite Across LGs: increased because clearer Depends on clear and proactive policy and allocation rules mechanisms May increase because LGs have different capacities Quality and integration Increased because local pressure and Districts have very different capacities and of care oversight weaker ones may lose out Increased because easier to coordinate More dif�cult to coordinate across levels horizontally of care Risk of fragmentation: LGs as islands Health status Increased because services are better Depends on factors beyond LG control focused on local needs and priorities Allocation may prioritize curative services rather than public health Source: Author’s elaboration. 30 A World Bank Study Notes 1. PHRPlus, September 2002, and United Cities and Local Governments, 2008. 2. Ahmad et al. 2005. 3. Between 1999 and 2002 LGAs were entitled to 20 percent of total revenue collected by the central government (the main �nancer of public services in Nigeria), but a signi�cant propor- tion is “deducted at source� for paying teachers of primary schools and other LGA-mandated expenditures. 4. An important and persistent problem in the Brazilian “municipalization� process is the small population size—and thus, the limited managerial and �nancial capacity: 81 percent of the coun- try’s municipalities have a population of under 30,000, and 46 percent have less than 10,000. 5. In the case of Ghana, the study looked at the main decentralization feature, which is the delega- tion of operational authority from MOH to GHS and the deconcentration of GHS toward district- level offices. 6. Even though devolution to municipalities is the main feature of the Brazilian reform, the state governments also have a constitutional mandate in the health system (mostly regional coordina- tion and referral systems), and �nancing is a joint responsibility. 7. Brazil has currently 7,400 hospitals, of which 1,800 are municipal, and 400 are owned by MOH or state governments; but most private hospitals care for “public� patients under a contract with government. CHAPTER 3 Regulatory and Policy Framework on Decentralization in Ghana S ince its �rst a empts at decentralization, Ghana has oscillated between two differ- ent models of decentralization: delegation and devolution. While some legal docu- ments de�ne the Ghanaian process as one of devolution to districts, the MOH itself has delegated the responsibility of managing its facility network to an autonomous entity, GHS. This section briefly reviews the main elements of the past and current decentral- ization legislative and regulatory framework in Ghana, focusing on (i) general regula- tions regarding GOG decentralization; and (ii) health-speci�c regulations and policies. General Legal Regulatory Framework1 Ghana has a relatively long history of decentralization, which goes back to colonial times, when it was characterized by a dual administrative system: a central government author- ity acting through its regional and local branches, and local government agencies based on—usually small—local authorities (councils). The move to decentralization since inde- pendence took place in several waves of different intensities, usually linked to particular governments or political regimes (Ahwoi, 2010); the main legislative and regulatory pieces of each wave are listed and summarized in Annex I. ■ First wave: 1957–1981—Initial steps ■ Second wave: 1982–87—Reflection and preparation ■ Third wave: 1988–1999—Legal basis for current policy ■ Fourth wave: 2000–present—Implementation2 The �rst wave of decentralization reforms started with independence and was marked with conflicting movements back and forth. Its major thrust was the set of legal docu- ments produced in 1972–74. It focused initially on regionalization (Constitutions of 1957 and 1960). The Local Government Act 54, of 1961 maintained the duality between cen- tral ministries acting through their local offices, and local governments with expanded roles but still limited decision powers. The functions of the District Councils (DCs) included: environment inspection, water and sanitation; protection and security (local police force); social services provision (education, health and welfare); and road infra- structure. However, the Act contributed, along with the weak �nancial and administra- tive capacity of local agencies, to duplication of responsibilities and services, regional disparities, and an overall bad image of local governments. In face of this rather confusing background, the military Government of the National Liberation Council (1966–1971) commissioned three reports in 1967–68, which pointed to the excessive centralization of authority and recommended a move toward devolution of central authority to the local level.3 Some of the Commissions’ recommendations were incorporated into the 1969 Constitution and the Local Administration Act of 1971, and 31 32 A World Bank Study 62 DCs were established. But the Act was criticized for “a empting to balance a system of quasi-autonomous elected councils and administration by agencies of central govern- ment� (Ahwoi 2010). For instance, the District Administrative Officer was a representative of the central government. In addition, initiatives at implementing that policy met with serious challenges. The Local Administration Act (359), passed in 1971, was only implemented in 1974, due to a change of government in 1972 and after signi�cant modi�cations. That reform sought to unify central government branches and local government agencies under the authority of the DCs, and most government functions were to be decentralized, including health; District Departments were to be created for that purpose. The regions were strengthened to provide support to local administrations. However, the 1974 reform, enacted mostly through the Local Administration Amendment Decree (NRCD 258) never took hold, because of several factors, among which are:4 ■ lack of political legitimacy and structure at the local level, since two-thirds of council members were to be appointed by the central government and one-third by local chiefs; ■ centralization of decision-making in the regions (which was supposed to be an intermediate stage but was never transferred to the districts); ■ no strengthening—or transferring—of capacity and competence at the local level; ■ policy contradictions, with the issuance of decrees which actually recentralized activities that had been under local responsibility;5 ■ lack of staffing and recruitment policies for DCs, which ended up hiring large numbers of staff, including many relatives of council members; ■ insufficient efforts at addressing the administrative and logistics needs of DCs; ■ enactment of a Financial Administration Decree (1979) which centralized all �s- cal controls in Accra, including payment of local civil servants. In fact, the 1974 reforms were characterized to some extent by a movement of recen- tralization of functions previously decentralized, to central ministries or newly created national entities, such as the Ghana Education Service (1976), and removal of local elec- tion of council members. The Third Republic Constitution of 1979 reinstated elections for local governments, empowered District Councils, and established a Development Fund to support district development through grants-in-aid, and established Regional Coun- cils with responsibilities of regional coordination, planning, supervision and regional development. It was followed in 1981 by a Local Government Amendment (Act 403), which amended parts of the 1971 Act and changed the composition and functions of the Regional Councils. The second wave of reforms (1982–87) was actually a moment of reflection and prep- aration for a more solid basis for decentralization (Ahwoi 2010). The PNDC government that took over in 1982 dissolved the recently elected councils and issued Policy Guide- lines and a new Local Government Law (PNDCL 14), seeking to restructure the Pub- lic Administration System and promoting a “fully decentralized government system�. Other policy proposals were made under the Public Administration Restructuring and Decentralization Implementation Commi ee (PARDIC), translating �rst into the 1987 Blue Book on “District Political Authority and Modalities for District Level Elections�, and then into the Local Government Law (PNDCL 207) of 1988. Three main obstacles Decentralization and Governance in the Ghana Health Sector 33 were identi�ed in the Blue Book to previous efforts at decentralization: literacy in Eng- lish (required to be elected to the councils, while 50% of the population was not literate in that language); poverty (50 percent of population lived on less than 1 US$ a day, and could not commit resources to be elected); and capture of local representation by urban elites or political parties. The Law (and related policies) sought to remedy these obstacles, and restrict cen- tral ministries’ role to policy planning coordination, monitoring and evaluation. It envi- sioned to “transfer functions, powers, means and competence to the District Assemblies (DAs)� and to a lesser extent to regional entities. Twenty-two departments from central ministries were to be transferred and established as DA Departments. DAs increased in number to 110, and were supposed to exercise state power—deliberative, legislative, executive and administrative—as the people’s local government (Ahwoi 2010). The law also promoted community participation in administration, facilitated joint ventures among districts, established a district composite budget, and provided a system of checks and balances between local authority and central power. This legal document provided the most comprehensive and clear vision of decentralization to date. Following the district level elections of 1988–89, a National Commission for Decen- tralization (NCD) was established to strengthen decentralization and “evolve a true democracy�. Regarding decentralization, the NCD Report acknowledged the weaknesses of local government systems port-independence and the incompleteness of decentral- ization: “low development capacity, weak resource and revenue base, poor �nancial administration and corruption, lack of technical expertise, poor quality staff, poor remu- neration, unclear delineation of functions, dysfunctional effect of partisan politics and gerrymandering with local government boundaries�. The reforms passed during this period resulted in the 1992 Constitution and the Local Government Act of 1993 (Act 462), the �rst major legal basis for implementation of decentralization. With the 1992 Constitution and its Article 240 begins the third wave of decentraliza- tion. The Constitution provides the key principle of the current model of decentraliza- tion in Ghana: “local government and administration should, as far as practicable, be decentralized�, to ensure “functions, powers, responsibilities and resources are at all times transferred from the Central Government to local government units in a coordi- nated manner.� The Local Government Act of 1993 (Act 462) mandated the transfer o functions and responsibilities to the DAs, and is to date the most comprehensive legal document governing decentralization. The act de�ned the political government entities at the local level to be districts, municipalities and metropolitan areas, and establishes the District (or municipal or metropolitan) Assembly as its highest political and admin- istrative authority, with deliberative, legislative and executive functions. 70 percent of DA members shall be elected by the local population and 30 percent appointed by the president. The DA exercises its administrative and executive functions through an Executive Commi ee, chaired by a District Chief Executive appointed by the president.6 The Exec- utive Commi ee shall have several sub-commi ees, including, Development Planning, Finance and Administration, Justice and Security, Works, and Social Services. The la er is responsible for social sectors planning, information, needs assessment, and integration with other district sectors and areas (Ahwoi 2010 and Act 462). A speci�c subcommi ee may be formed for health, including at times environment and sanitation. The DA—and district-level activities under its authority—is to be funded through its own sources of 34 A World Bank Study revenue (property rates, levies, fees, and licenses), in addition to transfers from the central government, and has authority to borrow. The act also establishes in each region a Regional Coordinating Council (RCC) and a Regional Planning Coordinating Unit (RPCU) and de�nes their composition and functions. It de�nes that existing local branches of central ministries and departments should be transferred, together with their staff, to the DAs (art. 161), and form part of a Local Government Service (LGS). The District Assemblies’ Common Fund Act (Act 455) de�ned the �nancing mecha- nisms for DAs. Finally, the National Development Planning (System) Act (Act 480 of 1994) instituted a decentralized planning system in which the DAs were key players of the planning process. The 1992 Constitution and these three legislative pieces form the legal basis for the current decentralization framework in the country. However, they only provide the general legal basis for the establishment, structure and functions of LGs. Actual establishment of DAs requires a speci�c legal instrument (LI) issued by the MLG for each district created. Therefore, in the late 1980s a flurry of LIs were issued for the establishment of 110 districts. Others were issued in the mid- 2000s for the 60 newly created districts. Among other things, these instruments provided a detailed list of 86 functions to be performed by the DAs. These functions may be as speci�c as “paint walls . . . and cut trees�, but in other areas they are very vague. Overall, these functions are not structured along functional or strategic areas, and focus on spe- ci�c activities to be performed rather than general responsibilities, objectives or results (those functions relating to health are discussed in a speci�c section later on). Legislation and regulations in the early 2000s have been more incremental and instru- mental. The new legislation focused on complementing the basic framework given by the Constitution of 1992 and the Local Government Act of 1993, by detailing regulations for supporting public sector operation under decentralization (�nancial management, auditing, local government service) or on providing guidance for the implementation of reforms. The main legislation pieces produced in that period include the Local Govern- ment Service Act, the Internal Audit Act and the Institute of Local Government Stud- ies Act, the Public Procurement Act (all from 2003), and the Operational Guidelines for DPCUs and RPCUs and the Creation of Districts Instrument of 2003). The la er created 19 new districts on top of the 110 de�ned by the Constitution. An important policy document was produced by the MLGRD in 2003: the National Decentralization Action Plan. It recognized the slow pace of progress in the institutional and legal framework, and the inconsistency and weak coordination of initiatives aimed at deepening the process, and proposed a more incremental and realistic approach to decentralization, with a strong emphasis on establishing systems and mechanisms for supporting decentralization, building consensus, convergence of the currently competing approaches, strengthen institutions especially at the local level, harmonization and coordi- nation of decentralization efforts and initiatives. The Plan aimed to “promote convergence of the decentralization efforts, consolidation of the processes of resource allocation and management, building capacities for poverty-targeted development and governance at the local level and promotion of partnership and participation between local government, civil society, the private sector and traditional authorities.� Within that broad objective, the document proposed to work along four strategic program areas: (i) policy and institu- tional arrangements for decentralization implementation, (ii) a district development fund- ing facility, (iii) capacity building and human resources development, and (iv) partnership and participation for accountable local governance. Additionally, it strengthened inter- Decentralization and Governance in the Ghana Health Sector 35 ministerial coordination on decentralization through cabinet-level structures (secretariat and commi ees) and processes. The objective was reinforced of merging local deconcentrated departments and agencies—often subsisting along strong hierarchical lines—into one administrative unit under the DA’s authority. This was also a key element in the Local Government Service Act of 2003 (Act 656), which established a separate service for local government civil servants. As a result, it was expected that 33,000 civil servants (78 percent of the total) would be redeployed to local governments starting January 1 2007. The Act established a single LGS separate from the Civil Service, covering employees of the RCCs, DAs, and sub-district structures. Rather than decentralizing personnel management to LGs, the Act has centralized it in a new bureaucracy, responsible—through its district offices—for staff appointment and promotion. The staff of the decentralized departments listed are assumed to be transferred to the DAs’ authority and, thus, become staff of the LGS. Both education and health staff, however, were left out of the Act and remained as separate services (GES and GHS). Act 656 therefore conflicts with Act 462, which had included health and education as decentralized departments. Finance of the DAs were re-central- ized into a general Consolidated Fund According to the Act. ACHIEVEMENTS AND SHORTCOMINGS Over these many years, decentralization reforms have advanced based on three main dimensions: political decentralization, administrative decentralization and decentralized planning, and �scal decentralization. Political decentralization has involved the establish- ment and demarcation of districts and other MMDs (metropolitan areas and municipalities), and the establishment of DAs and their supporting offices. This has been the main focus of much existing legislation. Administrative decentralization has focused on the transfer of some functions and powers, means and resources from central ministries to LGs, the de�nition of functions and responsibilities that were been devolved, rearrangements at the central ministries to accommodate their new roles, some transfer of staff and officers to LGs, capacity building at local level, and the set up of a decentralized planning and budgeting system. Fiscal decentralization has focused on establishing �nancing mechanisms for DAs to enable them the resources at the disposal of LGs to perform the decentralized functions and responsibilities, and de�ning their level of autonomy over such resources. Results from this process include the creation of 110 local government units (DAs)— later increased to 1707—the transfer of some authority and functions from the central to local government level, building of initial infrastructure and capacity a ached to the DAs, and the establishment of �nancial mechanisms for �nancing local governments (such as the DAs Common Fund). The general roles of the different levels of government—central (ministries), regional and local (districts)—have been reasonably de�ned in the legislation and are consistent with international models of decentralization: ■ Central ministries are responsible for policy making, national planning, moni- toring and evaluation, development of standards and indicators; ■ Regions, through the Coordinating Councils (RCCs) and the RPCUs, are respon- sible for regional planning, harmonization and coordination of districts inter- ventions, monitoring [and support to the districts); ■ Districts, through the partially elected DAs, are responsible for adapting national policies to local realities, district-level development planning, and implementa- tion of such plans and programs. 36 A World Bank Study The Development Funding for Districts has been investing signi�cant resources in the districts to strengthen local capacity, and donors’ initiatives have also contributed, through a Harmonized Capacity Building program. In spite of these advances, issues have been identi�ed in the implementation pro- cess, especially in policy documents produced in recent years:8 contradictions in sectoral approaches to decentralization, unclear role of the regions, slow or no transfer to DAs, slowness in implementing �scal decentralization, ineffectiveness of LG substructures, low capacity and motivation of DA members, weak popular participation, unclear rela- tions with traditional authorities. In addition, the political instability during most of the post-independence era reflected in discontinuous and often contradictory legisla- tive and regulation initiatives regarding decentralization and local government, which shared some common features, managed to implement some of them, but prevented effective decentralization to take hold. Several aspects of the general legal framework for decentralization and related docu- ments are interesting to note. First, the objectives of the process have often been de�ned more in terms of formal processes than in terms of expected impact, and theoretical rather than operational. The current legislation de�nes the objectives of decentralization as “To ensure an inclusive, participatory and democratic system of local government [. . .] an accountable Assembly [. . .] that would make sure that the felt needs and priorities of the electorate were articulated [. . .] provide for the decentralization of functions best performed at the local levels [. . .] ensure balanced and equitable development . . . construction of the national democratic governance system� (PNDC’s “Blue Book� in Ahwoi 2010). The 2007 Comprehensive Decentralization Policy takes a different approach, and clari�es the key objectives of decentralization in Ghana as: “(i) Strengthening and expanding local democ- racy, and (ii) promoting local social and economic development, thereby reducing poverty and increasing the choices of the people�. Second, most of the legal and regulatory pieces focused on the design and guid- ing principles of a decentralized system rather than its effective implementation. In that sense Ghana’s experience is no different from that of many developing countries, where the planning of the decentralization process has been absent or weak, with li le focus on concrete strategies to make it effective (see the section titled “Lessons Learned�). Also, the process was characterized by several contradictory moves back and forth. For instance, as summarized by Ahwoi (2010), education was decentralized in 1974 (NRCD 258), then recentralized in 1976 (under Ghana Education Service Decree), decentralized again in 1988 (PNDCL 207), and removed from the decentralized departments in 1993 (Act 462) and 2003 (Act 656). Health followed a similar path: it was decentralized in 1993 (Act 462) and then omi ed in 2003 as a decentralized sector. Some of this back-and-forth and the contradictions in legislation were due to the fact that the process has met over the years with strong resistance: in the words of the Ministry of Local Government in 1992, “Decentralization has not taken place in Ghana. The reason largely is that the bureaucracy . . . particularly the top management personnel . . . is not in favour of decentralization. Every impediment has been placed in the way of implementing the decentralization program. Top civil servants do not want to know. Some have deliber- ately confused it with an exercise in deconcentration� (Ahwoi 1992). Finally, several aspects of passed legislation and policies have not been implemented, partly because of political resistance, low capacity and weak implementation strategies. The idea of a “composite budget� integrating the different sectors and funds at the local Decentralization and Governance in the Ghana Health Sector 37 level has not yet been fully implemented. Another factor contributing to slow implementa- tion is the lack of clarity as to which functions are to be decentralized and which should remain centralized. A recent set of new policy and legal documents since 2007 testify to the government’s renewed interest in decentralization. Interestingly, most of these documents are policy oriented—proposals or reviews—that (i) acknowledge the slow pace and the many dif- �culties in the process of implementing decentralization; (ii) propose new approaches and strategies to address these issues in a more systematic way and accelerate implementa- tion. This recent policy orientation is very welcome, since it has been a weak point dur- ing previous efforts at implementing decentralization, and the international experience review showed that a well developed and clear policy is key to successful decentralization processes. A major weakness of decentralization in Ghana is that the basic law governing the process, the Local Government Act of 1993, has been implemented only partially. On the one hand, the districts and their boundaries have been de�ned, and their basic political structure—especially the DAs and a ached offices—has been implemented as mandated by the legislation. However, the number and boundaries of districts has not been stable, as their number has increased by 50 percent since the Act. Many new districts do not have their administrative and physical infrastructure in place yet, and lack the person- nel to manage the functions assigned to them. On the other hand, the units, facilities and staff of some larger line ministries have not been transferred to the DAs’ authority as mandated by the legislation. This is especially the case of the MOH/GHS district- level structures. Several reports have correctly identi�ed the main problems in the decentralization process, which are mostly still valid today (see for instance, PHRplus 2002; NDAP 2003): ■ Mismatch between authority (the legal ability to do) and responsibility (the assigned function of executing or implementing), associated to lack of clarity about and competing views on decentralization; ■ Tensions and conflicts among objectives (national policies versus local priorities or preferences; preventive versus curative care; and weak intersectoral coordination ■ Capacity gaps (administrative, information, �nancial) at all levels, but espe- cially at local level; ■ Weak policy implementation capacity at all levels, coupled with weak monitor- ing and supporting capacity; ■ Weak economic basis of many districts; ■ Tensions between vertical and horizontal integration (vertical programs VS integrated care at local level); ■ Political and process dimensions (interest groups, resistance to change, political will, weak stakeholders involvement and participation); ■ Political and administrative instability: since the 1992 Constitution—which kept the number of districts at 110—60 more districts were created (an increase of 55 percent), 28 in 2003 and 32 in 2007.9 A National Decentralization Action Plan (NDAP) was prepared and de�ned eight stra- tegic objectives for the period 2003–05: ■ To strengthen political leadership and intersectoral collaboration for decentralization. 38 A World Bank Study ■ Enhance decentralization policy management, implementation and monitoring. ■ Increase discretionary funding to DAs and consolidate the overall District Resource envelope. ■ Strengthen overall district level �nancial and human resource management and accountability. ■ Strengthen DAs’ functional and governance performance. ■ Strengthen decentralized coordination and M&E at regional level. ■ Enhance and strengthen the sub-district level. ■ Promote popular participation and deepen association and partnership between DAs, civil society, private sector, and traditional authorities. Later policy documents are consistent with the NDAP, although the challenges and pol- icy proposals set forth vary somewhat. DECENTRALIZATION POLICY REVIEW (2007) Taking a step further from the NDAP, the Decentralization Policy Review of 2007 put the decentralization reform within a broader objective of “enhancing good [public] gover- nance� and aimed at “formulating a comprehensive decentralization policy�. Seeking to strengthen and re�ne the decentralization policy framework, the document analyzed the progress and status of decentralization along �ve key areas considered as requirements for effective decentralization: (i) the basic legal and policy framework, (ii) the political and administrative framework, (iii) local government �nancing, (iv) local government human resources and (v) the arrangements for reform coordination and support. The main recommendations of the report correctly addressed the main challenges in the decentralization process: ■ In the area of policy and legal framework, develop a broad decentralization policy that would clarify the vision for decentralization and clearly de�ne the functions to be transferred to local governments; and develop a Decentralization Strategic Framework to guide implementation. ■ In the area of political and administrative framework, promote increased democ- ratization of MMDAs, clarify lines of accountability (prioritizing accountabil- ity to local residents), revise the status and role of regional administrations (as deconcentrated offices) and sub-districts. ■ In the area of �nance, assign expenditure responsibilities across government lev- els following the principle of subsidiarity; clearly de�ne revenue assignments aligned and commensurate with MMDAs functions; harmonize and streamline the intergovernmental transfer system; improve MMDA �nancial management; ensure �nancial systems promote downwards accountability and participation; coordinate reform initiatives related to �scal decentralization; estimate the cost implications of decentralization. ■ In the area of human resources management (HRM), clarify the vision for HRM and the degree of autonomy LGs would have in HRM; develop policies and measures to a ract and retain staff in disadvantaged districts; detail and clarify the plan for integration of decentralized departments at district level; develop performance standards and monitor performance. ■ In the area of reform coordination, establish at the level of President’s Office a commission on decentralization to coordinate the process and provide direction from the highest level. Decentralization and Governance in the Ghana Health Sector 39 DECENTRALIZATION POLICY FRAMEWORK (2007 AND 2010) These two recent policy documents build on the 2007 Policy Review, recognize the need for a comprehensive and coherent policy framework to guide and support strengthen- ing and accelerating the implementation of decentralization in Ghana. The �rst aimed to “deepen political, administrative and �scal decentralization in Ghana�, and more speci�- cally, “(i) harmonizing the legal and regulatory framework for decentralization and local governance, (ii) clarifying and strengthening mechanisms of accountability of sub-national public officials, and (iii) improving the allocation of resources to DAs to fully deliver the tasks and functions transferred to them.� New efforts at decentralization will follow four guiding principles: ensuring participation of the people, improving public sector account- ability, strengthening DAs, and increasing public sector effectiveness. Under these prin- ciples, the document de�nes new reform initiatives in the areas of: ■ Legal and policy framework, including harmonization of existing legislation, especially in regard to sectoral legislation ■ Intergovernmental relations and the Inter-Governmental Fiscal Framework, bringing together the assignment of functions, the allocation of resources and clari�cation of accountability mechanisms across different government levels ■ Public �nancial management, to improve DAs capacity for sound �nancial man- agement and strengthen the performance-based District Development Funding mechanism ■ Decentralized human resource management, with the transfer of civil servants to the LGS (Act 656) and effective local autonomy in managing these resources ■ Institutional arrangements for policy coordination and policy management, to ensure effective coordination among ministries/departments/agencies (MDAs) and increased participation of stakeholders. The document highlights as the main challenge for the full implementation of decen- tralization the divergence between political intentions, as set forth in the general legal and policy frameworks, and prevailing administrative practices, as reflected in fund- ing flows, functional assignments, accountability and reporting lines. This divergence is more apparent in sector ministries and agencies’ practices; local branches and staff have not yet been transferred to the DAs. A second challenge resides in the lack of a compre- hensive �scal framework, and current multiplicity and irregularity of funding channels and mechanisms. A third challenge relates to the lack of staff management mechanism at the local level and the difficulty for remote districts to a ract and retain staff. The revised policy implies, among other initiatives, amending existing legislation, revising the current mixed model of district representation, clarifying the assignment of functions across levels of government, implementation of the Local Government Ser- vice Act, strengthening the �scal framework and matching revenues with functions, the strengthening of coordination and support structures, and the strengthening of human resource management within district administrations. Finally, the document de�nes a new implementation plan for decentralization reform, structured in a preparatory phase (2007–08), a �rst implementation phase (2008–2012), and a second implementation phase (2012–2016), with main activities assigned to key stakeholders. The second document is based on similar principles and objectives, but broadens the policy areas to nine areas for priority action: (i) political decentralization and legal 40 A World Bank Study issues; (ii) administrative decentralization; (iii) decentralized development planning; (iv) spatial planning; (v) local economic development; (vi) �scal decentralization; (vii) popular participation; (viii) the social agenda; and (ix) involvement of non-state actors in local governance. The review of these policy documents and interviews with stakeholders point to a major conclusion, which relates to the limited implementation achieved so far in the decentralization reform as a whole. Nearly all assessments and strategy documents high- light similar shortfalls and difficulties, which relate closely to those from other develop- ing countries (see the section titled “Lessons Learned� for the main conclusions from the review of international experience): ■ Fragmented and confusing legal framework, with some major conflicts ■ Lack of a strong and clear policy framework and roadmap to guide imple- mentation ■ Lack of consensus among key stakeholders about the objectives and implica- tions of decentralization by devolution ■ Unclear assignment of functions across government levels and agencies ■ Weak local capacity for implementing and operating a decentralized system based on devolution ■ Limited capacity at the regional and central level to support, monitor and evalu- ate implementation ■ Unclear framework for the transfer of personnel to local governments ■ Unclear and complex �nancing flows to local governments, and the lack of a comprehensive framework for �nancing them. Nearly all of these issues apply to the health sector as well, and maybe in a more pro- nounced manner, as the following sections will show. Current Legislation and Policies for Decentralization in Health The general GOG policies regarding decentralization make up a consistent framework— though incomplete—for general decentralization of government functions. Recent policy papers show a clear understanding of the issues at stake and the challenges to imple- mentation, and provide some guidance on the way ahead. However, missing from the existing policy documents is a roadmap with clear guidelines and practical steps on how to adapt and implement the general policy in the health sector. Health systems present additional complexities in terms of decentralization, that have not been addressed in gen- eral or speci�c legislation. At this point, there is no strategic policy document outlining how devolution is to be implemented in the health sector. The current legal framework for decentralization in the health sector is quite limited, but rather confusing. It is composed of several separate pieces, which are inconsistent and conflict among themselves. First, general legislation and regulations are meant to apply—with some exceptions— to every sector including health. The Local Government Act of 1993 (Act 462) and related regulations determined the establishment of a District Health Department (or Metro- politan or Municipal), but did not specify any health-related function or responsibility, except for the presence of a Medical officer responsible for the control of infectious dis- eases. The responsibilities of the DA Social Services Sub-Commi ee are limited to plan- ning and coordination, and do not include executing the provision of health services. Decentralization and Governance in the Ghana Health Sector 41 Second, the Legislative Instruments establishing DAs list, among the 86 functions to be transferred to DAs, typical public health responsibilities, including health promo- tion (“promote and safeguard public health�), sanitary surveillance and inspection (of threats to people’s health), water and sanitation (provision of safe water supply, public latrines and lavatories, waste disposal), disease surveillance and control (“prevent and deal with the outbreak of any disease�). Again, the provision of health care services of any kind is omi ed. These LIs provide for the nomination by the MOH of a District Medical Officer and other staff to the district. Third, aiming at effecting the transfer of sector departments to the DAs, the Local Government (Department of District Assemblies) (Commencement) Instrument, 2009 (L.I. 1961) has de�ned the commencement dates of such transfer and their establish- ment as departments of the DAs. Certain departments are supposed to be transferred immediately upon issuance of the LI (December 2009). However, other departments, including health and education, were not considered as decentralized departments in Act 656 of 2003, because they were created by speci�c law enactments. Their transfer to the DAs—and the establishment of a District Health Department—would thus have to wait for an amendment of the enactments. However, differently from other departments to be transferred, the document does not specify that the existing department or entity will cease to exist upon transfer to the DAs. This leaves it open to the coexistence of two entities responsible for health at the district level: the District Health Department (when established) and the existing GHS DHA. The LI also lists 35 functions to be performed by the department. Most of them relate to general public health functions, but the instrument includes four health care functions: “(b) assist in the operation and maintenance of all health facilities . . . ,� “(d) coordinate works of health centers or posts or community based health workers,� “(l) advise on . . . supervision and control of all District health institutions�, and “(q) assist in efficient man- agement of clinical care, community health care and environmental health service. . . .� It is noteworthy that as it relates to the provision of health care services at the district level, the role of the department is mostly of advising, assisting and facilitating (this is also true for most public health activities). The question then arises of who is actually respon- sible and accountable for making decisions in these areas, make sure that health policies are implemented and that a consistent local health care network operates.10 The main health-speci�c legislation to date is the Ghana Health Service and Teach- ing Hospitals Act (Act 525) of 1996, which created GHS as a separate, autonomous entity and delegated to it the management and operation of nearly all subnational facilities and offices. GHS in turn has a vertical structure by level of government, and has gradually deconcentrated operational functions to its Regional Health Administrations (RHAs) and especially DHA offices. In spite of this deconcentration, the delegation of health service provision at the regional and local levels to an autonomous entity is inconsistent with the general legislation, as many reports have pointed out. The degree of managerial autonomy conferred to regional and district health offices will be assessed in the next chapter. Ghana Health Service and Teaching Hospitals Act (Act 525 of 1996) Act 525 establishes the GHS as an autonomous entity to manage most of the country’s facility network, and de�nes the facilities and staff previously under the authority of the MOH to be transferred to GHS. It is governed by the GHS Council appointed by the president. GHS has a deconcentrated hierarchical structure in three levels: central, 42 A World Bank Study regional and local. At the central level, the Service is chaired by a Director General and has �ve divisions at the national level: Public Health, Institutional Care, Policy, Plan- ning, Monitoring and Evaluation (PPME), Health Administration and Support Services, Supplies, Stores and Drug Management, Human Resource Development, Finance, and others as needed. GHS also has a Regional Director of Health Services advised by a Regional Health Commi ee (composed of a chairman and representatives from professional and commu- nity groups, in addition to the Regional Director). Clinical, Public Health and Adminis- tration Divisions make up the basic administrative structure at the regional level. A the district level GHS has a structure similar to the one at the regional level, though smaller; composed of a District Director of Health Services (DDHS), a District Health Commi ee, and two or more divisions (Clinical and Public Health). Many experts and public officers, as well as available documents, see the contradic- tion between a delegated and deconcentrated GHS structure, and a (yet to be) devolved health department, as the main issue today in the health sector decentralization. However, as shown above, the existing legal and regulatory framework regarding health sector decentralization is characterized by many more contradictions between legal documents, and an overall lack of clarity about the design of a decentralized health system, the func- tions to be assigned to the different levels of government, and the entities that will effec- tively be responsible for operating health care networks. There is no clear model or policy on how a devolved health sector would look like, and the policy debate within MOH and GHS about how to implement devolution has not been consistent or systematic. The current reality of the health sector is that a limited number of functions and responsibilities have effectively been transferred to the DAs, but the major part of the structures, staff and functions remain within the structure of the GHS. It was expected that integrated planning (or composite planning) and coordination between the DAs and the DHAs would gradually bring together these two institutions in each district. However, integrated planning has not taken hold yet, and coordination remains heavily dependent on personal relationships between DA and DHA authorities. DAs are supposed to have a District Medical Officer, and install a Social Sectors Commi ee, on which the GHS DDHS sits. In addition, DAs are supposed to use part of the �nancial resources received through the District Common Fund on health activities. Accordingly, many DAs are supporting the DDHS office in concrete ways: providing office space, hiring staff to complement DDHS own staff, contract maintenance or sup- port services (security, cleaning and others). However, the involvement of DAs in health activities appears to vary greatly across districts and not be systematic. Overall, under the current con�guration district-level provision of health services is funded through three different and parallel channels: (i) through the GHS structure of DHAs, which are responsible for most of health care activities and services provided at the district level; (ii) through MOH-managed vertical programs, which may involve DHAs and/or DAs but are managed centrally; (iii) through the DAs own structure and funding, but mostly as supporting activities to DHAs or the MOH rather than direct responsibilities. The following table (table 3.1) a empts to map out the degree of autonomy DAs enjoy with regard to health. This includes both their authority for policy making in health, and their ability to work alongside and supervise the District Health Adminis- trations. The table illustrates the conflict between DAs’ theoretical authority based on Decentralization and Governance in the Ghana Health Sector 43 Table 3.1: Mapping out the decision space for Ghana districts regarding health Functions Responsibilities Weak Average Strong Governance and Appointment of DHA director X (GHS) Accountability Collaboration with DHA X (variable) Accountability of DHA to DA X Accountability to Central level X Accountability to local X (committee) community Policy and Planning Policy formulation X Health planning X (�nancial constraints) Program design X Finance Revenue generation X (limited) X (fees, IGF) Budget preparation, allocation X (in place) Accounting and audit X Set user fees X Set up & manage insurance schemes Service organization De�ne service packages X (GHS) and provision Target service delivery X (dialogue) Set norms, standards, X (central) regulations Monitor and oversee providers X (central) Contract in/out X Human resources Planning, evaluating HR X Hiring, �ring civil servants X (Pres) Hiring, �ring other staff X De�ne salaries & bene�ts X (central) Training X Paying staff X (civil serv) X (other) Support services Procurement of drugs & X (some) supplies Manage drugs and supplies X Maintain vehicles & X (support) equipment Maintain facilities & structures X (support) Information systems Design HIS X Collect, process, analyze data X Disseminate information X Source: Adapted from Bossert (1998). 44 A World Bank Study current legislation, and the lack of integration with DHAs. This particular aspect will be further explored in the next chapter. Under current regulations and policies DHAs are supposed to submit their bud- get to the DAs and participate in the formulation of the composite budget. However, in the end DHAs have limited control over their budget for several reasons. First, the budget is approved by GHS in a hierarchical process (consolidated and possibly revised at the regional level, then by GHS headquarters). Second, staff remuneration, which accounts for the major part of the total budget, is paid directly by the central govern- ment to staff bank accounts. Most capital expenditure and expenses funded by donors are also centralized through MOH vertical programs. Third, budget funds are often released with substantial delays (of weeks or months), and actual payments may take even longer (with the exception of remunerations). In addition, composite budget is not yet operational, and budget discussions with DA staff depend very much on per- sonal relationships between the District Health Director and the Chief Executive. In general, DHAs are not accountable to the DA, at most they inform the DA of their budget and performance. Actual accountability remains to GHS higher levels (regional and headquarters). Authorization of district-level spending, depending on the source of funding, is split between several organizations: GHS headquarters for the larger part (DHA and DHO); DA (health allocation from its own funds); MOH (for vertical and donor-supported pro- grams); DHA and facilities (for user fees). DHA performance is monitored by the Regional Director of Health (RDH). The RDH has authority to transfer staff within his region, and may request �ring or transfer to other regions to headquarters and Civil Service. In mid-2010, a new General Health Service Bill was before Parliament. Its objective is to revise the law on GHS and Teaching Hospital Act (1996), address operational difficul- ties, reduce overlap of functions and contradictions (e.g., authority over staff between GHS and THO); and restructure different services in the health sector. In fact, it seeks to harmonize and consolidate several separate regulations. The bill covers �ve different areas: (i) organization of the Health Service (Part I), (ii) the creation of a new Teaching Hospital Authority (Part II), (iii) the establishment of the National Ambulance Service (Part III), (iv) the creation of a National Blood Service (Part IV), and (v) administrative and �nancial provisions (Part V). However, the bill does not seem to contribute in a substantial way to improve the current legal framework, and in fact may create more confusion. Some of the limitations observed in the bill are summarized below: ■ The bill appears to recreate the GHS under the new name of “General Health Service�, but without substantive changes in structure or functions apart from some name changing; ■ It does not clarify the distinction between functions assigned to MOH and those assigned to GHS, or those assigned to the different levels of the system; ■ It maintains the system of centralized appointments by the President of regional and district directors and council/board members, without specifying technical requisites for these posts; ■ It maintains some confusion between advisory functions and executive/ operational responsibilities; for instance Health Commi ees have both types of functions; Decentralization and Governance in the Ghana Health Sector 45 ■ It is not clear why the bill creates separate autonomous services for teaching hos- pitals and certain other services (ambulances, blood services); this segmentation along separate vertical services may further contribute to system fragmentation. ■ Even though it creates new autonomous services, the bill does not propose con- tractual arrangements between the different levels of the system and among these new bodies. ■ Most importantly, the bill does not take into consideration the ongoing decen- tralization process, and does not even mention it; the document actually main- tains confusion in the lines of authority and accountability: for instance, every Region and DDHS is answerable both to the GHS Director General and to the District Chief Executive, without clarifying who has the highest authority in case of conflict; Notes 1. This section draws signi�cantly from Ahwoi (2010). 2. The data limits for these waves is somewhat arbitrary, and take into account both the govern- ment or the regime in place and the main legislative pieces produced at different moments. 3. The Mills-Odoi Commission Report of 1967, the Siriboe Commission Report of 1968, the Akufo- Addo Constitutional Commission Report of 1968. 4. See Ahwoi (2010). 5. See, for example, the Omnibus Services Authority Decree (of 1972) and Amendment (of 1973), and the Ghana Education Services (Amendment) Decree of 1976. 6. More recently, the election of the chief executive has been proposed and is still the subject of debate. 7. The Constitution maintained the number of Districts at 110, but 28 additional districts were cre- ated in 2003, and 32 more in 2007, bringing the total to 170. 8. Such documents include the National Decentralization Action Plan (2003). 9. 124 are actually DAs, 6 are Metropolitan Assemblies, and 40 Municipal Assemblies. Sub-district structures do not correspond to political levels of government: Sub-Metropolitan District Councils (numbering 30), Urban Councils, Zonal Councils, Town/Area Councils, and Unit Commi ees. 10. This appears to be the case also for other decentralized departments, but we limit our discus- sion here to health. CHAPTER 4 Rapid Assessment of Local Capacity and Challenges G hana has been in the process of decentralization for nearly three decades, and the GOG is intent to deepen and speed up its implementation. It is important at this point to assess the degree to which general and health speci�c functions have already been decentralized to district level and are already carried out at that level, and under what conditions. A second issue in successful implementation of decentralization reforms is the capacity of subnational governments to carry on the new responsibilities de�ned in the decentralization policies. Finally, it is also important to be able to cor- rectly identify weaknesses and challenges faced by local governments in implement- ing decentralization, so that focused training and technical assistance can be offered to support the process. In order to assess the current status of decentralization, assess LG capacity for expanding and deepening the reform, and to identify obstacles to the process, a rapid assessment was undertaken in a sample of the country’s regions and districts. Approach and Methods For that purpose a one-day workshop was convened on May 26, 2010 in Accra, to which were invited 6 Regional Directors of Health Services, 24 District Directors of Health Services, and 3 facility directors. Overall 33 participants a ended the work- shop. In addition, information was gathered personally by this consultant in one addi- tional region (Eastern Region) and two of its districts. In all, information was obtained for 7 RHAs and 26 DHAs for this assessment, and the sampling strategy—focused on reflecting the diversity of local conditions at a minimal logistic cost—suggests that the results are fairly representative (though not statistically) of the country as whole (�gure 4.1). Information was gathered through the workshop in three ways. First, an instru- ment was designed with 78 questions regarding the different functions or dimensions of decentralization, as listed in table 4.1. The responses to these questions were then tabu- lated and analyzed in terms of frequency and distribution. Second, participants were encouraged to provide comments and details about the questions asked; and third, the discussions that took place during the group sessions and the �nal debate were orga- nized and summarized so as to highlight major �ndings. Size and Organization of District Health Systems Ghana’s regions and districts are relatively homogeneous in population size; their mean population, of 170,000 (145,000 if Accra and Kumasi are not considered), is relatively large to reduce diseconomies of scale. Just for comparison, the 5,600 municipalities in Brazil have an average population of 34,000. The size of local government units is 46 Decentralization and Governance in the Ghana Health Sector 47 Figure 4.1: Participants in the workshop 45.0% 40.0% 35.0% 30.0% RHA 25.0% 20.0% DHA 15.0% 10.0% 5.0% 0.0% ti fo ra ta rn n er an ol cc ha te th sh V es A A or A W er g N on at re Br G Source: Author’s survey of DHAs and RHAs. important for efficiency considerations: a World Bank study showed that in Indonesia, a district of 500,000 is twice as efficient administratively as one of 100,000 (Hofman and Kaiser 2002). The district-level facility network in each district includes, on average, 1 hospital— usually a District Hospital (mean size of 73 beds), or a mission hospital (102 beds on average); overall, (including all public hospital beds) each district has an average of 2 hos- pitals and 160 beds. 59 percent of districts have a District Hospital, but half of those have a mission hospital; however, 25 districts (18 percent of the total) have no hospital at all. Additionally, districts have on average, 9.4 health centers (health stations), and 2,1 CHPS (GHS, Summary of Health Facilities). This average network is reasonable and of manage- able size for local government health agencies. However, it shows large variations across Table 4.1: Structure and number of questions in the instrument Function or dimension Number of questions Identi�cation 9 Governance and structure 10 Planning and budgeting 10 Finance and budget execution 5 Human resources 12 Procurement 7 Accountability 4 Health service coordination and management 7 Knowledge and beliefs about decentralization 14 TOTAL 78 Source: Author’s survey of DHAs and RHAs. 48 A World Bank Study Table 4.2: Mean density of public health infrastructure per district Hospitals per Beds per Beds per CHPS/ Region District District 1,000 People HC+ / District District Western 1.15 156 0.79 10.69 4.31 Central 1.00 139 0.94 6.08 3.31 Greater Accra 2.33 767 1.07 12.17 0.67 Volta 1.33 148 1.14 13.47 1.27 Eastern 1.06 157 1.12 11.29 2.59 Ashanti 1.90 196 0.84 8.76 0.19 Brong Ahafo 0.89 97 0.81 7.26 0.58 Northern 0.72 73 0.56 8.22 0.56 Upper East 0.75 102 0.81 8.75 6.88 Upper West 0.75 88 1.05 9.25 4.88 Ghana 1.17 160 0.91 9.41 2.07 Source: GHS (List of Facilities and Bed Statement Returns 2008). regions and even more so across districts, especially in the number of community health units (CHPS): as shown in table 4.2, their density is much higher in the Upper East and West regions, as well as the Western and Central regions. Governance and Accountability of DHAs The lines of authority and accountability of DHAs appear to reflect an intermediate stage of decentralization, but also some duplication or lack of clarity of authority lines: more than three-fourths of the DHAs surveyed mentioned reporting to both the District Chief Executive (81 percent) and the Regional Director of Health Service (�gure 4.2). Regional Health Directorates also indicated unclear lines of authority, with 43 percent Figure 4.2: Administrative lines of authority 90.0% 80.0% 70.0% 60.0% RHA 50.0% 40.0% DHA 30.0% 20.0% 10.0% 0.0% District District Regional Regional Ministry of Other Don't Know Chief Assembly Minister Director of Health HQ Executive Health Service Source: Author’s survey of DHAs and RHAs. Decentralization and Governance in the Ghana Health Sector 49 indicating that they report to the Regional Minister, and 4 RHAs reporting to a District Chief Executive, the central Ministry of Health, or others. Even though it is expected in a country in transition to full devolution, that local health offices relate simultaneously with the central ministry and to the local government, this is not a desirable or sustain- able situation: in case of conflicting policies or guidance, one entity should take �nal responsibility, and this line of authority should be clear to all. The technical responsibility for supervising and overseeing district health systems is much clearer, with all DHAs reporting to the Regional Director of Health Service. However, for RHAs the supervising authority does not seem to be very clear, with an almost even split among alternatives (GHS headquarters, MOH headquarters, the Regional Director of Health Services; 40 percent reported “other� or “don’t know�). In a similar pa ern as in Figure 4.2, regional and district health officers participating in the survey described confusing lines of accountability: they indicated to be account- able both horizontally (to the government authority at their level) and vertical (to the higher government authority), as shown in Figure 4.3. More speci�cally, RHAs reported being accountable to both the regional government/Regional Coordination Council (57 percent) and to MOH/GHS headquarters (43 percent); 14 percent also indicated to be accountable to a health commi ee. As for DHAs, 87 percent indicated to be accountable to the RHA, 31 percent to the District Assembly, and 12 percent did not know who they were accountable to. Interestingly, RHAs appear to be—or feel—more accountable to regional govern- ment authorities than to MOH/GHS headquarters, while DHAs are—or feel—more accountable to their hierarchical superior authority (the Regional Health Directorate or Administration) than to LGs. Even though subnational health offices may report to one authority from a techni- cal point of view and another from a political/administrative perspective, the perceived multiple lines of accountability reported in the survey seem to indicate more than that, namely blurred or unclear lines of authority and accountability. This is likely to result Figure 4.3: Lines of accountability WHO IS YOUR OFFICE ACCOUNTABLE TO? 100.0% 80.0% RHA 60.0% DHA 40.0% 20.0% 0.0% al ovt th ovt / ttee now ntr eal mi HC e al G lH ct G om n'tK ion na tri C Do MO Re g Re gio Dis alth A He Source: Author’s survey of DHAs and RHAs. 50 A World Bank Study from the decentralization process of the Ghanaian health system which is actually stuck midway between deconcentration (within GHS) and devolution (with district govern- ments taking over some but not all authorities and responsibilities). This important issue will be further discussed in the next section. A survey of 17 DAs by Ahwoi (2010) found similar results regarding “the dual report- ing responsibility and divided loyalty of the Departments [transferred or to be transferred to] the District Assembly . . .�: no real administrative decentralization; the departments actually performed as de-concentrated offices of central ministries; these pa erns are likely to also reflect some lack of clear understanding from DA officers of the implications of decentralization. All RHAs, and 87 percent of DHAs, reported having established a commi ee to oversee or support the management of the office. The main role of this commi ee is, in both cases, advisory or consultative, followed by oversight and monitoring (�gure 4.4). The commi ee is usually not involved in decision making. Most health offices also reported having a management team, but it appears that respondents could not clearly distinguish between the responsibilities of this team from those of the commi ee (or understood them as being the same thing). As shown in �gure 4.5, all RHAs and 63 percent of DHAs reported having no contractual arrangement with their supervising office or higher level entities, did not respond, or did not know. 2 DHAs indicated having a performance-based contractual agreement with MOH/GHS headquarters, and 4 others some arrangement with other entities, whether performance-based or not. These answers are somewhat puzzling, since from 1998 service performance contracts are supposed to be established between operational BMCs and their supervising BMCs; this consultant was presented with an instrument (a memorandum of agreement) at the DHA level that would constitute a contractual arrangement, although a somewhat weak one. The responses seem to indi- cate both a lack of clarity about what constitutes a contractual arrangement (including Figure 4.4: Role of the committee WHAT IS THE MAIN ROLE OF THIS COMMITTEE 100.0% 90.0% 80.0% RHA 70.0% 60.0% DHA 50.0% 40.0% 30.0% 20.0% 10.0% 0.0% ry e / t t ng g so tiv icy en en isi rin ow dv i lta P ol em em Ra ni to Kn su ag ag ’t A Co n On an an nd ,M o Do n on nM M Fu ht c isi O ing ig De on tat ers isi ili Ov ec Fac D Source: Author’s survey of DHAs and RHAs. Decentralization and Governance in the Ghana Health Sector 51 Figure 4.5: Contracting arrangements with other public entities 60.0% 50.0% RHA 40.0% DHA 30.0% 20.0% 10.0% 0.0% . . e... ct se ow ct.. ct.. tra on anc Kn tra tra n sp Co on on rm n’t Re eC eC rfo No Do No -pe anc anc on orm orm r, N erf erf the s, P s, P sO Ye Ye Ye Source: Author’s survey of DHAs and RHAs. MOUs) and its purpose and implications, and that existing instruments are not enforced or systematically used. In fact, existing evidence suggests that accountability and transparency throughout the system are weak. Overall, health administration office and facility managers are not held accountable to results in a signi�cant way, in spite of the existence of a system for measuring performance. Enforcing performance assessment and using it appears to depend heavily on the personal commitment of regional (and to a lesser extent, local) managers. The planning and budgeting process, although decentralized in principle, ends up lacking transparency due to the weak relation between original plans and bud- gets (responsibility of local managers) and the effective flow of funds (de�ned by the central government—see the section on planning and budgeting below). The unreliabil- ity of existing or reported data on health spending at the district level (as evident from the recent PER) contributes to the problem. Community participation and voice is usually seen as a major objective and expected impact of decentralization. In the Ghana health system, health councils and commi ees have been established at the different levels—regional, district, facility, and sub-district—and are usually functional. However, the effectiveness of these participa- tion channels is unclear, and evidence of capture by local elite or interest groups has been found. Structure and Staff Most of the RHAs and DHAs surveyed appear to also have established a functioning administrative structure: more than half of them mentioned operating with three or four divisions or units as mandated by GHS establishment act (Act 525 of 1996). The most common were: Administration, Public Health, Clinical, and Finance/Budget (�gure 4.6). 52 A World Bank Study Figure 4.6: Administrative units present at RHAs and DHAs 100.0% 80.0% RHA 60.0% DHA 40.0% 20.0% 0.0% l t t l r ca ealth ation ning dge men ne ation he ow ini lan e/Bu cure on Ot t Kn Cl lic H nistr P ers alu ’ b i an c Pr o /P Ev Do n Pu Adm Fin HR & M Source: Author’s survey of DHAs and RHAs. It is worth noting that in many cases, these units were not formally established, and were basically working teams mobilized for a given technical area. The number of staff working in these units varied greatly across RHAs (between 1 and 103, with a median of 15), but not so among DHAs, where 49 percent of these units employed less than 3 pro- fessionals and 70 percent less than 5. The quali�cation and training of the DHAs’ teams appears weak, with 53 percent of them having only 1 or 2 professionals with appropri- ate training, and fewer still having a university degree. The quality of staff is likely to be much lower in remote regions and areas, which were not captured in the survey. Mana- gerial and �nancing capacity is thus likely to be quite variable, and very weak in many districts, which often suffers from the lack of quali�ed personnel. Recent reports point to a good capacity for �nancial management at the central level (MOH-HQ, GHS, Teaching Hospitals) but poor capacity at the district and sub-district- levels (Independent Health Sector Review 2008); this is mostly associated with poor record-keeping and availability of staff. Many sub-districts are reported to have diffi- culty in hiring and retaining �nance staff. Claims preparation by facilities and process- ing by district-level insurance schemes also have been reported to be weak, implying in substantial delays, and overall poor capacity for tracking and reporting resource flows (World Bank and MoH, 2009). Planning and Budgeting Planning is a regular activity of both regional and district health offices. All RHAs and 80 percent of the DHAs surveyed are responsible for their own planning, and all have an annual plan. The majority also prepare a multiannual strategic plan (71 percent of RHAs and 62 percent of DHAs. Budget preparation also is decentralized, and is usually formulated by the Health Administration offices themselves; however, the central level often alters the proposed budgets signi�cantly before �nal approval. Also, budget funds are often transferred with important delays, which greatly reduce the effectiveness of the formal planning and budgeting systems in place. Decentralization and Governance in the Ghana Health Sector 53 Most health facilities have their own budget, including 100 percent of Regional Hos- pitals, 94 percent of District Hospitals and the majority of Health Centers. In most cases, the facilities themselves are responsible for preparing their own budget (71 percent of Regional Hospitals and 69 percnet of District Hospitals). However, respondents indi- cated that nearly one-half of Health Centers had their budget prepared by the DHAs. Integrated planning (also known as composite planning) is not yet a standard fea- ture, although a signi�cant number of RHAs indicated participating in regional planning (43 percent), DHAs’ planning (29 percent) and facilities planning (14 percent). In spite of the mandated composite planning, only 24 percent of DHAs reported participating in district level planning (with the DA), 21 percent in facilities planning, and 12 percent in regional-level planning (�gure 4.7). It is surprising that so few DHAs report participat- ing in the planning and budgeting of health facilities under their authority. Although District Hospitals should, from a local health system perspective, be within the authority of DHAs, most often do not report to them (see below), and thus do not integrate their planning and budgeting with them. Most health offices reported participating in some sort of policy formulation, gen- erally within GHS (71 percent of RHAs and 81 percent of DHAs), and 56 percent of DHAs indicated that they participated in policy formulation jointly with the DAs. These responses are surprising, given that policy formulation is mostly a function of central level of MOH and GHS, and health policy making is unlikely to be a common feature as yet of local governments, given the weak integration between DAs and DHAs. For planning and budgeting purposes, GHS has been structured into Budget Man- agement Centers (BMCs). Over 300 BMCs were created: one for GHS headquarters, 10 RHAs, 8 Regional Hospitals, 110 DHAs, 95 District Hospitals, and 110 Sub-District BMCs. These BMCs have their own budget allocation and are in principle responsible for de�ning budget allocation and executing the budget. They are therefore, by design, an important mechanism for increasing �nancial autonomy at the local and facility Figure 4.7: Health Of�ce involvement in integrated planning 50.0% 40.0% RHA 30.0% DHA 20.0% 10.0% 0.0% District Health Facilities District Regional Don't Know Office Plan Plan Gov't/ Plan Assembly Plan Source: Author’s survey of DHAs and RHAs. 54 A World Bank Study level, and their establishment was a major step in decentralized �nancial management. However, various factors reduce the level of real autonomy local instances actually are empowered with, as discussed below. Finance and Expenditure Budget execution is seen as the responsibility of the health administration offices them- selves, but it is often a function of a specialized unit within RHAs (57 percent), while it is undertaken by the top management of DHAs (94 percent), as shown in �gure 4.8. This suggests that many DHAs perform their functions with a low level of administra- tive structuring and segregation of functions; this is probably due to the small size—in physical space and personnel—of many of these offices. In addition, it is worth not- ing, as mentioned in the review of the regulatory framework, that the larger part of district-level health expenditure is actually not de�ned or executed at this level. Person- nel Emoluments—which make up the largest part of the budget—are determined and executed centrally, and paid directly to staff bank accounts. Investment expenditure, and most of donor-funded expenditures, are also de�ned and executed centrally, the la er through MOH vertical programs. Autonomy for reallocation of budget funds appears reasonable: the vast majority of RHAs (86 percent) and DHAs (69 percent) reported having the authority to reallocate within line items, mostly without authorization from higher levels (86 and 94 percent respectively). In addition, 29 percent of RHAs and 19 percent of DHAs indicated that they may reallocate across items. However, one should take into account that this flex- ibility is limited to a small proportion of DHAs’ budget. Budget recording and reporting systems are widely used. Both RHAs and DHAs use the general government system for that purpose; but a few of them reported using also another system, often developed by the office itself. Most RHAs and DHAs reported using regular systems for monitoring and accounts rendering. The main instruments for render- ing accounts are: for RHAs, “another instrument� (57 percent); a monthly budget report (29 percent) and an annual management report (14 percent); for DHAs, a monthly budget report (87 percent), and an annual management report (43 percent). Further, all RHAs and 75 percent of DHAs surveyed indicated using a general GHS monitoring and evaluation Figure 4.8: Responsibility for budget execution 100.0% 80.0% RHA 60.0% DHA 40.0% 20.0% 0.0% Division Within Office Top Regional This office Management Health Office Source: Author’s survey of DHAs and RHAs. Decentralization and Governance in the Ghana Health Sector 55 system; 14 percent of RHAs and 62 percent of DHAs indicated that they also use other M&E systems or instruments, usually their own internal systems. Interestingly, in spite of using different M&E systems, only one DHA mentioned using a DA-based monitoring system. These �ndings suggest that regular, formal reporting and M&E systems are in place and used, which indicates a relatively advanced—at least formally—information and reporting structure. However, the number of health offices indicating multiple M&E sys- tems is cause for concern. This flexibility to use locally de�ned information and reporting systems has advantages and drawbacks. Being able to improve and expand on official systems is a positive feature, but multiple parallel and nonstandardized information sys- tems make consolidation and comparison difficult, and suggest unnecessary duplication and waste of efforts. Standardization and consolidation of the various systems in use would likely streamline the information and accounts rendering process, reduce waste, and strengthen the reliability and comparability of data across regions and districts. On paper, the country’s accountability mechanisms put in place for monitoring local expenditure and management are interesting: BMCs accounts are subject to independent auditing, and have to submit regular �nancial statements of revenues and expenditures. These statements are consolidated at the regional level and sent to GHS headquarters. However, these mechanisms appear not to be used appropriately, since the central level does not have readily available reliable data on health expenditure by each district. Even though DHAs collect and submit routine data on district expenditure, these data were not available for this report. And a Public Expenditure Review supported by the World Bank collected new data that contained substantial errors and had to be revised system- atically and validated. The survey also highlighted another important weakness regarding accountability. Even though structured management and information systems are in place, along with performance-based incentives schemes (see below), these systems are not clearly linked to a functional system of contracting or management agreement between health offices and their higher government levels or other government entities. As shown in �gure 4.9, Figure 4.9: Contracting arrangements with other public entities 60.0% 50.0% 40.0% 30.0% RHA 20.0% DHA 10.0% 0.0% Performance Based No Response Don't Know Contract With Other No Contract Contract With MOH/ Yes Other, Non- Yes, Performance Yes, Performance Contract Entity GHS Source: Author’s survey of DHAs and RHAs. 56 A World Bank Study all RHAs and 63 percent of DHAs reported having no contractual arrangement, did not respond, or did not know. Two DHAs indicated having a performance-based con- tractual agreement with MOH/GHS headquarters, and four others some arrangement with other entities, whether performance-based or not. These answers are somewhat puzzling, since this consultant was presented with an instrument (a memorandum of agreement) at the DHA level that would constitute a contractual arrangement, although a somewhat weak one. The responses may indicate either a lack of clarity about what constitutes a contractual arrangement (including MOUs) and its purpose and implica- tions, or that existing instruments are not enforced or systematically used. Human Resources Authority for determining, hiring, �ring staff seems to be unclear to local health offices, whether because of weak or unclear policies or multiple staff regimes. While most RHAs indicated that authority for determining staff numbers and composition belongs to GHS headquarters, DHAs reported an unclear or dispersed assignment of this responsibility or unclear policies (�gure 4.10).1 A similar pa ern appears when respondents were asked about the authority for determining size and composition of staff of facilities within their area. An even greater dispersion of responses shows up with respect to authority for hir- ing and �ring staff, but about half of the respondents indicated that this authority was centralized at MOH or GHS headquarters (�gure 4.11). The dispersed pa ern of authority relating to staff management seems to be at least partially associated with the existence of multiple contractual regimes. The majority of health offices indicated employing staff under “another public sector regime� (71 per- cent of RHAs and 62 percent of DHAs), while 14 percent and 18 percent respectively indicated that staff was under the general civil service regime. On the other hand, the majority of health offices hire temporary workers (57 percent of RHAs and 81 percent of DHAs), for which they have authority to hire and �re. All respondents reported main- taining a database of staff working in the facilities under their jurisdiction. It should also be noted that DAs sometimes hire or redeploy staff to perform supporting activi- ties at the DHA or particular facilities. Even though all respondents informed keeping a Figure 4.10: Authority for determining staff numbers and composition 80.0% 70.0% 60.0% 50.0% RHA 40.0% DHA 30.0% 20.0% 10.0% 0.0% This Office District Regional GHS HQ Not Defind Itself Health Health Formally Office Office Source: Author’s survey of DHAs and RHAs. Decentralization and Governance in the Ghana Health Sector 57 Figure 4.11: Authority for hiring staff 80.0% 70.0% 60.0% 50.0% RHA 40.0% DHA 30.0% 20.0% 10.0% 0.0% f ce y ce cu Q Q ce ly w el bl H H no al ffi ffi ffi /rp Its m rm S H K O O O se cc H ce O Fo ’t lth lth ic as t/r G M ffi on bl 't/ ov ea ea d O Pu D in ov tH H G is ef tG er al Th al D ric th on on ric ot ist O gi gi N ist D Re Re D Source: Author’s survey of DHAs and RHAs. staff database, it is not clear whether this database covers all the different regimes and arrangements for hiring staff. Staff incentives are a relatively common—though not standard—feature of health offices at the sub-national level: 57 percent of DHAs and 44 percent of DHAs indicated using such a system based on staff performance. Moreover, most health offices directors believe such incentives schemes are likely to promote performance of staff and admin- istrative units alike. The vast majority of respondents reported having authority to take disciplinary actions on staff, and nearly 70 percent of both RHAs and DHAs reported having taken such actions at least once in the last year. Procurement Based on the respondents to the survey, authority for procurement of supplies and ser- vices to Regional Health Administrations lies with the RHAs themselves; 29 percent of RHAs indicated this responsibility rested with a speci�c unit within the RHA office, while the others indicated the office as a whole as having this authority. 75 percent of DHAs also indicated their office were responsible for procurement, but a signi�cant number provided an array of different answers: the District Assembly (1 DHA), the Regional Coordinating Commi ee, RPCC or office (1), MOH headquarters (1), another public office (1), and 4 did not know or did not provide an answer (�gure 4.12). This dispersion signals that either policies and rules regarding procurement are unclear to health office managers, or that DHAs obtain their supplies and services from different channels (or both). This may also be related to the availability of different sources of funding which may be linked to different procurement practices (see section on district expenditures below). A similar pa ern—though with less dispersion—was observed with respect to responsibility for procurement of drugs. Respondents also indicated that even though most facilities have their own budget, most depend at least partially on procurement from DHAs and RHAs. Maintenance and 58 A World Bank Study Figure 4.12: Responsibility for drugs procurement 45.0% 40.0% 35.0% 30.0% RHA 25.0% 20.0% DHA 15.0% 10.0% 5.0% 0.0% ce ce ce Q Q e ce ns H H ffi ffi ffi ffi po S H O O O O H s O Re is lth lth G is M Th Th ea ea o N tH H in al ith ric on W ist gi D t Re en tm ar ep D A Source: Author’s survey of DHAs and RHAs. support services for RHAs and DHAs are usually procured by the offices themselves, with few reporting that they depend on higher level offices (RHAs in the case of DHAs, or MOH/GHS headquarters). Drugs and supplies are usually stocked at RHAs and DHAs themselves, but 43 per- cent of respondents did not know or did not respond. The relatively large number of “don’t know� and no responses in this topic suggests that procurement responsibilities and flows may be more diverse or nonstandardized than is assumed. Nearly all health offices reported preparing an annual procurement plan (100 per- cent of RHAs and 81 percent of DHAs). Health Care Management With respect to coordination and supervision of health service provision, the survey indicated a dispersion of responsibilities at the regional level: while 57 percent of RHAs reported being themselves responsible for this activity, another 57 percent responded that headquarters (of GHS and/or MOH) were responsible, and 14 percent did not know. The pa ern was somewhat clearer among DHAs, where 87 percent indicated that this responsibility rested with the DHAs themselves; however, nearly one-third indicated the regional health office or MOH/GHS headquarters as responsible, and another 12 per- cent did not know. Monitoring, supervision, and the overall interaction of the DHAs with District Hos- pitals are problematic in many districts. Both RHAs and DHAs interact directly with DHOs, and district health managers often complain that DHOs, though operating at the district level, actually report to RHAs and are not integrated in the district health system. Figure 4.13 shows that both RHAs and DHAs supervise DHOs through occa- sional visits or meetings and monitoring of hospital indicators. The proportion of RHAs involved in these activities is higher than that of DHAs, con�rming the relatively low Decentralization and Governance in the Ghana Health Sector 59 Figure 4.13: Supervision of and interaction with District Hospitals 90.0% 80.0% 70.0% 60.0% RHA 50.0% 40.0% DHA 30.0% 20.0% 10.0% 0.0% s s s g eds s er nse now ing ing ing tor nin Oth Ne spo ica eet eet eet tK lan Ind Re Of rM rM rM n' ly P Do ent No Of sO sO sO ear ssm ng isit isit isit nt Y ori sse yV ly V al V Joi nit nt A ekl ion nth Mo We Joi cas Mo Oc Source: Author’s survey of DHAs and RHAs. interaction between DHAs and DHOs. Further, most often interaction appears to hap- pen only occasionally and not systematically: only 29 percent of RHAs and 31 percent of DHAs indicated that they made regular routine visits to hospitals (at least monthly). Joint involvement in planning or needs assessment was mentioned by few DHAs. The interaction and supervision with sub-district health services are lower for RHAs, but stronger for DHAs, when compared to DHOs. This is expected, since sub- district services (provided by Health Centers and Community Health Services) are the full and main responsibility of DHAs. As shown in �gure 4.14, regular and frequent visits and meetings from DHAs are common—though not the generalized practice that would be expected. Joint planning and needs assessment is also more frequent than with District Hospitals. Surprisingly, RHAs are also involved in sub-district supervision and monitoring in a signi�cant manner (86 percent reported making occasional visits and meetings, and 57 percent monitor sub-district indicators). It is not clear if the RHAs’ involvement operates as support to DHAs’ supervision activities or duplicates them; but the apparent duplication of efforts deserves some a ention. Knowledge and Perceptions of Decentralization Awareness of the government’s decentralization policy is limited at the local level: while 43 percent of RHAs reported being pre y much aware, and 57 percent somewhat aware of this policy, 37 percent of DHAs indicated having li le or no knowledge of it. In addi- tion, only 25 percent of DHAs and 14 percent of RHAs have a clear vision of what would be their responsibilities under the decentralized system (�gure 4.15). However, when asked what would be their responsibilities, health offices direc- tors provide a list that is inconsistent with the likely distribution of responsibilities in a devolved system (�gure 4.16). Both RHAs and DHAs indicated nearly all functions as 60 A World Bank Study Figure 4.14: Supervision and interaction with sub-district health services 90.0% 80.0% 70.0% 60.0% 50.0% 40.0% 30.0% 20.0% 10.0% 0.0% s s s g eds s her nse ing ing ing tor nin Ot Ne spo ica eet et eet lan Me Ind Re of rM rM yP ent or No of so so arl sits sm ng Ye isit isit ses ori Vi yV al V nt As nit Joi ly RHA ekl ion nth Mo nt We Joi cas Mo DHA Oc Source: Author’s survey of DHAs and RHAs. their responsibility, although with differing emphasis. For instance, both indicted coor- dination of primary care services as their responsibility in equal proportion, when PHC would expectedly be the sole responsibility of the districts. DHAs also give relatively li le emphasis on planning and monitoring activities. RHA and DHA officers were also asked what they knew and thought about GOG’s policy of devolution of the health system to local governments. Such devolution implies the transfer of local level facilities and offices to the DAs’ authority, as stated in several legal and policy documents on decentralization (see Section 3). In this regard, the survey Figure 4.15: Awareness of role and responsibilities under decentralization 0 Not at all 100% 0.1875 90% No, very little 80% 70% 0.25 Yes, somewhat 60% 0.857142857 Yes, pretty much 50% 40% 0.375 30% 20% 0.142857143 0.1875 10% 0% RHA DHA Source: Author’s survey of DHAs and RHAs. Decentralization and Governance in the Ghana Health Sector 61 Figure 4.16: Expected main responsibilities under decentralization 120.0% 100.0% 80.0% 60.0% 40.0% 20.0% 0.0% n ion n on s re s re ent ces nt ea ff ug lie tio tio Sta ctu Ca me Ar uti gem uat rvi Dr pp ara ara tru uip xec ary the of Su Se al of p rep ana ras Ev Pre Eq ing rim et E of in of ent et P Inf ff M nd Fir in are ent ent fP ing em dg ga dg nt in ll C em no em g& Bu nn Sta cur me Bu nt rin Pla cur tio cur fA Pro me rin est ito ina Pro Pro Hi no est Inv on ord Inv nM tio Co RHA ina Pla ord DHA Co Source: Author’s survey of DHAs and RHAs. shows a substantial lack of clear information and understanding of the official government policy, and—especially at the district level—some resistance to the idea. To some extent, a split between RHAs and DHAs’ perceptions is visible. On the issue of transferring local health offices (DHAs) to DA authority, 71 percent of RHAs and 100 percent of DHAs were unsure or disagreed with that policy (�gure 4.17). As shown in �gures 4.18 and 4.19, most RHAs see the transfer of Health Centers and Dis- Figure 4.17: Perceptions about transfer of DHAs to DA authority (percentage of agreement) 60.0% 50.0% 40.0% 30.0% RHA DHA 20.0% 10.0% 0.0% Yes, Somewhat Unsure No, Somewhat Don't Know Agree Disagree Source: Author’s survey of DHAs and RHAs. 62 A World Bank Study Figure 4.18: Perceptions about transfer of facilities to DAs: Health Centers (percent of agreement) 70.0% 60.0% 50.0% RHA 40.0% DHA 30.0% 20.0% 10.0% 0.0% Yes, Yes, It Doesn't Unsure No Don't Know Certainly Probably Matter Source: Author’s survey of DHAs and RHAs. trict Hospitals to DA authority as positive, while most DHAs do not have a clear opinion on that. The la er actually seems to hide an unformulated resistance to the transfer. However, 69 percent of DHAs and 43 percent of RHAs perceive district health offices as best prepared to organize and manage health care services. Surprisingly, the majority of RHAs do not trust DHAs to perform that function, and see GHS headquarters as bet- ter prepared, in a rather central-focused vision. Overall, RHAs are more optimistic on the feasibility and impact of decentralization than DHAs; as shown in �gures 4.20 and 4.21, 94 percent of the la er either disagree or are unsure about feasibility of devolution (compared with 57 percent of RHAs) and 81 percent are unsure or negative about its impact (against 29 percent of RHAs). This Figure 4.19: Perceptions about transfer of facilities to DAs: Hospitals (Percent of agreement) 70.0% 60.0% 50.0% RHA 40.0% DHA 30.0% 20.0% 10.0% 0.0% Yes, Yes, It Doesn't Unsure No Don't Know Certainly Probably Matter Source: Author’s survey of DHAs and RHAs. Decentralization and Governance in the Ghana Health Sector 63 Figure 4.20: Perceptions on the feasibility of decentralization 70.0% 60.0% 50.0% RHA 40.0% DHA 30.0% 20.0% 10.0% 0.0% Yes, Yes, Unsure No, Don't Know Totally Somewhat Somewhat Agree Agree Disagree Source: Author’s survey of DHAs and RHAs. skepticism about the feasibility and impact of decentralization constitutes a potentially important obstacle to the devolution policy of GOG. Insufficient information and dis- cussion on the policy is likely to contribute to this skepticism, although when asked about the advantages and disadvantages of decentralization, respondents gave consis- tent responses. The most cited advantages are: decision power closer to users (100 percent of RHAs and 81 percent of DHAs), be er adaptation to local needs (100 percent and 69 percent), greater flexibility and agility (86 percent and 62 percent), and easier cross-sector integra- tion (86 percent and 50 percent). Among the disadvantages or threats, the most cited were: insufficient funding (100 percent and 60 percent); weak local management capac- Figure 4.21: Expectations on positive impact of decentralization on health system performance 80.0% 70.0% 60.0% RHA 50.0% DHA 40.0% 30.0% 20.0% 10.0% 0.0% Yes, Unsure No, Somewhat Don't Know Somewhat Disagree Agree Source: Author’s survey of DHAs and RHAs. 64 A World Bank Study Figure 4.22: Rating of capacity of DHAs (top) and DAs (bottom) 0.0% 100% 18.8% 42.9% 80% 25.0% 60% 14.3% 31.3% 40% 42.9% 20% 25.0% Very 0% Weak RHA DHA Weak Average Strong 100% 14.3% 50.0% 80% 42.9% 60% 40% 37.5% 42.9% 20% 12.5% 0% RHA DHA Source: Author’s survey of DHAs and RHAs. ity (71 percent and 60 percent), difficult sector fragmentation (86 percent and 53 percent), and health system fragmentation (57 percent and 53 percent).2 The technical and managerial capacity of RHAs, DHAs and DAs was rated by respondents as average or below (�gure 4.22). RHAs rated themselves higher (43 percent strong and 14 percent average) than DHAs (44 percent weak or very weak). However, both rated the DAs’ capacity to absorb health functions much lower; RHAs rated them as weak (43 percent) or very weak (14 percent); but DHAs, which have closer and more frequent contact with DAs structure and staff, rated them as very weak (50 percent) or weak (37 percent); no RHA or DHA rated any DA as having a strong capacity. This per- ception of local governments’ capacity con�rms and explains the skepticism with which subnational entities and especially DHAs view decentralization, but may also be biased by the respondents’ clear preference for maintaining the current system. Respondents were also asked to rate the DAs capacity for a number of typical health related functions (�gure 4.23). The only strength perceived in DAs is the capacity to plan and prepare a budget, rated as very strong by both RHAs and DHAs. Functions related to procurement, human resource and health care management were rated as very weak. Decentralization and Governance in the Ghana Health Sector 65 Figure 4.23: Mapping of DAs capacity for typical functions or dimensions 0.0 1.0 2.0 3.0 4.0 Planning Budget preparation Percentage of respondents who reported Financial management Procurement process Budget execution/spending RHA Number of existing staff Qualification of existing staff DHA Human resource management Political will to engage in health Clarity of role and functions Monitoring & evaluation Health care/system knowledge Funding level/sufficiency Source: Author’s survey of DHAs and RHAs. This pa ern seems to reflect the main responsibilities already transferred to local gov- ernments (mostly budget preparation and management). Because this survey was limited in scope and sample size, and did not include a sample of DAs, these �ndings are preliminary and mostly qualitative. Extending this type of capacity mapping to all districts—and a sample of DAs—could help identify more precisely existing strengths and weaknesses and thus support the development of a focused strategy for increasing local capacity in key areas. Conclusion Decentralization—especially in the form of devolution—in health should not be seen as a packaged deal, because local capacity varies substantially across different functions and responsibilities, and other factors will influence the feasibility and impact of decen- tralizing speci�c functions, such as: the efficiency arising from economies of scale, the technical content of each function, and local characteristics. Table 4.3 provides a prelimi- nary assessment of strengths and weaknesses for decentralizing particular functions to different levels of the health system, based on the rapid assessment presented above. Pursuing this exercise based on more detailed data regarding DAs would be quite useful in be er planning and implementing the decentralization process. This survey did not include a sample of DAs, so an objective assessment of their capacity is not possible at this point. However, the 2006 Public Expenditure Tracking Survey (PETS) found various problems in decentralized resource management at the central and DA levels, including inconsistencies in �nancial data regarding transfers from central government to DAs, substantial delays in effecting these transfers,3 incon- sistent record keeping at the DA level, poor accountability of funds received with sig- ni�cant leakages, and weak overall managerial capacity. The PETS �ndings suggest that 66 A World Bank Study Table 4.3: Weaknesses in capacity and process by level of government and function Capacity Level Intensity Present Gap Governance Central +++ +++ + Regional + ++ 0 District +++ + +++ Facility ++ ++ + Accountability Central ++ ++ + Regional + ++ 0 District +++ + +++ Facility +++ ++ ++ Policymaking Central +++ ++ ++ Regional + 0 ++ District ++ 0 +++ Facility + 0 ++ Planning & Budgeting Central +++ +++ + Regional ++ + ++ District ++ + ++ Facility + NA ++ Monitoring & Evaluation Central +++ + ++ Regional ++ + ++ District + 0 ++ Facility + NA + Care coordination Central + + + Regional ++ + ++ District +++ + +++ Facility ++ + ++ HR Management Central +++ ++ ++ Regional + + + District ++ 0 +++ Facility + + + Procurement Central + ++ 0 Regional ++ ++ + District +++ + +++ Facility + + + Finance & Execution Central ++ ++ + Regional + + + District +++ + +++ Facility + NA ++ Source: Author’s survey of DHAs and RHAs. Decentralization and Governance in the Ghana Health Sector 67 local governments suffer from signi�cant weaknesses regarding the already devolved responsibilities. It should be noted that similar problems were found in the deconcen- trated systems and flows (including NHIS), as encountered in the PETS and noted in the section titled “Data sources and issues.� Within the devolution design, sub-districts are supposed to be strengthened as administrative units and become the main level for provision of primary health care services. The key element of this strategy is the establishment of Community Health Planning and Services (CHPS). However, only 285 sub-districts had been formally implemented in 2007, 83 percent of which in �ve regions (Western, Upper East, East- ern, Central and Upper West). Furthermore, in most cases sub-districts have li le or no autonomy or resources of their own. Although 110 BMCs have been established at the sub-district-level, according to the 2009 Public Expenditure Review, only the Ashanti Region reported �nancial allocation to sub-districts separated from the DHA (see �g- ure 5.10 below). Facilities also enjoy some level of formal autonomy. Most hospitals and many health centers have a commi ee a ached to them, which has mostly an advisory role. All hospitals have been made into BMCs, which means they are planning and budgetary units. National hospitals are autonomous units with a governance board. However, as reflected in the survey of RHAs and DHAs, facility autonomy is limited by the central- ization of decision-making and payment process for staff, some procurement and invest- ment. Overall, the larger part of facilities’ expenditure is controlled from the central government. Notes 1. The sample for this section was signi�cantly smaller than intended: �ve of the six RHAs, and 16 of the 24 DHAs invited a ended the workshop. No DHA from the Northern Region was present. This may have introduced a bias toward more developed districts of the coastal and forest zones (�gure 4.1). 2. The fact that DHAs indicated multiple responses, adding up to more than 100 percent is also suggestive of unclear responsibilities in that ma er or unclear rules. 3. These alternatives were offered in the questionnaire and may thus have influenced the responses; open responses were not cited as frequently. 4. Transfers were often sufficiently large as to reduce substantially the amount of budget funds released during the year. CHAPTER 5 Analysis of Regional Resource Allocation T his chapter looks at expenditure pa erns and some aspects of �scal decentralization at the district level in Ghana. The next section discusses data sources and issues encoun- tered in the collection and analysis of relevant data. The section titled “Some Evidence on Fiscal Decentralization and Health� presents some evidence of �scal decentralization in other countries. The main �nancial flows and pa erns to and within district level are sum- marized in the following section. The section titled “Resource allocation pa erns analyzes expenditure pa erns at the district level of GHS structure. The section titled “Regional Variation in Health Expenditure per Capita� compares per capitation health expenditure across regions and districts, and discusses the possible factors affecting it. The section titled “Summary of Findings from the Resource Allocation Analysis� summarizes the main �nd- ings of this chapter. Data Sources and Issues The data sources used for this chapter include GHS Consolidated Statement of Revenue and Expenditure for 2008, expenditure data by BMC from the Public Expenditure Review (PER) exercise (PER 2010), the GHS list of facilities and Bedstate Returns by district, and population data for the same year. We decided not to use budget allocation data because of the frequently large differences between budget allocation and actual expenditure.1 The process of gathering and analyzing data on resource distribution at the local level highlighted important problems in data collection and management that reflect on the ability to analyze health expenditure. First, there is a problem of inconsistency in the number, classi�cation and names of local governments. The number of districts has been rising steadily in recent years, and at the same time some of them were transformed into municipalities or metropolitan areas. However, different data sources have not kept pace with this process; for instance, some new districts may have facilities allocated to them, but budget and population estimates refer to the original district before separa- tion. Adjustments were made in the data to account for these changes, to the extent that information was available. Furthermore, existing data sources and documents use dif- ferent names for the same LG unit, or different classi�cations; this may be due to delays in updating databases, or to inconsistent writing of district names. These inconsistencies between different documents and databases were found systematically when matching population data, budget �gures, and expenditure data. Obtaining district expenditure data was problematic, in spite of the existing system for collecting, reporting and consolidating of health expenditure within the MOH/GHS sys- tem. Such data is collected at the district level manually, and sent to the Regional Health Administration offices for consolidation; RHAs then enter the data into an electronic system (ACCPAC); consolidated data are then sent electronically to GHS headquarters. 68 Decentralization and Governance in the Ghana Health Sector 69 However, it was not possible to obtain the data produced in that system, for two rea- sons. First, as explained by GHS officers, the ACCPAC system records only consolidated data by region, and disaggregated data by district apparently cannot be retrieved once entered. Second, requesting them from regional or district offices proved very cumber- some. Third, substantial differences—possibly data errors—were found in the succes- sive partial data obtained, suggesting low reliability of existing data. Fortunately, at the same time this study was performed, a new Public Expenditure Review exercise, funded by the World Bank, was under way. A preliminary version of the PER data showed substantial data errors, which were later revised and corrected. Another data issue is the inconsistency in the organization and presentation of exist- ing data—especially �nancial data—with li le information provided on de�nitions and scope or aggregation of each �gure provided. This makes it difficult to compare �gures for budget allocation with actual expenditure. Available data indicate wide variations that seem to relate to discrepancies in classi�cation or aggregation rather than delays in disbursement of funds or year-to-year variations. For example, the 2009 budget allo- cation gives a �gure of GHC 17 million for DHAs, while expenditure data from Con- solidating Statements for 2008 give 195 million. Another example is that a signi�cant number of BMCs appear as not having personnel expenditure; this is most likely an issue of reporting error, although it might be related in some cases to the division of old districts into new ones. Finally, discrepancies were noted also in the number of facilities reported and incon- sistencies in their classi�cation. Different sources (GHS website, budget documents, and others) seem to work with different classi�cations and numbers of facilities, without making clear how the different types of facilities were aggregated. In summary, structured information systems are in place, which should provide valuable information for analysis; but the inconsistencies and reporting errors, with insufficient data monitoring and checking, result in unreliable or imprecise data for criti- cal information. As a result, some uncertainty exists regarding the precise value com- puted for per capita expenditure in particular districts. However, the general pa ern observed and described below is unlikely to be greatly affected by these issues. Some Evidence on Fiscal Decentralization in Health The degree of �scal decentralization varies substantially across countries. If measured by the proportion of local government expenditure relative to total government spend- ing, it is highest in Western Europe and the United States (mean value of 26 percent), average in Asia (15–20 percent) and Latin America (11 percent), and lowest in Africa (5 percent)2 and the Middle East (United Cities and Local Governments 2009). In Scan- dinavian countries, local governments’ expenditure reaches 52 percent, and some Asian countries such as Japan, China, and the Republic of Korea come close to that value. Among African countries, only in one—South Africa—this proportion reaches 10 per- cent; in most other countries, it varies between 3 and 6 percent. The �gure for Ghana was 6.2 percent in 2005 (Decentralization Policy Review 2008). This suggests that in spite of recent efforts, African countries have achieved a much lower degree of decentralization than other regions, if expenditure responsibility is used as a proxy.3 With respect to the health sector, similar variation across countries is observed in the existing arrangement for �nancing and providing health services, and the extent 70 A World Bank Study to which responsibilities have been transferred to local governments. Of the 21 African countries reviewed in United Cities and Local Governments (2009), in 19 basic health services were provided at the local level. In the Asia-Paci�c region, the proportion was of 8 out of 12 (in Australia, Malaysia, New Zealand and Thailand, LGs do not provide such services). In some federal countries (such as Canada and Australia), health expenditure is an exclusive responsibility of states or provinces, even though the federal government �nances a signi�cant part of the total. In regional countries, such as Spain and Italy, health services provision is a joint responsibility of regional and local governments, and are jointly �nanced, but in different proportions (Bordignon and Turati 2003). It is worth noting that several countries have maintained a centralized health system, with (rela- tively) good results; the United Kingdom and France (the la er through its Social Secu- rity system) are two examples among developed countries. The different pa erns of health �nancing are presented in Figure 5.1 for a few illus- trative countries. The �gure shows that four different pa erns arise: (i) countries where health expenditures are fully or mostly decentralized to local governments (Scandina- vian countries and Italy); (ii) countries which have decentralized mostly to state or pro- vincial governments (this is the case in federal states such as Spain, Australia, Canada and Swi erland); (iii) countries where health expenditure remains fully or mostly cen- tralized (United Kingdom, USA and Israel); and (iv) countries where a Social Security system is the major �nancer of the health system (mostly western Europe countries as France, Germany, Netherlands and Belgium). Brazil in Latin America, Tanzania and Uganda in Africa, are among developing countries with the highest proportion of local expenditure (around 30 percent), but in all these countries the largest share comes from the central government. Brazil is atypical among federal states in its decentralization Figure 5.1: Fiscal decentralization in health across countries 100% 90% 80% 70% 60% SocSec 50% Local 40% State 30% Central 20% 10% 0% Sw rk n Fin ly d Au ain C lia ite itzerl a ing d m A ael Ge nce y Be nds rea m oth ilip zil es Ind ndia Ar esia Ta tina Ug ia a Sw anad and ede lan d K an Ne rman US n Ita do Ko lgiu a pin a Bra Isr Sp nza nm ,S Fra rla str gen on I the De Ph Un Source: OECD (Government Expenditure by level and function, 2006–2008) and NHA for individual countries (reference year varies). Note: Private expenditure is excluded. Decentralization and Governance in the Ghana Health Sector 71 to local government. The �gure also illustrates the variety of �nancing arrangements across countries, and highlights the fact that no one single arrangement is best: countries known for having a (relatively) successful health system have very different forms of centralization/decentralization. The Financing of District Health in Ghana The health sector in Ghana is �nanced by four different sources: the central government budget (which may be executed at different levels of government), local government funds (whether from their own revenues or general transfers from the center), donors (with a large part of donor funding channeled through the budget), and households out-of-pocket expenditure; spending by private enterprises is quite small. The relative importance of the different sources and channels of �nancing is not known precisely in Ghana. The country’s �rst and only National Health Accounts (NHA) exercise was conducted in 2006, based on 2002 data, and its estimates have not been updated since then. That study estimated that Out-of-Pocket (OOP) household expenditure was the largest source of funding for the health sector, with 24 percent of Total Health Expendi- ture (THE), while MOH/GHS, DAs and donors channeled 20 percent each (Ghana NHA 2002). The overall composition is shown in �gure 5.2. The World Health Organization monitors health expenditure of each country based on data informed by countries’ government and supplemented by NHA assessments when available. The composition for 2002 and 2008, given in �gure 5.3, is quite different from that of the NHA data. However, the comparison between the two years highlights Figure 5.2: NHA expenditure data for 2002 Donors Other NPISH (Donors) 1% 1% 6% MOH 20% MOE Households OOP 2% 24% MLG&RD 6% Regional Governments 8% Private Corporations 7% District Assemblies Quasi-Govt 19% 6% Private insurance Social Security 0% 0% Source: Ghana National Health Accounts 2002. 72 A World Bank Study Figure 5.3: Health expenditure composition by WHO estimates 100% NGOs 90% Private insurance 80% 70% User fees (IGF) 60% Out-of-Pocket 50% 40% Foreign Assistance 30% 20% Social Security 10% Local Governments 0% 2002 2008 Central Govt/MOH Source: WHO Ghana Health Expenditure data. one of the two main trends in Ghana’s health expenditure in the last decade: the growth of national health insurance from its inception in 2003. The �gures suggest a strong impact of the NHIS on households’ expenditure, including both through user fees (the “cash-and-carry� part of IGF) and out-of-pocket expenditure.4 A recent report from the World Bank indicates that the proportion of OOP as a proportion of private expenditure decreased from 79.5 percent in 2000 to 78.8 percent in 2006 (Beciu and Haddad 2009), a very small change. However, one should note that the NHIS membership picked up after 2006, and that the impact of insurance protection on actual OOP may take some time to materialize. The second trend is an important result of the decentralization process. Decentral- ization has increased the proportion of expenditure spent at the district level. Alloca- tion to district-level facilities and services has increased from 22.8 percent in 1996 to 34 percent in 1997 and 50.7 percent in 2007 (Financial Statements); regional spending increased from 17 to 25 percent but fell back to 15 percent in 2007. District-level health services and programs in Ghana are funded through �ve main sources and channels: (i) Central government budget allocated to local facilities, programs and administra- tive offices; (ii) Internally Generated Funds (IGF); (iii) central government transfers to DAs; and (iv) local government own revenues. However, it is difficult to disentangle the different sources of �nancing for these district-level expenditures. Budget and report- ing documents disaggregate expenditure by level of government and BMC, but do not distinguish district-level expenditure funded by MOH/GHS budget, by transfers from MLGRD, or by DAs funds. Some approximation is, however, possible. The Controller Accountant General’s Office (CAG) pays salaries and bene�ts directly to civil servants’ accounts working at the district level, and Personnel Emoluments (Item 1) represent 54 percent of recurrent expenditures at the district level (and 40 percent of total expenditures in 2008), and over Decentralization and Governance in the Ghana Health Sector 73 90% of Central Government recurrent disbursements for health. A second flow regards investment expenditure (Item 4), also spent directly by the central government, usually by MOFEP; Item 4 expenditure accounted for 15 percent of public expenditure on health in 2007 and 48 percent in 2008 (the la er is due to a heavy investment in regional hos- pitals). The third flow relates to central government expenditure—funded by general state revenue—on Items 2 (Administration) and Item 3 (Service), including MOH direct spending through its vertical programs, and MLGRD transfers for capacity building and other initiatives. This flow is of unknown magnitude, but likely to be relatively minor, since over 90 percent of central government recurrent health spending relates to personnel. A fourth �nancing flow to district health is funded by Development Partners and channeled through MOH (or MOFEP in the case of investment expenditure). These funds accounted for 4 percent of all district-level recurrent expenditures but the majority of Service Expenditure (Item 3). They usually are passed through to MOH budget, and then to facilities and administrative offices at regional or district levels. The largest part of MOH programs are funded in that manner. Internally Generated Funds (IGF) consti- tute the �fth flow, and include revenues from user fees charged at the facility level for certain health services and goods; they accounted for 23 percent of district-level expen- diture in 2008. Finally, payments from the National Health Insurance System for services ren- dered to its members by local public facilities amounted to 6 percent of total expendi- ture (MOH Audited Financial Statements 2008). NHIS provides health care coverage to an increasing proportion of the population—15.8 percent in 2006, and 35 percent in 2008 (DHS 2008)—but constitutes a modest source of revenue for the public health system.5 The main flows for �nancing district health services are shown in Figure 5.4, with its approximate percentage composition. The �gures for 2007, which are not distorted by the heavy investment expense of 2008, are: GOG budget 49.6 percent; IGF 1799 percent; donor funds 17.5 percent (including Health Fund/Budget Support 6.9 percent), MOH Programs 2.8 percent and HIPC 3.0 percent); National Health Insurance payments 5.9 percent; Financial Credits 4.7 percent; and other sources. DAs’ expenditure from their own and transferred revenues constitute another �nancial flow for which no estimates are available. According to the LG Act of 1993, DAs own revenues include: Basic, Special and Property Rates, Licenses, Fees, Royal- ties, Rent and other sources. In addition, the 1992 Constitution mandated, and the Dis- trict Assemblies Common Fund Act (1993) established, that the (central) government should allocate at least 5 percent of its total revenues to a District Assemblies Common Fund (DACF),6 aimed at strengthening structures and capacity at the district level. The DACF has thus instituted a new flow of funds transferred by the central government to DAs, and represents the main channel of �scal decentralization in Ghana so far.7 Successive legislations have required that the DAs allocate such funds to speci�c func- tions and purposes—including health and education—according to the responsibilities devolved to these local governments. According to the PETS 2007, in 2005 and 2006 the DAs received from DACF about GHC 800 million each year, of which 42 percent was earmarked for education (PETS 2007).8 However, as mentioned above, the degree to which DAs are performing the health functions already devolved to them is unclear, and so is the amount DAs are spending on health. 74 A World Bank Study Figure 5.4: Financing flows for district health services GOVERNMENT OF GHANA MOH PROGRAMS DEVELOP. MOFEP MLGRD MOH & BUDGET SUPPORT PARTNERS NHIF PERSONNEL EMOLUMENTS GHS INVESTMENT SERVICE AND ADMINISTRATION NHIF SUBSIDY DISTRICT ASSEMBLIES 4% 40% 27% ?% 6% DMHIs SUB- ?% DHA DHO DISTRICT ?% F IG 23% Source: MOH Audited Financial Statements 2008. A recent report on Ghana budgeting process (Abeka-Nkrumah et al. 2009) has found signi�cant weaknesses in the process, in a pa ern that corroborates the �ndings in this report: ■ Weak �nancial management capacity at all levels of the health system ■ Weak reporting practices associated with important delays and information quality ■ Low use of instruments and tools for budget and �nancial management at the decentralized levels ■ Substantial budget and cash flow variations that imply redoing the budget sev- eral times ■ Bureaucratic procedures that are followed without a focus on actual results ■ Low effective local authority over budget allocation and spending. In particular, the authors found that “the budget preparation process at BMC level has to be done twice or even more. The �rst budget which is prepared is based on the ceil- ing provided by the MOH and the policy priorities for the coming year. This budget is submi ed to the regional and central levels together with all other budgets submi ed to MoFEP. But the actual resources allocated to the BMCs differ in general substantially Decentralization and Governance in the Ghana Health Sector 75 from the original ceilings. These irregular budget cuts mean that the BMCs have to pre- pare their budgets again based on actual resources received. This tends to undermine the budget preparation process and make it difficult for a BMC to plan ahead. As a result, some BMCs do not put any effort in the �rst budget preparation process knowing that after receiving their actual ceiling they have to do it all over again.� Resource Allocation Patterns Expenditure on district-level facilities and services have been around 50 percent of MOH expenditure in the last three years (�gure 5.5), which would place Ghana among the top countries in terms of �scal decentralization in health (see data for other countries in the section titled “Data sources and issues� above). However, this includes expenditure spent on behalf of the local level, but not controlled by local authorities. The degree of real �scal decentralization is limited by several factors. First, more than 50 percent of district-level expenditure (two-thirds in 2008) is executed centrally on behalf of district- level offices and facilities (staff payment and investment expenditure). Second, both DAs’ spending and MOH non-personnel recurrent expenditure are limited by centrally de�ned guidelines for required allocations. These earmarked allocations—including donor sponsored programs and DACF resources—constitute the larger part of service expenditure. The 2007 PETS indicated that 42 percent of DACF funds were retained at the central level for several earmarked programs or initiatives (some of them related to health).9 Third, transfers to local level have not been made in a regular and timely fashion, and cash releases have been short of planned. A new District Development Fund Facility has been proposed to consolidate and increase �nancing streams used for strengthening district development, and provide incentives for local officials to fol- low GOG policies and processes to ensure accountability and transparency. Overall, the Figure 5.5: Distribution of recurrent health expenditure by type and level, 2006–08 100% 90% Subventions, etc 80% District Health Services 70% Regional Health 60% Services 50% Psychiatric 40% Hospitals 30% Teaching 20% Hospitals 10% GHS Headquarters 0% 2006 2007 2008 Source: MOH Financial Statements 2006–08. 76 A World Bank Study Figure 5.6: National and district health spending by head item, 2008 Item 4: Item 1: Personnel 25% 28% Item 1: Personnel Item 4: 41% 48% Item 3: Service Item 3: Service 20% 28% Item 2: Administration 4% Item 2: Administration 6% Source: MOH Audited Financial Statements 2008. real local authority over �nancial resources, both within the current MOH/GHS system and within Das’ responsibilities, is much more limited than the high proportion of fund transfers suggest. Consolidated public expenditure (including IGF and donors’ funds) on health in Ghana in 2008 has emphasized capital investment, and personnel expenditures have represented only 28 percent of the total (�gure 5.6a). But as a proportion of recurrent expenses, personnel represented over half of the total, which is consistent with interna- tional evidence. However, at the district level, personnel represents a proportion well above the national �gure: as shown in �gure 5.6b, personnel expenditure accounted for 41 percent of total expenditure, but a proportion of recurrent expenditure similar to the country average (MOH Audited Financial Statements 2008). The different expense categories (head items) are funded by different sources, in a pa ern similar to national health expenditure. As shown in �gure 5.7, there is a great deal of specialization in �nancing sources. Personnel (Item 1) at the district level is almost exclusively funded by GOG budget (2.5 percent are funded by IGF). Administra- tion (Item 2) is funded mostly by IGF, with a 12 percent contribution from GOG. Ser- vice (Item 3) is funded by a mix of IGF (42 percent), donor-sponsored MOH programs (33 percent) and NHIS reimbursements (24 percent). As for investment (Item 4), it is almost exclusively funded by Financial Credits. It is worth noting that Item 3, which represents the major part of non-personnel recurrent expenditure at district level, is fully funded by non-GOG sources. The detailed distribution of health expenditure by level of care and type of facility appears in �gure 5.8. It shows that in 2008 the largest share of total recurrent expen- diture was allocated to District Health Administration offices. According to the PETS 2007, this is so because DHAs often spend on behalf of their sub-districts (health centers and CHPS) or other facilities without recording that these expenditures are for those units. The 2007 PER seems to con�rm this: �gure 5.9 shows the distribution of sub- national expenditure by type of facility and office, and indicates that most district-level expenditure is allocated to district hospitals (DHOs—52 percent of subnational spend- ing) and sub-district facilities (Health Centers and Clinics) and programs (CHPS), with 19 percent. Other discrepancies between the two sources (MOH Audited Statements and Decentralization and Governance in the Ghana Health Sector 77 Figure 5.7: Financing sources by head item in district expenditure 200 NHIS 180 160 HIPC 140 Million GHC 120 F/Credits 100 80 MOH Prog 60 HFund 40 20 IGF - Item 1 - Item 2 - Item 3 - Item 4 - GOG Personnel Administration Service Investment Source: MOH Audited Financial Statements 2008. PER) are apparent, and reveal the lack of consistency in expenditure classi�cation and/ or possible data errors. Figure 5.8 also shows that MOH Headquarters report an important expenditure in the Service category, which appears to be related to the ministry’s vertical programs. We computed the mean expenditure value, for the different types of facility and unit, and compared it across different sources. Table 5.1 below shows sizeable differences across types of facilities, with Teaching Hospitals receiving the largest budget allocation Figure 5.8: Health expenditure by level and type of service, 2008 200 180 160 140 Million Cedis 120 100 80 60 40 20 - Q s s s s s s Cs BV O A O TI A O -H s/H H RH RH H H SU H s/P D D PC O O M TH Item 3: Service Item 2: Administration Item 1: Personnel Source: MOH Audited Financial Statements 2008. Recurrent expenditures are shown on top and investment below. 78 A World Bank Study Figure 5.9: Expenditure allocation at regional and district levels RHD SDG 19.1% TI 0.7% RHO 7.0% 9.9% MHO DHA 2.9% 8.0% DHO 52.3% Source: PER 2009. by far, and District Hospitals the smallest. RHAs also receive a much greater allocation than District offices. But the table also shows substantial differences across data sources. Even though it is well known that actual expenditures can be different—and usually lower—than budget allocations, and that the table compares two different years, some of the differences are too large to be explained by the expected variation year-to-year or between budget and execution. They are more likely related to differences in classi�ca- tion and data errors. To check whether the observed differences among facility types were justi�ed, we also compared the main features—size, production and expenditure—for different categories of hospitals. The results, shown in table 5.2, suggest that District Hospitals receive a proportionately larger allocation relative to their production, and thus have a substantially higher cost per bed or per patient day. A more detailed analysis of hospital Table 5.1: Mean expenditure* by type of facility or unit Facility/Unit Budget 09 Expenditure 08 Expenditure 08 (PER) Teaching Hospital 26,432,260 27,767,450 NA Psychiatric Hospital 4,008,386 NA Regional Hospital 3,382,482 1,558,141 1,810,741 District Hospital 1,008,196 969,755 667,617 Sub-district (HC+) 125,199 54,918 172,002 CHAG 160,348 112,721 928,548** RHA 2,003,876 5,329,608 114,776 DHA 124,892 1,419,988 173,400 Teaching Institution 312,564 373,228 248,895 Source: 2009 Appropriation Bill Detailed (Budget 09), GHS Consolidating Statement of Revenue and Expenditure by BMC Group, and PER 2009. *All �gures in Cedi. **Mission hospitals only. Decentralization and Governance in the Ghana Health Sector 79 Table 5.2: Mean features of hospitals by category Mean Mean Type of Mean No of Mean No. Patient- Recurrent Mean Mean Hospital N Beds Admissions Days Expenditure* Expend/Bed Expend/PDay Teaching 3 935 41,700 250,723 26,432,260 28,270 105.42 Psychiatric 3 395 2,209 146,060 4,008,386 10,139 27.44 Regional 8 218 12,640 55,886 1,558,141 7,143 27.88 District 99 73 4.578 15,081 969,755 13,270 64.30 CHAG 59 102 5,474 23,113 687,790 6,720 29.76 Quasi-Public 24 70 2,078 7,448 NA NA NA Muslim 4 66 3,920 32,479 NA NA NA Source: GHS Bedstate Repports 2008, Budget 2009 (for teaching and psychiatric hospitals), PER 2009. *All expenditures in Cedi. costs and a revision of criteria for allocation of budget and other resources would clarify the cause of this apparent distortion. Regional Variation in Health Expenditure per Capita The PER study showed reasonable variation among regions, but wide variation across districts, in the distribution of decentralized health spending per capita. Figure 5.10 shows that national spending at subnational levels was GHC 5.52 in 2008, with rela- tively li le variation across regions, within a range of 7.86 in Brong Ahafo to 3.23 in the Northern Region. The �gure also shows that the largest contribution to that level Figure 5.10: Variation in sub-national health expenditure per capita across regions, 2008 9 8 MHO 7 Expend Cedis per Capita 6 RHO Expend 5 RHD 4 Expend 3 DHO Expend 2 SDG 1 Expend DHA Expend A fo rn ti t ta rn ra l st n es ra er an Ea N ol cc ha ste te nt W th A sh V es A A Ce Ea er or H er A W er g pp G N pp on at U U re Br G Source: PER 2009. Note: Includes spending by Regions and Districts (but not by MOH and GHS headquarters) “Other� includes mostly mission hospitals and training institutions. SDGs allocation was not available for all districts, and was often included in DHA expenditure. 80 A World Bank Study Figure 5.11: Total Health Expenditure per capita by District, 2008 45 40 Cedis per capita 35 30 25 20 15 10 5 - Accra Metro Shama Ahanta Ga East Kwabre Sunyani Techiman Amenfi West Nzema East Nkwanta L. Manya Krobo Tolon-Kumbungu Ejisu-Juaben Aowin Suaman Bolgatanga Mun Jomoro Dangbe West Twifo Hemang L. Assin North Akwapim North Asunafo North East Mamprusi Adaklu Anyigbe Asikuma Odoben Atiwa Krachi West Wenchi Tain Pru Krachi East Asunafo South Jaman North Tano North Nanumba South Karaga Source: PER 2009. Note: Figures do not include expenditure on mission hospitals, but do include training institutions. Expenditure by other MDAs, especially DAs direct spending, are not included. of spending was services provided by District Hospitals (DHOs), followed by regional- level expenditure. District level primary care services (reflected in spending by DHAs and sub-districts-SDGs) contributed the smallest part, except in the Eastern Region. Spending by central level offices and facilities (headquarters and teaching and special- ized hospitals), are concentrated in three regions—Greater Accra, Ashanti and Northern— where offices and hospitals are located. Total health expenditure per capita—including all expenditure by MOH and GHS on decentralized administrative offices, health facilities and training institutions— varied enormously across districts, as shown in �gure 5.11. The mean value was GHC 5.52, but 7 districts had a value of GHC 20 or more, with Kintampo South leading at GHC 41.40, while 31 districts showed a value lower than 1 Cedi.10 The distribution of total health expenditure is not clearly related to district population, as apparent from �gure 5.9, and is li le influenced by regional differences, since no regional pa ern is apparent from the graph (�gure 5.10 also showed much smaller variations across regions). Different hypotheses might explain variation of government health expenditure. First, population size; but the variation in health expenditure per capita show that this factor is not important. Second, poverty, as existing budget allocation rules recommend that budget allocation should take into account poverty or population needs; however, �gures 5.12 and 5.13 show no association pa ern between expenditure per capita and a welfare indicator constructed from the GLSS5 survey.11 District-level expenditure—expenditure allocated to DHAs and DHOs, includ- ing sub-districts—shows similarly large variation across districts. The only difference from the distribution in �gure 5.11 is the exclusion of regional-level spending, in other words, Regional Health Directorates, Regional Hospitals and most training institutions. This affects only regional capitals, where most regional expenditure concentrates. For Decentralization and Governance in the Ghana Health Sector 81 Figure 5.12: Expenditure per capita and welfare across regions 3500 9 WELFARE INDEX IN THOUS. CEDIS HEALTH EXP /CAPITA IN CEDIS 8 3000 7 2500 6 2000 5 1500 4 3 1000 2 500 1 0 0 er n lta n n nt i ra l Vo ter ter a nt rth es as sh Ce No W E A WELFARE INDEX HEALTH EXP /CAPITA Source: PER and GLSS5. Figure 5.13: Health expenditure per capita and welfare across districts 5000 35 WELFARE INDEX IN THOUS CEDIS 4500 HEALTH EXP /CAPITA IN CEDIS 30 4000 3500 25 3000 20 2500 15 2000 1500 10 1000 5 500 0 - DISTRICTS WELFARE INDEX HEALTH EXP /CAPITA Source: PER and GLSS5. 82 A World Bank Study Figure 5.14: Health expenditure by level in regional capitals 35 30 Cedis per capita 25 20 15 10 5 - ro ro a ro i o n un t al an as nt H e ip et et et ab aM Co ha ny ic M iM eM Ju aA Su un pe g ra as ew an al aM cc Ca am um m at N A Ta lg Sh W K Bo Regional-Level Expenditure District-Level Expenditure Source: PER 2009. regional capitals, regional spending accounts for the majority of total expenditure, as illustrated in �gure 5.14. In the smaller capitals, regional spending amounts to a high level per capita, making these MMDAs among the highest spenders overall. Overall, the variation observed across districts was substantial. Another hypothesis for such variation would be that spending is related to the existing health infrastructure. To investigate this hypothesis, �gures 5.15 and 5.16 show the variation across districts of Figure 5.15: Primary care expenditure per capita (DHAs and SDGs) across districts, 2008 20 Asunafo South 18 19.19 Atiwa 16 19.81 Kintampo South 14 17.69 Cedis per capita 12 Yilo Krobo 10 12.38 Kwahu West 8 7.40 6 4 2 - Accra Metro Shama Ahanta Ga East Kwabre Sunyani Techiman Amenfi West Nzema East Nkwanta L. Manya Krobo Tolon-Kumbungu Ejisu-Juaben Aowin Suaman Bolgatanga Mun Jomoro Dangbe West Twifo Hemang L. Assin North Akwapim North Asunafo North East Mamprusi Adaklu Anyigbe Asikuma Odoben Atiwa Krachi West Wenchi Tain Pru Krachi East Asunafo South Jaman North Tano North Nanumba South Karaga Source: PER 2009. Decentralization and Governance in the Ghana Health Sector 83 Figure 5.16: Hospital expenditure per capita across districts, 2008 40 Kwabre 35 35.52 30 Asante Akim South Kintampo South Cedis per Capita 25 25.88 23.72 Sunyani 20 22.04 15 10 5 - Accra Metro Shama Ahanta Ga East Kwabre Sunyani Techiman Amenfi West Nzema East Nkwanta L. Manya Krobo Tolon-Kumbungu Ejisu-Juaben Aowin Suaman Bolgatanga Mun Jomoro Dangbe West Twifo Hemang L. Assin North Akwapim North Asunafo North East Mamprusi Adaklu Anyigbe Asikuma Odoben Atiwa Krachi West Wenchi Tain Pru Krachi East Asunafo South Jaman North Tano North Nanumba South Karaga Source: PER 2009. the two components of local-level expenditure: primary care services, generally provided by sub-district facilities and teams (SDGs) and to some extent by DHAs themselves (for management and support activities); and hospital services, provided by District Hospi- tals (DHOs) and Regional Hospitals (RHOs) (mission hospitals were not included in the analysis due to the paucity of data in the PER study). Regional Hospitals and District Hospitals were pooled together in this analysis because it is assumed that the former treat mostly patients from the locality where they are located, in spite of their role as referral facilities. Even if that assumption is relaxed, and only DHOs are considered, it would affect only regional capitals and not the general pa ern across 170 districts. The two �gures show that large variations across districts are found in both catego- ries, with the absence of a clear pa ern. In �gure 5.13, Primary Care spending per capita varies a lot, in a pa ern similar to that for Total Expenditure (�gure 5.11). Apart from �ve outliers—Atiwa, Asunafo South, Kintampo South, Yilo Krobo and Kwahu West— the range of values is between GHC 4.22 and 0.07. Even assuming that the outliers are data errors, the remaining variation is still quite large for a type of spending that should be closely related to districts’ population. Hospital spending per capita also varies a lot (Figure 5.16), ranging from GHC 35 to 0. Four outliers were encountered: Kwabre, Asante Akim South, Kintampo South and Sunyani. This variation is likely to reflect the presence and size of a hospital (or more than one) in the district. However, even among districts with one or more hospitals, variations are quite large. Such variation would be expected to relate not only to the number of hospitals in the district, but also to the size of the hospital(s).12 Indeed, the mean expenditure by a DHO varies greatly, and some district hospitals spend more than most regional hospitals. Table 5.2 showed the mean size and expenditure by type of facility; but there is a wide variation in the size, production and expenditure of hospitals within the same category. 84 A World Bank Study Some DHOs may actually have the size to be a regional hospital, and the inverse is also true (see �gure 5.17 for a comparison of mean sizes and variation by type of hospital). This variation suggests that the de�nition, requisites, and functions of the two categories of hospital—regional and district—may not be standardized or clearly de�ned. Summary of Findings from the Resource Allocation Analysis The analysis in this chapter brought a few insights into the �nancial and �scal dimension of decentralization in Ghana. First, a short review of international country data revealed that the degree of �scal decentralization in health—in other words, the degree to which expenditure on health has been decentralized to local governments—varies widely across both developed and developing countries. And the existing evidence indicates that organized health systems can function well with quite different levels and forms of decentralization, including with centralized systems. Second, the analysis of recent trends in the health sector indicated important trends in health care �nancing. First is the rapid growth of the National Health Insurance Sys- tem, which is reducing out-of-pocket expenditure and the importance of Internally Generated Funds. Additionally, Ghana shows a high proportion of �nancial resources allocated to decentralized levels of the health system, but most of this expenditure is in fact controlled and executed by the central government. Therefore, the degree of �scal decentralization associated with health is much more limited than the simple resource allocation would suggest. Third, the �nancing and expenditure pa erns at the district level show a high degree of specialization of �nancing sources regarding expenditure categories: GOG budget �nances nearly all personnel expenditure, and li le else, while non-personnel recurrent Figure 5.17: Mean and range of size by type of hospital (in No. of beds) 1400 1200 1000 800 600 400 200 0 ) ) 4) 7) ) ) ) ) 4) (3 (9 24 (6 67 (3 10 (5 s( t( t( ng al lim ric er t( G on ov ric hi at th A ric us gi ac G hi ist CH eO M ist Re si- yc Te .D D at ua Ps iv iv Q Pr Pr Max Min Mean Source: PER 2009. Decentralization and Governance in the Ghana Health Sector 85 expenses are funded by non-government resources (Development Partners funds, IGF and, increasingly, NHIS reimbursements. The study also highlighted important variations in public expenditure across dis- tricts, both for primary care and for hospital services; these variations are not related to population size and do not seem to reflect rational criteria for geographical resource allocation. Even though the size and concentration of hospitals is an important factor in determining health expenditure levels, the observed differences are highly indicative of substantial inequalities. The existing formula for resource allocation to districts does not seem to be working adequately, or is not effective. Notes 1. The Decentralization Policy Review of 2007 indicated that in a survey of three districts—Gush- egu, Ajumako/Esiam and Accra Metro—over three years deviation between budgeted expendi- tures and actual expenditure varied between -28 percent and +64 percent. Such deviations are much smaller for central government, but are still substantial. 2. Sources: United Cities and Local Governments (2009); and OECDStats (2008). Note that these aver- age values are approximated, because only few countries with data available were considered. They refer to local governments only, and exclude state or provincial level governments. 3. Tidemand et al. (2007 forthcoming), cited in the Decentralization Policy Review of 2007, gave a much higher proportion for Uganda in 2004–05: 25.5 percent. 4. However, it should be noted that a large part of the NHIS funding comes actually from central government general taxes (rather than speci�c social security contributions as in European coun- tries) through subsidies for vulnerable groups. 5. NHIS reported in 2008 that 54 percent of the country’s population was enrolled, and 45 percent had valid ID card holders (MOH 2009—Independent Review). In 2006, paying members repre- sented 27,5 percent of all members, and exempt members 72,5 percent (Health Sector Review 2006). NHIS includes four different types of insurance schemes, both public and private: (i) a mandatory scheme for those involved in the formal sector, (ii) district mutual insurance schemes (one per district), (iii) private mutual schemes, and (iv) private for-pro�t insurance schemes. Of all services provided through NHIS, 50% were provided by public facilities, 30 percent by CHAG and 30 per- cent by private providers (NHIA, Banking on Health 2008). 6. The Ghana Poverty Reduction Strategy (GPRS) pledged to increase the proportion of govern- ment revenues to 7.5 percent. 7. Funds are distributed to all DAs annually according to a formula approved by parliament and transferred on a quarterly basis. Transfers are made against submission of annual action plans and budgets (DACF 1994). 8. Surveyed DAs indicated spending on average 5.9 percent of the funds received on education, but a large number of DAs in the sample did not inform whether and how much they were investing in education. 9. DACF earmarked funds include 1 percent for malaria prevention and allocations to capacity building (2 percent) and other items which may impact indirectly on health); they amount to 41 percent of total DACF transfers, leaving 59 percent to cover all responsibilities mandated for district governments. 10. In these calculations, all decentralized MOH/GHS expenditure (to regional offices, facilities, training institutions—have been allocated to the district where the unit is located; for instance, the cost of regional hospitals was allocated to the regions’ capitals. While this is not entirely correct, the distortion is unlikely to be important, since the proportion of a hospital’s users coming from the district where the hospital is located is usually quite high. Also, It is important to note that the exact �gures for individual districts may be distorted by the fact that some newly established districts may not yet have separate budget allocation, population or facilities. These values do not include public expenditure on mission hospitals, because only 5 of these hospitals reported expenditure 86 A World Bank Study data in the PER exercise, and including those would distort the �gures for a small number of dis- tricts relative to the others. 11. This welfare indicator is computed from household consumption expenditure, adjusted by household size and composition and geographic price variation; it is expressed in consumption expenditure per equivalent adult divided by a relative price index (Ghana Statistical Service, 2007: Pa ern and Trends of Poverty in Ghana—1991–2006). One should note that district-level analysis of GLSS data should be interpreted with caution, since the GLSS5 sampling framework was strati- �ed by region, ecological zone and urban/rural, but not by district. 12. It should be noted that few mission hospitals provided data to the PER survey, and a number of districts were a mission hospital is important or the only hospital, would have its expenditure level distorted. But this is unlikely to have altered the general pa ern signi�cantly. CHAPTER 6 Main Challenges and Policy Options W ithin a long process of decentralization that goes back to independence, the Gov- ernment of Ghana has de�ned a form of decentralization by devolution to Dis- tricts. This choice is reflected in nearly all legal and policy documents produced so far. Even though there are advantages and disadvantages to each modality of decentraliza- tion, this report takes it as given that the country has made its choice regarding the form of decentralization, and discusses how to best plan and implement such choice in the health sector. Decentralization is a complex process that does not necessarily result in efficiency or equity improvements. In some cases, if structural, organizational and �nancial issues are not correctly addressed in the design and implementation process, decen- tralization may actually result in lower efficiency and greater inequalities. While decentralization in general is a complex and lengthy process in any context, decen- tralization of the health sector has proven to imply additional challenges, due to the nature and characteristics of health systems. For instance, the need for a coordinated multi-level health system that ensures referral mechanisms to higher-level facilities, and the highly technical nature of some health functions and programs, does not make it simple to fully devolve health functions and responsibilities. A careful discus- sion and clear policies on the different dimensions and implications of health system decentralization is key to build a functional decentralized health system. Overall, Ghana has over the years established several of the building blocks needed for a successful decentralization, but these efforts lack cohesion and unity of purpose. The GOG needs to bring together these many policies, instruments and systems and make them work for effecting decentralization. This study has identi�ed a number of the key issues and challenges; they are summarized below. ■ Several of the basic elements of administrative decentralization are already in place, in the form of the district political and administrative units, their DA with a signi�cant, though still weak, management structure, and GHS District Administration offices. If brought together under a single authority, these struc- tures make for the basic structure of a decentralized health system. ■ A number of useful information systems and management tools have been devel- oped and implemented, including planning and budgeting systems, reporting and information systems, performance measurement, and �nancial transfer mechanisms to local governments, among others; however, their effectiveness is limited by the many overlaps and duplications, the fragmentation among sys- tems, and especially their inability so far to produce reliable information. ■ The existing regulations, assessments and policy documents have correctly iden- ti�ed the main challenges in implementing effective government decentralization 87 88 A World Bank Study in the form of devolution to local governments. But only in recent years has MOH produced consistent policy documents addressing the speci�c issues of decentralization in the health sector. A comprehensive and clear policy frame- work to guide implementation of devolution in health is still lacking. ■ The existing legal framework concerning health is confusing and contradic- tory; existing laws and regulations offer changing and conflicting views of what decentralization should look like in the health sector, and are quite vague as to which functions are to be devolved. The often highlighted contradiction between GOG’s policy of devolution and MOH/GHS model of delegation cum deconcentration is only one among several conflicts in regulation and policy. ■ The �nancing framework for local governments is complex and confusing: dif- ferent funding sources specialize in �nancing speci�c line items or programs, and the DAs resource allocation to sectors is not transparent. In addition, �s- cal decentralization in Ghana is more apparent than real: over 50 percent of public health expenditure is allocated to the district level, but the larger part of these resources are allocated and controlled by the central government; local authorities—whether DAs or GHS District Offices and facilities—have li le real decision power on resource allocation. ■ In addition, substantial delays in the transfer and release of funds, both by GOG and NHIS, have hampered the functioning of local governments and local facili- ties and programs alike. ■ Another contradiction in general legislation relates to staff management: broader legislation calls for “full devolution to local governments�, but man- agement of local staff has been centralized in a parallel LGS, which in effect withdraws from LG authority over the major resource they need for managing local services. ■ Capacity for implementing and managing a truly decentralized health sys- tem is low, not only at the district level but across all levels of government, although the weaknesses may be different at each level; one of the main bo le- necks is the lack of reliable information for decision making, monitoring and evaluation. ■ Efforts at decentralization have happened in waves over many years, and have lacked continuity and consistency; they have contributed to the current frag- mentation in the health system, with unclear distribution of functions and responsibilities and unclear lines of authority. The two key players in local health systems—the DA and DHAs—have an insufficient level of collaboration and integration, even though formal mechanisms have been established, such as the composite planning. ■ As a result of decentralization efforts, participatory mechanisms—usually in the form of local councils and commi ees—appear to have quite variable effec- tiveness; in many areas, they have not been functioning as an effective channel of participation in decision making and planning. ■ A major obstacle to effective implementation in the health sector has been that many stakeholders have a limited understanding of the process objectives, pre- requisites and implications. This is especially true in the health sector, as the survey of regional and district officers clearly showed, and makes it difficult to build consensus and support for the process. Decentralization and Governance in the Ghana Health Sector 89 The main issue most often highlighted in decentralization of the health sector in Ghana is the current contradiction between GOG’s model of devolution, and GHS deconcen- trated structure. However, this study found that other issues also contribute to hamper real decentralization in the health sector, some of them residing in the general legislation and reflecting directly on health decentralization. Among them is the low level of con- trol local authorities have over budget and expenditure: most of the resources allocated to local facilities and services is actually executed centrally on behalf of local offices, or earmarked from the center to speci�c programs or initiatives. This clearly runs counter the idea of full devolution. On the other hand, the contradiction between a deconcentrated GHS and devolution to DAs is not necessarily critical or does not necessarily constitute an obstacle to effective devolution. In fact, GHS deconcentrated district-level structures and facilities, with its signi�cant—though not homogeneous and without weaknesses—managerial and �nan- cial autonomy, may be seen as an important intermediate step towards full devolution. GHS deconcentration has promoted the establishment of local health administrative and technical units, which can easily constitute the core of a health sector department in the DAs structure once these units are transferred to DAs’ authority. Some of the existing legislation and policies de�ne exactly that, but have not been implemented yet. The following sections present a number of policy options to address the challenges summarized above, and thus strengthen and accelerate effective decentralization in health. Some of these options have already been considered or decided, but have not been effectively implemented. Capacity Strengthening The planning and implementation of decentralization would greatly bene�t from a sys- tematic assessment of DHAs’ and DA’s capacity and conditions for taking responsibil- ity for particular functions; this report provides some insights on this issue based on a sample of DHAs, but this assessment should be performed systematically for each and every district, and should focus on identifying speci�c needs for technical support and strengthening; this assessment should produce a comprehensive mapping of needs for training and technical assistance, that would lead to the development of a comprehen- sive and prioritized plan for LG capacity building in health, and would integrate the MOH, GHS and development partners efforts in capacity building. This assessment, and the overall process of capacity building, would be be er coordinated by the coordinat- ing commi ee suggested below—or, be er still, by one of its sub-commi ees. Coordination Mechanism The discussion and de�nition of a decentralization policy framework for the health sector would have to mobilize all stakeholders, both for strengthening technically the �nal pro- posal that will emerge, and to build consensus and support around this proposal. Such an endeavor will require strong and commi ed leadership from the part of the overseeing institution in the health sector, the Ministry of Health. An MOH-led Commi ee could be established to coordinate the formulation and implementation of decentralization in the health sector; it would be best placed under the direct authority of the Minister of Health (the highest authority in the sector), be chaired jointly by MOH and MLGRD, and include representatives from the key stakeholders, namely GHS, NHIA, MLGRD, 90 A World Bank Study MOFEP, regional health officers, DAs and DHAs, civil society organizations and develop- ment partners. The commi ee would be a high-level policy-making body, with top-level representatives from each institution, and would be supported by one or more technical sub-commi ees with technical people to actually develop the technical proposals. To disseminate ideas and gather support for decentralization, this commi ee could organize regular (say, quarterly) workshops or consultative meetings, which would bring together all the main stakeholders to discuss policy-relevant issues on decentral- ization. This approach would greatly contribute to clarify concepts and policies relat- ing to decentralization, seek inputs from diverse sources, build consensus, and overall advance the reform agenda. Policy Framework A health system decentralization framework is greatly needed, and would further clar- ify and detail the responsibilities and functions of each government level and agency; the de�nition of functions to be decentralized should take into account factors such as: economies of scale (especially in procurement of drugs and other strategic supplies and services) and the highly technical nature of some functions and services. International experience suggests that not all health-related functions and respon- sibilities can be successfully decentralized. Some typical public health activities, such as disease surveillance, do not work well at the local level. Other responsibilities, such as procurement for some critical supplies (prescription drugs are the best example) would be be er performed centrally, because it allows for economies of scale and standardization of supply items, or because decentralization would imply unmanageable fragmentation of certain programs or activities (such as disease surveillance and immunization campaigns, as mentioned in the review of international experiences). A few mixed strategies are dis- cussed in the section titled “Procurement Process� below. The design of a strong policy framework should thus encompass a detailed techni- cal discussion on which functions and responsibilities should be decentralized to local (or regional) level, and which would remain centralized. Table 6.1 provides an illustrative and preliminary example of the prospects for decentralization of some typical health functions. This exercise should be systematically carried on until a clear and consensual view emerges on which functions would be best performed at which level of government. It is important to note that this strategic discus- sion may lead to questioning of some general legislations regarding staff, procurement, �nancial management, and others. A strategic approach could be to develop a detailed plan for phased implementation of decentralization, in which LGs with a stronger capacity receive technical assistance and support as needed to effectively implement a �rst group of decentralized LGs in health. Learning from these experiences would then allow �ne tuning and correcting implementing strategies for a second round of districts to implement health decentral- ization, and so on. Integrated Planning and Budgeting As most assessments and policy documents have pointed out, the current “composite budget� policy has not yet taken root. Though, it is a critical step in moving toward full devolution, and needs to be strengthened in the short term. This could be done by the Decentralization and Governance in the Ghana Health Sector 91 Table 6.1: Illustrative assessment of prospect for decentralization by function Function Prospect for Decentralization/Devolution General management and support functions Planning and budgeting Good Implementation and execution Good Monitoring and evaluation Average (all levels) Procurement of drugs Poor (selective) Procurement of supplies Average (economies of scale) Procurement of non-technical services Good Financing Poor (dependence on central transfers Financial management Average Technical functions/services Highly technical public health activities, e.g. Disease Poor surveillance Public health outreach activities Good Curative primary care Good Curative secondary/tertiary care Poor (to the regions OK) Health care networks coordination (local) Good Needs identi�cation Good Source: Author’s elaboration. development of practical guidelines for effective joint planning and budgeting in health at the district level; this means revising and strengthening existing guidelines for the “composite budget� and actual practices, detailing or adapting them to the health sector, in order to promote effective participation and involvement of local stakeholders in the discussion and preparation of the district health plans and budget. Following the strategic approach suggested for the policy framework, integrated planning and budgeting could be implemented stepwise, �rst in the districts with greater installed capacity, and then to other districts. Strengthening of DAs’ Structure and Capacity Several aspects are to be considered in strengthening LG capacity. First, once effectively devolved, the different responsibilities will be carried out in different places by differ- ent groups of staff within the district government. It is thus necessary to break down the general capacity assessment and planning and clearly identify who would be doing what, and thus the different types of capacity that will be needed within the local gov- ernment. Current legislation not only gives conflicting views (as mentioned above) but also does not give a clue on how a devolved health system would be structured and would function at the district level. Some responsibilities will be carried out at the facil- ity level, others by the (yet to be transferred) local health offices, and others at the DA and its management commi ee and staff. Even though LGs need signi�cant autonomy and flexibility in allocating responsibilities according to local reality, it is important to map out health responsibilities within LG in order to be er identifying and targeting capacity building needs and efforts. 92 A World Bank Study The mandated Social Services Sub-Commi ee does not seem to function effectively in many districts, and in some cases DHA officials do not participate regularly. On the other hand, the District Department of Health has yet to be formally established once DHAs are transferred to DA authority. However, the future functions of these two instances are unclear and appear duplicated. The Sub-Commi ee is likely to be responsible for sector planning and monitoring and intersectoral coordination, while the Health Department is likely to focus on the actual management and implementation of health services and programs and supervising of district level facilities. But an alternative approach would be to merge all coordination and supervision functions in one body. As mentioned in the preceding chapters, the current regulation does not clarify this distribution of responsi- bilities within the district government level. It would also be very useful to revise the proposed structure of DAs support offices so as to include a speci�c health commi ee which should include representatives of the DHA, local health professionals, and local communities, and would necessarily meet on a regular basis to prepare the annual plan and budget, have it passed by the DA, monitor its implementation through prede�ned performance indicators, and yearly evaluate the impact of the plan. Within this approach, it would be useful to assess the availability of technical (that is to say, health-related) staff with the required quali�cation for carrying out the speci�c tasks flowing from devolution. At this point it would be necessary to identify clearly if such staff is already present at the current DHA offices—and, thus, the issue is one of integration and transfer of staff—or if some required staff or capacities are altogether missing from the local level. This is likely to vary across regions and districts, with larger and more developed ones having more and quali�ed staff available, and deprived zones likely not to have them. Two types of capacity building are needed at the DA level. One focuses on gen- eral management and �nancial capacity for local governments, including general plan- ning, budgeting, procurement, human resource management, among others. The other relates to the capacity for planning, coordinating monitoring and supervising health- speci�c activities undertaken at the district level. The �rst will need to be built in the general management support offices such as Planning, Procurement, Human Resources, Finance, and other functions. The second would be built within the sectoral commi ee of the DA structure and the future Department of Health—which is likely to be based on the DHA structure. Management at the Facility Level Many countries have decentralized health system functions to local or regional govern- ments, but have not increased the degree of managerial autonomy at the facility level. In that case, decentralization seems to stop midway, because executing agencies (health facilities) have li le or no autonomy to carry out their responsibilities, and given the often weak management capacity of LGs, provision of health services may be signi�- cantly hampered by lengthy or inefficient processes at the level of LG headquarters—in the Ghana case, DAs and their supporting management infrastructure (management commi ee or departments). Therefore, as part of the discussion of capacity and autonomy of LGs, health authori- ties at all levels need to discuss and de�ne what degree and form of managerial autonomy Decentralization and Governance in the Ghana Health Sector 93 will be given to what types of health facilities. GHS has over the years deconcentrated some responsibilities to facilities, but the result appears unsystematic and unclear, and—as shown in the rapid assessment of Chapter 4—activities actually performed at the facility level vary signi�cantly across facilities of similar type and size. According to the rapid assessment, most facilities, especially larger ones, have already established a facility board or commi ee, but they tend to have more of an advisory role. To enhance governance, accountability and community participation, these facility boards could take a greater role and become the �rst level of accountability for facility managers. Financing Framework for Decentralization Development of a clearer �nancing framework for LGs on the health sector would be a great contribution to decentralization implementation; this framework should seek to streamline the multiple existing flows and funds, clarify responsibilities for �nancial management, and take advantage of the District Financing Fund to consolidate these flows. As a �rst step, it would be important to estimate �nancial needs (expenditures) to upgrade local governments capacity (as described above), and to meet the devolved responsibilities. This estimate can be based on existing facility expenditure reports, cost information of health programs, services and facilities, including recent PERs and PETS.1 Then, based on �nancial needs associated with speci�ed decentralized functions, the framework would de�ne �nancing sources and flows for decentralized levels, including the incentives structure needed to promote effective implementation and a ract staff to more remote areas. As part of this framework, the current policy and formula for budget allocation across regions and districts would be revised, so as to emphasize the objective of equitable redistribution of funding that is part of the decentralization process and do so based on transparent rules. Different reports and assessments have pointed out a serious bo leneck in the bud- get flows to DA, that is, the frequent delays in fund transfer and differences in actual funds transferred relative to planned or budgeted. This bo leneck needs to be resolved in the short term, to ensure predictability and stability in LG �nancing. It is not clear from the reports what drives these delays and shortfalls, and if necessary an in-depth assessment should be undertaken to clearly pinpoint where the causes of the problem lie and how to resolve them. Opportunities for testing and implementing performance-based �nancing (PBF) schemes should be seriously considered, as international experience has shown that such schemes can provide a proper incentive structure for improving performance. The review of international experience has shown some successful examples of performance- based �nancing (such as the case of the Family Health Program in Brazil), and Ghana itself has implemented a performance assessment system, which however needs to be beefed up and revised if it is to be the basis for a PBF system. Procurement Process The existing procurement system regulations would bene�t from a broad revision, in order to standardize processes at different government levels while ensuring flexibility to address local needs and characteristics. New initiatives may include the establish- ment of an online price dataset that can be accessed by any public entity to inform each public sector procurement process; this dataset would be fed by information from 94 A World Bank Study every procurement done by any government agency. Such price databases have been successfully used in Brazil and other countries to inform public sector procurement (at all government levels) and reduce the average costs of public purchases. It would also be useful to pilot projects of pooled procurement among small dis- tricts and/or facilities, and “framework contracts�, as implemented in other countries. Some countries have adopted a system of centralized purchasing for some items coupled with decentralized “drawing rights� by individual districts and/or facilities, by which these districts and facilities can draw from the supplies procured centrally, based on annual plans of needs and use. Other countries have established a procurement system by which local governments or individual facilities pool their procurement needs at the regional level or sub-regional level; in this case, the procurement process is performed at the regional or sub-regional level, but pulled by local governments and/or facilities. In Brazil, the MOH has set up and maintains an online database of suppliers and prices for a signi�cant number of supply and service items based on recent purchases by public sector organizations; this “price data bank� is now been replicated in other countries. Given the current limited capacity of local governments in Ghana, such pooled procure- ment schemes are likely to be the best option for some time. Human Resources Management Given the apparent weak standardization of HR management policies and processes shown in the Rapid Assessment, the country would greatly bene�t from revising and/ or de�ning regulations and policies regarding human resource management in a decen- tralized system, and formulate a clearer regulatory framework that at the same time homogenizes processes and provides transparent basic standards, without limiting local governments’ autonomy to manage its own staff; this framework should necessarily include provisions for transferring staff from the central to local level, and a structure of incentives to encourage staff to transfer. Important issues deserving careful discussion include the de�nition of the regime for hiring staff at the district level, including remuneration levels and career path. A related issue would be to reexamine the usefulness of having a separate and central- ized LGS in face of devolution—the LGS Act states that all personnel from decentral- ized departments should automatically transferred to LGS—since it greatly reduces LG authority over their own staff. Information and M&E Systems This study has shown that a major issue is the duplication, lack of standards, and over- all poor quality of information systems, which in turn greatly reduce the effectiveness of M&E systems. In order to be er promote and support the decentralization process, and monitor and evaluate the decentralization process and its impact, the capacity of central and regional levels for M&E would have to be strengthened—in terms of human resources, systems and instruments; this activity could take advantage of several impor- tant initiatives regarding M&E and performance assessment that have been adopted in recent years, and adapt them to a decentralized system. But existing systems would need to be standardized (to facilitate comparison) and streamlined (to avoid duplica- tions and conflicting information). Improving information systems involves identifying strategic information to monitor and evaluate, revise and streamline information flows, Decentralization and Governance in the Ghana Health Sector 95 standardize and enforce concepts and de�nitions, and extensive training. GOG perfor- mance assessment system could be strengthened based on the experience and lessons learned from other countries. Some countries for instance have implemented systems of district scorecards. These performance-focused M&E systems would then support the performance-based �nancing schemes proposed in the section titled “Financing Frame- work for Decentralization� above. Legal Framework As a result and consolidation of these efforts, the �nancing and functional decentraliza- tion policies would in a second phase be consolidated into a coherent legal framework for health system decentralization; this legal framework could be prepared once policy documents have de�ned all main dimensions and aspects of decentralization, to avoid contradictions and reduce the need for later revisions. It is the view of this consultant that the legal framework for decentralization in health should follow the de�nition of a clear policy framework, for several reasons. First, it is difficult to issue a good and stable legislation if one is not clear about where to go and does not have a clear vision of how a decentralized health system should function. Second, as clearly shown in this volume, successful decentralization in health depends on revising and improving the general legal framework for decentralization and solving its many duplications and inconsistencies. Third, legally harmonizing the contradiction between current devolution policy and GHS mandate and structure could be left for a later phase, so that it does not confuse the process of strengthening LG structures, capac- ity and practices for a decentralized system. In the meantime, strong incentives and clear policies should be put in place—using as starting points existing systems such as performance assessment and current �nanc- ing mechanisms—to produce effective coordination between the existing DAs and GHS’ district-level structures. For instance, a �nancial reward could be built into the �nancing flows to both DAs and DHAs contingent on the presentation of a joint district health plan and a composite/integrated planning. Note 1. Health PETS are available for 2007, and a new PER should be officially released soon based on 2009 data. APPENDIX A Main Legislation and Policy Documents Regarding Decentralization and Local Government in Ghana* 1957–1966: First Republic  1) Constitution of 1957  2) Constitution of 1960  3) Local Government Act, 1961 (Act 54) 1966–1971: NLC Government and Second Republic  4) NLC Commissions reports (Mills-Odoi 1966; Siriboe 1968; Akufo-Addo 1968)  5) Constitution of 1969  6) Local Administration Act, 1971 (act 359) 1972–1981: NRC/SMC Government  7) Local Administration (Amendment) Decree, 1972 (NRCD 138)  8) Local Administration (Amendment) Decree, 1974 (NRCD 258)  9) Local Government (District Councils) (Establishment) Decree, 1974 (NRCD 290) 10) Local Government (District Councils) (Establishment) Decree, 1975 (NRCD 352) 11) Local Government (Amendment) Decree, 1976 (SMCD 15) 12) Local Government (Amendment) Decree, 1978 (SMCD 196) 13) Local Government (Amendment 2) Decree, 1979 (SMCD 219) 1979–1981: Third Republic (PNP Government) 14) Constitution of 1979 15) Local Government (Amendment) Act, 1980 (Act 403) 1982–1992: PNDC Government 16) Local Government (Interim Administration) Law, 1982 (PNDCL 14) 17) Local Government Law, 1988 (PNDCL 207) 18) NCD Report “Evolving a True Democracy�, 1991 19) Constitution of 1992 *Source: GOG/MLGRD, Draft Comprehensive Decentralization Policy Framework, 2007; Ahwoi 2010. 96 Decentralization and Governance in the Ghana Health Sector 97 1992–1999: Fourth Republic 20) District Assemblies’ Common Fund Act, 1993 (Act 455) 21) Local Government Act, 1993 (Act 462) 22) Civil Service Law of 1993 (PNDC Law 327) 23) Financial Administrative Act, 1993 (Act 654) 24) National Development Planning System, 1994 (Act 480) 25) Local Government Establishment Regulation of 1994 26) Ghana Health Service and Teaching Hospitals Act (525) of 1996 27) Ghana Audit Service Act, 2000 (Act 584) 28) Legislative Instruments to establish DAs 29) Local Government Service Act, 2003 (Act 656) 30) Internal Audit Agency Act, 2003 (Act 658) 31) Institute of Local Government Studies Act 647 of 2003 32) Creation of Districts Executive Instrument, 2003 33) Public Procurement Act, 2003 (Act 663) 34) National Decentralization Action Plan, 2003 35) Financial Administration Regulations, 2004 (LI 1802) 36) Operational Guidelines for DPCUs and RPCUs. Recent documents (2007–2010) 37) Decentralization Policy Review, 2007 38) Draft Comprehensive Decentralization Policy Framework, 2007 39) National Report on Regional Consultations, 2009 40) General Health Service Bill, 2009 41) Draft Decentralization Policy Framework, 2010 The Local Government Act, 1993 (Act 462) and the National Planning (System) Act, 1994 (Act 480): Section 3 of Act 462 makes the MLGRDE responsible for the establish- ment of District, Municipal and Metropolitan Assemblies by Legislative Instruments. The same Act 462 also spells out the responsibilities of the DAs in their areas of jurisdic- tion. It establishes them as “Planning Authorities� with scope for managing the overall development within the districts. Speci�cally, it outlines the framework for the DAs to exercise their executive, deliberative and legislative functions by specifying the opera- tions of the general assembly, planning functions, �nancial ma ers, rating responsibili- ties and auditing requirements among others. The planning functions of the DAs are further elaborated by the National Planning (System) Act, 1994 (Act 480) which provides the framework for decentralised develop- ment planning of DAs and the planning functions of the Regional Coordinating Coun- cils (RCCs) and Ministries, Departments and Agencies (MDAs). To enable the DAs and RCCs to possess the necessary technical capacity to ful�l their mandates, Act 462 establishes the District and Regional Planning Coordinating Units (DPCUs&RPCUs). The National Development Planning Commission (NDPC) and MLGRDE in 2003 jointly issued guidelines to operationalise the DPCUs and RPCUs. The document details the composition, roles and responsibilities of these Units in ful�lment of assigned functions contained in Acts 462 and 480. The guide also speci�es an annual planning and budgeting cycle that links the DAs process to the national budgeting cycle. 98 A World Bank Study The RPCUs in their mandate have responsibility to monitor and coordinate all activities of local government authorities in their regions of operation. Financial Administration Act 2003 (Act 654) and Financial Administration Regu- lations, 2004 (LI 1802): The �nancial accountability system for the MDAs and DAs is covered by The Financial Administration Act 2003 (Act 654) and Financial Administra- tion Regulations, 2004 (LI 1802). Additionally, Act 654 provides direction and control for overall �nancial administration in Ghana. It makes the Ministry of Finance and Economic Planning (MoFEP) the authority for the preparation of the �scal policy of government for presentation to Parliament (Act 654 Section 2a). The Controller and Accountant Generals Department (CAGD) by the law is the primary disbursement agency of government. The Act and its regulations, generally spell out modalities for preparation of MDA and DA budgets, ensuring accountability for �nancial commitments and maintaining appropriate records in line with professionally accepted accounting standards and norms. The CAGD through various national, regional and district offices provides the necessary support for managing disbursements, payments, receipts and record keeping as vehicles for the sub- mission of �nancial statements to government within agreed timelines. Public Procurement Act, 2003 (Act 663): Procurement of goods, works, and ser- vices constitutes one of the most important activities in the operations of MDAs and DAs. In 2003, the Public Procurement Act, 2003 (Act 663) was enacted to outline the structure, methods and tendering procedures of procurements and the threshold for review/approval authorities, modalities for disposal of plant/equipment, and other mis- cellaneous provisions. This Act complements and reinforces other Acts especially Act 654. Acts 654 and 633 provide the necessary framework to DAs to contract services to implement their Annual Action Plans once they are developed and funds are available. District Assemblies Common Fund Act, 1993 (Act 455): In line with the constitu- tional provision [Article 252 (2)] that Parliament shall annually allocate not less than �ve percent of total revenues of Ghana to the District Assemblies for development, the District Assemblies Common Fund Act, 1993 (Act 455) was enacted. The Act established the Office of the Administrator of the District Assemblies Common Fund as well as the structure and responsibilities for ful�lling the obligations of the office. The District Assemblies Common Fund (DACF) is allocated to each DA annually based on a formula approved by Parliament. It is designed to be disbursed to DAs on a quarterly basis, but a quarter in arrears. The DACF is available to the DAs only for investment expenditure and is a sure source of funding for investment projects. Utilisation of the funds is informed by guidelines approved yearly by Parliament. The DAs gain access to their allocation only subsequent to the submission of Annual Action Plans and Budgets to the Office of the Administrator of the District Assemblies Common Fund. Accordingly the DAs are required to forward returns on utilisation of the fund to the Office of the Common Fund Administrator monthly. The Audit Service Act, 2000 (Act 658): outlines the task of the Auditor- General in respect of audit of public accounts and the audit limits. The requirement is that the audits must be completed within six months after the close of each �nancial year. The thrust of the operations is to establish whether the accounts have been well kept, rules and procedures followed. Other operations include whether funds have been appro- priately expended, records maintained, assets protected and �nancial operations con- ducted with due regard to efficiency and effectiveness. Decentralization and Governance in the Ghana Health Sector 99 Local Government Service Act, 2003 (Act 656): The Local Government Service Act establishes the Local Government Service and provides for the objects, functions, admin- istration of the Service and related concerns. The functions of the Service include the provision of technical assistance to the DAs and RCCs to enable them to effectively per- form their functions and discharge their duties; conduct organisational and job analysis; design and coordinate management systems and processes. Others include, the execu- tion of management audits of RCCs and the DAs to improve overall performance of the Service and assist the RCCs and DAs to perform their statutory as well as related responsibilities. Internal Audit Agency Act, 2003 (Act 658): The Internal Audit Agency Act estab- lishes a body to coordinate, facilitate and provide quality assurance for internal audit activities within the MDAs and DAs. Section 3 (1) enjoins the agency to set standards and procedures for the conduct of internal audit activities Section 3 (4) enjoins the Agency to monitor, undertake inspections and evaluate the internal auditing of the MDAs and DAs. The Guidelines for the Operationalisation of RPCUs and DPCUs (November 2003), �nally, have been issued jointly by MLGRDE and NDPC to provide operational guid- ance for district level �nancial management and administration. The National Health Insurance Act of 2003 (Act 650) establishes the National Health Insurance System (NHIS), under the coordination of a National Health Insurance Coun- cil, and the National Health Insurance Fund; it also regulates and licenses all health insur- ance schemes, including the District Mutual Insurance Schemes, and the accreditation of health care providers. Enrollment with an insurance scheme—mutual or private—is mandatory, except for members of the armed and police forces. District Mutuals are funded by membership contributions paid by enrollees living in the district, and subsi- dies from the National Health Insurance Fund. Such subsidies are distributed to District Mutuals according to a formula approved yearly by Parliament. References Agyepong I. A. 1999. “Reforming Health Service Delivery at District Level in Ghana— the Perspective of a District Medical Officer.� HealthPolicyandPlanning 14(1): 59–69. Abekah-Nkrumah G., Dinklo T., and Abor J. 2009: “Financing the Health Sector in Ghana—A Review of the Budgetary Process.� European Journal of Economics, Finance and Administrative Sciences, Issue 17 (2009), EuroJournals, Inc. Ahwoi K. 2010: Local Government & Decentralisation in Ghana. Unimax Macmillan, Accra. Ahmad J., Devarajan S., Khemani S., and Shah S. 2005. “Decentralization and Service Delivery.� Policy Research Working Paper 3603, World Bank, Washington, DC. Akin J., Hutchinson P., and Strumpf K. 2001. “Decentralization and Government Provi- sion of Public Goods: The Public Health Sector in Uganda.� Working Paper, Mea- sure Evaluation Project, USAID. Arredondo A. and Orozco E. 2006. “Effects of Health Decentralization, Financing and Governance in Mexico.� Rev. Saude Publica 2006:40(1):152–60. Bach S., Blöchliger H. and Wallau D. 2009. The Spending Power of Sub-Central Governments: A Pilot Study. OECD Network on Fiscal Relations Across Levels of Government, OECD. Bahl, R. (n.d). Worldwide Trends in Fiscal Decentralization. Bordignon M. and Turati G. 2003. “Bailing Out Expectations and Health Expenditure in Italy.� CESifo Working Paper No. 1026 (Public Finance). Bossert T. 1998. “Analyzing the Decentralization of Health Systems in Developing Coun- tries: Decision Space, Innovaion and Performance.� Soc.Sci.Med. Vol. 47, No. 10, pp. 1513–1527. Elsevier Science Ltd. Bossert T. 2000. “Guidelines for Promoting Decentralization of Health Systems in Latin America.� DDM Project, Harvard School of Public Health, Cambridge, MA. Bossert T. and Beauvais J. 2002. “Decentralization of Health Systems in Ghana, Zambia, Uganda and the Philippines: A Comparative Analysis of Decision Space.� Health Policy and Planning 17(1): 14–31. Bossert T., Bowser D., and Amenyah J. 2004. Ghana: Decentralization and the Health Logis- tics Systems. Deloi e and Touche. Campbell T. 2009. “Decentralization and Democracy: A Global Perspective in 2007.� In United Cities and Local Governments, 2009: Decentralization and Local Democracy in the World. First Global Report by UVLG. Collins C., Araujo J., and Barbosa J. 2000. “Decentralizing the Health Sector: Issues in Brazil. Health Policy,� 52 (2000) 113–127. Crivelli E., Leive A., and Stratman T. 2010: “Subnational Health Spending and Soft Bud- get Constraints in PECD Countries.� Working Paper WP/10/147. International Monetary Fund, Washington DC. Dafflon B. (n.d.) Fiscal Decentralization. Dunlop D. W. (n.d.) Decentralization in the Health Sector: the Ugandan Experience. AusAID. 100 Decentralization and Governance in the Ghana Health Sector 101 Egbenya G. R. K. 2010. The Effectiveness of Decentralization Policy in Ghana: a Case Study of Komenda-Edina-Eguafo-Abrim (KEEA) and Abura-Asebu-Kwamankese (AAK) Districts in Ghana.� African Journal of Political Science and International Relations, Vol. 4(1), 13–28. Farrant M. and Clarke J. (n.d.) PRSPs in Decentralized Government Systems. DFID. Fleury S., Belmartino S., Lobato L. V. C., and Tamez S. 1999. Salud—Reshaping Health Care in Latin America; a Comparative Analysis of Health Care Reform in Argentina, Brazil and Mexico. IDRC. Fran B., Mpuku H., and Wright J. 2004. Supporting the Decentralized Delivery of Health Services—USAID and the District Basket in Zambia. Ghana (Government of), 1993: Local Government Act (Act 462). Ghana (Government of), 1996: Ghana Health Service and Teaching Hospitals Act (Act 525). Ghana Ministry of Health. May 2009. Pulling Together, Achieving More. Independent Review—Health Sector Programme of Work 2008. Ghana Ministry of Health. October 2009. Programme of Work 2009. Ghana Ministry of Health. 2008. Programme of Work 2008. Ghana Ministry of Health. 2002. National Health Accounts 2002. Ghana Ministry of Local Government and Rural Development. September 2003: National Decentralization Action Plan (NDAP). Ghana Ministry of Local Government and Rural Development, 2007: Draft Comprehen- sive Decentralization Policy Framework. March 2007. Ghana Ministry of Local Government and Rural Development and Institute of Local Government Studies, Jan. 2009: National Report on Regional Consultations— Regional Consultative Process for the Acceleration of Decentralization in Ghana. Ghana Ministry of Local Government and Rural Development, 2007: Draft Comprehen- sive Decentralization Policy Framework. Ghana Ministry of Local Government and Rural Development, 2010: Draft Decentraliza- tion Policy Framework—“Accelerating Decentralization and Local Governance for National Development.� Ghana Steering Commi ee for 2007 Public Expenditure Tracking Survey, 2007: Ghana— PublicExpenditureTrackingSurvey 2007—Education and Health. Gomez E. J. 2008. “A Temporal Analytical Approach to Decentralization: Lessons from Brazil’s Health Sector.� Journal of Health Politics, Policy and Law, Vol. 33, No. 1, Feb 2008. Duke University Press. Government of Ghana and Development Partners, 2007: Decentralization Policy Review—Final Report. Gyapong J., Garshong B., Akazili J., Aikins M., Agyepong I., and Nyonator F. 2007. Criti- cal Analysis of Ghana Health System, with a focus on equity challenges and the National Health Insurance. SHIELD Work Package Report Hofman B. and Kaiser K. 2002. The Making of the Big Bang and its Aftermath: A Political Economy Perspective, Jakarta, World Bank; 2002. Khemani S. 2005. “Local Government Accountability for Health Service Delivery in Nigeria.� Journal of African Economies, Vol 15, No. 2, 285–312. Kojuro K. N. 2003. “Results-Oriented Public Expenditure Management: Case Study of Ghana.� Working Paper 208, Overseas Development Institute. Kruse I., Pradhan M., and Sparrow R. 2009. Health Spending and Decentralization in Indonesia. 102 A World Bank Study Lieberman S. S. (n.d.) “Decentralization and Health in the Philippines and Indonesia— an Interim Report.� East Asia Human Development, Washington, DC: World Bank. Lister S. and Betley M., June 1999: Approaches to Decentralization in Developing Countries. Paper prepared as background reading for the MEFMI/DFID workshop on Good Practice in Public Expenditure Management, Capetown June 1999. Litvak J., Ahmad J. and Bird R. 1998. “Rethinking Decentralization in Developing Coun- tries.� Sector Studies Series 21491, PREM. World Bank, Washington, DC. Mills A., Vaughan J.P., Smith D.L., and Tabibzadeh I. (eds), 1990. Health System Decentral- ization—Concepts, Issues and Country Experience. WHO, Geneva. Ministry of Health. 2007. Service Availability Mapping. MOH/DFID/WHO. April 2007. Molina G. M. 2009. “An Integrity Perspective on the Decentralization of the Health Sector in Colombia.� Advances in Qualitative Research in Ibero-America, Vol 10, No. 2. Oates W. 1972. Fiscal Federalism. New York: Harcourt Brace Jovanovich. Pallai K. 2001. “The Framework for Decentralization.� Presentation prepared for the World Bank Institute, August 2001. Philippines (Government of the). 2000. Health System Reform Agenda. PHRPlus (Partners for Health Reformplus). 2002. “Decentralization and Health System Reform. Insights for Implementers Series.� No. 1, Sep 2002. Abt Associates Inc. Rondinelli D. A., Nellis J. R., and Cheema G. S. 1983. “Decentralization in Developing Countries – A Review of Recent Experience.� Staff Working Papers No. 581, World Bank, Washington DC. United Cities and Local Governments. 2009. “Decentralization and local democracy in the world: First Global Report 2008� UCLG and the World Bank, Washington DC. World Bank. 2008. Ghana2007 External Review of Public Financial Management—Vol 1. World Bank and Ministry of Health, May 2009. Investing in Health in Ghana: A Review of Health Financing and the National Health Insurance Scheme. World Bank and the International Monetary Fund, 2004: Public Expenditure Manage- ment Country Assessment and Action Plan. Washington, D.C. World Health Organization Ghana Country Office. 2005: Annual Report 2005. WHO. Ye X. and Canagarajah S. 2002. “Efficiency of Public Expenditure Distribution and Be- yond: A Report on Ghana’s 2000 Public Expenditure Tracking Survey in the Sectors of Primary Health and Education.� Africa Region Working Paper Series No. 31, World Bank, Washington, DC. ECO-AUDIT Environmental Bene�ts Statement The World Bank is commi ed to preserving In 2010, the printing of endangered forests and natural resources. this book on recycled paper The Office of the Publisher has chosen to saved the following: print World Bank Studies and Working • 11 trees* Papers on recycled paper with 30 percent • 3 million Btu of total postconsumer �ber in accordance with the energy recommended standards for paper usage • 1,045 lb. of net greenhouse set by the Green Press Initiative, a non- gases pro�t program supporting publishers in • 5,035 gal. of waste water using �ber that is not sourced from endan- • 306 lb. of solid waste gered forests. For more information, visit www.greenpressinitiative.org. * 40 feet in height and 6–8 inches in diameter D ecentralization and Governance in the Ghana Health Sector is part of the World Bank Studies series. These papers are published to communicate the results of the Bank’s ongoing research and to stimulate public discussion. Ghana’s government has embarked on a decentralization process since the 1980s, but devolution of the health system faces important challenges and weaknesses. Ghana has made significant progress and, along the years, put in place several important building blocks for a truly decentralized health system. However, these efforts have been hampered, and their effectiveness diminished, by the absence of a strong regulatory and policy framework for health, by regulatory conflicts and duplications accumulated by several waves of regulations, by weak capacity to coordinate and manage a devolved health system, and by fragmented management systems regarding staff management, procurement, budgeting, and financing. Political instability and the resulting lack of continuity and consistency in the process during most of the period have also prevented these efforts from taking root and have contributed to the current fragmentation in the health system. International comparisons indicate that Ghana shows a narrow effective decision space with respect to most health functions and sub-functions when compared with several other developing countries. The decentralization of the health system could be made more effective and consistent with the Government of Ghana’s (GOG) general policy if the government takes advantage of the structures and features that have been put in place, strengthens the policy framework, ensures effective coordination, and addresses the weaknesses highlighted above. This study analyzes the strengths and weaknesses of the decentralization of the Ghanaian health system and presents a few recommendations and strategies for strengthening the process. It is relevant and timely as the GOG seeks to strengthen and deepen decentralization of the health system; it is also useful for other middle- and low-income countries that are pursuing some form of decentralization of their health system to improve their efficiency and effectiveness. World Bank Studies are available individually or on standing order. This World Bank Studies series is also available online through the World Bank e-library (www.worldbank.org/elibrary). ISBN 978-0-8213-9589-9 90000 9 780821 395899 SKU 19589