66724 A WORLD BANK STUDY Making Health Financing Work for Poor People in Tanzania Dominic Haazen W O R L D B A N K S T U D Y Making Health Financing Work for Poor People in Tanzania Dominic Haazen © 2012 International Bank for Reconstruction and Development / International Development Association or The World Bank 1818 H Street NW Washington DC 20433 Telephone: 202-473-1000 Internet: www.worldbank.org 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 volume is a product of the staff of The World Bank with external contributions. The �ndings, interpretations, and conclusions expressed in this volume 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, denominations, and other information shown on any map in this work do not imply any judg- ment on the part of The World Bank concerning the legal status of any territory or the endorsement or acceptance of such boundaries. Rights and Permissions The material in this work is subject to copyright. Because The World Bank encourages dissemina- tion of its knowledge, this work may be reproduced, in whole or in part, for noncommercial purposes as long as full a ribution to the work is given. For permission to reproduce any part of this work for commercial purposes, please send a request with complete information to the Copyright Clearance Center Inc., 222 Rosewood Drive, Danvers, MA 01923, USA; telephone: 978-750-8400; fax: 978-750-4470; Internet: www.copyright.com. All other queries on rights and licenses, including subsidiary rights, should be addressed to the Office of the Publisher, The World Bank, 1818 H Street NW, Washington, DC 20433, USA; fax: 202-522- 2422; e-mail: pubrights@worldbank.org. ISBN (print): 978-0-8213-9473-1 ISBN (electronic): 978-0-8213-9474-8 DOI: 10.1596/978-0-8213-9473-1 Library of Congress Cataloging-in-Publication Data Haazen, Dominic, 1954- Making health �nancing work for poor people in Tanzania / Dominic Haazen. p. ; cm. Includes bibliographical references. ISBN 978-0-8213-9473-1 -- ISBN 978-0-8213-9474-8 I. World Bank. Africa Regional Office. Human Development Dept. II. Title. [DNLM: 1. Health Expenditures--Tanzania. 2. Health Policy--Tanzania. 3. Healthcare Disparities-- economics--Tanzania. 4. Insurance, Health--economics--Tanzania. 5. Organizational Case Studies-- Tanzania. 6. Poverty--Tanzania. W 74 HT3] 362.109678--dc23 2012000724 Cover photo: Tanzanian women and their children walking in the school yard of a rural village. Photo by Isabelle M. Haazen. Contents Acknowledgments .................................................................................................................. vii Abbreviations, Acronyms, and Currencies ..........................................................................ix Executive Summary ..................................................................................................................xi 1. Introduction ......................................................................................................................... 1 2. Socioeconomic Context ..................................................................................................... 4 The Tanzanian Economy and Poverty Reduction........................................................... 4 Population Dynamics and Demographic Changes ........................................................ 6 3. Diagnosis and Evaluation of Current Health Financing System ............................ 10 Overview of the Health System and Health Financing in Tanzania .......................... 10 Trends in Health Financing ............................................................................................. 13 Overview of the Functioning of Current Prepayment Mechanisms.......................... 27 4. Experiences from Other Countries................................................................................ 35 Thailand .............................................................................................................................. 35 Vietnam ............................................................................................................................... 37 China ................................................................................................................................... 39 Ghana .................................................................................................................................. 40 Rwanda ............................................................................................................................... 42 Kenya ................................................................................................................................... 43 Colombia ............................................................................................................................. 45 Mexico ................................................................................................................................. 48 Summary of Lessons from International Experience ................................................... 50 5. Key Policy Issues and Reform Options ....................................................................... 53 Areas To Be Considered in the Health Financing Strategy ......................................... 54 Economic and Financial Impact of Reform Options .................................................... 61 Summary and Next Steps ................................................................................................. 69 Appendix A. Health Financing Systems in Other Countries .......................................... 71 Thailand .............................................................................................................................. 71 Vietnam ............................................................................................................................... 73 China ................................................................................................................................... 74 Ghana .................................................................................................................................. 76 Rwanda ............................................................................................................................... 79 Kenya ................................................................................................................................... 82 Colombia ............................................................................................................................. 86 Mexico ................................................................................................................................. 89 iii iv World Bank Contents Study References.................................................................................................................................. 93 Boxes Box 4.1: Key Messages from Thailand ............................................................................ 37 Box 4.2: Key Messages from Vietnam............................................................................. 38 Box 4.3: Key Messages from China ................................................................................. 40 Box 4.4: Key Messages from Ghana ................................................................................ 41 Box 4.5: Key Messages from Rwanda ............................................................................. 43 Box 4.6: Key Messages from Kenya ................................................................................ 45 Box 4.7: Key Messages from Colombia .......................................................................... 47 Box 4.8: Key Messages from Mexico ............................................................................... 49 Figures Figure 1.1: Framework for Health Financing Reform Options ..................................... 2 Figure 3.1: Organization of the Health System in Tanzania ........................................ 11 Figure 3.2: Public Health Spending as a Percentage of Total Government Expenditure ........................................................................................................................ 15 Figure 3.3: Household Spending on Health, as Percentage of Expenditure ............. 20 Figure 3.4: Kakwani Indices of Health Care Financing Sources ................................. 22 Figure 3.5: Household Health Care Financing as a Proportion of Income ............... 23 Figure 3.6: Total Income and Consumption Taxes as a Proportion of Income ......... 23 Figure 3.7: Bene�t Incidence for Various Health Interventions .................................. 25 Figure 3.8: Share of Bene�ts versus Need by Wealth Quintile ................................... 26 Figure 3.9: Health Care Use by Insured and Uninsured Persons ............................... 26 Figure 3.10: Member and Bene�ciary Growth .............................................................. 28 Figure 3.11: Expenses as Percentage of Revenue .......................................................... 28 Figure 3.12: Costs as Percentage of Premium Revenue ............................................... 29 Figure 5.1: Coverage with Prepaid Health Insurance .................................................. 62 Figure 5.2: Service Growth in Public Facilities .............................................................. 63 Figure 5.3: Modeled Health Expenditure as a Percentage of GDP............................. 64 Figure 5.4: Modeled Health Expenditures by Financing Source ................................ 64 Figure 5.5: Public Health Expenditures as Percentage of GDP and Government Budget .......................................................................................................... 65 Figure 5.6: Effect of Efficiency Gains on Expenditures as Percentage of Government Budget .......................................................................................................... 66 Figure 5.7: Sensitivity Analysis—Impact of 4 Percent Real Economic Growth ..................................................................................................... 68 Figure 5.8: Sensitivity Analysis—Real Unit Cost Growth from 2 Percent to 4 Percent ........................................................................................................ 69 Figure A.1: Insurance Coverage by Income, Quintile and Year.................................. 88 Figure A.2: Enrollment in Seguro Popular by Income Quintile .................................... 91 Contents Making Health Financing Work for Poor People in Tanzania v Tables Table 1: Summary of Current Health Financing Approaches in Tanzania ................xi Table 2.1: Key Economic Indicators .................................................................................. 4 Table 2.2: Population and Demographic Indicators and Projections ........................... 6 Table 2.3: Employed Population Characteristics............................................................. 8 Table 3.1: Total Health Spending by Source of Funds ................................................. 13 Table 3.2: Public Health Spending by Financing Sources ............................................ 14 Table 3.3: Changes in Key Health Sector Indicators ..................................................... 16 Table 3.4: NHIF Income and Reimbursement, 2001/02 to 2009/10 ............................. 17 Table 3.5: Summary of Prepayment Plans, 2010............................................................ 18 Table 3.6: Poverty Impact of Health Expenditures ....................................................... 21 Table 4.1: Summary of Health Financing Approaches in Thailand ........................... 36 Table 4.2: Summary of Health Financing Approaches in Vietnam ............................ 38 Table 4.3: Summary of Health Financing Approaches in China ................................. 39 Table 4.4: Summary of Health Financing Approaches in Ghana................................ 41 Table 4.5: Summary of Health Financing Approaches in Rwanda............................. 42 Table 4.6: Summary of Health Financing Approaches in Kenya ................................ 44 Table 4.7: Summary of Health Financing Approaches in Colombia .......................... 46 Table 4.8: Summary of Health Financing Approaches in Mexico .............................. 48 Table 5.1: Incremental Costs by Type ............................................................................. 67 Acknowledgments T his report was prepared by Dominic Haazen (Lead Health Policy Specialist, Africa Human Development Sector Department). Research for the report was contributed by Alexander Hayer (summer intern) and Dr. Emmanuel Malangalila (Consultant). The author is grateful to peer reviewers John Langenbrunner (Lead Health Economist, East Asia and Paci�c Human Development Department) and Christoph Kurowski (Sector Leader, Latin America and Caribbean Human Development Department) for their com- ments. Special thanks are due to Josephine Borghi (Senior Research Scientist, Ifakara Health Institute, Dar es Salaam, Tanzania) for her comments on several drafts and her collaboration on the use of the Strategies for Health Insurance for Equity in Less Devel- oped Countries (SHIELD) model for the �nancial and economic modeling included in the report. The support of the Pensions and Health Financing Technical Assistance team, led by Michel Noel (Lead Financial Specialist, AFTFE), as well as Chris Walker (Cluster Leader, AFTHE, retired) and John May (Acting Cluster Leader, AFTHE), is appreciated. vii Abbreviations, Acronyms, and Currencies BMI Basic Medical Insurance CAG Comptroller and Auditor General CFS Consolidated Fund Services CHF Community Health Fund CMS Cooperative Medical Scheme CSMBS Civil Servants’ Medical Bene�t Scheme CSSC Christian Social Services Commission DHS Demographic and Health Survey FBO faith-based organization FCFCU6 Free Care for Children under 6 FNG National Guarantee Fund FSD District Solidarity Fund FY �scal year GBS general budget support GDP gross domestic product GNI gross national income HCFP Health Care for the Poor HIV/AIDS Human Immuno-De�ciency Virus/Acquired Immune De�ciency Syndrome HSS health systems strengthening HSSP Health Sector Strategic Plan IDRC International Development and Research Centre ILFS Integrated Labour Force Survey IMSS Mexican Social Security Institute ISSSTE Social Security Institute for Government Workers M&E monitoring and evaluation MA Medical Assistance MHI medical health insurance MHO mutual health organization MMAM Primary Health Care Development Program (in Kiswahili) MMI military medical insurance MoDLA Ministry of Decentralization and Local Affairs MoH Ministry of Health MoHSW Ministry of Health and Social Welfare NAO National Audit Office NBS National Bureau of Statistics NGOs nongovernmental organizations NHIA National Health Insurance Authority NHIF National Health Insurance Fund NHIS National Health Insurance Scheme NHSO National Health Security Office NRCMS New Rural Cooperative Rural Medical Scheme ix x World Bank Study Abbreviations, Acronyms, and Currencies NSSF National Social Security Fund OOP out-of-pocket PBF performance-based �nancing PEPFAR (U.S.) President’s Emergency Program for AIDS Relief PHC primary health care PIN personal identi�cation number PIT personal income tax PMI (U.S.) President’s Malaria Initiative PMO-RALG Prime Minister’s Office, Regional Administration and Local Government RAMA Rwanda Health Insurance Company (Rwandaise d’Assurance Maladie) RC contributory scheme (Régimen Contributivo) RHMT regional health management team RS subsidized scheme (Régimen Subsidiado) SCHIP Strategis Community Health Insurance Plan SHI Statutory Health Insurance SHIB Social Health Insurance Bene�t SHIELD Strategies for Health Insurance for Equity in Less Developed Countries SISBEN Selection System of Bene�ciaries for Social Programs SP Seguro Popular SSF Social Security Fund SSRA Social Security Regulatory Authority SSS Social Security Scheme TB tuberculosis TC-SWAP Technical Commi ee-Sector Wide Approach THE total health expenditure TIKA Tiba kwa Kadi, the urban form of the CHF TPH Tropical and Public Health (Swiss Institute for) UC universal coverage UMASITA Tanzania Informal Sector Community Health Fund USAID U.S. Agency for International Development VAT value added tax VHI Voluntary Health Insurance VIBINDO umbrella organization of informal sector operators in Dar es Salaam region VSS Vietnam Social Security Agency WHO World Health Organization Currencies B Thai baht T Sh Tanzania shilling US$ U.S. dollar Executive Summary T his policy note is designed to support the development of the health �nancing strat- egy in Tanzania. It is directed at decision makers in the areas of health and social policy as well as the Ministry of Finance, which will play a crucial role in integrating the �nancial implications of this note into the overall �scal situation in Tanzania. It is also hoped that this note will stimulate debate among interested stakeholders on the best funding modalities for health and the most appropriate ways to integrate those modalities. On the basis of the data and options described in this policy note, the World Bank will work with authorities and other interested stakeholders to develop a �nancing program to support the needed reforms in these sectors. The health �nancing system in Tanzania is highly fragmented, with many �nanciers and modes of �nancing. Out-of-pocket payments account for about half of total health spending, with an increasing portion of that total being channeled through various pub- lic and private prepayment schemes. Table 1 provides an overview of current health �nancing approaches. Public �nancing includes a large proportion of external �nancing, with a signi�cant part of that being off-budget. This fragmentation results in substan- tial inefficiencies in the use of resources and often-conflicting incentives for the various actors in the health system. Given the limited funding that is available, it is clear that a rationalization of these various modes of �nancing must take place to ensure improved access to services, improved quality, and ongoing sustainability. Moreover, the existing public �nancing mechanisms must be reviewed to ensure that they promote access to services for the poor and vulnerable. Table 1: Summary of Current Health Financing Approaches in Tanzania Formal Formal Informal public private Informal rural/ Dimension sector sector Poor urban farmers Prepayment scheme NHIF NSSF-SHIB CHF/TIKA TIKA CHF Coverage (number) 2.0 million 51,300 1.7 million Coverage (% of 4.6% 9.2% 3.9% population) Source of revenue 6% payroll Part of 20% NSSF Councils pay Individual contribution of T Sh 5– contributions, half contribution through premium for 15,000 matching grant from Health employer, half payroll; SHIB not poor Basket Fund employee differentiated Revenue collection Payroll deduction Payroll deduction Collected at health facility level, remitted to district method and submission and submission to NHIF to NSSF Number of risk pools One One One per district, but funds held at facility level so they are effectively the risk pool Payment methods Fee-for-service Capitation, some Essentially capitation, as there is no payment for fee-for-service services; facilities use CHF revenue to support service delivery Benefit package Full range of Broad range of Mostly public primary health care and some hospital services services services, within district (Table continues on next page) xi xii World Bank Study Executive Summary Table 1: Continued Formal Formal Informal public private Informal rural/ Dimension sector sector Poor urban farmers Facility coverage Extensive network Smaller network Mostly the facility where the subscriber registers; little of over 5,500 of 264 facilities portability to other districts or other district facilities health facilities Regulatory oversight Separate Board Social Security MoH and NHIF of Directors Regulatory Authority Other financing Government and nonprofit facilities available upon payment of user fees (some exemptions); how- methods ever, problematic availability of drugs and medical supplies Other insurance Private insurance Private insurance, Not Urban-based Rural and FBO schemes purchased by including purchased applicable micro-insurance micro-insurance employees by employers schemes schemes Source: Author. Notes: CHF = Community Health Fund; FBO = faith-based organization; MoH = Ministry of Health; NHIF = National Health Insurance Fund; NSSF = National Social Security Fund; SHIB = Social Health Insurance Bene�t; TIKA = Tiba kwa Kadi, the urban form of the CHF. Because the various parts of the system are closely connected, a comprehensive and sustainable health �nancing strategy must address each part of the system. Speci�c at- tention must be focused (a) on the regulatory system for both insurers and providers and (b) on speci�c strategies to a ract and enroll the poorest in prepayment schemes. The analysis of the approaches in different countries suggests a number of recurring and reinforcing themes: ■ Special efforts are required to reach the poor because they cannot afford ser- vices and often cannot afford prepaid health insurance. ■ Such special efforts to serve the poor and other members of the informal sector require incremental subsidies. ■ Such incremental subsidies can increase public spending by as much as 1.0 to 1.8 percent more than existing levels expressed as a share of gross domestic product (GDP), although this increase can be largely offset by reductions in out-of-pocket payments. ■ Capped, capitation, case-based, or a combination approaches are essential for cost control, and fee-for-service should be avoided if at all possible. ■ If one is to maintain a sustainable system, it is critical to match bene�ts to rev- enue, even if this approach means starting with a small bene�t package or dif- ferent packages for different groups. ■ The �scal transfers required and the redistributive effects of major health �- nancing reforms make political leadership essential, as shown in the Thailand and Ghana examples. ■ The need for cost controls makes a strong regulator essential for reform, as shown in the Thailand and Rwanda examples. ■ Although small risk pools can enhance community ownership, the Ghana and Rwanda examples show that those pools can also create sustainability problems People in Tanzania Making Health Financing Work for PoorExecutive Summary xiii and administrative capacity issues, unless explicit cross-subsidization or rein- surance mechanisms or support systems such as consolidated revenue collec- tion or claims management are put in place. It is clear that health reforms require trade-offs that must be acceptable to the major- ity of the population. For example, the goal of enrolling 45 percent of the population by 2015 must confront the costs of identifying, registering, and serving the informal sector and the poor. Increasing the enrollment in such groups will require raising the aware- ness and understanding of the concept of risk pooling and insurance generally, as well as building trust in the institutions that offer insurance so that those who sign up are con�dent that the bene�ts offered will be received when needed. It is also likely that a broader range of bene�ts will need to be offered, especially low-use and high-cost ser- vices such as inpatient care, because the a ractiveness of the Community Health Fund (CHF) bene�t package is limited by the relatively low cost of primary care. A number of areas have been identi�ed that should be pursued and could be in- cluded in the Health Financing Strategy. They are organized here along the lines of the health �nancing functions: Service Provision ■ Reduce existing inefficiencies and inequities in the health system. ■ Recognize the need to review the management and governance structures of health facilities. ■ Supply the improved mechanisms that are needed for the allocation of drugs and medical supplies to facilities. Purchasing ■ Evaluate how alternatives to fee-for-service could be implemented within the National Health Insurance Fund (NHIF), as well as other insurers, and then develop action plans for shifting to an alternative approach over time. ■ Decide whether to use per diem or case-based reimbursement for inpatient hos- pital services, and develop an appropriate action plan. Risk Pooling ■ Change the NHIF to a scheme �nanced only by subscribers, thereby redirecting the government contribution toward improving social health protection for the poor. ■ Work with the NHIF to use T Sh 250 billion of the estimated T Sh 300 billion in accumulated NHIF reserves to help strengthen the health system. ■ Remove the Social Health Insurance Bene�t (SHIB) from the National Social Security Fund (NSSF), and provide coverage through the NHIF with a total premium of 2.5 percent paid by employees. ■ Have CHFs focus on community mobilization and revenue raising, and have the NHIF use its comparative advantage to act as the claims processor, card is- suer, and funds manager. ■ Establish uniform CHF rates at T Sh 30,000 for urban and T Sh 15,000 for rural. xiv Executive Summary World Bank Study ■ Set matching funds at 100 percent for urban and 150 percent for rural to rec- ognize the higher poverty rates in rural areas, and use savings from the NHIF changes to provide the matching funds and to increase coverage for the poor. Revenue Generation ■ Recognize the need for flexible payment approaches for membership in the CHF and TIKA (Tiba kwa Kadi, the urban form of the CHF) to reflect the ir- regular income pa erns of informal sector workers and subsistence farmers. ■ Leverage existing budget resources to promote access to services, thus �nanc- ing outputs rather than inputs. ■ Look at other sources of sustainable �nancing for the health sector, such as an earmarked portion of value added tax (VAT) revenue. ■ Examine the future role of user fees, and determine appropriate approaches for identifying the poor for the purposes of exemptions, waivers, or premium subsidies. The Regulatory Framework and the Political Environment ■ Put in place a comprehensive regulatory framework to govern the overall health �nancing system. ■ Garner the appropriate political commitment for the Health Financing Strategy. To assess the �nancial and economic effects of reform options, simulations were run using a model developed by the Ifakara Health Institute for the Strategies for Health Insurance for Equity in Less Developed Countries (SHIELD) project. The following op- tions were simulated: ■ A status quo option used no major changes in the parameters of health �nancing and no signi�cant initiatives to encourage enrollment in prepaid health insur- ance schemes. This option involves current health systems strengthening (HSS) efforts, including investments in the development of health human resources, monitoring and evaluation (M&E), and infrastructure. ■ A 45 percent target option (Option 1) simulated the effect of reaching the stated objective of covering 45 percent of the population with prepaid health insurance by 2015. The option includes various changes to the NHIF, SHIB, and CHF/ TIKA coverage and premium levels, as well as utilization changes for CHF/ TIKA and NHIF members to reflect a changing bene�t package and a higher number of accredited facilities. HSS efforts are intensi�ed to keep pace with increasing service demand. ■ A universal coverage option (Option 2) used the same basic parameters as Op- tion 1 but assumed (a) that the entire informal sector be covered through either the CHF (rural) or TIKA (urban) scheme and (b) that 77 percent of the private formal sector be covered through either private insurance or the NHIF. HSS investments are also scaled up in this option. The simulations show that the extension of prepaid health insurance coverage to a larger segment of the population should not cause signi�cant increases in either the percentage of GDP spent on health services or the proportion of the total government People in Summary Making Health Financing Work for Poor Executive Tanzania xv budget devoted to such services, as long as robust economic growth continues. A lower growth assumption results in health expenditures as a percentage of GDP increasing from 4.5 percent to 5.0 percent in the universal coverage option and increasing as a per- centage of the government budget from 9 percent to 13 percent, still well below the Abuja target. It would take a substantial slowing in economic growth (to 4 percent annu- ally throughout the projection period) to push health expenditures to 6 percent of GDP and 15 percent of government expenditures. Even in the absence of any reforms, there will be a signi�cantly increased demand for health services as a result of population growth, with about 53 percent more services projected by 2025 than are currently being provided. This forecast underscores the need for continued a ention to health system strengthening, especially in terms of health hu- man resources. It also highlights the fact that efficiently using the human and �nancial resources that are already available will become increasingly important. The simulation results indicate that efficiency improvements can help to mitigate the cost implications of scaling up prepaid health insurance coverage. Given the strong commitment of a number of development partners to improving social health protection, it should be possible to generate interest in helping to �nance the transition and perhaps other costs associated with this expansion of coverage. Of course, a credible, sustainable action plan and detailed costing will be needed to gar- ner such support, not just from development partners, but also from the government, which will likely bear the greatest proportion of any cost increases. The health �nancing strategy and the accompanying action plan should provide the basis for these types of discussions. This policy note provides background information, cross-country examples, and policy options, which can all be incorporated into the development of a comprehensive health �nancing strategy. It also provides a framework for looking at the various ele- ments of the health �nancing system, and it explores the �nancial, economic, and health system implications of a number of the options. A key aspect in moving forward is to determine the acceptability of the options from a political economy perspective, as well as to consider in depth their feasibility and expected costs. This effort should assist in �nalizing the health �nancing strategy and should lead to the identi�cation of speci�c policy actions and reforms that will be needed to set into operation the strategy and its various components. Given the nature of such policy actions and the reforms they would support, it may be possible to consider them for inclusion in future development policy operations, either at the general budget support (GBS) level or as a separate sectoral development policy operation. Drawing on this analysis, speci�c support could be targeted to assist with the transition costs to universal coverage, notably in the following areas: ■ Set in place an appropriate regulatory framework to monitor and provide over- sight to prepaid health insurance operations (both public and private), bearing in mind the evolving role of the Social Security Regulatory Authority (SSRA). ■ Address the role and future of the SHIB, which should be consistent with the pension reforms that are also being pursued. ■ Make substantive changes to the �nancing framework for the NHIF, along the lines described in this note. xvi World Bank Study Executive Summary ■ Focus on the organizational and operational changes required for the CHF/ TIKA to enable them to play the role envisioned in this policy note. ■ Address the need for increased efficiency and provider autonomy. Health �nancing reform of any signi�cance will create major changes in the way in which the health sector operates. Those changes make it important to fully discuss with all relevant stakeholders the overall vision for health �nancing, its effect on the health sector, and the speci�cs of any proposed changes before proceeding. The experience of other countries, most notably Kenya, shows that such reforms can be doomed to failure if proposed changes are not fully explained and discussed. It is recommended, therefore, that consultation and advisory mechanisms be put in place immediately to facilitate this essential dialogue. The explicit signaling of high-level support for health �nancing reform is also a critical next step in the process. CHAPTER 1 Introduction T he vision of health policy in Tanzania is to improve the health and well-being of all Tanzanians, especially those at risk, and to enable the health system to be more responsive to the needs of the people. To achieve this vision, the health sector seeks to facilitate the provision of equitable, quality, and affordable basic services that are gen- der sensitive and sustainable and that are aimed at achieving improved health status. Tanzania’s latest Health Sector Strategic Plan (HSSP III, 2009–15) focuses on the need for effective partnerships—with public and private health facilities, development partners, and other stakeholders—to contribute to the achievement of the Millennium Develop- ment Goals. An emerging goal of a national health �nancing system is to provide universal coverage and social health protection. The objective is to improve access to services by removing barriers to care—especially for poor people and those in rural and remote areas—and to ensure that sufficient resources exist to enable health care providers to de- liver a basic package of high-quality health care services. A key component of this effort is measuring and shaping the performance of this system. An important prerequisite for this is a comprehensive health �nancing strategy, which will help Tanzania (a) analyze the adequacy, organization, distribution, and effects of its health investments on its over- all health goals; (b) plot a path for sustainable health �nancing into the future; and (c) guide future mobilization, allocation, polling, and use of resources. This note is designed to serve as a key input to the development of this �nancing strategy by highlighting the key parameters and issues in the health �nancing system and suggesting options that might be pursued during the implementation of the strat- egy. It is directed at decision makers in the areas of health and social policy and at the Ministry of Finance, who will play a crucial role in integrating the �nancial implications of this note into the overall �scal situation in Tanzania. It is also hoped that this note will stimulate debate among interested stakeholders on the best funding modalities for health and the most appropriate ways of integrating these modalities. Reform options must address key aspects of revenue generation and expenditure management, as well as provider organization and service delivery arrangements. Such options also need to take into account the critical issues of service quality (and the mechanisms for ensuring it) and regulation of the system (including both insurers and providers). Figure 1.1 provides a framework for looking at health �nancing reform options. It shows the interrelationship of the revenue collection, pooling, purchasing, and provi- sion functions within the overall health system. The system itself operates within a regu- latory framework (the shaded blue area), the most salient features of which are shown next to the applicable functions. 1 2 World Bank Study Because the various parts of the system are closely connected, a comprehensive and sustainable health �nancing strategy must address each part of the system. This frame- work can be applied to any type of system, from a totally tax-based system to fully premium-based social health insurance, although in the former model, the central gov- ernment is essentially responsible for revenue collection, pooling, and purchasing and may provide services as well. Figure 1.1: Framework for Health Financing Reform Options Source: Adapted from Ku in 2000. HSSP III expresses the view that implementation of cost-sharing and prepayment schemes has a great potential for raising additional revenues for the health sector and providing flexible funding to health facilities. The government of Tanzania prefers to improve health insurance schemes rather than increase out-of-pocket expenditure by patients, and it would like to increase social health insurance coverage to eventually reach universal coverage. To this end, the government has set a target of enrolling 45 percent of the population in prepayment schemes by 2015. However, to do this, policy makers must put regulatory and �nancing mechanisms and speci�c strategies in place to a ract and enroll the poorest in prepayment schemes. Such strategies will need to be carefully developed, and the potential �nancial effects must be closely studied. Effective regulatory mechanisms to guide and supervise health insurance schemes must also be established. On the expenditure side, sustainable approaches must promote the use of funds received by health care facilities to ensure that facility managers—with increased com- munity input—are able to make the decisions necessary to allow them to provide high- quality services. In practical terms, all resources a facility receives should be free of speci�c conditions or encumbrances, and a common budgeting and approval process Making Health Financing Work for Poor People in Tanzania 3 should be established. Moreover, to the extent possible, health facilities should be able to retain and use the revenue that they generate from prepayment schemes and user fees. The structure of the note is as follows. Chapter 2 describes the socioeconomic con- text in Tanzania. Chapter 3 provides a diagnosis and evaluation of the current health �nancing system, focusing on the overall health �nancing situation, the trends in health �nancing, and the contribution of the major �nanciers within the system. It also includes a speci�c examination of the various prepayment schemes that currently exist. Chapter 4 describes the efforts of other countries in reforming their health �nancing systems to provide examples of what can be done given proper planning and political will. Finally, chapter 5 discusses a number of potential reform options and assesses the �nancial and health system effects of each option, as well as the implications of each option relative to the framework described in this chapter. CHAPTER 2 Socioeconomic Context The Tanzanian Economy and Poverty Reduction1 Tanzania experienced economic growth of between 5 and 7 percent per year during the period from 2000 to 2008, until the global �nancial crisis hit the economy in 2009 (table 2.1). Sustained economic reforms and macroeconomic stability, coupled with a favorable external environment, contributed to this growth. Although the government had carried out an expansionary �scal policy since 2000 with a fast rise in government expenditures, the �scal environment was overall on a prudent track, sustained by strong growth in tax revenues, large debt relief, and signi�cant growth in foreign aid. Inflation was initially kept in check but accelerated starting in 2005, and it reached double digits in 2008, driv- en by adverse regional food supply shocks and challenges in managing large liquidity coming from public and private capital flows. The real effective exchange rate converged to a more competitive level after being overvalued at the beginning of the decade but it has started to appreciate again in real terms, mainly because of higher domestic inflation. The current account balance has deteriorated since 2006, driven by high imports (high imported oil prices and capital import growth). It is now no longer fully covered by foreign aid and, therefore, is more dependent on foreign direct investment flows. Table 2.1: Key Economic Indicators Actual Estimated Projected 2005 2006 2007 2008 2009 2010 2011 2012 2013 GDP (US$ million, 14,251 14,797 17,305 19,017 21,628 23,724 25,738 27,980 30,586 current prices) GNI per capita 378 383 387 442 460 475 490 510 550 (US$, Atlas method) GDP growth (%) 7.4 6.7 7.1 7.4 6.0 6.0 6.0 6.0 7.4 Annual inflation 5.0 7.2 7.0 10.2 11.9 9.9 9.0 8.5 7.5 (% CPI, end of year) Current account 6.8 8.8 7.8 12.5 10.6 9.1 9.3 10.0 10.3 deficit (% of GDP) Exchange rate 1,165 1,261 1,132 1,280 1,320 1,440 n.a. n.a. n.a. (T Sh/US$) (Table continues on next page) 4 Making Health Financing Work for Poor People in Tanzania 5 Table 2.1: Continued Actual Estimated Projected FY05/ FY06/ FY07/ FY08/ FY09/ FY10/ FY11/ FY12/ FY13/ 06 07 08 09 10 11 12 13 14 Domestic revenue 12.4 14.1 15.9 16.2 15.4 16.1 16.3 16.5 16.7 to GDP Overall budget −4.8 −4.9 0.0 −4.5 −6.4 −6.5 −6.0 −5.2 −4.1 balance to GDP (after grants) Domestic borrowing to GDP 1.5 1.2 −1.5 1.0 1.8 1.0 1.0 1.0 1.0 Source: World Bank 2011a, 7, 58, 59. Notes: n.a. = not applicable; FY = �scal year; GNI = gross national income. The global �nancial and economic crisis has led to a growth slowdown in Tanzania. Real GDP growth in 2009 was estimated to have dropped to about 6 percent from 7.4 percent in 2008. The crisis has affected Tanzania through the export channel (mainly tourism, cash crops, and regional manufacturing exports) and through lower capital flows (foreign assistance and private investment flows). Manufacturing, wholesale and retail trade, transport, communication services, and construction are the most affected sectors. The decline of international commodity prices and the reduction of food shortages in the subregion have yet to result in lower inflation- ary pressures. Inflation has been mainly driven by food prices and reflects the impact of the drought in parts of northern Tanzania and neighboring countries. Despite the solid growth record, �ndings from 2009 analysis of the Household Bud- get Survey 2007 by the National Bureau of Statistics (NBS) show a mixed picture of the country’s progress in poverty reduction over the past eight years (NBS 2009). Lower relative prices from imported goods and increased government spending led to tangible improvements in some areas, including ownership of consumer durables, housing qual- ity, and some social indicators, (such as mortality of children under �ve years of age and enrollment in primary education). Nevertheless, progress was rather limited in terms of basic needs income poverty, ownership of productive assets in rural areas, and some other social indicators (such as maternal mortality and access to safe water). Between 2001 and 2007, the incidence of income poverty fell slightly in mainland Tanzania, as did the depth and intensity of poverty. However, the size of those decreases is small. Per capita consumption measured through the Household Budget Survey in- creased by only 5 percent in real terms between 2000–01 and 2007. With such a small change in consumption, the poverty head-count ratio dropped only slightly, from 35.6 in 2001 to 33.6 in 2007. Because the head count fell by only a li le while the population continued to grow rapidly, the absolute number of poor people increased by 1.3 million in the period. In rural areas, the change in poverty was not statistically signi�cant. Even though relative inequality remained more or less unchanged, consumption declined for a small fraction of households at the very bo om of the income distribution. There was also a marked urban-rural gradient in poverty incidence, with the poverty headcount ratio in Dar es Salaam calculated at 16.4, compared with 24.1 in other urban areas and 37.6 in rural parts of Tanzania. 6 World Bank Study Population Dynamics and Demographic Changes Changes in population numbers and demographics are important because they indicate the changing requirements for various types of services and infrastructure, even if there is li le or no change in living standards. With a population of approximately 43 million in 2010, Tanzania is the seventh most populous country in Africa. At 2.9 percent, the an- nual population growth rate is high, and it is projected to remain high into the foresee- able future, dropping to only 2.73 percent by 2025. According to the National Bureau of Statistics Population Projection, this relatively stable growth rate is the result of a signi�- cant decline in the total fertility rate, from 5.1 to 3.4, as well as a reduction in the crude death rate, from 13.3 to 9.7. The expected decline in the infant and under-�ve mortality rates contribute to the overall decline in the death rate. Despite the expected decline in the total fertility rate, the signi�cant (almost 50 per- cent) increase in the number of women of childbearing age means that the total number of babies born each year will remain more or less constant over the projection period. As a result of those dynamics, Tanzania’s population is projected to almost double from 33.7 million in 2000 to 65.3 million by 2025. This projection is important for policy making and planning for future health �nancing options, because the size and structure of the population has an effect on what might be feasible. Table 2.2 shows the historical and potential future characteristics of the population. Table 2.2: Population and Demographic Indicators and Projections Historical NBS Projection 1980 2000 2010 2015 2020 2025 Total population (million) 18.67 33.71 43.19 49.86 57.10 65.34 Women (aged 15 to 49) 4.15 7.74 10.37 11.82 13.44 15.45 Percent 22.2 23.0 24.0 23.7 23.5 23.6 Children (aged 0 to 14) 8.78 15.48 19.17 22.23 25.27 28.18 Percent 47.0 45.9 44.4 44.6 44.3 43.1 Working age population (aged 15 to 59) 8.83 16.77 22.02 25.43 29.31 34.18 Percent 47.3 49.7 51.0 51.0 51.3 52.3 Older population (aged 60 or more) 1.06 1.46 2.00 2.20 2.52 2.97 Percent 5.7 4.3 4.6 4.4 4.4 4.5 Dependency ratios (%) Young 0.99 0.92 0.87 0.87 0.86 0.82 Elderly 0.12 0.09 0.09 0.09 0.09 0.09 Total 1.11 1.01 0.96 0.96 0.95 0.91 Population growth rate 3.03 3.00 2.51 2.92 2.75 2.73 Number of births (thousands) 896.1 1,396.0 1,678.3 1,704.8 1,697.9 1,660.2 Number of deaths (thousands) 360.6 555.9 573.2 589.1 621.7 636.7 Crude birth rate (per 1,000 population) 48.01 41.41 38.86 34.19 29.73 25.41 Crude death rate (per 1,000 population) 19.32 16.49 13.27 11.81 10.89 9.74 Total fertility rate (births per woman) 6.84 5.74 5.10 4.50 3.90 3.40 Infant mortality rate (per 1,000 births) 128.6 82.2 80.5 72.2 64.7 56.8 Life expectancy (years) 47.3 46.3 55.0 57.0 58.0 61.0 Sources: Historical data are from the U.S. Census Bureau International Database (1960 population is 10.26 million); projections are from the Tanzania National Bureau of Statistics. Making Health Financing Work for Poor People in Tanzania 7 The impact of population growth on the need for infrastructure and staffing will be considerable because by 2025, health services will need to be delivered to more than half again as many people as in 2010. As will be shown, the system is already having difficulty coping with the current service needs, so gearing up for such a large increase in just 15 years will be a signi�cant challenge. It is worth noting that the relatively stable number of births projected from the NBS model means that maternity and obstetrical care will need to focus on ensuring that services are available for current levels of deliv- ery and that those services are, in fact used. Other factors are also important when evaluating health �nancing reform options. Dependency ratios are particularly important, because they reflect the ability of the workforce to cover the costs of those who are not economically active. Overall, there will be a slight decline in the dependency ratio, mostly as a result of a reduction in the percentage of young people in the overall population, as those currently under 15 years of age enter the workforce. Changes in household size are also important. Overall house- hold size dropped from 5.7 in 1991/92 to 4.9 in 2001/02 and 4.8 in 2007. At an average of 5.1 adult equivalents in 2007, household size in rural areas is considerably higher than in urban areas; the decline in size in those areas has not been as great (the average size in 1991/92 was 5.9). In contrast, there has been a continuing drop in household size in Dar es Salaam, from 4.6 in 1991/92 to 4.3 in 2000/01 and 3.7 in 2007. The change in household size in other urban areas has not been quite as dramatic, going from 4.9 in 1991/92 to 4.4 in 2007. The proportion of female-headed households also increased considerably, from 18 percent in 1991/92 to 23 percent in 2000/01 and 25 percent in 2007. Note that these �gures are taken from the 2007 Household Budget Survey and are expressed in terms of “adult equivalents,� not just the number of family members. Another important element is the extent to which jobs will be available for these new workers. According to the Integrated Labour Force Survey (ILFS 2006), labor force participation rates for women and men are high (88.8 percent and 90.5 percent, respec- tively) even though there is a high degree of informality in the workforce and a sub- stantial proportion of the total workforce is engaged in agriculture. In 2006, 40 percent of all households engaged in informal sector activities, with the proportion reaching 57 percent in Dar es Salaam and 54–55 percent in urban areas (NBS 2007: 43). The propor- tion in rural areas was 33 percent. Between 2001 and 2006, the level of informal activities decreased in urban areas but increased in rural areas, growing overall from 35 to 40 per- cent. The overall number of households estimated to engage in informal sector activities grew by 65 percent, from 2.0 million to 3.3 million households. About three-quarters of all employed individuals work informally in agriculture and �sheries, with this proportion increasing to almost 80 percent for women and decreasing to 70 percent for men. Naturally, the proportion in rural areas is much higher than in the cities: 88 percent versus 13 percent in Dar es Salaam and 43 percent in other urban areas (ILFS 2006, 26). An important consideration for any health �nancing option is the fact that the incomes of those engaged in agriculture are considerably lower than incomes in other occupations. For example, paid employees in agriculture, hunting, and forestry have a mean monthly income of just T Sh 39,731 (T Sh 44,896 for men, T Sh 25,009 for women), while self-employed individuals in the same sector have a slightly be er mean income of T Sh 56,186 (T Sh 67,457 for men, T Sh 43,896 for women), although they may also receive additional income in kind. Median incomes are even lower, at T Sh 14,000 8 World Bank Study for paid employees and T Sh 20,000 for self-employed workers (ILFS 2006, 82, 85). Such low levels of cash income for such a large proportion of the workforce severely constrain the potential options for health �nancing, especially in terms of revenue generation. Table 2.3 highlights relevant labor force composition and income �gures. It does not differentiate between the formal and informal sector, which is important because the formal sector has a higher likelihood of securing regular wage increases. For example, the imputed average wage of formal public sector workers was about T Sh 182,000 per month in 2006, based on NHIF premium revenue and enrollment data, which is consis- tent with the �gures in table 2.3. This �gure had increased to T Sh 312,000 by 2009. Also, the recently released Tanzania Pension Policy Note shows an imputed average monthly wage for formal private sector workers of more than T Sh 300,000 in 2009, implying sig- ni�cant wage growth in the formal sector as well.2 It is not known whether informal sector wages—especially for those working in the agriculture sector—have grown to the same extent, but with the volatility in many commodity prices this is unlikely and further disparities between formal/informal and urban/rural incomes can be expected when updated information becomes available. The table also shows quite a high labor force participation rate (85–90 percent), variable em- ployment ratios (highest for rural areas, lower for other urban areas, and lowest in Dar es Salaam—owing to higher unemployment and lower female participation), and infor- mal sector participation levels that are much higher in Dar es Salaam and other urban areas, and lower in rural areas. Table 2.3: Employed Population Characteristics Currently Employed Mean Population by Area (%) Monthly Income Dar es Other Self- Salaam Urban Rural Total Paid employed Agriculture, hunting, forestry 13.6 43.6 88.7 75.3 39,173 56,186 Fishing 0.8 1.0 0.3 1.2 52,539 94,861 Subtotal agriculture and fishing 14.4 44.6 89.0 76.5 39,383 56,793 Mining and quarrying 0.3 0.9 0.4 0.5 70,609 108,492 Manufacture 7.7 6.3 1.3 2.6 87,085 62,700 Electricity, gas and water 0.6 0.3 0.0 0.1 284,592 61,826 Construction 4.0 2.5 0.5 1.1 78,925 99,858 Subtotal industry 12.6 10.0 2.2 4.3 87,675 77,510 Wholesale and retail trade 29.0 19.6 2.8 7.6 73,806 98,682 Hotel and restaurants 6.3 5.2 0.8 2.0 74,130 65,603 Transport, storage, communications 7.6 2.9 0.5 1.5 131,830 116,979 Financial Intermediation 0.7 0.2 0.0 0.1 294,582 63,690 Real estate 2.5 0.9 0.1 0.5 117,654 173,385 Subtotal services 46.1 28.8 4.2 11.7 85,061 98,267 (Table continues on next page) Making Health Financing Work for Poor People in Tanzania 9 Table 2.3: Continued Currently Employed Mean Population by Area (%) Monthly Income Dar es Other Self- Salaam Urban Rural Total Paid employed Public administration 5.1 2.9 0.3 1.1 194,294 98,545 Education 3.0 2.7 0.9 1.4 181,208 124,189 Health 1.8 1.6 0.3 0.6 158,738 75,191 Subtotal public sector 9.9 7.2 1.5 3.1 181,502 105,606 Other community 2.4 1.9 0.2 0.7 82,798 158,831 Private household 14.6 7.3 1.9 3.8 23,505 50,456 TOTAL 100.0 100.0 100.0 100.0 98,454 75,693 Labor force participation rate 85.8 87.2 90.8 89.6 Current employment ratio 58.7 72.7 84.0 79.2 Households with informal sector activities 57.0 54.0 33.0 40.0 Source: Integrated Labor Force Survey 2006. Notes 1. This section is based largely on the Tanzania Country Assistance Strategy (World Bank 2011a). 2. See World Bank (2011b, 17) and author’s calculations. CHAPTER 3 Diagnosis and Evaluation of Current Health Financing System Overview of the Health System and Health Financing in Tanzania1 Before trade and economic liberalization in the 1980s, the Tanzanian health care system was solely in the hands of the central government and was �nanced through the national budget. There was some recognition of religious and other nonpro�t facilities, which were �nanced through donations from their sponsoring institutions and some service fees from clients. Those facilities were heavily concentrated in areas where government facilities were not well established, especially rural and remote areas. Mainland Tanzania is divided into 21 administrative regions and 113 districts, which are represented in 133 councils. There are about 10,342 villages. Services are organized in a pyramidal structure with dispensaries and health centers at the base, followed by district hospitals, regional hospitals, and national referral hospitals as one moves toward the apex. Both public and private providers operate dispensaries, health centers, and at least one hospital at the district level. There are 4,679 dispensaries and 481 health centers throughout the country. About 90 percent of the population lives within 5 kilometers of a primary health facility (Tanzania MoHSW 2008a). The 55 district hospitals are owned by the government, and 13 designated district hospitals are owned by faith-based organizations (FBOs). There are also 86 other hos- pitals at the �rst referral level (owned by the government, parastatals, and the private sector), as well as 18 regional hospitals, which function as referral hospitals for the district hospitals, plus 8 consultancy and specialized hospitals. Compared to the exist- ing government staffing norms, only 35 percent of positions are �lled with quali�ed health workers. Although the staffing norms are being revised to take into account the increased density of health facilities under the Primary Health Care Development Pro- gram (“MMAM� in Kiswahili), there is no doubt that signi�cant human resource short- ages exist. It is generally perceived that the quality of care in FBO facilities is relatively be er than that in public facilities, although there is li le empirical evidence to substantiate this perception. The government is revising its support to the Christian Social Services Commission (CSSC) and other FBOs through formal public-private partnership arrange- ments, including the introduction of service agreements that will specify contractual obligations in return for government funds and will put the relationship between CSSC facilities and the government on a more businesslike basis (Tanzania, MoHSW 2008 (a), 25, 34, 41). 10 Making Health Financing Work for Poor People in Tanzania 11 The Tanzania Mainland health system is administered by the Ministry of Health and Social Welfare (MoHSW)—which provides overall policy direction and quality control, and runs regional and national referral hospitals—and by the Prime Minister’s Office, Regional Administration and Local Government (PMO-RALG), which oversees the dis- trict system of district hospitals and primary health care centers and dispensaries. Those facilities are owned and operated by local government authorities. The organization structure for the publicly �nanced health sector is shown in �gure 3.1. Figure 3.1: Organization of the Health System in Tanzania Source: Tanzania, MoHSW 2008, 12 Notes: MOHSW = Ministry of Health and Social Welfare; PMO-RALG = Prime Ministers Office, Regional Administration and Local Government; RHMT = regional health management team. A series of HSSPs have been produced to guide the development of the health sys- tem in Tanzania. The latest, HSSP III, was launched by the president of Tanzania on June 30, 2009. It outlines 11 strategies and six cross-cu ing issues that encompass the three main levels of care in the health sector (district, regional, and national), and the �ve types of services (dispensary, health center, district hospital, regional hospital, and national hospital). The 11 strategies focus on the following areas: ■ District health services ■ Referral hospital services ■ Central support ■ Human resources for health ■ Health care �nancing ■ Public-private partnerships ■ Maternal, newborn, and child health ■ Disease prevention and control ■ Emergency preparedness and response ■ Social welfare and social protection ■ M&E and research 12 World Bank Study The health care system is still run under the implicit assumption that a major part of the �nancing of health care facilities runs through the national budget (�nanced partly by taxes and partly by money from development partners). In fact, as will be shown, a signi�cant portion of total health care �nancing comes from other sources. The propor- tion of total funding provided by development partners is also growing, in both on- budget and off-budget �nancing. The largest portion of the off-budget �nancing is for HIV/AIDS and malaria programming. Although the MoHSW is responsible for licensing health facilities, there is no formal mechanism for accreditation, except by the NHIF and the SHIB of the NSSF. The NHIF mechanism is explained the later section describing prepayment mechanisms. Because public �nancing for health care providers in 1993 was not sufficient to maintain existing health care structures, the �nancing system was overhauled and user fees were introduced as an additional �nancing source to the national budget. User fees were generally conceived as a means to increase the �nancial resources available for the health care system, but they were also intended to (a) reinforce the notion that there is no free health care, (b) reduce informal payments, and (c) help to avoid unnecessary visits to health care facilities. The user fees made access to health care services relatively more expensive for poorer people than for richer people. Such regressive effects were accepted as a precondition for improving the quality of health care services, which had been dete- riorating signi�cantly under previous budget �nancing mechanisms. Although exemptions and waivers were put in place to ensure access to essential health services, especially by the poor and vulnerable, several reviews of those policies indicate that they are not having the desired effect (Mtei and Mulligan 2007, 47–48). The reviews found that exemption systems favor the be er-off people more than the poor because most of those exempted belong to households that are able to pay the CHF membership fees. A household survey conducted by the SHIELD project estimated that 44 percent of those who were eligible for exemptions pay user fees for outpatient care; 70 percent pay for inpatient care. For those not eligible for exemptions, 89 percent pay for outpatient care and 86 percent for inpatient care (Borghi, Mtei, and Ally 2011, 39). Thus, exemptions are somewhat effective in reducing out-of-pocket payments. With respect to waivers, the poorest often do not have access because of a lack of information, denial of the waiver by a provider, loopholes that allow misuse and sometimes abuse of the system, and a lengthy and cumbersome identi�cation process that often deters people from applying for waivers (Mtei and Mulligan 2007, 6). Identi�cation of the poor is a major problem, because there are no clear guidelines and because no money is set aside to compensate facilities for waivers. In addition to exemptions and waivers, other systems were developed to address the highly regressive aspects of user fees, thus enabling people to access medical treatment when needed through prepayment and risk pooling. The mechanisms included voluntary public community health funds organized at the district level, voluntary micro-health schemes organized at the community level, private health management organizations, private health insurance, public mandatory social health insurance through the NHIF, and voluntary health bene�ts through the NSSF. Whereas the NSSF itself is mandatory, members must sign up separately for the SHIB health bene�t, but most do not. Voucher schemes have also been implemented in a limited way, although the scheme for anti-malaria bed nets has been found to be very effective. This scheme was superseded to a large extent by the “under-�ve� and “universal� bed-net campaigns Making Health Financing Work for Poor People in Tanzania 13 that were funded by the Global Fund to Fight AIDS, Tuberculosis, and Malaria; the U.S. President’s Malaria Initiative (PMI); Swi erland; and the World Bank. Vouchers were also examined by the NHIF to encourage facility-based delivery, although the current thinking is to provide insurance cards rather than vouchers. In summary, the health �nancing landscape in Tanzania is quite fragmented, with a large number of funding sources covering different aspects of the costs of health ser- vices. The HSSP III has as a major priority the development of a comprehensive health �nancing strategy to determine the appropriate mix of �nancing modalities and to de- velop a road map for improving �nancial and physical access to health services by all Tanzanians. The next section reviews the trends in health �nancing, both in total and in different areas. Trends in Health Financing Overall Health Expenditure Health expenditures involve central government budgets through the MoHSW, local government budgets through PMO-RALG, off-budget payments from development partners, cost-sharing revenue, revenue from public and private insurance, and house- hold out-of-pocket expenditures. Although a new National Health Accounts report is under development, the latest comprehensive �gures are from 2005/06. Table 3.1 shows that there were signi�cant changes in both the size and the composi- tion of health expenditures between 2002/03 and 2005/06, with total health expenditure more than doubling over the three years in question. The greatest increase (in both ab- solute and percentage terms) was seen in donor funds, and signi�cant increases were also seen in public funding. Although the proportion of private expenditure on health decreased considerably, from 47.1 percent to 27.8 percent, actual private spending still increased by T Sh 88 billion, or 38 percent, with most of this increase coming from house- hold out-of-pocket payments. Table 3.1: Total Health Spending by Source of Funds T Sh million Distribution (%) Increase Increase 2002/03 2005/06 2002/03 2005/06 T Sh million (%) National expenditure on health 496,030 1,159,217 100.0 100.0 663,187 133.7 Distribution by financing source (%): Public 126,140 325,740 25.4 28.1 199,600 158.2 Donors 136,061 511,215 27.4 44.1 375,154 275.7 Private 233,829 322,262 47.1 27.8 88,434 37.8 Household out-of-pocket 201,388 267,779 40.6 23.1 66,391 33.0 Household other 6,944 17,388 1.4 1.5 10,444 150.4 Private nonhousehold 25,496 37,095 5.1 3.2 11,599 45.5 Source: Tanzania National Health Accounts 2008. Both donor and public �nancing have continued to increase since 2005/06, so the results of the current National Health Accounts round will be quite important, especially in determining whether private funding has increased again as a share of total health expenditure or has continued its downward trend. Those �gures will have a signi�cant effect on the assessment of potential future policy options for health �nancing. 14 World Bank Study Public Financing This section examines public �nancing, which includes government spending and on- budget spending by development partners. Table 3.2 shows funding from public sources by year, including per capita spending and the proportion of government spending. As can be seen, total public health sector spending more than tripled between 2004/05 and the 2010/11 estimates. The largest increase has been in foreign non-basket funding, al- though a substantial part of this increase is the result of Global Fund grants (included in “non-basket� �nancing) moving from off-budget to on-budget. The table also shows that there are considerable problems with budget execution for non-basket funds. A considerable amount of off-budget foreign funding (most notably funding from the U.S. government) is not shown in the table. With the emergence of the U.S. Presi- dent’s Emergency Program for AIDS Relief (PEPFAR) and the PMI, signi�cant off-bud- get resources have started flowing into the health sector. For example, PEPFAR funding in 2006/07 amounted to US$205.5 million and increased to US$356.2 million by 2008/09. The la er �gure represented 72 percent of total HIV/AIDS funding in �scal 2009 and was greater than the actual expenditures by the MoHSW for that year.2 The PMI accounted for another US$35 million in �scal 2009 and increased its contribution to US$52 million in �scal 2010. Table 3.2: Public Health Spending by Financing Sources 2004/05 2005/06 2006/07 2007/08 2008/09 2009/10 2010/11 Source of Actual Actual Actual Actual Actual Actual funds expendi- expendi- expendi- expendi- Approved expendi- Approved expendi- Approved (T Sh million) ture ture ture ture estimates ture estimates ture estimates Government 206,554 296,819 348,890 378,113 459,496 461,504 548,658 578,682 643,011 funds Foreign funds Basket 91,777 68,299 103,204 80,956 99,730 85,401 121,640 128,796 161,804 Non-basket 2,896 61,257 61,512 112,003 204,368 154,168 254,801 200,049 387,908 Total Foreign 94,673 129,555 164,716 192,959 304,098 239,569 376,441 328,845 549,712 Off-budget 3,384 3,363 2,964 15,289 5,858 10,784 Total 304,612 429,738 516,570 586,361 763,594 706,931 925,099 918,311 1,192,723 Real spending 304,612 403,693 460,825 478,347 566,980 524,907 640,851 636,148 791,153 (FY05=100) Real per capita 8,328 10,707 11,856 11,939 13,728 12,709 15,052 14,941 18,074 (T Sh) Real per capita 7.51 8.98 9.49 9.46 10.40 9.63 11.34 11.26 12.31 (US$) Total as % GoT 11.3 14.1 13.3 12.2 11.1 12.1 11.6 12.9 12.1 (excl. CFS) Total as % GoT 10.1 11.9 11.8 11.0 10.0 10.8 9.7 9.8 10.3 (incl. CFS) Percentage of total Government funds 67.8 69.1 67.5 64.5 60.2 65.3 59.3 63.0 53.9 Basket 30.1 15.9 20.0 13.8 13.1 12.1 13.1 14.0 13.6 Non-basket 1.0 14.3 11.9 19.1 26.8 21.8 27.5 21.8 32.5 Total foreign 31.1 30.1 31.9 32.9 39.8 33.9 40.7 35.8 46.1 Sources: Tanzania Ministry of Health and Social Welfare, Public Expenditure Review, 2010; WDI (for GDP deflator). Making Health Financing Work for Poor People in Tanzania 15 Government �nancing is decreasing as a proportion of total funding—from 69.1 percent in 2005/06 (actual) to 53.9 percent for 2010/11 (estimated). However, the large ab- solute increases in both government and external �nancing over this period suggest that there may be not be a large amount of additional room for further signi�cant increases in health �nancing from these sources. By contrast, at the recent United Nations Millen- nium Development Goals summit in New York, the president of Tanzania reaffirmed his government’s commitment to achieving the Abuja target of 15 percent of total govern- ment spending going to health. Figure 3.2 highlights the recent experience in relation to this target. It shows that actual expenditures in 2009/10 reversed the general downward trend since 2005/06 and that the 2010/11 estimates seem to continue this reversal. The previous trend appears to have resulted from a combination of robust growth in general government revenue and several other high priorities for government, such as agriculture, education, and infrastructure. The 2010/11 estimates include major increases in Global Fund grants (cur- rently on-budget), which more than doubled from T Sh 152.7 billion (or 31.9 percent of the total MoHSW budget) in 2009/10 to T Sh 333.7 billion in 2010/11 (or 49.3 percent of the budget). Excluding the Global Fund, the overall development budget increased by just 6.3 percent, from T Sh 107.9 billion to T Sh 114.7 billion, while the total budget in- creased by just 5.0 percent, from T Sh 326.3 billion to T Sh 342.5 billion. In addition, the recurrent budget increased by just 4.3 percent, which is barely above population growth. Figure 3.2: Public Health Spending as a Percentage of Total Government Expenditure Sources: Tanzania MoHSW 2008, 2009, and 2010. Notes: CFS = Consolidated Fund Service; Est. = estimated. Those �gures emphasize that the amount of government revenue available for the health sector has been limited. It is not likely that this limitation will change in the fore- seeable future, especially if external �nancing continues to be available and if the other demands on government revenue continue to increase. The combination of shrinking 16 World Bank Study pledges to global health initiatives and the rapid use of the funds from existing Global Fund grants could result in an absolute decrease in the health sector budget in �scal 2012, thereby raising questions about the extent to which the government will step in to keep the overall level of health �nancing stable. However, the scope for increased �nanc- ing is limited by the other pressures mentioned earlier, as well as the current �scal situ- ation, in which an overall budget de�cit of 6.1 percent of GDP (12.1 percent before exter- nal �nancing) is expected in 2009/10, up from 4.5 percent (9.4 percent) the previous year. This �nding suggests that the �scal space in terms of budget resources is very limited. Global Fund grants, as well as most external �nancing other than the Health Basket Funds, are earmarked for speci�c purposes, which results in extremely limited discre- tionary funding for the health sector. This limitation means that signi�cant increases in those funds are not translated into general increases in health service delivery. To have such an effect, either the Health Basket Funds or the allocation from the government’s resources would need to increase. So the question may be asked, “What has been achieved by the signi�cant increase in the overall level of health sector �nancing?� The Annual Health Sector Performance Pro�le Reports give some indication of the answers to this question. Table 3.3 shows some of the key performance indicators that are tracked as part of the HSSP process. Table 3.3: Changes in Key Health Sector Indicators Change, Change, Indicator 1999 2004/05 2009/10 2004–09 (%) 1999–2009 (%) Infant mortality rate 99 68 51 −25.0 −48.5 Under-five mortality rate 147 112 81 −27.7 −44.9 Maternal mortality ratio 529 578 454 −21.5 −14.2 Births in health facilities (%) 36 46 51 +10.9 41.7 TB treatment completion rates (%) 81.0 82.6 88.0 +6.5 +8.6 Outpatient visits per capita 0.71 0.78 0.74 −5.1 +4.2 Clinical staff per 10,000 populationa 4.30 4.60 4.93 +7.1 +14.7 Source: Annual Health Sector Performance Pro�le Report, various years. Note: TB = tuberculosis. a Includes physicians, assistant medical officers, nurses, and nurse-midwives. These data show that although there have been impressive improvements in key health outcomes, the progress on process indicators has been relatively modest. In the case of outpatient a endance, visits per capita have actually declined over the past �ve years. This result is consistent with the recent performance audit by the Comptroller and Auditor General (CAG, the government’s independent auditor), which concluded that—at the local level—health facilities “are not efficiently managed and are funded without proper consideration of service demands and performance� (NAO 2008, 2). Spe- ci�cally, it found that 80 percent of clinical staff members of the selected sample of health centers and dispensaries see fewer than 10 patients per day. Further, it found that health centers often tended to have the same allocation of drugs, even though there might be a 20-fold difference in the number of patients seen. This disparity clearly indicates that there is substantial room for efficiency improvements to make the maximum use of the available funds, regardless of the level of funding. There are also signi�cant regional variations in the allocation of funds, with the highest-resource regions ge ing 2.6 times Making Health Financing Work for Poor People in Tanzania 17 the amounts per capita of those with the least resources. All of these data suggest that improving efficiency and equitable allocation of resources should be a major thrust of any health �nancing strategy. NHIF Expenditure Another important source of funding for the health sector is the NHIF. Fortunately, cur- rent data are available for the NHIF expenditure, which is a relatively small proportion of total health expenditure although the contributions collected represent more than 10 percent of total public resources. As shown in table 3.4, while there has been signi�cant growth in contributions and expenditure, the proportion of total income paid out as bene�ts had been persistently less than 15 percent until 2007/08, after which it increased to 15.6 percent in 2008/09 and then jumped to almost 20 percent in 2009/10. However, a collective 20 percent return on investment for NHIF members may still represent a relatively poor value proposition, which will need to be addressed if the government is to realize its growth goal of enroll- ing 45 percent of the population in prepayment schemes. Table 3.4: NHIF Income and Reimbursement, 2001/02 to 2009/10 2001/02 2002/03 2003/04 2004/05 2005/06 2006/07 2007/08 2008/09 2009/10 Income (T Sh million) Premium revenue 11,388 13,866 20,310 24,700 29,847 45,516 55,472 73,283 85,826 Investment income 36 577 1,568 3,501 6,800 10,333 16,343 16,317 16,997 Total income 12,240 14,535 21,938 28,610 37,256 56,884 72,178 89,943 102,884 Premium income as 93.0 95.4 92.6 86.3 80.1 80.0 76.9 81.5 83.4 % total Investment income as 0.3 4.0 7.1 12.2 18.3 18.2 22.6 18.1 16.5 % total Paid amounts for claims 247 1,346 3,808 4,072 4,918 8,269 10,188 14,052 20,057 (T Sh million) % of premium income 2.2 9.7 18.8 16.5 16.5 18.2 18.4 19.2 23.4 % of total income 2.0 9.3 17.4 14.2 13.2 14.5 14.1 15.6 19.5 Sources: NHIF Annual and Statistical Reports. Table 3.4 shows a large absolute increase in revenue from premiums, increasing from T Sh 11.4 billion in 2001/02 to T Sh 85.8 billion in 2009/10. This revenue growth is a function of the increase in the number of bene�ciaries—from 692,000 people in 2001/02 to 1.97 mil- lion in 2009/10—as well as more than a three-fold increase in the premium per subscriber. The overall revenue composition has also changed considerably over time, with the revenue from premiums dropping from 95 percent in 2002/03 to just 77 percent in 2007/08, and then increasing again to 83 percent in 2009/10. There has also been sig- ni�cant growth in investment income, from just 4 percent of total income in 2002/03 to 23 percent in 2007/08, although it declined to 17 percent in 2009/10. In 2008/09, this investment income exceeded the amount spent on bene�ts. If this income stream can be maintained, it could be used to help improve the level of bene�ts for NHIF members. NSSF and CHF The SHIB program, which has been operating since January 2006, is one of the seven bene�ts of the NSSF. It is �nanced through the combined social security contribution of 18 World Bank Study 20 percent of wages, shared between the employer and the employee. In the year ending June 30, 2010, the SHIB paid out T Sh 1.2 billion, which is a 30 percent increase over the T Sh 923 million paid out the previous year. However, it is important to note that only 9.2 percent of the NSSF members are registered with the SHIB, and major initiatives to increase this percentage are under way. If all 348,000 NSSF members joined the SHIB and used bene�ts in the same proportion as those currently enrolled, the total SHIB expenditure could be as much as T Sh 13 billion, representing 6.4 percent of total NSSF contributions. Of course, it is unlikely that all members would be registered, because many have private health insurance that is provided through their employers. As of April 30, 2010, 99 councils from 20 regions had established a CHF. Another 9 new councils from 8 additional regions had been sensitized for CHF establishment; 38 councils from 16 regions had established a CHF but were not currently active. A total of 18 municipal and town councils from 18 regions had been sensitized to start operating TIKA. Data on the total amount of funding made available through the CHF are not readily available, but a rough calculation can be made. According to the 2009 Public Expenditure Review, CHF revenue accounted for an average of 18.7 percent of the complementary �nancing for the 12 councils sampled. If those councils can be taken as a representative sample, this �nding implies that CHFs collected a total of T Sh 3 billion in 2008/09. It is not clear, however, whether this revenue would include the T Sh 1 billion that councils received as matching contributions. It is important to note that the matching funds re- sulted from applications by just 34 councils (23 were approved), which suggests that signi�cant matching grant revenue is being left on the table. Another way to calculate the amount of funding is through the number of people enrolled in CHFs (currently 1.68 million), although this calculation is complicated some- what by the fact that the annual payment ranges from T Sh 5,000 to T Sh 30,000 per household, depending on the district. However, if an average household size of 5 people is used and an average payment of T Sh 10,000 is assumed, the total revenue would be T Sh 3.3 billion, which is close to the Public Expenditure Review estimate. Table 3.5 sum- marizes the coverage of publicly administered prepayment mechanisms. It shows that despite the progress that has been made, population coverage is still less than 9 percent, although the revenue collected is less than 10 percent of total publicly �nanced health spending and the actual bene�ts paid are about 2.5 percent of that amount. Table 3.5: Summary of Prepayment Plans, 2010 Revenue Revenue per Benefits Benefits per Number of Share of collected beneficiary paid (T Sh, beneficiary Plan beneficiaries pop. (%) (T Sh, million) (T Sh) million) (T Sh) NHIF 1,971,251 4.6 85,826 43,539 20,057 10,175 CHFa 1,678,734 3.9 3,009 1,792 2,106 1,792 SHIBb 51,306 0.1 1,200 23,389 1,200 23,389 Total 3,701,291 8.6 90,035 24,325 24,266 6,556 Source: Author’s calculations. a Revenue collected is kept and used at the district; there is no direct payout of bene�ts; 30 percent is as- sumed to be used for administration (per CHF guidelines). b SHIB is funded out of the general NSSF contribution, so no revenue is speci�cally a ributed to the program. Making Health Financing Work for Poor People in Tanzania 19 Summary of the Key Issues and Problems in Health Financing Although the total resources available for health have increased considerably over the past number of years, many issues remain. For on-budget �nancing (government and donor), the level of discretionary funding is low, with most of the funds earmarked ei- ther for salaries or for speci�c donor programs such as activities �nanced by the Global Fund. Because the overall budget envelope for health is �xed, increases in earmarked donor funds tend to reduce the discretionary budget even further. In addition, there is a signi�cant amount of off-budget funding; the ability of the government to influence spending priorities for this funding is even more limited. The public consistently expresses concerns about access to quality services, citing absence of staff members, medical supplies, drugs, and equipment, as well as the high cost of out-of-pocket payments. The poor quality of services provided by accredited health facilities is also a key complaint of those enrolled in various health insurance schemes (NHIF, SHIB, and CHF). Members of those schemes feel that the bene�ts they receive are not commensurate with their contributions. The data generally support this contention, although the situation has improved in recent years for the NHIF. By contrast, health care providers feel that they are not adequately reimbursed for the level of expected care, despite the increased money from user fees and insurance schemes. This perception is exacerbated by the lack of understanding within medical fa- cilities about payment methods; that lack of understanding leads to facilities not submit- ting insurance claims and not requesting the resulting funds from the local authorities who are paid by the insurers. Similarly, the resources from CHF contributions (which are matched by MoHSW/Health Basket Funds) remain unused or are used unevenly by different local authorities. Speci�cally with respect to the CHF, a study by the Swiss Institute for Tropical Med- icine and Public Health (described later) found a lack of payment to providers for ser- vices rendered (Swiss TPH 2010). Another key issue is the lack of feedback that members of the NHIF can provide: often the only recourse left to them is voting with their feet by switching providers or discontinuing service. The NHIF has a very high level of unused reserves. According to its latest annual report, the NHIF has more than T Sh 300 billion in investments and reserves, represent- ing about 15 years worth of claims at current expenditure levels. Because this funding is tied up and cannot be used to improve services or bene�ts to subscribers, it represents a signi�cant opportunity cost to the health system as a whole. The issue of investing reserves prudently also needs to be addressed and regulated. The issue of regulation needs signi�cant a ention. Each prepayment scheme oper- ates under its own enabling legislation and regulatory authority. An overall regulatory framework is needed to de�ne the rules for all players, thereby enhancing transparency regarding the roles, rights, and responsibilities of each stakeholder and, in turn, ensur- ing that stakeholders are held accountable for their actions. Such a regulatory frame- work should extend to private sector insurers as well. To this end, a Social Security Regulatory Authority (SSRA) has been established, but it is not clear which aspects of the health insurance market, other than the NSSF, it will be able to regulate. To date, the SSRA Act and regulations are silent on critical areas such as tariff de�nitions, deter- mination of the bene�t package, �nancial solvency, governance and organization, and contribution rates. 20 World Bank Study Medical equipment such as microscopes and drugs is often in short supply or lack- ing altogether at public health facilities. This is a common complaint among both NHIF and CHF members; however, this problem adversely affects CHF members more be- cause their health insurance in most cases offers treatment only at public primary facili- ties within their home district. In rural areas, it is often necessary to travel long distances to reach a referral hospital. For the majority of the rural population, especially those working in the informal sector (primarily in subsistence agriculture), it is difficult to cover the transport cost. As a result, rural dwellers do not access referral facilities even if they are members of �nancing schemes that offer this bene�t.3 According to the Household Budget Survey of 2007, average household spending on health services was just 1.9 percent of household expenditure for Tanzania as a whole, with relatively li le variation between the poor (2.1 percent) and nonpoor (1.8 percent). However, because of the nature of health expenditures, averages are a poor way to mea- sure their effect. The Household Budget Survey data provides two other ways to look at these data. One way is to calculate the incidence of catastrophic health expenditures. There is no generally accepted de�nition of the level of spending that is considered catastrophic, so this analysis looks at various levels: 5 percent, 10 percent, 20 percent, and 30 percent of to- tal household spending. The results are shown in �gure 3.3. The graph in panel A shows that overall, health-related spending accounts for more than 5 percent of total spending for 9 percent of households, with a higher proportion of households in Dar es Salaam spending 5 percent or more of their income on health than those in rural areas. The pro- portion of the total population that spends more than 10 percent of income on health is about 2.6 percent, with li le variation between areas; the proportions spending more than 20 percent and 30 percent of their income on health are 0.4 percent and 0.2 percent. Figure 3.3: Household Spending on Health, as Percentage of Expenditure A. Based on total spending B. Based on nonfood spending Source: Special tabulations from the Household Budget Survey (2007) data set. Making Health Financing Work for Poor People in Tanzania 21 However, when one looks at the bo om graph, it is clear that the burden on the poor and extreme poor is proportionately greater than on the nonpoor. Health expenditures account for at least 5 percent of total household spending for more than 17 percent of the extremely poor (those beneath the food poverty line), with 8 percent spending 10 percent or more, 2.6 percent spending 20 percent or more, and 0.5 percent spending 30 percent or more. Clearly, health spending is a signi�cant burden for the one-sixth of the population who are extremely poor. The negative impact is even more dramatic when nonfood spending is used as the basis for the analysis (panel B). Health-related items make up 5 percent or more of nonfood spending for 37 percent of all households and 46 percent of extremely poor households. About 18 percent of households allocate 10 percent or more of their nonfood spending to health, increasing to almost 28 percent for the extremely poor. More than 15 percent of the extremely poor spend 20 percent or more, and 8.4 percent spend more than 30 percent on health. For the population as a whole, 6.2 percent allocate 20 percent or more and 2.8 percent allocate 30 percent or more of their nonfood spending to health. Another way of looking at the poverty impact of health expenditures is to examine the poverty head count with and without health spending. The results of this analysis are shown in table 3.6. Table 3.6: Poverty Impact of Health Expenditures Poverty Poverty without Poverty head Percent due head health count due to to health count expenditure health expenditure expenditure Dar es Salaam 16.40 15.15 1.25 7.6 Other urban 24.10 23.06 1.04 4.3 Rural 37.60 36.10 1.50 4.0 Total 33.60 32.19 1.41 4.2 Source: Household Budget Survey (2007), special tabulation by World Bank. This table shows that health expenditures are a major contributor to poverty in Tanzania, accounting for 4.2 percent of the poverty head count. Even more important and perhaps surprising, health expenditures account for 7.6 percent of the total poverty headcount in Dar es Salaam and 4.3 percent in other urban areas. Those �gures suggest that improving access to prepaid health insurance and reducing out-of-pocket spending could have a major effect on poverty, especially in Dar es Salaam and other urban areas. As part of the SHIELD project—which is �nanced by the International Development and Research Centre (IDRC) and the European Commission and is looking into health insurance options and experience in Tanzania, Ghana, and South Africa—the Ifakara Health Institute has been systematically looking at various aspects of the health �nanc- ing situation in Tanzania. Two particular aspects have been examined: (a) the �nancial effect of health expenditures on different income groups and (b) the bene�ts of health expenditures on these income groups. In terms of �nancial effect, the project evaluated the level of progressivity of various types of health �nancing mechanisms, including taxes, direct out-of-pocket payments, the NHIF, and the CHFs. It used the 2000/01 Household Budget Survey as well as a SHIELD-�nanced survey in 7 districts in Tanzania, which covered 2,234 households and 12,201 individuals in 4 rural districts and 3 urban councils. 22 World Bank Study Figure 3.4 summarizes the results of the progressivity analysis using the Kakwani index, which is de�ned as the difference between the concentration index of health care payments and the Gini index of income distribution. The index is negative if the �nanc- ing source is regressive, positive if it is progressive, and equal to zero if it is proportional. The Gini index is a measure of income/wealth inequality. A low Gini coefficient indicates a more equal distribution while a zero value indicates a proportional distribution (Mtei and Borghi 2010, 5). Figure 3.4 also shows that the health �nancing system overall is marginally progressive, with taxes and the NHIF being progressive and with a Kakwani index of 0.419. The CHF and out-of-pocket payments are both regressive, with indices of -0.078 and -0.493. Figure 3.4: Kakwani Indices of Health Care Financing Sources Source: Mtei and Borghi 2010, 30. Note: OOP = out-of-pocket. Figure 3.5 shows the effect by income group. The richest 20 percent of the popula- tion pay the highest proportion of their income to �nance health care, while the poorest 20 percent bear the second highest proportion. The middle-income quintile bears pro- portionally the smallest proportion of the health �nancing burden. Making Health Financing Work for Poor People in Tanzania 23 Figure 3.5: Household Health Care Financing as a Proportion of Income payments as a % of income income quintile Source: Mtei and Borghi 2010, 28. Note: OOP = out-of-pocket. If one is looking at income and consumption taxes as a whole, �gure 3.6 shows the extent to which they are progressive. The lowest burden is borne by the poorest quintile, with just under 9 percent of income paid as taxes; larger shares are paid by each succes- sive income quintile, reaching almost 22 percent of income paid as taxes for the richest quintile. The major reason for the jump in the richest quintile is that this group is the only one that pays any signi�cant level of personal income tax. Figure 3.6: Total Income and Consumption Taxes as a Proportion of Income payments as a % of income income quintile Source: Mtei and Borghi 2010, 27. Notes: PIT = personal income tax; CIT = corporate income tax. 24 World Bank Study After it examined the sources of �nancing for the health system and how much the various income groups paid relative to each source, the SHIELD project also looked at the incidence of bene�ts for the system as a whole and its component parts for different wealth quintiles. Again the special survey was used, supplemented by cost information from the NHIF, the National Health Accounts, and the MoHSW Health Management Information System. In the core analysis, NHIF costs were used because using other cost estimates produced li le difference in the results. Figure 3.7 shows the relative share of health services for various types of care across income groups. Although public primary health care facilities are pro-poor, the pro-rich orienta- tion of outpatient care (panel B) at district and referral hospitals leads to the lowest and highest income groups ge ing a proportionately smaller share of the bene�ts, while the middle-income group gets the largest share. Similarly, publicly provided hospital care (panel D) is skewed toward speci�c income groups, such as the second poorest quintile for health centers and district hospitals and the middle quintile for regional and referral hospitals. Private inpatient care is predictably skewed toward the rich, while the inpa- tient distribution for FBOs is much more evenly distributed. The study concludes (as shown in panel A) that “When combining bene�ts from out- patient and inpatient care from public providers, the overall distribution is more or less even, although the poorest 20 percent are ge ing less than 10 percent of all bene�ts…. Overall bene�ts from private for-pro�t providers, including drug shops, are strongly pro-rich, with the richest 20 percent ge ing more than 45 percent of the bene�ts. Over- all bene�ts from faith-based providers are evenly distributed …� (Makawai and others 2010, 24). A �nal area examined by the SHIELD study produced a comparison of the bene�ts received versus need, with need being de�ned on the basis of responses to the survey with respect to self-assessed health status. Figure 3.8 shows that the poorest quintile has the greatest share of need but receives the lowest share of total health care bene�ts. All other quintiles receive a larger share of bene�ts than their share of need, and all—except the poorest 20 percent—get more than 20 percent of the total bene�ts. Both the Financial Incidence Analysis and the Bene�t Incidence Analysis need to be borne in mind in developing any health �nancing options. Further, a recent study by the Ifakara Health Institute looked at the use of health services by people who did and did not have prepaid health insurance. Figure 3.9 shows that although outpatient health care use (that is, using any outpatient services in the previous month) among the insured relative to the uninsured increases across all wealth quintiles, the percentage growth in the poorest groups is greater, with 87 percent and 83 percent growth in the �rst and second quintiles, compared with 44 percent, 51 percent, and 56 percent respectively for the upper-income quintiles. Further, the ratio of health care use between the highest and lowest quintiles shows a signi�cant decline: from 1.57:1 for the uninsured to 1.32:1 for the insured group. This �nding clearly shows that health insurance can have a pro-poor effect. As a result of this and similar studies, as well as the commonly held view that pre- payment plans can potentially play an increasingly important role in improving social health protection, the next section delves deeper into the key prepayment schemes that are currently in place in Tanzania. Making Health Financing Work for Poor People in Tanzania 25 Figure 3.7: Bene�t Incidence for Various Health Interventions A. Total bene�ts B. Public outpatient C. Private outpatient care D. Inpatient care Source: Makawai and others 2010, 21–25. 26 World Bank Study Figure 3.8: Share of Bene�ts versus Need by Wealth Quintile % of total income quintile Source: Makawai and others 2010, 26. Figure 3.9: Health Care Use by Insured and Uninsured Persons % using outpatient services income quintile Source: Tanzania MoHSW 2010, presentation by Ifakara Health Institute. Making Health Financing Work for Poor People in Tanzania 27 Overview of the Functioning of Current Prepayment Mechanisms National Health Insurance Fund (NHIF)4 The NHIF began operations in July 2001 and, aside from out-of-pocket payments, is currently the main source of health service �nancing after the government funds. As shown in table 1, the NHIF covers the health insurance costs of public servants. It covers spouses and up to four children or legal dependents or a combination thereof per mem- ber. Membership is compulsory. NHIF premiums are 6 percent of members’ income and are deducted from the em- ployees’ salary and remi ed to the NHIF. This premium is split 50/50 by the member and employer (the government in the case of public servants). About 2 million people are currently covered, which is substantial but still less than 5 percent of the 38 million Tanzanians. There is only one risk pool to cover all NHIF members. To increase member- ship, the NHIF has extended coverage to pensioners, police, prison staff, immigration officers, and �re and rescue service staff members, and it has allowed subscribers to pay extra for insuring family members beyond the numbers included in the basic package. NHIF is also exploring mechanisms to enroll the private formal sector and the “orga- nized� informal sector. Providers are paid on a fee-for-service basis. Members must go to an accredited public or private health facility or provider to receive health services. The NHIF requires accreditation for health care providers who want contracts. All public providers are accredited regardless of quality; thus selective accreditation applies only to FBO and private facilities. Private pharmacies are also accredited. The accreditation process uses predetermined criteria and is based on MoHSW standard guidelines, which include the following: (a) availability of human resources, equipment, and facilities in accordance with MoH guidelines; (b) acceptance of a formal program of quality assurance pre- scribed by the NHIF; (c) acceptance of NHIF standard payment mechanisms and fees; (d) adherence with NHIF referral guidelines; (e) acceptance of reporting requirements; and (f) recognition of the rights of the patient. As of June 30, 2010, a total of 5,576 health facilities—68 percent of all health facilities in Tanzania—were accredited by the NHIF to provide services. Of the total, public facilities accounted for 91 percent of the dispen- saries, 77 percent of the health centers, and 53 percent of the hospitals. The scheme is managed by a board of directors appointed by the government. The NHIF offers a wide range of bene�ts, including basic diagnostic tests, drugs, outpatient services, inpatient services, and minor and major surgery. The NHIF cur- rently reimburses 90–95 percent of the claims presented to it by providers, which is a signi�cant increase over the 50 percent reimbursement rate paid when the NHIF was launched. At the same time, the expenditure of the NHIF is quite low. Despite continu- ing efforts, claim reimbursements still account for less than 23 percent of total NHIF in- come. Those claims account for about 2 percent of the total national health expenditure. 28 World Bank Study Figure 3.10: Member and Bene�ciary Growth Sources: NHIF Annual Reports and Statistical and Actuarial Bulletins, various years. Figure 3.10 shows the signi�cant growth in NHIF membership over the past 10 years: increasing from 691,714 (164,708 members with 527,006 dependents) in 2001/02 to 1,971,251 (373,326 members with 1,597,925 dependents) in 2009/10. Although this in- crease is still a small percentage of the total population (5.2 percent, based on a popula- tion of 38 million), it is encouraging to see such growth. Figure 3.11: Expenses as Percentage of Revenue Sources: NHIF Annual Reports and Statistical and Actuarial Bulletins, various years. Making Health Financing Work for Poor People in Tanzania 29 Figure 3.11 shows that expenses as a percentage of total revenue are increasing. Starting in 2005/06, there has been steady growth, to 27.1 percent of revenue in 2008/09 and a sudden jump to 35.7 percent of revenue in 2009/10. This increase is a result of higher reimbursement, increased administrative cost, and a signi�cant increase in other expenses (see �gure 3.12). Figure 3.12: Costs as Percentage of Premium Revenue Sources: NHIF Annual Reports and Statistical and Actuarial Bulletins, various years. In �gure 3.12, the amounts paid out for claims compared with premium revenue show a slight increase from 16 percent to 19 percent between 2004/05 and 2008/09, fol- lowed by a jump to 23.4 percent of premium revenue in 2009/10. The 2009/10 result is a change from the previous trend of relatively li le change in the bene�ts paid out year over year when compared with premium revenue. Although this trend is a very positive, it must be noted that the be er-performing health insurance schemes in Eastern Europe (for example, Estonia and Romania) spend about 95 percent to 98 percent of their pre- mium revenue on bene�ts to their members (Estonian Health Insurance Fund 2010, 9; CNAS 2010, 1). Another trend to watch closely is the increase in the administration cost and other expenditures as a percentage of premium revenue; such �gures increased from 8.4 per- cent and 1.2 percent in 2004/05 to 12.0 percent and 7.5 percent in 2009/10. Although some of this increase is the result of NHIF funding of medical equipment and facility rehabili- tation—which arguably should not be funded by a health insurer—the costs must be closely tracked to ensure that they remain within acceptable levels for a national health insurance program. A study by the World Health Organization (WHO) indicated that the average spent on administration for mature health insurance systems is 4.2 percent, with a range of 2 percent to 6.6 percent. Estonia spends 0.96 percent and Romania spends 1.7 percent of premium revenue on administrative expenses (WHO 2004, 40). Even when one allows for the challenges of communications and overall capacity within Tanzania, it should be 30 World Bank Study possible to keep administrative and other expenses to signi�cantly less than 10 percent of premium revenue. The “other expenses� for 2009/10 include T Sh 1.84 billion provided for medical equipment to accredited facilities, as well as T Sh 906 million provided for facility re- habilitation. Although these are necessary expenses, they do not fall within the types of expenditures that should normally be incurred by a health insurer. The fact that the NHIF is being called on to fund such items highlights the signi�cant funding pressure on the system from the central MoHSW, which would usually be expected to provide the funds. Recently, the NHIF has been tasked with providing administrative, coordination, and technical assistance support to CHFs. This support has included the development of a CHF action plan and inclusion of CHF issues in the second NHIF Strategic Plan (2010–15). National Social Security Fund 5 The SHIB of the NSSF was established in July 2006 for the bene�t of private sector work- ers. SHIB members bene�t from health services through the �nancing of their 20 percent contributions to the NSSF, which are collected through payroll deduction (Mtei and oth- ers 2007, 35). Health insurance is one of seven bene�t categories under the NSSF, although the main bene�t is pensions. All bene�ts are funded out of a single risk pool. Health insur- ance operations commenced 10 years after the NSSF began operating, and the SHIB is a standard component of the overall NSSF bene�t package. However, enrollment into the SHIB requires a separate procedure that results in SHIB bene�ciaries representing just 9.2 percent of total NSSF membership. This proportion translates into roughly 31,000 individuals. There are also about 20,000 dependents, which amounts to a much lower rate of dependents than as part of the NHIF (0.66 dependents per subscriber for SHIB compared to 4.28 dependents per subscriber for the NHIF). Various factors can contribute to the low enrollment rate into the SHIB. They include private sector employers providing their own health bene�ts separate from the NSSF to employees, as well as lack of public knowledge about the health scheme. NSSF repre- sentatives believe the Fund’s single contribution rate, which covers all categories, may lead members to fear that other bene�ts such as pensions could be decreased if they take advantage of health insurance. Finally, in some areas, there is a lack of accredited health facilities, which leads members not to enroll because they would not be able to access health services. Overall, 264 hospitals, health centers, and dispensaries are accredited to the SHIB, which is a small proportion of the total health facilities in the country. NSSF pensioners are also entitled to membership in the SHIB, but the contribution rate is 6 percent of the gross pension. To boost membership, the NSSF is reviewing its health care strategy and examining approaches for increasing the uptake of the SHIB. In contrast to the NHIF, the SHIB makes capitation payments to accredited health facilities and health providers for health services rendered to members. Using this approach, the member must select a single facility, and the facility is then paid a flat amount per year to provide health services to that member. The capitation amount is not based on any speci�c information because the SHIB does not collect information on the costs incurred by members or on utilization rates. Some facilities, especially in Dar es Salaam, refuse to take capitation, and the NSSF makes special fee-for-service arrangements with those facilities. Making Health Financing Work for Poor People in Tanzania 31 Health bene�ts provided by NSSF include a broad range of services, among them the following: ■ Outpatient services: consultations, basic and specialized diagnostics, simple and specialized procedures, and drugs on the National Drug List ■ Inpatient services: admission (overnight stay), consultations, simple and spe- cialized procedures, and referrals to a higher level and specialized hospitals CHF/TIKA (Community Health Fund and TIKA [Tiba kwa Kadi])6 The CHF was established as an alternative to user fees at the point of service. The idea is that district residents (usually informal workers and farmers) can join a CHF on a voluntary basis and can get access to health care without paying user fees. Currently, 1.7 million people are registered with CHFs, representing 3.9 percent of the population. The contribution is paid by the member; those who cannot afford the membership fee can theoretically bene�t from an exemption policy (Kamuzora and Gilson 2007, 4). Contribu- tions are between T Sh 5,000 and T Sh 15,000 per household per year. The funds raised are doubled by a “matching grant� from the national budget (Health Basket Funds), sub- ject to a particular district CHF raising T Sh 5 million or more in annual contributions. Currently, 72 of 132 districts have a functioning CHF. The goal is to enroll 85 percent of employees of the informal sector and of those living in rural areas (Kamuzora and Gilson 2007, 1). However, districts are at different phases of implementation. According to reports, members in only 29 of 72 districts had access to services in 2007. Of the 29, enrollment varied between 4 and 20 percent of the population (Sendro 2007, 12). The trend of enrollment differs from district to district. Some districts are slow to register new members and are poorly motivated. CHFs have the problems of many voluntary health insurance schemes, including a poor provider structure. Membership cards are typically honored only in the district where the CHF is located, and members are typically restricted to the facility where they registered. People tend not to re-enroll. CHFs also suffer from limited management capacity and a lack of proper data management, although since the NHIF got involved, this problem has improved. A recent situation analysis by the Swiss Institute for Tropical Medicine and Public Health found that CHFs had issues in the areas of design, enrollment, ser- vices, and sustainability. It also found that the district medical officer played a key role in determining the viability of a fund (Swiss TPH 2010). Studies have identi�ed issues regarding a lack of accountability at both the district and the ward levels, with managers interpreting policy on an individual basis. This is- sue is exempli�ed by households that would normally qualify for an exemption being denied such exemptions owing to either (a) the interpretation of the exemption policy by individual managers or (b) a complete lack of a de�ned exemption policy or of ear- marked funds to �nance the exemptions (Kamuzora and Gilson 2007; Mtei and Mulligan 2007, 6). Kamuzora and Gilson (2007) found that “inability to pay membership contributions was the most important barrier, whereas poor quality of care, non-acceptance of the need to protect themselves against the risk of sickness, and lack of trust in CHF manag- ers ma ered more to average and wealthy community members� (Mtei and Mulligan 2007; Kamuzora and Gilson 2007, 8). A study by Kuwawenaruwa, Macha, and Borghi (2010, 16) on willingness to pay for voluntary health insurance found that “People are 32 World Bank Study willing to join and pay for health insurance if they are made aware of the principles of in- surance and properly understand the concept of risk pooling.� The study also found that there is a general unwillingness to pay more than a minimal amount for health insurance coverage. Willingness to pay was higher for those districts with a higher bene�t pack- age, although few indicated that they were prepared to pay more for increased bene�ts (Kuwawenaruwa, Macha, and Borghi 2010, 14–15). Clearly, such issues would need to be addressed in any potential scaling up of CHF/TIKA. The Swiss study also found issues with respect to access for the poor, as mentioned earlier, as well as adverse selection and high dropout rates during drought seasons. It also found a policy of le ing potential applicants sign up at their leisure rather than ac- tive enrollment seeking. This study, as well as the review by Mtei and Mulligan, did �nd a number of best practices that could help to increase enrollment. They include improv- ing the bene�t package, clearly determining the ability or willingness of potential mem- bers to pay, strategically using sensitization, and promoting flexibility in membership fees in terms of both payment timing and contribution rates. For example, councils could consider two payments per year, monthly payments, or seasonal payments (based on harvest times) or even payments for multiple years (for example, following good harvests). Enrollment could also be increased through col- laboration with micro-�nance institutions to help fund the premiums. At the facility level, district drug buffer stocks could be established for supplementary drug supply (�nanced through cost-sharing funds) to ensure that services are available when CHF members need them, bank accounts could be established at the health facility level to allow facilities to access CHF funds, and the emphasis on supporting pro-poor funding mechanisms within districts could be encouraged in order to enroll more people from vulnerable groups. Support to CHF management was also mentioned as being critical (Mtei and Mulligan 2007, 11–13). Both the MoHSW and the NHIF provide regulatory oversight to CHF/TIKA. The NHIF has developed a CHF Action Plan (2009–12) that outlines how the newly estab- lished linkage between the NHIF and CHF/TIKA is going to work. This plan notes several advantages of this linkage, including (a) providing an entry portal for wider health insurance coverage, (b) providing a cost-effective coordination mechanism for CHF/TIKA at the national level, (c) promoting the harmonization of services and claims mechanisms, (d) facilitating the extension of coverage to the informal sector, (e) provid- ing professional management of CHF/TIKA revenue, and (f) facilitating the portability of services (Tanzania Nation Health Insurance Fund 2009a, 13–14). Speci�c interventions mentioned in this plan include reviewing CHF bene�t packages, registration, and collec- tion mechanisms; introducing claims management and information systems at the CHF level; undertaking capacity building of CHF operations; promoting increases in user fees to encourage CHF membership; and establishing a Risk Equalization and Reinsurance Fund (Tanzania Nation Health Insurance Fund 2009a, 17–18). Nongovernment Nonpro�t (Micro-Insurance)7 Health micro-insurance schemes in Tanzania are typically sponsored by religious groups, informal groups, and associations. Micro-insurance schemes in the Mainland, such as UMASITA (Tanzania Informal Sector Community Health Fund) and VIBINDO (the umbrella organization of informal sector operators in the Dar es Salaam region), seek to strengthen informal sector communities by providing be er access to health care, Making Health Financing Work for Poor People in Tanzania 33 improved quality of care, and ways to promote comprehensive health care services at af- fordable prices. Micro-insurance for health care is still in its infancy. Most of the schemes enroll groups rather than individuals (for example, all market vendors are required to join), but each group operates as a separate risk pool, causing potential �nancial sustain- ability problems. Dar es Salaam has three major initiatives. The �rst, organized under VIBINDO, was facilitated by the International Labour Organization through small business operators in the second half of the 1990s. This initiative has remained small, with a very low rate of registration and renewal of membership. The second initiative, registered by UMASIDA, is now known as UMASITA. This scheme, like VIBINDO, is facing the challenge of registering new members. UMASITA members include small-scale market retailers, tinsmiths, cobblers, stone crushers, and food vendors. VIBINDO covers about 1,102 people out of about 40,000 VIBINDO soci- ety members. UMASITA had up to 40,000 people enrolled, although it recently stopped functioning as a result of issues related to revenue collection and management, service utilization, and continuity of enrollment. Together the micro-insurance schemes covered substantially less than 1 percent of the total population. The UMASITA bene�t package included maternal and child health; voluntary counseling and testing; and treatment of common diseases such as malaria, pneumonia, diarrhea, and sexually transmi ed infections. Surgical services are provid- ed at government facilities, and the user fee is paid by the scheme. The VIBINDO bene�t package includes primary health care services, reproductive health care services, some referral services, minor surgery, and limited hospitalization. Neither scheme required a copayment (McIntyre and others 2008, 24–25). The third type of initiative is the church-based initiatives led by the Catholic Church, Lutherans, and other denominations in different parts of Tanzania. A new initiative that started in September 2010, sponsored by the Anglican Health Network, had a goal of registering 40,000 people within the �rst six months of operations and going nationwide within three years. Although the various churches have differing reasons to start such initiatives, there are two underlying themes. First, they seek to help alleviate the �nanc- ing problems of the existing church-run health facilities; second, they seek to help the poor access health care by creating risk-pooling mechanisms. This approach has seen membership problems similar to those experienced by other initiatives. One of the key problems contributing to the low uptake by the population is the general lack of knowledge of the fundamentals of insurance operations. As well, there is a failure to explain these fundamentals in the local, simpli�ed context where the ma- jority of target groups that would be interested exist. Another major issue is �nancial sustainability because such schemes operate separate risk pools and suffer from cost escalation (Jamu and others 2009, 32). Private Health Insurance8 As health sector reform took place in the mid- to late-1990s, private health insurance became popular with most private companies. According to company representatives, approximately 120,000 people are covered by private health insurance. This number represents only a small percentage of the overall population of Tanzania. On average, premiums represent an estimated 8 percent of payroll. 34 World Bank Study Less than 1 percent of the total expenditure within the national health system comes from the private health insurance market. In 2002, T Sh 7.2 billion was the total amount of private insurance premiums paid (other than for life insurance, but including health coverage). It is to be hoped that the new National Health Accounts process that is under way will provide more up-to-date information on this sector. These private insurance premiums indicate that the degree of health protection of- fered by private insurance companies is quite small. Interviewed representatives noted two factors for the lack of growth of private health care in Tanzania. First, the growth of private health insurance is difficult because of the poor level of health care in many parts of the country. Second, the bureaucracy surrounding private insurance is overwhelming and keeps companies from developing new initiatives. Private health insurance compa- nies do not report to the MoHSW but—together with other forms of private insurance— are supervised by the Insurance Supervisory Department. A private micro-insurance initiative was recently started in Tanzania, �nanced in part by the government of the Netherlands through PharmAccess and The Health In- surance Fund. The scheme is directed at entrepreneurs who receive micro-credit from the Pride Bank. Pride has partnered with the Tanzanian insurer Strategis to provide the entrepreneurs with a health insurance package. Called the Strategis Community Health Insurance Plan (SCHIP), participants pay approximately 10 percent of the premium as a copayment. The other part of the insurance premium is provided by The Health Insur- ance Fund, which is an international organization based in the Netherlands. The Fund works with a grant from the Dutch government; other funders such as the U.S. Agency for International Development (USAID) are also reported to have joined in the program. The longer-term sustainability of this approach—once the primary funders leave the sec- tor—will be an important element in the evaluation of these approaches. Notes 1. This section draws heavily from Lankers and others (2008, 12–13). 2. For additional information, see TACAIDS (2010). 3. See Jamu and others (2009, 33–37) and Makawia and others (2010, 28–29). 4. The text in this section draws heavily from Lankers and others (2008, 21–22). 5. This section draws heavily from Lankers and others (2008, 22–23). 6. This section draws heavily from Lankers and others (2008, 23–24). 7. This section draws heavily from Lankers and others (2008, 24–25). 8. This section draws heavily from Lankers and others (2008, 25–26). CHAPTER 4 Experiences from Other Countries E xperience gleaned from looking at other countries can suggest what might be pos- sible and the types of challenges that can be expected. The key is to carefully de- termine whether what worked in other countries is applicable in the local context. The comparison countries were selected for the following reasons: Thailand, Vietnam, and China were visited by a Tanzanian delegation as part of a Southeast Asia study tour organized by the World Bank. Ghana and Rwanda have had great success in increasing coverage and face different sustainability challenges. Kenya has gone through the process that Tanzania is just starting, although discus- sions continue. It is interesting to see the results of the discussions thus far, and there are also similarities in the organization of health services. Mexico and Colombia have made signi�cant progress in expanding health insur- ance coverage from the formal sector to the informal sector (including farmers) and have established effective mechanisms for identifying and reaching the poor. Each of the following sections contains a summary table of the key aspects of health �nancing in the country (similar to table 1 in the Executive Summary), focuses on �nan- cial protection for the informal sector, and ends with some key messages that are impor- tant for Tanzania. Further details on each country appear in appendix A. Thailand1 The move to universal coverage in Thailand followed campaign promises during the 2001 general election. Known as the “30 baht� scheme (representing the amount of the copayment, equivalent to US$0.75), it used existing health budgets and some additional �nancing to expand coverage to virtually all people not covered through the Civil Ser- vants’ Medical Bene�t Scheme (CSMBS) and the Social Security Scheme (SSS) for formal private sector workers. Transition toward universal coverage started in April 2001. Door-to-door approach- es were used to sign up bene�ciaries, reaching 44.5 million people by April 2002. There was some resistance from health care providers, but this resistance did not hold up the implementation of the scheme. The major design principles include (a) ge ing to universal coverage as quickly as possible for those not already covered by existing schemes; (b) ensuring that cost control mechanisms are built into the universal coverage scheme; and (c) a minimal payment toward each patient’s care (30 baht per visit; note also that Thailand is a middle-income country), although this minimum has since been dropped with minimal effect on utiliza- tion. Table 4.1 shows the key characteristics of health �nancing in Thailand. 35 36 World Bank Study Table 4.1: Summary of Health Financing Approaches in Thailand Formal public Formal private Informal Informal Dimension sector sector Poor urban rural/farmers Prepayment scheme CSMBS SSS Universal coverage scheme Coverage (number) 5.7 million 10.1 million 47.2 million Coverage (% of 9 16 75 population) Source of revenue Government 1.5% each from Government budget budget employer, employees, and government Revenue collection General taxation Payroll contribution General taxation method Number of risk pools One One One Payment methods Fee-for-service Capitation contract Capitation for outpatient care, global budget capped for inpatient and for inpatient and diagnostic-related groups for inpatient care outpatient outpatient Benefit package Extensive inpatient Extensive inpatient Broad benefit package for inpatient and outpatient, not and outpatient and outpatient as extensive as CSMBS or SSS (average of services (average services US$60/person/year) US$400/person/year) Facility coverage Free choice of Extensive coverage All public and private providers willing to accept the facilities agreed tariff Regulatory oversight Run by Ministry of Run by Ministry of Scheme run by National Health Security Office (NHSO) Finance Labor, Social Security Office Source: Author. It is important to note the different provider payment methods for each scheme. The SSS uses a capitation contract model for both inpatient and outpatient care, while the CSMBS uses fee-for-service reimbursement extensively. The use of this payment method is felt to be a contributor to the high cost of this scheme (the cost of bene�ts for CSMBS is equivalent to about US$400 per person while bene�ts for universal coverage are about US$60 per person). The universal coverage scheme uses capitation for primary care and global budget-capped diagnostic-related groups for hospital care. Originally, there was a copayment of B 30 for each service under the universal coverage scheme, although this copayment was subsequently eliminated. This elimination has resulted in no discernible long-term increase in utilization. There are no deductibles for any of the schemes. The poor are automatically eligible for the universal coverage scheme. Research has shown that the scheme is particularly effective in targeting the poor, with 44 percent of the enrollees in this scheme belonging to the �rst two income quintiles. It has also been successful in improving access to health services and in reducing impoverishment resulting from illness, which has declined by about 75 percent for families in the �rst income quintile and by about 64 percent overall. Impressively, the results have been achieved without any signi�cant increase in to- tal health expenditure, indicating that—to a large extent—universal coverage bene�ts have replaced out-of-pocket payments, and the costs have been accommodated within GDP growth. However, sustainability is an ongoing concern, especially with the recent Making Health Financing Work for Poor People in Tanzania 37 expansion of the universal coverage bene�t package, and increased �nancial access to services has strained the ability of the health system to cope with increasing service demand. The payment methods for the CSMBS scheme have led to continuing �nancial pressure. Box 4.1 highlights the key takeaway messages for Tanzania from the Thailand experience. Box 4.1: Key Messages from Thailand • Political will and popular support are essential. • Rapid scale-up is possible if planned properly. • Plans should avoid fee-for-service payment methods if possible. • It is essential to plan for resulting utilization growth. • Additional government (or development partner) financing will be needed in order to cover the poor. Vietnam2 Health �nancing reform in Vietnam was driven by the rising costs of health care and by decreasing accessibility of the population, especially those in rural areas and the poor. The consolidation of the various schemes into a single administrative arrangement under the Vietnam Social Security Agency (VSS) and risk pool resulted in administra- tive efficiencies and increased solidarity between the Statutory Health Insurance (SHI) and Voluntary Health Insurance (VHI) programs. A high level of enrollment has been achieved by explicitly targeting the poor for coverage. Table 4.2 provides details on the keys aspects of health �nancing in Vietnam. The main design principles appear to be (a) to sign up as much of the population as possible, (b) to provide the same level of coverage for both SHI and VHI (with cross-subsidization if necessary), and (c) to focus on administrative efficiency by merging existing arrange- ments into just two programs and by having both programs administered by the same organization, that is, the VSS. About half of the population (48 million people) is now covered by either SHI or VHI. Providers are mostly reimbursed using fee-for-service, although some districts are experimenting with capitation arrangements. The fee-for-service approach is leading to concerns about �nancial sustainability. Providers are mostly public, although private providers are able to participate if they are willing to accept the VSS tariff. Referral re- quirements are enforced so that if people go to a higher-level facility without a referral le er, they have to pay the cost themselves. Since 2001, there have been moves to increase hospital autonomy, giving more con- trol over staffing, organization and management, and �nancial management. Copay- ments of 20 percent are standard, although there are exemptions for the poor. There are no deductibles, but there are maximum payments for high-tech procedures (US$1,250 equivalent). The Ministry of Labor and Social Welfare does an assessment to determine who is poor. 38 World Bank Study Table 4.2: Summary of Health Financing Approaches in Vietnam Formal public Formal private Informal Informal Dimension sector sector Poor urban rural/farmers Prepayment scheme SHI SHI, criteria VHI US$11/month rural, US$14/ month urban Coverage (number) 11.8 million 18.2 million 12 million Coverage (% of 14 21.7 14.3 population) Source of revenue 2% paid by employer, 1% by 3% of minimum Subscribers, local government employee–total 3% salary paid by subsidy for near-poor government Revenue collection Payroll deduction General VSS contracts with local collection method taxation agents (10,000) Number of risk pools Single risk pool with allocations to regional and county Single risk pool with allocations to branches of VSS regional and county branches of VSS Payment methods Mostly fee-for-service, some districts experimenting with capitation arrangements Benefit package Same package: inpatient, outpatient, diagnostics, rehabilitation, drugs, transport for the poor and those in rural areas; standard copayment of 20%, exemptions for the poor Facility coverage Mostly public, but private can participate if they will accept the VSS tariff Regulatory oversight Both schemes run by VSS Source: Author. A number of issues and challenges have been identi�ed: (a) controlling abuse of the system; (b) dealing with adverse selection (people take coverage only when they are sick); and (c) dealing with the de�cit, which reached 33 percent of SHI/VHI revenue in 2008 through both revenue and expenditure measures. There is a need to make VHI more affordable, especially for the “near poor� (those just above the poverty line); to sign up the other 50 percent of the population; and to improve the quality of health services. Box 4.2 provides the key takeaway messages from Vietnam. Box 4.2: Key Messages from Vietnam • There are potential efficiencies from shared administration. • Plans should avoid fee-for-service payment methods if possible. • It is essential to match the benefit package to available revenue. • There is a place for both public and nonpublic providers within a national health insurance arrangement. • The reach of the formal national health insurance organization can be extended by employing community-based contractors to collect premiums. Making Health Financing Work for Poor People in Tanzania 39 China3 A number of initiatives were put in place in China to address declining social health protection—especially for the poor—and poor resource allocation. Table 4.3 highlights the various initiatives. Table 4.3: Summary of Health Financing Approaches in China Formal Formal Formal Informal public private sector, Informal rural/ Dimension sector sector Poor rural urban farmers Prepayment scheme Basic Medical Insurance Medical New Cooperative Rural Medical (BMI)–Urban Assistance (MA) Scheme (NRCMS) Coverage (number) 223 million 51.9 million 830 million Coverage (% of 16.8 3.9 62.7 population) Source of revenue 6% employer, National Payment per person per year, shared 2% employee government between individual (20%), local (40%), and national government (40%) Revenue collection Paid by employers to Social General Respective levels of government, method Security Department revenue NRCMS county level offices, or local agents Number of risk pools One risk pool at each county or National risk One risk pool at each county or prefecture, also employee health pool prefecture (large counties) accounts (4.2% of payroll) at discretion of employee Payment methods Plan that pays a portion of medical As SHI Global budget for outpatient services, expenses inpatient per bed-day, reimbursement Plan that allows subscriber to use rate that varies by type of service saving account to pay copays or (considering appropriate referral patterns) other costs Benefit package Better than NRCMS Started relatively small, expanded as client and district or region demand increased Facility coverage All contacted service providers to be As SHI criteria All contacted service providers to be eligible eligible Regulatory oversight General oversight by national and provincial governments Source: Author. The major success factors included (a) having the highest-level political commit- ment to these programs at all levels (national, provincial, local); (b) piloting to prove the concepts but not waiting to achieve perfection in the pilots before scaling up; and (c) beginning with a small premium and bene�t level that would quickly demonstrate the value of the scheme, thereby leading to both consumer demand and increased political and �nancial commitment by governments to improving the level of bene�ts. The central government sets rules and provides overall leadership, while local gov- ernments implement. BMI workers are part of risk pooling at the county or prefecture level, but each employee is also his or her own risk pool. A “debit card� is issued that can be used at the employees’ discretion to pay some or all of the deductible or copayment portions for which employees are responsible for when they receive care. 40 World Bank Study Management of all funds is at the county level (Social Security Department for BMI, and Health Authority for the NRCMS). There is some discussion about pooling at the prefecture level as a result of variations in county revenue-raising capability and the problems of smaller risk pools. All schemes pay only a portion of the total medical expenses (30–70 percent based on the type of services), with deductibles, copayments, and ceilings on the maximum amount of reimbursement. The levels of deductibles, copayments, and ceilings are set at the local level in accordance with local conditions and �nancial availability, and they are revised on a regular basis. Reimbursement rates encourage the use of local medical facilities. The Civil Affairs agency conducts annual poverty level assessments. Income levels of US$100–150 equivalent per person per year are considered poor, and income lev- els under US$100 are considered very poor. The poor are automatically enrolled in the NRCMS and eligible for the MA scheme. MA bene�ts are in addition to those under NRCMS. The main messages for Tanzania are shown in box 4.3. Box 4.3: Key Messages from China • There is value in starting with a small benefit package and increasing both the benefits and funding according to consumer demand. • Risk pools and administration at lower levels of government require capacity at that level, and smaller risk pools can be problematic. • There is always a requirement for strong central regulation to ensure consistency in administration and delivery at the local level. Ghana4 The National Health Insurance Scheme in Ghana is one of the legacies of the John Ku- fuor administration. A National Health Insurance Authority (NHIA) licenses, monitors, and regulates the operation of health insurance schemes in Ghana. There are three main categories of health insurance in Ghana. District mutual health insurance schemes are operational in every district. This is the public or noncommercial scheme, and anyone who is a resident in Ghana can register under this scheme. The second category comprises private commercial health insurance schemes that do not receive National Health Insurance Scheme subsidies and must pay a security deposit before they start operations. The third category is known as the private mutual health insurance scheme. Any group of people, such as members of a church or social group, can come together and start making contributions to cater for their health needs, thus providing for services approved by the governing council of the scheme. Private mutual health insurance schemes do not get a subsidy from the National Health Insurance Scheme. The details are highlighted in table 4.4. Making Health Financing Work for Poor People in Tanzania 41 Table 4.4: Summary of Health Financing Approaches in Ghana Formal public Formal private Informal Informal Dimension sector sector Poor urban rural/farmers Prepayment scheme District medical health insurance schemes Coverage (number) 12.5 million (2008) Coverage (% of 54 population) Source of revenue Premiums of US$8 to US$53 per year for the informal sector, plus 2.5% points from VAT and 2.5% points from Social Security contributions for the formal sector Revenue collection District fund for premiums, and central government for VAT and Social Security contributions; method significant exemptions from premium payment (for example, under 18, elderly) Number of risk pools One per district (145 districts) Payment methods Diagnostic-related groups for inpatient, capitation, or fee-for-service for outpatient Benefit package Full inpatient, outpatient, maternity, emergency care, and essential drugs Facility coverage Member with a card in any scheme to have access to any provider Regulatory oversight NHIA that licenses, monitors, and regulates the operation of health insurance schemes Other insurance Private schemes and private mutual health insurance schemes (any group of people, such as schemes members of a church or social group) that are alternatives to MHI Source: Author. Coverage is mandatory for both the formal and informal sector, although it is hard to enforce for the informal sector. People who register under any of the schemes are given a card that can be used to seek treatment in any hospital in the country. There are no copayments unless extra services, such as a private ward, are used. The card can also be used to buy prescribed drugs at accredited pharmacies or licensed chemical shops. There is an extensive bene�ts package. There have been delays in processing claims and making payments, which have led to reports of hospitals and pharmacies turning patients away. A number of reforms are being implemented to deal with such challenges and the issue of cost escalation gener- ally, including (a) the establishment of a consolidated premium account to collect all premiums, (b) the establishment of consolidated claims management centers to process claims, (c) the use of a standardized prescription form, (d) the introduction of capitation for outpatient services, (e) the implementation of a unitary system of contracting with service providers, and (f) several other initiatives. A one-time premium payment was also promised some time ago, but it has not yet been implemented. Box 4.4 highlights the main takeaway messages from Ghana. Box 4.4: Key Messages from Ghana • Although small, district-level risk pools are politically attractive, the realities of health insurance operations require specialized expertise in revenue collection, contracting, and claims management. • Expanding the revenue base through earmarked taxes (for example, VAT) can be an effective strategy for ensuring financial stability and coverage of the poor. • Fee-for-service should be avoided if at all possible. • It is essential to match the revenue received to the benefit package. • Reaching the poor requires special efforts and approaches. 42 World Bank Study Rwanda5 Rwanda has pioneered major programmatic, organizational, and health �nancing re- forms aimed at improving the quality of care and, ultimately, the health status of the population with a particular focus on the most vulnerable segments. Prepaid �nancing was organized in the form of mutual health organizations (MHOs). From only one initia- tive in 1998, these schemes have expanded to cover virtually the entire country. Table 4.5 highlights the key features of the current arrangements. Table 4.5: Summary of Health Financing Approaches in Rwanda Formal public Formal private Informal Informal Dimension sector sector Poor urban rural/farmers Prepayment scheme RAMA and MMI MHO or RAMA MHO (Mutuelles) schemes Coverage (number) 297,000 (2006) 7.6 million (2008) Coverage (% of 3.3 85 population) Source of revenue RAMA–15% Member contribution of US$7.60 per year for up to seven in a family, plus (shared equally); contributions from government and donors for those who cannot afford this MMI–22.5% (17.5% amount government) Revenue collection Payroll deduction Collected by MHO method Number of risk pools One each for RAMA One per district (approximately 392), but National Guarantee Fund and District and MMI Solidarity Fund have been created to provide equalization and reinsurance support Payment methods Fee-for-service Some capitation and fee-for-service; output-based payment methods that have also been implemented for some services Benefit package Full range of Preventive and curative services, prenatal care, delivery care, laboratory services exams, drugs on the MoH essential drug list, ambulance transport to hospital, limited district hospital services Facility coverage Own facilities plus Contracts with district health center and surrounding hospitals; recent changes contracts with that have allowed subscribers to obtain service at any health facility public/FBO Regulatory oversight RAMA/MHI Boards MHOs that are nonprofit, self-administered organizations; policy direction that comes from MoH Source: Author. Notes: FBO = faith-based organization; MHI = medical health insurance; MHO = mutual health organiza- tion; MMI = military medical insurance; MoH = Ministry of Health; RAMA = Rwanda Health Insurance Company (Rwandaise d’Assurance Maladie). After a pilot implementation, MHOs were adapted to �t within the decentraliza- tion model that was being developed in Rwanda, speci�cally involving the Ministry of Decentralization and Local Affairs (MoDLA) and its agencies in promoting MHOs. This adaptation anchored them in the community and facilitated the mobilization of local authorities in the various administrative districts and district subdivisions. This involve- ment also resulted in the involvement of nongovernmental organizations (NGOs) and religious leaders, which raised the population’s awareness of the importance of enroll- ing in MHOs. Leadership at the central level was also mobilized to ensure the backing of the highest authorities in government. Making Health Financing Work for Poor People in Tanzania 43 In mid-2006, bene�t packages were expanded, and coverage for the indigent, vulner- able groups and for people living with HIV was institutionalized by the government and foreign partners. The bene�t packages now covered primary health care, secondary care, and tertiary care, which dramatically improved the price-quality ratio for MHO services. Recognizing the potential problems involved in small risk pools, Rwanda estab- lished a National Guarantee Fund (FNG) and a District Solidarity Fund (FSD) to bol- ster �nancing mechanisms for MHO expansion in the country. The FNG/FSD system strengthens equity of access and �nancing by harmonizing MHO bene�ts with those received by the bene�ciaries of other social health insurance systems and by providing care for the indigent. The FNG is �nanced through contributions from the government, Rwanda Health Insurance Company (RAMA, or Rwandaise d’Assurance Maladie), mili- tary medical insurance (MMI), private insurance systems, and foreign partners, includ- ing the Global Fund. The FSD is �nanced by the contributions of MHOs, administrative districts, transfers from the FNG, and contributions from the development partners that are involved at the district level. The main takeaway messages from Rwanda are shown in box 4.5. Box 4.5: Key Messages from Rwanda • Strong government leadership, vision, and a regulatory framework at all levels, foster the short- and long-term sustainability. • Strategies must be adapted to a changing macro and health sector environment. • Independent controls and quality checks are essential for monitoring and evaluating health facility performance. • Cultural and social factors, particularly solidarity within communities, contribute to the success of several health service delivery innovations. • Provisions are needed for financial protection and other support for indigent popula- tions. • Government coordination of donor funding is critical to ensure that aid is used effectively and aligned with national priorities. Kenya6 The previous examples highlight reforms that are already in place; Kenya presents some of the evolving thinking on health �nancing in a system that is similar to Tanzania in many respects. The process for the Kenya strategy drew heavily on the previous fail- ure at health �nancing reform in 2005 and is the result of a broad consultation process, which included a wide variety of stakeholders and drew on both extensive analytical work and reviews of health �nancing systems in other parts of the world. The key ele- ments of the strategy are highlighted in table 4.6. 44 World Bank Study Table 4.6: Summary of (Proposed) Prepayment Schemes in Kenya Formal public Formal private Informal Informal Dimension sector sector Poor urban rural/farmers Prepayment scheme National Health Insurance Fund (NHIF) or NHIF private insurers offering benefit package Coverage (number) 6.8 million including informal sector Not known (See left.) Coverage (% of 18 (including informal) Target 100% Currently small but increasing population) proportion of population Source of revenue Payroll contribution based on income, Government or Negotiated contributions (for groups), ranging from US$22.50 to US$300 annually development premium subsidies for near-poor partners Revenue collection Deduction from payroll paid directly to NHIF NHIF paid Through community health funds or method from budget associations Number of risk pools One Payment methods Proposed case based (capped) for inpatient care and capitation for outpatient care Benefit package To be determined–but to start with modest package and expand as resources allow Facility coverage All facilities (public, FBO, NGO, private) willing to accept tariff Regulatory oversight Proposed Benefits and Tariffs Board for health-related issues (including quality) and financial oversight by Ministry of Finance Other insurance Private schemes that can provide the basic benefit package if they are willing to charge the same schemes premiums and pay the same tariffs (subscriber choice) and that can also offer private supplementary insurance for services outside the basic benefit package Source: Author. Improving efficiency is an important focus of the strategy, including maximizing the value from the current resources and the functioning of the health system generally. The role of the Ministry of Health (MoH) should be as the steward of the health system, thus providing overall direction and guidance and se ing the overarching policies of the system but not being involved in day-to-day management of health service provid- ers. Making the MoH both the regulator and a service provider will inevitably lead to conflicts of interest. Strong quality assurance and accreditation mechanisms are extremely important, and strengthening the referral system will ensure that each level of care provides the most appropriate services efficiently and effectively. Separating the purchaser from the provider of services will also enhance efficiency, but this move will require increased autonomy for public health care providers, so that they have the ability to manage their facilities in the most cost-effective way. The strategy proposes the creation of a Bene�ts and Tariffs Board to work with pur- chasers, providers, and other stakeholders to develop appropriate reimbursement levels, as well as the basic bene�t package. Although this package will initially be a limited set of services, as the economy grows or as the total resources available for health increase or both, the package would be reviewed and expanded. An important feature of these processes will be transparency and availability of information. For services not included in the package, options would exist for people to take out supplementary private health insurance. Because this option would be an important element of the overall health care �nancing and delivery system, it will be necessary to develop effective �nancial and quality oversight structures to ensure that private insurance remains sustainable. Making Health Financing Work for Poor People in Tanzania 45 If one is to reach informal workers and farmers, smaller risk pools in the form of community health funds may be established. Moreover, the NHIF would essentially act as a reinsurer for those funds, covering the full range of services for members regardless of the claims activity. Capacity will need to be developed in actuarial analysis to deter- mine the appropriate and minimum premium levels for such schemes, as well as for the general population of insured people covered through formal employment or informal group arrangements. Protection of the poor is central to the overall health care �nancing strategy. A critical issue, therefore, will be the timely and accurate identi�cation of the poor. An important element is to make sure that health service providers do not face disincentives to treating the poor. Thus, it will be important to compensate providers for lost user-fee revenue when they treat poor people. The registration card that members receive will both signal to the provider that no user fee should be collected and signal to the purchaser that the re- imbursement should be augmented by the user fee that would otherwise have been paid. To involve development partners, the strategy suggests the creation of an Access and Equity Fund, which would receive funding from both government and develop- ment partners and would use those funds to expand coverage through premium subsi- dies for the poor and near-poor. Box 4.6 highlights the key messages from the proposed Kenya strategy. Box 4.6: Key Messages from Kenya • It is essential to improve the functioning of the health system, including efficiency. • Conflicts of interest from having the MoH as both regulator and provider must be addressed. • The regulatory framework must include independent and transparent institutions. • Protection of the poor requires careful planning, including identification, incentives for providers to treat the poor, and funds that will allow premiums to be covered. • Development partners’ support should follow the shift from input- to output-based financing. Colombia7 Before 1993, inefficiency, badly targeted public subsidies, and fragmented markets result- ed in only one Colombian in �ve having any protection against the �nancial risk of health shocks from serious illness. Moreover, only those with more �nancial resources could af- ford to join social security schemes or pay out of pocket for health care. Households con- tributed a signi�cant portion of health care �nancing, and cost was the most signi�cant barrier to health care, with only one individual in six who fell ill seeking medical care. Law 100 of 1993 mandated the creation of a new national health care system with universal health insurance coverage and reorganized �nancing and delivery. A key fea- ture of this law was that public subsidies would go directly to individuals rather than to institutions. The effects of the changes on the overall framework of social health protec- tion are shown in table 4.7. 46 World Bank Study Table 4.7: Summary of Health Financing Approaches in Colombia Formal public Formal private Informal Informal Dimension sector sector Poor urban rural/farmers Prepayment scheme Contributory scheme (RC) Subsidized RC if able to contribute (over scheme (RS) US$170/month), otherwise RS Coverage (number) RC–21.5 million (2007) RS–18.2 million (2007) Eligibility determined through proxy means test Coverage (% of 50 42 population) Source of revenue 11% of payroll For RS, national and local tax revenue and 1.5% payroll tax (solidarity contribution from RC) Revenue collection Submitted by employers to Social Security National and local governments and employers (for method Fund (SSF) solidarity contribution–to SSF); insurers paid Insurers paid US$244/person/year, on US$137/person/year, on average average Number of risk pools Subscriber choice of 21 insurers Subscriber choice of 43 insurers Payment methods Contracted providers paid mostly capitation Contracted providers paid in the same way as RC for primary care and preventive services, Public providers funded through historical budgets to fee-for-services, or case-based for specialist provide services not covered by RS benefit package or care to the uninsured Benefit package RC–comprehensive benefit package RS–low-complexity care, catastrophic illness, only including all levels of care limited coverage for most hospital care, no short-term disability coverage Facility coverage Insurers that contract with a network of public, private, or own service providers (owned by the insurers) Insurers in RS that must contract with public providers for at least 40% of premiums Regulatory oversight Premiums and benefits that are determined by the government; insurers that compete on quality Other insurance Partial subsidy scheme for those who do not fully qualify for RS; separate plans for police and military schemes and some other groups (5% of population) Source: Author. The reform introduced four main elements to reach the poor: ■ Having a proxy-means testing to target public health subsidies to the neediest (SISBEN, or Selection System of Bene�ciaries for Social Programs) ■ Transforming traditional supply-side subsidies, which �nanced the public health care network, into demand-side subsidies, which subsidize individual insurance premiums for the poor ■ Having an equity fund in which revenue from payroll contributions and trea- sury resources cross-subsidize insurance premiums for the poor ■ Contracting for health service delivery from both the public and private sectors The new system is a universal health insurance coverage plan with two schemes: ■ A contributory scheme (RC) to cover formally employed and independent workers who contribute to the scheme, with contributions collected by the in- surer of choice ■ A subsidized scheme (RS) to cover the poor and indigent individuals who could not afford to make any insurance contribution Making Health Financing Work for Poor People in Tanzania 47 Payroll contributions go into a national health fund with four separate accounts. The fund �nances insurance premiums for all enrollees in the RC. In the RC-RS cross- subsidization process, a share of the contributions is allocated to �nance the RS, together with treasury transfers to the territories. Those who are eligible for enrollment in the RS but are still uninsured are to rely on public hospitals for care. Every insured individual is free to choose an insurer and to consult any provider in the insurer’s network. Both schemes have a basic bene�ts package, but the RC package includes every level of care while the RS package has to be complemented with services provided by public hospitals and �nanced through supply-side subsidies. The design provides that the supply-side subsidies would be turned into demand-side subsidies (�- nancing services provided rather than inputs) over time to achieve universal insurance coverage with the same package in both schemes. The reform increased �nancial protection for all, but especially for the poor and ru- ral population. For Colombians as a whole, �nancial protection increased from 23 to 62 percent, but insurance coverage among the poorest group increased from 9 to 49 percent, compared with a much smaller increase from 60 to 82 percent for the wealthiest group. Formal insurance in Colombia reduced out-of-pocket expenditures on ambulatory care by between 50 percent and 60 percent. The poor in the RS spent about 4 percent of their income on ambulatory care, but the uninsured poor spent more than 8 percent. Out-of-pocket expenditures on hospitalization among the uninsured poor absorbed more than 35 percent of their income in 2003. The poor in the RC spent a smaller propor- tion of their income on inpatient care than the poor enrolled in the RS. Further, a health shock requiring hospitalization pushed 14 percent of those hospitalized and uninsured below the poverty line compared with only 4 percent of inpatients covered by the RS. Access to preventive care was also improved. Despite positive results, the Colombian social insurance scheme has been criticized for not having achieved universal coverage and for �nancing a less comprehensive bene�ts package for the poor (RS) than for the wealthy (RC). The slower-than-planned transformation in changing supply-side subsidies (which �nance public hospitals) into demand-side subsidies so they �nance health insurance for the poor has also come un- der �re for slowing down the expansion of the RS bene�ts package. The key messages from the Colombian experience are summarized in box 4.7. Box 4.7: Key Messages from Colombia • Explicit cross-subsidies can be an effective way of financing the poor, but financing from government will still be required. • Different benefit packages for contributory and subsidized plans can help to contain costs but may be attacked on equity grounds. • Health insurance improves access to both curative and preventive services and reduces both out-of-pocket and catastrophic health-related payments. • Long transitions from supply-side subsidies to performance-based payments can be problematic—a more direct and immediate approach may be preferable. • Existing or consistent mechanisms for identifying the poor should be used if possible. 48 World Bank Study Mexico8 Mexico established social security institutions for formal private and public employ- ees some time ago, but the poorer half of the population faced problems with access to services, quality of care, and affordability. There was a great risk of impoverishment because of health spending: 2 million to 4 million households spent 30 percent of their annual disposable income on health. In 2000, a new health insurance program, Seguro Popular (SP), was introduced for those not covered by existing plans. As a voluntary program, it provided bene�ciaries with subsidized insurance coverage comparable to formal sector plans. Table 4.8 shows the key features of the various plans. Table 4.8: Summary of Health Financing Approaches in Mexico Formal public Formal private Informal Informal Dimension sector sector Poor urban rural/farmers Prepayment scheme Social Security Mexican Social SP–state SP Institute for Security Institute governments Government Workers (IMSS) to identify and (ISSSTE) enroll Coverage (number) 9.6 million 42.6 million 41 million Coverage (% of 9 40 42 population) Source of revenue 8% of payroll (2% 8% of payroll (2% Government Premiums (waived for first two income worker, 6% employer, worker, 6% employer, revenue deciles, US$56 for decile 3, rising to plus US$11.10/month plus US$11.10/month US$887 for decile 10) plus national from government) from government) and state government revenue Revenue collection Employers and Employers and Premiums and government funding collected by state method government pay to government pay to health insurance agencies ISSSTE IMSS Number of risk pools One One 32–one per state Payment methods Capped budgets Diagnostic-related Capped budgets groups for inpatient, capitation for outpatient Benefit package Extensive package Extensive package 266 primary/secondary plus 17 tertiary interventions Facility coverage Own network: 1,123 Own network: 1,077 Specified institutions, mostly state government primary health care primary health care facilities, plus tertiary care at a network of higher-level units and 95 units, 215 general, tertiary care institutions operated by the federal and hospitals 41 specialty hospitals state governments Regulatory oversight Part of overall Part of overall IMSS Separation of insurers (payers) and providers at state ISSSTE scheme scheme tripartite level, leading to active purchasing with nonpublic Board includes board–labor, providers government and management, worker members government Source: Author. Making Health Financing Work for Poor People in Tanzania 49 The health ministries of Mexico’s 32 state governments are responsible for identify- ing, enrolling, and serving eligible SP participants. SP was designed to focus on the poor- est families �rst. Premium payments are subsidized on a sliding scale, and families from the poorest quintile do not pay. Participants are entitled to treatment at no additional cost through some 250 interventions at speci�ed institutions (mostly state government facilities). SP participants are also eligible for 17 interventions at a network of higher-level tertiary care institutions operated by the federal and state governments. The gap between premium revenue and the program’s total cost is covered by government subsidies. Most of the subsidies come from the federal to state governments, with federal funding cover- ing 83 percent of the premiums. There is also equalization between states. The program has required signi�cant increases in federal government resources allocated to health. Between 2004 and 2008, government health expenditure increased about 23 percent in real terms, while health spending as a percentage of GDP rose about 13 percent. The program design includes three related measures to increase the focus on the poor. First, premium rates are graduated by economic status, with exemptions for the poor. Second, mechanisms are in place to identify poor families and to determine pre- mium subsidies or exemptions. Third, there is a system for paying implementing agen- cies that gives them an incentive to focus on enrolling poor families. Families pay up to 5 percent of disposable income (de�ned as total spending after basic needs are covered). Further down the economic ladder, the premium falls and the government subsidy rises. Identi�cation of the poor uses mostly existing proxy-means test programs at the discre- tion of the state. About 40 percent of those enrolled are in the poorest quintile, compared with less than 3 percent in the population’s richest quintile. Thus, Yazbeck (2009, 252) notes that the SP program is very pro-poor. SP participants in need of health services were signi�cantly more likely to obtain them than were those who did not participate. Challenges for the SP program include the ongoing commitment and requirement for federal funds, and ways to encourage those above the poorest income quintile to pay the premiums required to participate in SP. Those people will constitute an ever-increas- ing proportion of the additional people who will have to be enrolled in the coming years if SP is to meet its objective of universal care. Box 4.8 highlights the main takeaway mes- sages from the Mexico experience. Box 4.8: Key Messages from Mexico • Substantial and sustained government financing will be needed to cover the poor. • It is a challenge to get those above the lowest income quintile to pay premiums. • Different benefit packages for contributory and subsidized plans can help to contain costs but may be attacked on equity grounds. • Capped budgets and case-based payment methods can help control costs. • Existing or consistent mechanisms for identifying the poor should be used if possible. 50 World Bank Study Summary of Lessons from International Experience This section summarizes the lessons learned from international experience by organiz- ing the speci�c lessons from each country into the key health �nancing functions of the collection of funds, the pooling of risks, and the purchasing of services, as well as the various regulatory issues. Regulation and the Political Environment Thailand shows that rapid scale-up of health insurance is possible if political will and popular support exist. However, it must be planned and executed properly. This lesson can also be drawn from Rwanda, where we learn (a) that strong government leadership and vision are essential to the short- and long-term sustainability of such reforms and (b) that strategies must be adapted to a changing macro- and health sector environment. Cultural and social factors, particularly solidarity within communities, also play a role in the acceptability and success of health insurance initiatives, although the initiatives are quite difficult to influence, which suggests that the health insurance approach se- lected must be adapted to such factors rather than the other way around. However, the Rwanda example also shows that any health insurance initiative must be supported by a strong regulatory framework at all levels. The China case reinforces the need for strong central regulation to ensure consistency in administration and delivery at the local level, and the Kenya case shows that the regulatory framework must include independent and transparent institutions. The Vietnam example shows that there can be efficiencies from shared administration between health and other social insurance programs. Various case studies, including Colombia, Ghana, Kenya, Mexico, and Rwanda, show that reaching the poor requires special efforts or approaches. The Colombia and Mexico examples suggest that, where available, existing mechanisms should be used for identi- fying the poor and that those mechanisms should be applied in a consistent manner. The Kenya case study shows that protection of the poor requires careful planning, including the identi�cation of the poor, incentives for providers to treat the poor, and funds to en- able premiums for the poor to be covered. The Rwanda example shows that provisions are needed for �nancial protection and other support for indigent populations. Revenue Collection The Colombia example shows that explicit cross-subsidies can be an effective way to �nance the poor but that �nancing from government will still be required. This message also emerges from the Mexico example, where substantial and sustained government �nancing is needed, especially because it is a challenge to get those above the lowest income quintile to pay premiums. The Ghana example suggests that expanding the rev- enue base through earmarked taxes (for example, VAT) can be an effective strategy for ensuring �nancial stability and coverage of the poor, but that it is essential to match the revenue received to the bene�t package. From Thailand we learn that although a rapid scale-up is possible and additional �nancing will be needed to cover the poor, this approach should not necessarily increase total health expenditures. Rwanda is a good example of the mobilization of donor �- nancing to support prepaid health insurance. However, it also shows that government coordination of donor funding is critical to ensure that aid is used effectively and aligned with national priorities. Making Health Financing Work for Poor People in Tanzania 51 In terms of improving revenue collection, the Vietnam example shows that the reach of the formal national health insurance organization can be extended by employing com- munity-based contractors to collect premiums. Pooling The Colombia example shows that health insurance improves access to both curative and preventive services and reduces both out-of-pocket and catastrophic health-related payments. Those results are echoed by the other case studies. From the Ghana example, we learn that although small, district-level risk pools are politically a ractive, the realities of health insurance operations require specialized expertise in revenue collection, contracting, and claims management. The results are consistent with the experience in Tanzania with respect to CHF/TIKA. The China ex- ample teaches us (a) that risk pools and administration at lower levels of government require capacity at that level and (b) that smaller risk pools can be problematic in terms of sustainability. Issues related to the identi�cation of the poor, the coverage of their premiums, and the inclusion of the poor in the risk pool are all important in the overall consideration of risk-pooling arrangements. Examples from Colombia, Ghana, Kenya, Mexico, and Rwanda all underscore this message. Purchasing The China example shows us that there is value in starting with a small bene�t package and increasing both the bene�ts and funding on the basis of consumer demand. Dif- ferent bene�t packages for contributory and subsidized plans can help to contain costs but may be a acked on equity grounds. This approach is used in a number of coun- tries, including China, Colombia, Mexico, and Thailand. However, regardless of the ap- proach chosen, the Vietnam example and that of other countries show that it is essential to match the bene�t package to available revenue. In terms of purchasing methods, Ghana, Thailand, and Vietnam all show that the use of fee-for-service payment methods should be avoided if at all possible, particu- larly for informal sector schemes where premiums are lower. Further, as demonstrated in Mexico and Thailand, capped budgets and case-based payment methods can help control costs. However, the example of Colombia shows that long transitions from sup- ply-side subsidies to performance-based payments can be problematic and that a more direct and immediate approach may be preferable. Regardless of the payment methods chosen, the Kenya and Rwanda examples suggest that development partner support should follow the shift from input- to output-based �nancing. Service Provision The experiences of Thailand and other countries suggest that it is essential to plan for the utilization growth resulting from the introduction of expanded health insurance cov- erage. Further, the Kenya example indicates that the functioning and efficiency of the health system must be addressed as an essential element of any scale-up strategy if one is to ensure that maximum value is obtained from the funds that already exist in the system. This example also suggests that conflicts of interest from having the MoH as both regulator and provider must be addressed. From Ghana, we learn that independent 52 World Bank Study controls and quality checks are essential for monitoring and evaluation of health facil- ity performance. Finally, the Vietnam example indicates that there should be a place for both public and nonpublic providers within a national health insurance arrangement. These various lessons should be borne in mind as health �nancing options for Tan- zania are examined in more detail in the next chapter. Notes 1. For further details, see appendix A. 2. For further details, see appendix A. 3. For further details, see appendix A. 4. For further details, see appendix A. 5. For further details, see appendix A. 6. For further details, see appendix A. 7. For further details, see appendix A. 8. For further details, see appendix A. CHAPTER 5 Key Policy Issues and Reform Options We summarize the analysis as follows: ■ Special efforts are required to reach the poor because they cannot afford ser- vices and often cannot afford prepaid health insurance. ■ Such special efforts to serve the poor and other members of the informal sector require incremental subsidies. ■ Such incremental subsidies can increase public spending by as much as 1.0 to 1.8 percent more than existing levels expressed as a share of GDP, although this percentage can be largely offset by reductions in out-of-pocket payments. ■ Capped, capitation, or case-based approaches or a combination are essential for cost control; fee-for-service should be avoided if at all possible. ■ If a country is to maintain a sustainable system, it is critical to match bene�ts to revenue, even if this approach means starting with a small bene�t package or different packages for different groups. ■ Given the �scal transfers required and the redistributive effect of major health �nancing reforms, political leadership is essential, as shown in the Ghana and Thailand examples. ■ Given the need for cost controls, a strong regulator is also essential for reform as shown in the Rwanda and Thailand examples. ■ Although small risk pools can enhance community ownership, the Ghana and Rwanda examples show that they can also create sustainability problems and administrative capacity issues, unless explicit cross-subsidization or reinsur- ance mechanisms or support systems such as consolidated revenue collection or claims management are put in place. It is clear that health reforms require tradeoffs that must be acceptable to the major- ity of the population. For example, meeting the goal of enrolling 45 percent of the popu- lation by 2015 requires confronting the costs of identifying, registering, and serving the informal sector and the poor. Increasing enrollment in those groups will require raising the awareness and understanding of the concept of risk pooling and insurance generally, as well as building trust in the institutions that offer insurance so that those who sign up are con�dent that the bene�ts offered will be received when needed. It is also likely that a broader range of bene�ts will need to be offered, especially low-use and high-cost services such as inpatient care, because the a ractiveness of the CHF bene�t package is limited by the relatively low cost of primary care. 53 54 World Bank Study Public and political support is required in other areas as well, whether it is (a) of- fering different bene�t packages for the formal and informal sector, (b) accepting in- creased autonomy for health care providers and allowing “money to follow the patient,� or (c) simply commi ing to the fundamental requirements of expanding coverage (for example, increased funding and effort). Therefore, although a number of options or sug- gestions are put forward, they need to be carefully reviewed to determine what is ac- ceptable within the Tanzanian context. This review will be an important next step in the development of the overall Health Financing Strategy. Areas to be Considered in the Health Financing Strategy With these words of caution in mind, a number of measures can be taken on the revenue and expenditure sides to improve health �nancing in Tanzania. This section considers the �ve health �nancing functions discussed above (service provision, purchasing, pool- ing of funds, revenue collections, and regulation and the political environment). Service Provision Three critical areas must be addressed on the service provision/expenditure side. First, existing inefficiencies and inequities in the health system must be reduced. Inefficiencies include (a) productivity losses caused by absenteeism and other factors, (b) supply chain problems, and (c) the uneven distribution of resources. The potential effect of cu ing the inefficiencies is large. For example, Kurowski and others (2003, 34–36) estimated that the productivity of front-line health workers was 57 percent, with patient productivity at only 37 percent. The potential productivity gain was 26 percent, which could translate into another 20 million outpatient visits on the basis of Tanzania’s current outpatient utilization. As noted in chapter 3, the CAG report showed signi�cant variation in staff productivity, with 80 percent of facilities seeing 10 or fewer patients per front-line health worker per day, thus suggesting excess capacity despite the overall lack of staff mem- bers in Tanzania (NAO 2008, 19). The Joint External Evaluation reported that “all [health facilities] still experience major delays in supply. In all visited regions and councils, the situation in hospitals has not improved as much as in [health centers] and dispensaries� (Denmark, Ministry of Foreign Affairs 2007, 69). This situation persists. For example, a report of the meeting of the Technical Commi ee of the Sector-Wide Approach (TC-SWAp 2011, 2) on February 9, 2011, stated, “Health facilities have been experiencing stock-outs of essential medi- cines and medical supplies a ributed by delayed delivery, nondelivery of right quanti- ties, and delivery of expired products or almost expiring.� Added to those issues are the current inequities in the distribution of resources. For example, in �scal 2010, the ratio of per capita budget allocations between the highest and the lowest funded regions was 2.6:1 (JAHSR 2010, slide 7-8 and author’s calculations). The CAG performance audit indicates that facility allocations for drugs and medical supplies are unrelated to the number of patients seen, with the average allocation per patient ranging from T Sh 51 to T Sh 1,218 (US$0.03 to US$0.81) (NAO 2008, 20). The �ndings suggest signi�cant potential for both stock-outs in some facilities and excess drugs in others, leading to wastage if drugs expire before they are used. The examples suggest that efforts to improve efficiency could increase the value and quantity of health services by 25–30 percent. This �nding is consistent with the �ndings of the World Health Report (WHO 2010, xvii). If the lower end of this range of efficiency Making Health Financing Work for Poor People in Tanzania 55 gains could be achieved, this accomplishment would create savings of T Sh 150 billion (US$100 million) of the government’s resources and T Sh 240 million (US$160 million) in total. Because both the causes and the potential remedies are already well known, what is needed now is both detailed planning and renewed commitment by all stakeholders to addressing efficiency issues. The elements should be reflected in any comprehensive health �nancing strategy. A second area of efficiency gains on the service provision or expenditure side is the need to review the management and governance structures of health facilities. Effective management and accountability structures are essential if any of the efficiency gains mentioned in the previous paragraphs are to be realized. In terms of other health �nanc- ing options, the ability of individual managers and health facility boards to respond to the challenges and opportunities of “money following the patient� will be closely cor- related with the success of these options. This approach means that oversight structures, including the MoHSW, Ministry of Finance, PMO-RALG, President’s Office–Public Ser- vice Management, regions, and districts, will all need to be prepared to let go of some of their current role in day-to-day management of health care service delivery at the facility level and to shift from “rowing� to “steering� the health system. Day-to-day decision making—including some control over staffing, allocation of funds to different types of expenditures, and even hours of operation—must be left to health facility managers and their board structures. Only by making this shift can facilities respond to the decisions made by patients enrolled in prepayment schemes regarding where they obtain their care. Facilities that offer be er services and are conse- quently preferred by patients should be able to respond to patient demand by changing their staffing mix or opening hours or by providing performance bonuses to their staff members. A third area of service delivery or expenditure is mechanisms for the allocation of drugs and medical supplies to facilities, so that the distortions mentioned in the CAG report do not continue. Those issues will become more important over time, as more people register with prepayment schemes in the expectation that drugs and medical supplies will be available. Lack of supplies at lower level public facilities deters CHF membership. Therefore, mechanisms need to be developed to translate the registrations into the availability of drugs and medical supplies. As a minimum, the allocations to facilities for the supplies should be related to the workload: either the number of patients seen or the number of patients registered with the facility, or a combination of both. One approach may be to extend the concept of the matching grant to drugs and to medical supplies as well, so that each person or household registered with a CHF would generate an allocation to his or her home facility (either an actual funding transfer or notional allocation that can be drawn upon). This allocation would be similar to the “fund-holding� approach that has been used in Latvia and the United Kingdom. The small size of risk pools (individual facilities) could increase costs, but this situ- ation could be mitigated by establishing a mechanism that would enable facilities to continue to get drugs even if their allocation was exhausted. There would also need to be the opportunity to top up the allocation from the central budget with the facility’s own resources, whether from user fees, CHF prepayment revenue, Health Basket Fund allocations, or other sources. This approach implies a much more interactive relationship between health facilities and the Medical Stores Department, in terms of both the flow of 56 World Bank Study funds and the ordering and delivery of supplies. Of course, in addition to incorporating such provisions into the mechanism for �nancing drugs and medical supplies, health fa- cilities would also need to have the discretion to make such allocations, which links this issue directly to the one of management and governance discussed earlier. They would also need their own bank accounts so that they could readily access the available funds. Purchasing The case studies show that fee-for-service models should be avoided if possible. In the current health �nancing structures, neither the CHF nor the NSSF use this approach, whereas the NHIF does. Yet the NHIF represents the largest insurer in terms of both population coverage and funding. It is, therefore, important (a) to evaluate how alterna- tives to fee-for-service could be implemented within the NHIF, as well as other insur- ers, and (b) to develop action plans for shifting to an alternative approach over time. Although this approach would generate signi�cant resistance from providers, interna- tional experience shows that moving away from fee-for-service is an important element in ensuring the long-term sustainability of the health �nancing system. Similarly, the effect of the per diem system used by the NHIF for inpatient care should be evaluated, and the feasibility and potential bene�ts of moving to case-based reimbursement should be reviewed. Following this examination, a decision should be made about whether to use per diem or case-based reimbursement for inpatient hospi- tal services, and an appropriate action plan should be developed. Considerations here would include not only the incentives for over-provision of care inherent in the per diem system and the extent to which the incentives affect care, but also the ability of facilities providing inpatient care to cope with case-based approaches. Appropriate regulations and administrative mechanisms would be needed to ensure cost containment and to set suitable tariffs for both public and private facilities. Although a diagnostic-related group system such as those used in Europe and North America is probably not feasible, there are examples of more simple systems such as the one currently used in Ghana or that in Bulgaria (Georgieva and others 2007, 101–3). The incentives inherent in case-based sys- tems (re-admission, unnecessary admission) would also need to be considered. Risk Pooling The risk pools in the current health �nancing system are fragmented, with different schemes for the formal public sector, the formal private sector, and the informal sector. In addition, the current government contribution to prepaid health �nancing currently goes to those who are relatively be er off than the general population. In this section, op- tions are explored for reducing this fragmentation and for directing government funds to areas where they more clearly bene�t the poor and vulnerable. As the largest single risk pool, the NHIF can serve as the anchor of any new risk- pooling regime. However, for it to assume this role, its revenue and expenditures need to be rationalized. Currently, the sum of the bene�ts paid plus administrative costs is only 35 percent of the premium revenue collected. Although bene�ts could be increased, this approach would have a distortionary effect relative to other schemes, perhaps lead- ing to a situation such as that in Thailand, where the bene�t packages for the formal sec- tor schemes are signi�cantly more generous than the universal scheme, with resulting budget pressures and inequities. Even with the recent increases in the payout percent- age, the matching contribution is not required to provide current bene�ts. Making Health Financing Work for Poor People in Tanzania 57 One approach would, therefore, be to change the NHIF to a subscriber-only �- nanced scheme, with a contribution rate of 2.5 percent of salary. Subscribers would, therefore, save 0.5 percent of salary, while the government would save its entire contri- bution (about T Sh 47.4 billion in 2010/11), which could be applied to matching funds to help support the expansion of CHF membership. Although this application would still provide the NHIF with a bit of a cushion to absorb future utilization growth, the NHIF would need to strictly control its operating costs and to stop �nancing noninsurance- related activities (for example, purchasing equipment and �nancing renovations) out of operating income. There are signi�cant needs for improvement in health infrastructure and other HSS investments. The NHIF currently makes some of these types of investments, but the efforts are problematic in two respects. First, they are not strictly related to the core business of the NHIF, which is providing prepaid insurance to its subscribers. Second, the extent to which the investments are coordinated with the government’s overall HSS efforts is not clear. Thus, while funding such investments out of operating funds is not advisable, the accumulated surplus represents a unique opportunity to support the overall HSS initiative, thereby improving health facilities for NHIF members and non- members alike. One way to do this would be to redirect T Sh 250 billion of the estimated T Sh 300 billion in accumulated surplus toward a special Health System Strengthening Account within the NHIF, which could help �nance the improvements and expansion of the health system that will be needed to accommodate improved insurance cover- age. Planning for the use of the funds could be done through a joint process that would include the NHIF, the MoHSW, the PMO-RALG, the development partners, and other stakeholders. The account could also receive contributions from other sources, including development partners. Such an account would represent a signi�cant legacy by past and current NHIF members to the health system as a whole, as well as a tribute to the �rst 10 years of NHIF operations. The remaining T Sh 50 billion would still be sufficient to cover 2.5 years of claims at current expenditure levels and should represent a suitable reserve, given that the NHIF is backed by the government and is a going concern. It is worth noting that established health insurance funds, such as that in Estonia, have reserves of just 8 percent of rev- enue, “legal reserves of 6 percent of its budget, and risk reserves of 2 percent� (Go ret, Schieber, and Waters 2008, 244). The prepayment scheme for the formal private sector is currently being run by the NSSF as one of the seven bene�ts that it provides. A review of the pension system done by the World Bank determined that the current pension system was unsustainable in its present form and that signi�cant reforms would be needed. One option is to separate the �nancing of the pension and non-pension-related bene�ts. For the NSSF, this option would include the health insurance bene�ts (SHIB). The latest actuarial review of the NSSF determined that the current SHIB could be �nanced through 2.5 percent of pay- roll, which is consistent with what has been proposed for the NHIF (see earlier; also see Tanzania NSSF 2010). Rather than continuing to run the SHIB as a separate program within the NSSF, one approach would be to remove the SHIB from the NSSF and to provide coverage through the NHIF, with a total premium of 2.5 percent of salaries paid by employees. This ap- proach would make the scheme for all formal sector workers the same and should result 58 World Bank Study in administrative efficiencies, especially if participation in the NHIF was made mandato- ry for such workers. As noted in the previous section, a review of the provider purchas- ing approaches should determine which purchasing method is the most cost-effective for different types of care. The CHF/TIKA is the other main prepayment scheme at present, with a number of issues as described in chapter 3. Initiatives are under way as part of the CHF Action Plan and, through various development partners, to train district-level staff members in health insurance operations, but another option is proposed in this policy note. Overall, it may be more cost-effective and appropriate to have CHFs focusing on community mobilization and raising revenue and to have the NHIF act as the claims processor, card issuer, and funds manager. Such an option will allow both the NHIF and the CHF to focus on their areas of comparative advantage, and the option should increase the value of CHF membership to community members, especially if provisions for portability of bene�ts between facilities and eventually between districts and regions can be built into the overall structure. One of the difficulties with the present CHF approach is that the premium payments and bene�ts vary by district. This variance could be addressed by establishing uniform CHF rates at T Sh 30,000 per household per year for urban members and T Sh 15,000 for rural members. The rates reflect both the differing ability to pay (as measured by aver- age income levels) and the differences in potential bene�ts that would accrue to urban and rural residents. To justify the new rates, the bene�t package for CHF/TIKA would be expanded to include �rst-level inpatient (for example, district hospital), as well as outpatient care in public facilities. To help offset some of the ability to pay differences, matching funds could be set at different levels, perhaps 100 percent for urban members and 150 percent for rural members. The matching grants could be used to (a) top up the contribution of paying subscribers (resulting in total revenue of T Sh 48,000 per house- hold for urban and T Sh 22,500 per household for rural) and (b) pay the contributions of the lowest 20 percent (urban) or 40 percent (rural) of the population (percentages that are based on the poverty incidence in urban and rural areas). Savings from the NHIF changes could be used to provide the matching funds. Initial calculations reveal that this provision would be sufficient to achieve 45 percent health insurance coverage for the entire population (see information that follows). As in the current arrangement, the matching grant would be driven by the number of people who voluntarily enroll, but the grant would bene�t not just the household that enrolls but other poor households in the same community. This arrangement provides some type of “micro-solidarity� (“if I enroll, some of my poorer neighbors will also be able to be enrolled�), which could help to encourage the overall enrollment efforts, especially if explicit, uniform, and transparent ground rules are in place. The creation of larger, more stable risk pools should help to ensure the ongoing sustainability of the CHF/TIKA- based approach. Revenue Generation Special efforts are needed to raise the participation of the informal sector, including subsistence farmers, in prepayment schemes. This need is due to (a) the poverty in the informal and rural sectors, which prevents people from making payments; (b) the high vulnerability to adverse events such as drought and crop failure, which require frequent Making Health Financing Work for Poor People in Tanzania 59 use of �nancial reserves and prevent many from participating in prepaid programs; and (c) the lack of understanding of prepayment schemes and of the bene�ts of risk pooling. From the revenue generation standpoint, an important element is the need for flex- ible payment approaches for CHF/TIKA membership, which reflect the irregular income pa erns of informal sector workers and subsistence farmers. In addition, there is a need for a signi�cant amount of awareness raising and communications regarding the chang- es and the bene�ts that can be obtained by enrolling in a prepayment scheme (as well as the concept of risk pooling) to ensure there is an appropriate understanding of how funds will be managed and used. This scheme should be an essential element of the overall implementation of the Health Financing Strategy. Another potential revenue generation option would be to use budget funds to sub- sidize premiums for the informal sector and the working poor, thereby allowing them to enroll in prepayment schemes and leveraging existing budget resources to promote access to services. If one is to �nance outputs rather than inputs and to have the money follow the patient, it should be possible to obtain be er value from the funds that are already spent in the health sector and to encourage the efficiencies and quality improve- ment mentioned in the section on service provision. Health providers would get paid only when they provide services or encourage subscribers to sign up with their facility, so this approach to �nancing should encourage health facilities to provide high-quality services. Of course, the changes in the management and accountability structures men- tioned earlier would need to be in place to allow providers to respond to those incen- tives. Complementary pay-for-performance schemes at the health worker level—such as those currently being piloted—could also be useful in promoting effective use of service and improved quality. Despite improved efficiency and the rationalization of existing revenue sources, including the NHIF, the health system will likely require additional funding at some point. It would, therefore, be useful to look at other sources of sustainable �nancing for the health sector. Although the discussion on �scal space concluded that there would be limited capacity for increasing government funding for health care, the realities of im- proving social health protection clearly indicate that additional funding will be needed. Other countries have looked (a) at targeted taxes such as “sin taxes� on cigare es and alcohol or (b) at earmarking a speci�c proportion of particular taxes to support sustain- able social health protection improvements. As noted earlier, Ghana currently allocates 2.5 percentage points of the VAT to the NHIS, and the amount generated goes up automatically when spending on goods and services increases. The VAT revenue is the major source of funding for the NHIS. Simi- larly, Latvia allocates a �xed proportion of income tax to the health insurance scheme, which does not charge premiums to subscribers. In the Tanzanian context, 2.5 percent- age points of the VAT is equivalent to approximately T Sh 200 billion, which would go a long way toward ensuring universal health coverage for the population. Further, as seen in �gure 3.8, the VAT is progressive in Tanzania and would not place the entire burden on the formal sector. Any revenue raised through this source could help increase cover- age, starting with the poor. Finally on the revenue side, there is a need to review user-fee policies to be er tar- get the poor for exemptions, waivers, or premium subsidies. User fees are currently a signi�cant source of discretionary revenue for health facilities, but they are also clearly a major barrier to accessing care, especially for the poor for higher-level services (see 60 World Bank Study �gure 3.11). It is, therefore, essential that the effect of user fees on the poor be mitigated as much as possible, but such fees should also be set to promote enrollment in prepaid schemes by those who are not exempted. Other countries have put in place clear and transparent means to identify the poor (such as using existing targeting mechanisms for social bene�ts in China, Colombia, Mexico, and Vietnam). Similar approaches need to be considered in Tanzania. The new National Social Protection Framework, as well as the conditional cash transfer scheme currently being piloted under the Tanzania Social Action Fund, provide both an opportunity and some potential lessons for developing harmonized targeting mechanisms, which should then also be used to identify the poor and near-poor for premium waivers or subsidies. It is also important that exemptions and waivers be “revenue neutral� to the facility, so that a facility’s �nancial position is not adversely affected by applying the user-fee policy appropriately. This approach means that—to the extent that user fees are retained as a means of raising revenue—facilities should be compensated for the exemptions and waivers that they provide. Of course, this plan would require both objective, veri�able criteria to be used and ex post veri�cation to be performed. Because exemptions and waivers are part of the overall social protection system, this compensation should be provided by the government, possibly through the reallocation of existing resources (again shifting from �nancing inputs to outputs) or with the support of development partners. The role of user fees in an environment of increasing prepaid health insurance coverage also needs to be determined. Regulation and the Political Environment A comprehensive regulatory framework must be in place to govern the health �nancing system. Although the SSRA Act has been passed and a director has now been appointed, the extent to which this authority will interact with the health sector in general and health insurance in particular (either public, private, or both) has yet to be determined. The Technical Working Group on Health Financing is conducting further analysis of the regulatory framework, including the following areas: ■ Premiums and subsidies ■ Reserve requirements and investment regulation ■ Contracting, including the selection of providers ■ Bene�t package determination, tariff se ing, and provider payment methods ■ Licensing, accreditation, and governance for health providers ■ User-fee levels, waivers, and exemptions The various provisions should be integrated into a comprehensive and mutually supportive framework of legislation to guide the overall operation of the health �nanc- ing system. A critical element of health �nancing reform is political commitment to change. Such political leadership was instrumental in the successes of some of the case studies in chapter 4, especially in Ghana and Thailand. As shown in �gures 3.9 through 3.11, some elements of society are bene�ting more than others from the status quo, which means that any changes to increase equity and coverage could be seen as coming at the expense of those who are currently bene�ting. For example, even though the funding that is cur- rently being raised is not used for providing health services, the potential changes iden- ti�ed for the NHIF may be seen by some members as making them worse off. Therefore, Making Health Financing Work for Poor People in Tanzania 61 it is clear that garnering the appropriate political commitment is a critical element of any health �nancing strategy. Economic and Financial Impact of Reform Options Some of the reforms described here were assessed using a “scaling up� model developed in the SHIELD project.1 Three options were simulated using the model. First, a status quo option was calculated based on the assumptions that (a) there would be no major changes in the parameters of health �nancing, and (b) there would be no signi�cant initiatives to encourage enrollment in prepaid schemes. This option therefore shows the effect of population growth alone on the demand for health services. This information provides an important baseline because the population is expected to grow by 51 percent or 22 million people between 2010 and 2025, reaching 65.3 million by 2025. This option also assumes that the current efforts at HSS—including investments in the development of health human resources, M&E, and infrastructure—would be suf- �cient to accommodate the growth in population and services. Such efforts are expected to cost T Sh 843 billion over the projection period, with most of the investments front- loaded into the �rst �ve years of the simulation. Of the total amount, T Sh 321 billion is expected to be provided by income from foreign donors, including the Global Fund, the United States, Canada, the Netherlands, Norway, and Japan, while T Sh 250 billion is expected to be provided from the NHIF surplus, and the balance (T Sh 272 billion) would come from the government’s contribution to the MMAM and other infrastructure development. A second simulation was done to determine the �nancial and utilization effect of reaching the stated objective of covering 45 percent of the population with prepaid health insurance by 2015. This “45 percent target option� (also referred to as Option 1) incorporates the changes mentioned earlier in terms of (a) premium levels and incidence for the NHIF, (b) use of the NHIF surplus, (c) replacement of SHIB, and (d) CHF/TIKA coverage and premium levels. In addition to these changes in the �nancing parameters, it was assumed that inpa- tient utilization by CHF/TIKA members would grow as such services are added to the bene�t package and that both inpatient and outpatient utilization for NHIF members would grow as the number of accredited facilities increased. Compared to the status quo, this option recognizes that this projected increase in services will require additional investment. Thus, a total of T Sh 1.8 trillion is provided over the 15 years of the simula- tion (average of T Sh 120 billion per year), essentially extending the level of HSS in the �rst �ve years of the status quo option throughout the life of the projection. Finally, a universal coverage option (Option 2) was run, again using the same ba- sic parameters. However, in this option, it is assumed that the entire informal sector is covered through either the CHF (rural) or the TIKA (urban) scheme, while 77 percent of the private formal sector is covered through either private insurance or the NHIF. The increases in coverage will be phased in over 10 years. Speci�c utilization assumptions mirror those of Option 1. As will be shown next, this option will generate signi�cant increases in services. To accommodate them, one must assume that the total HSS invest- ment in this option will be T Sh 2.6 trillion between 2011 and 2015, or an average of T Sh 170 million per year. 62 World Bank Study Each option is assessed using different assumptions regarding economic and gov- ernment budget growth, with mid-range, low-growth, and high-growth scenarios. For the ease of presentation, the results shown next include only the mid-range and low- growth results, thereby providing an expected and a pessimistic scenario of the eco- nomic and �nancial effects. Population Coverage Figure 5.1 shows the expected total population coverage with the three scenarios, as- suming a 5-year scale-up for Option 1 and a 10-year scale-up for Option 2. As shown, Option 1 tops out at 45 percent of the total population, while Option 2 reaches almost 90 percent. The status quo option keeps total coverage for about 10 percent of the popula- tion throughout the simulation period. Figure 5.1: Coverage with Prepaid Health Insurance Source: Author’s calculations using Ifakara “scaling up� model. In terms of absolute numbers, enrollment in the NHIF and CHF/TIKA is expected to grow from 2.0 million and 1.5 million in 2010 to 3.2 and 3.3 million for the status quo option, 5.7 million and 23.8 million for Option 1, and 11.3 million and 46.7 million for Option 2. Health Care Services Figure 5.2 shows that the total number of outpatient visits in public facilities will in- crease from 65 million to almost 100 million in the status quo option, 160 million in Option 1, and 227 million in Option 2, while the number of required beds in public hospitals is expected to grow from 26,000 currently to 40,300 in the status quo option, Making Health Financing Work for Poor People in Tanzania 63 52,600 in Option 1, and 63,200 in Option 2. The number of required beds was estimated using an assumed average occupancy of 70 percent and current average lengths of stay. In each option, there would be an additional 17,200 FBO and private beds, increasing to 25,000 for the status quo option, 22,700 for Option 1, and 19,700 for Option 2. The lower requirements for nonpublic beds in Options 1 and 2 reflect the assumption that the CHF/ TIKA bene�t package will cover public facilities only. Although it would be possible to use FBO and private facilities for some of the increased utilization, issues such as the higher cost structure of such facilities and the scope for facility expansion would need to be addressed. Figure 5.2: Service Growth in Public Facilities Source: Author’s calculations using Ifakara “scaling up� model. Overall Economic Impact The total health expenditure covered by the model is estimated at 4.5 percent of GDP in 2010. Because there are expenditures in other areas not covered by the model, such as off-budget donor �nancing, a direct comparison between this percentage of GDP and the National Health Accounts data is not possible. However, the �gures can be reli- ably compared year over year as shown in �gure 5.3. Assuming that current economic growth projections continue, all options show an initial bump in expenditure as the HSS funding kicks in and then a gradual decline in the percentage of GDP spent on the ser- vices to a level below 4.5 percent. In fact, even with lower economic growth, all options except for Option 2 result in health expenditures as a percentage of GDP returning to current levels by 2020. 64 World Bank Study Figure 5.3: Modeled Health Expenditure as a Percentage of GDP Source: Author’s calculations using Ifakara “scaling up� model. However, the shift to prepayment schemes with signi�cant support for the poor through the matching grant will greatly reduce direct out-of-pocket expenditure. As can be seen in �gure 5.4, private payments are expected to decline as a percentage of total expenditures while public expenditures (including on-budget donor �nancing) are pro- jected to increase. Figure 5.4: Modeled Health Expenditures by Financing Source Source: Author’s calculations using Ifakara “scaling up� model. Making Health Financing Work for Poor People in Tanzania 65 In the status quo option, private expenditure is expected to increase from 37 percent to 40 percent, while public expenditures are expected to decrease from 63 percent to 60 percent. In Option 1, by contrast, private expenditures are projected to decline to 30 per- cent, while Option 2 estimates that private expenditures will be just 18.7 percent of total health expenditures. The public share is estimated to be 70 percent for Option 1 and 81.3 percent for Option 2. Government Budget Impact The increasing share of total health expenditures flowing through the public purse is examined next (�gure 5.5). The status quo option indicates that despite a slight increase in public health care expenditure as a percentage of GDP and the total government bud- get as the HSS investments kick in, both the mid-range and low economic growth op- tions see expenditures quickly return to 2010 levels and dip even more throughout the projection period. Public health care expenditures are expected to dip below 2 percent of GDP and 6 percent of government spending, assuming mid-range economic growth, and then to drop to 2.2 percent of GDP and 7 percent of government spending under a lower-growth scenario. Looking at Option 1, public health care expenditure tops out at 3.5 percent of GDP under the lower-growth assumption and 3.2 percent under the mid-range assumption, while spending as a percentage of the government budget is expected to reach a maximum of 10.8 percent for the lower-growth scenario and 10.1 percent in the mid-range scenario. In both scenarios, spending as a percentage of both GDP and the government budget de- creases to close to or below 2010 levels by 2025. As can be seen from �gure 3.2, this level of variation is within the range of movement experienced over the past �ve years or so. With a signi�cant amount of both service growth and support for health insurance premiums, Option 2 results in higher sustained levels of �nancing, staying around 3.5 percent of GDP and 11 percent of government expenditure for the mid-range economic growth scenario (average real GDP growth of 6.6 percent) and 4.0 percent of GDP and 13 percent of government expenditure in the lower-growth scenario (average of 5.6 percent). It is worth noting that both of those levels of �nancing are still below the Abuja targets. Figure 5.5: Public Health Expenditures as Percentage of GDP and Government Budget A. Public as share of GDP B. Public as share of government Source: Author’s calculations using Ifakara “scaling up� model. 66 World Bank Study It has been mentioned frequently that there is signi�cant scope for efficiency gains in the delivery of publicly �nanced services. Figure 5.6 shows the effect of various effi- ciency scenarios on health care expenditures as a percentage of the government budget. In Option 1, even a 10 percent efficiency gain will result in expenditures returning to the current level by the end of the projection period on the basis of lower economic growth; this level will be reached by 2018 if mid-range growth is achieved. An efficiency gain of 20 percent will result in current expenditure levels being achieved in 2022 in the mid- range growth scenario and 2017 in the lower growth scenario. Because the health system strengthening and matching grant funds are not affected by efficiency gains, the initial increase in expenditures is inevitable regardless of the efficiencies obtained. Panel B shows the potential effect of efficiency gains on Option 2. Here both the pro- longed HSS requirements and the substantial matching grant requirements limit what can be done, although the effect relative to the base case can still be signi�cant. For ex- ample, using a lower economic growth scenario, health as a percentage of government spending would be reduced by up to 1.1 percentage points if 20 percent efficiency gains are achieved and up to 1.6 percentage points if 30 percent efficiency gains are possible. Translated into real numbers, this reduction would mean between T Sh 220 billion and T Sh 315 billion in savings annually once the full efficiency gains were realized in 2020 (a 10-year phase-in period is assumed). Even a 20 percent efficiency gain in a lower growth scenario will result in health expenditures being kept in the 11–12 percent range seen in �gure 3.2 for a number of years. Figure 5.6: Effect of Ef�ciency Gains on Expenditures as Percentage of Government Budget A. Effect on Option 1 B. Effect on Option 2 Source: Author’s calculations using Ifakara “scaling up� model. Incremental Costs of Each Option Table 5.1 highlights the additional costs to the government of Tanzania for each of the two scaling-up options. This �nding assumes that the current level of donor support grows by an average of 3 percent annually in real terms and that the already pledged HSS funds materialize. In the absence of speci�c additional funding commitments, all of the remaining costs are a ributable to the government. Making Health Financing Work for Poor People in Tanzania 67 Table 5.1: Incremental Costs by Type Total for Period ($ million) Annual Average for Period ($ million) 2011-2015 2016-2020 2020-2025 2011-2015 2016-2020 2020-2025 Option 1 Matching Grant 0.0 0.0 41.8 0.0 0.0 8.4 Systems Strengthening 90.7 270.7 390.7 18.1 54.1 78.1 Subtotal Transition costs 90.7 270.7 432.4 18.1 54.1 86.5 Service Delivery 214.6 998.5 1,344.0 42.9 199.7 268.8 Total 305.3 1,269.2 1,776.4 61.1 253.8 355.3 Option 2 Matching Grant 0.0 132.5 432.1 0.0 26.5 86.4 Systems Strengthening 90.7 421.0 641.2 18.1 84.2 128.2 Subtotal Transition costs 90.7 553.5 1,073.3 18.1 110.7 214.7 Service Delivery 62.6 1,474.0 2,676.6 12.5 294.8 535.3 Total 153.3 2,027.5 3,750.0 30.7 405.5 750.0 Source: Author’s calculations using Ifakara “scaling up� model. This table provides U.S. dollar estimates of both the �ve-year interval and average annual costs, and it separates the costs for each option into transition costs and service delivery costs, with transition costs including the incremental matching grant and the portion of the HSS costs not covered by donors or other sources. The estimates take into account both the use of the saved NHIF premiums for the matching grants and the use of the excess NHIF reserves for HSS. In the absence of those funds, the funding require- ments would be signi�cantly higher. Table 5.1 shows that—even in Option 2—the transition costs for the next �ve years would be relatively modest, although the use of the NHIF funds is largely responsible for this result. Beyond 2015, annual transition costs increase to an average of US$54 to US$118 million per year for Options 1 and 2 until 2020, and then US$87 million to 215 million per year for the last �ve years of the projection. Clearly, the “transition costs� do not necessarily have to be borne solely by the government, because both overall develop- ment partner interest in increasing �nancial health protection, as well as the time-limited nature of the investment costs, could make the costs a ractive investments for increased donor �nancing. This approach would help to improve the overall affordability of this option to the public purse. Sensitivity Analysis This section presents a sensitivity analysis with respect to economic growth assumptions and real unit-cost growth assumptions. In the model, economic growth is estimated at 6 percent in 2010, 7 percent in 2011, 8 percent for 2012 and 2013, 7 percent from 2014 to 2018, and 6 percent for the remainder of the projection period. This estimate is referred to as the “mid-range projection.� The lower-growth scenario reduces the growth estimates by one percentage point below the mid-range level each year. For the purposes of this sensitivity analysis, real economic growth was set to 4 percent per year. 68 World Bank Study Figure 5.7 shows the effect of this change with total health expenditures as a per- centage of GDP (panel A) for the status quo option being 1.34 percentage points higher than the mid-range projection, with Option 1 being 1.57 percentage points higher, and with Option 2 being 1.77 percentage points higher and heading to almost 6 percent of GDP by 2021. The effect on health as a percentage of government expenditures is simi- larly dramatic, with Option 2 reaching 15 percent (the Abuja target) in 2021 and staying there. For Option 1, health expenditure still hovers around the 11 percent mark for most of the projection period, and lower economic growth keeps the percentage of the govern- ment budget around the current level for the status quo option, once the HSS activities are completed. Figure 5.7: Sensitivity Analysis—Impact of 4 Percent Real Economic Growth A. Effect on THE as % of GDP B. Effect on health as % of government spending Source: Author’s calculations using Ifakara “scaling up� model. Note: THE = total health expenditure. The �nal area to be examined in terms of sensitivity analysis is the cost inflation for health care services. The base case model assumes that annual unit cost increases of 2 percent per year in real terms. The sensitivity analysis doubles this assumption to 4 percent annually. Figure 5.8 shows that unit cost changes also have an effect on pub- licly �nanced health expenditure as a percentage of the total government budget. By the end of the projection period, the status quo options would be expected to require 1.93 percentage points more of the government budget than the base case scenario, all other things being equal. The effect for Option 1 is an increase of 2.38 percentage points, while the effect for Option 2 is 2.83 percentage points. Each 1 percent increase in real unit cost growth will require an average of a 15.1 percent larger share of the government budget for the status quo option compared with the base case scenario, a 13.8 percent larger share for Option 1, and a 12.9 percent larger share for Option 2. The lower effects for Options 1 and 2 are a result of the signi�cant investment in HSS, which is not affected by the service unit cost growth. Making Health Financing Work for Poor People in Tanzania 69 Figure 5.8: Sensitivity Analysis—Real Unit Cost Growth from 2 Percent to 4 Percent Source: Author’s calculations using Ifakara “scaling up� model. Summary and Next Steps The simulations show that the extension of prepaid health insurance coverage to a larger segment of the population should not cause signi�cant increases in either the percent- age of GDP spent on health services or the proportion of the total government budget devoted to such services, as long as robust economic growth continues. A lower growth assumption results in health expenditures as a percentage of GDP increasing from 4.5 percent to 5.0 percent in the universal coverage option and increasing as a percentage of the government budget from 9 percent to 13 percent, still well below the Abuja target. It would take a substantial slowing in economic growth (to 4 percent annually throughout the projection period) to push health expenditures to 6 percent of GDP and 15 percent of government expenditures. Even in the absence of any reforms, there will be signi�cantly increased demand for health services as a result of population growth, with about 53 percent more services projected by 2025 than are currently being provided. This projection underscores the need for continued a ention to HSS, especially in terms of health human resources, but it also highlights that efficiently using the available human and �nancial resources will become increasingly important. The simulation results indicate that efficiency improve- ments can help to mitigate the cost implications of scaling up prepaid health insurance coverage. Given the strong commitment of a number of development partners to improving social health protection, it should be possible to generate interest in helping to �nance the transition and perhaps other costs associated with this expansion of coverage. Of course, a credible and sustainable action plan and detailed costing will be needed to garner such support, not just from development partners but also from the government, which will likely bear the greatest proportion of any cost increases. The health �nancing 70 World Bank Study strategy and the accompanying action plan should provide the basis for these types of discussions. This policy note provides background information, cross-country examples, and policy options, which can all be incorporated into the development of a comprehensive health �nancing strategy. It also provides a framework for looking at the various ele- ments of the health �nancing system and explores the �nancial, economic, and health system implications of a number of the options. A key aspect in terms of moving forward is determining the acceptability of the various options from a political economy perspective, as well as more in-depth con- sideration of the feasibility and the expected costs. This examination should assist in �nalizing the health �nancing strategy and lead to the identi�cation of speci�c policy actions and reforms that will be needed to operationalize the strategy and its various components. Given the nature of the policy actions and the reforms they would support, it may be possible to consider them for inclusion in future development policy opera- tions, either at the GBS level, or as a separate sectoral development policy operation. By drawing on the analysis just described, speci�c support could be targeted at the transi- tion costs to universal coverage, notably the following: ■ Se ing in place an appropriate regulatory framework to monitor and provide oversight to prepaid health insurance operations (both public and private), bearing in mind the evolving role of the SSRA ■ Addressing the role and future of the SHIB, consistent with the pension reforms that are also being pursued ■ Making substantive changes to the �nancing framework for the NHIF, along the lines described in this policy note ■ Focusing on the organizational and operational changes required for the CHF/ TIKA to allow them to play the role envisioned in this policy note ■ Addressing the need for increased efficiency and provider autonomy Health �nancing reform of any signi�cance will create major changes in the way in which the health sector operates. Because of this phenomenon, it is important that the overall vision for health �nancing, its subsequent effect on the health sector, and the spe- ci�cs of any proposed changes are fully discussed with all of the relevant stakeholders before proceeding. The experience of other countries, most notably Kenya, shows that without fully explaining and discussing proposed changes, such reforms can be doomed to failure. It is recommended, therefore, that consultation and advisory mechanisms be put in place immediately to facilitate this essential dialogue. The explicit signaling of high-level support for health �nancing reform is also a critical next step in the process. Notes 1. See Borghi, Mtei, and Ally (2010), for key assumptions, model structure, etc. APPENDIX A Health and Financing Systems in Other Countries Thailand1 In Thailand, health �nancing reform became a key issue in the 2001 general election campaign, and it has been reported to be an important factor in the election of the Thak- sin government. In addition to strong political leadership, there was also strong civil society support for what was known as the “30 baht� scheme (representing the amount of the copayment, equivalent to US$0.75). Thailand had developed a good geographical coverage of health care infrastruc- ture nationwide, primarily through facilities �nanced by the Ministry of Health, and a number of public �nanced schemes had been in operation as far back as 1974. The key milestones included (a) fee exemptions for the poor and a medical scheme implemented for private sector workers in 1974, (b) the development of a low-income scheme for the poor in 1975, (c) a medical scheme for civil servants (CSMBS) in 1985, (c) a social security scheme (SSS) with health insurance components in 1990, and (d) a health care scheme for the near-poor in 1991. By October, 2000, public �nancing covered 70 percent of the population with the four main schemes, and there was a great deal of accumulated ex- perience in the management of the various health insurance schemes, including (a) local fund management, (b) public capitation contracts, and (c) development and testing of a diagnostic-related group system. Implementation of the transition period toward universal coverage (UC) started in April 2001 with the enactment of the National Health Security Act using the Ministry of Public Health mechanisms and infrastructure. This implementation included realloca- tion of the existing health budgets with an additional B 10 billion from government for scheme implementation. Door-to-door approaches were used to sign up bene�ciaries, and the new scheme was scaled up nationwide within one year, going from 1.39 million enrollees in April 2001 to 44.5 million in April 2002. There was some resistance from health care providers, but it did not hold up implementation of the scheme. The major design principles include (a) ge ing to universal coverage as quickly as possible for those not already covered by existing schemes, (b) ensuring that cost control mechanisms are built into the UC scheme, and (c) including a minimal payment toward each patient’s care (B 30, or US$0.75 per visit—Thailand is a middle-income country), although this copayment has since been dropped, with minimal effect on utilization. Premiums for SSS are collected through payroll deduction and submission of both the employer and employee contributions to the Ministry of Labor Social Security Of- 71 72 World Bank Study �ce. Funds from the Ministry of Finance for their contribution are also paid into the SSS account. There are no premiums for either the CSMBS or the UC schemes, and the necessary allocations are transferred directly to the plan administrators by the Ministry of Finance. Different mechanisms exist for risk pooling in each of the schemes. The SSS uses a single risk pool consisting of 10 million people. Similarly, the CSMBS is a single risk pool of 5.7 million people, and the risk pool for the UC scheme is 47 million. Payment methods vary by scheme and type of provider. The SSS uses a capitation contract model for both inpatient and outpatient care, while the CSMBS uses fee-for-service reimburse- ment extensively. The use of this payment method is felt to be a contributor to the high cost of this scheme (the cost of bene�ts for CSMBS are equivalent to about US$400 per person while bene�ts for UC are about US$60 per person). The UC scheme uses capitation for primary care and uses diagnostic-related groups that are global budget capped for hospital care. Public health care providers are predominant, although the share of private provid- ers is increasing. Private hospitals constituted about 24.4 percent of total hospitals and 20.2 percent of total hospital beds in 2005. Both public and private hospitals can partici- pate in the UC scheme, providing that they are willing to accept the agreed tariff. Pub- lic providers also receive funding directly from the MoH for salaries and related costs, although the NHSO adjusts for this funding when calculating the payment amount. As already mentioned, there was originally a copayment of B 30 for each service under the UC scheme, although this co-payment has subsequently been eliminated. This change has resulted in no discernible long-term increase in utilization. There are no deductibles for any of the schemes. The poor are automatically eligible for universal coverage. Research has shown that the scheme is particularly effective in targeting the poor, with 44 percent of the enrollees in this scheme belonging to the �rst two income quintiles. Even more important, the scheme has been successful in improving access to health services, with improvements in the concentration index, especially for inpatient services at district and regional hos- pitals. The incidence of impoverishment caused by illness has declined about 75 percent for families in the �rst income quintile and about 64 percent overall, and the bene�t in- cidence of government subsidies for quintiles 1 and 2 have increased by �ve percentage points to 55 percent between 2001 and 2003. Moreover, the scheme has accomplished this improvement without showing any signi�cant increase in overall total health ex- penditure, indicating that to a large extent UC bene�ts have replaced out-of-pocket pay- ments, and the costs have been accommodated within GDP growth. Although the system has shown signi�cant progress in improving accessibility and �nancial protection, it faces a number of acknowledged challenges. Sustainability is an ongoing concern, especially because the bene�t package for UC has recently been ex- panded to include both kidney dialysis and antiretroviral therapy. The payment meth- ods selected as well as the monopsony position of the NHSO have helped to ensure that costs remain under control, and efforts at compulsory licensing have helped to contain drug costs. Quality and effectiveness have been addressed through disease management programs in areas such as diabetes and TB. The increased �nancial access to health services has also strained the ability of the health system to cope with the additional utilization, which has ampli�ed the effect of Making Health Financing Work for Poor People in Tanzania 73 human resources shortages and has led to discussions regarding the public-private mix of facilities and the speed at which the decentralized environment can cope with increas- ing the availability of services. Finally, the differences in the bene�t packages between the two schemes, as well as the high cost of the CSMBS relative to the other schemes, will make for some interesting policy dialogue in the future. Vietnam2 In Vietnam, the key driver for reform seems to have been the rising costs of health care and the decreasing accessibility by the population, especially those in rural areas and the poor. The consolidation of various schemes into a single administrative arrangement under the Vietnam Social Security Agency (VSS) and into a risk pool resulted in adminis- trative efficiencies and increased solidarity between the Statutory and Voluntary Health Insurance (SHI) and Voluntary Health Insurance (VHI). Explicit targeting has resulted in a high level of enrollment by the poor in the VHI scheme. The key milestones include (a) experimentation with health insurance and health sector reforms as a result of the economic crisis (1989–92); (b) introduction of health in- surance at the national level, including SHI for employees (public and private), a Health Care for the Poor (HCFP) scheme, Free Care for Children Under 6 (FCFCU6), and VHI for farmers and the self-employed (1992–97); and (c) integration of HCFP and FCFCU6 into SHI (2005–09). The main design principles appear to be (a) to sign up as much of the population as possible with health insurance, (b) to provide the same level of coverage for both SHI and VHI (with cross-subsidization if necessary), and (c) to focus on administrative ef- �ciency by merging existing arrangements into just two programs and then having both programs administered by the same organization—that is, the VSS. Currently, about half of the population (48 million people) is covered by either SHI or VHI. SHI covers 14 percent of the population with a premium of 3 percent of payroll (2 percent employer, 1 percent employee). SHI also covers the poor (another 21.7 percent of the population) through a premium of 3 percent of the minimum salary, which is paid by the govern- ment. VHI covers informal workers and farmers (14.3 percent of the population) with a premium of US$12–18 per year, which is based on urban or rural location. The premi- ums are partially subsidized by local governments. All funds are collected by VSS, either through payroll deduction and submission to VSS offices or through collection of VHI contributions by either VSS offices or one of 10,000 agents at the village level. There are single risk pools for SHI and for VHI with allocations to regional and county branches of VSS. Branches are expected to live within their allocations, although the mechanisms for imposing budget discipline are unclear. Fee-for-service is mostly used, although some districts are experimenting with capi- tation arrangements. The fee-for-service approach is leading to concerns about �nancial sustainability. Providers are mostly public, although private providers are able to par- ticipate if they are willing to accept the VSS tariff. Referral requirements are enforced so that if people go to a higher-level facility without a referral le er, they will have to pay the cost themselves. Since 2001, there have been moves to increase hospital autonomy, thereby giving more control over staffing, organization and management, and �nancial management. 74 World Bank Study Copayments of 20 percent are standard, although there are exemptions for the poor. There are no deductibles, but there are maximum payments for high-tech procedures (US$1,250 equivalent). The Ministry of Labor and Social Welfare does an assessment to determine who is poor whereby it uses criteria of about US$11 per month equivalent in rural areas and US$14 month equivalent for urban residents. A number of issues and challenges have been identi�ed, including the following: ■ How to ensure compliance with VSS requirements and control abuse of the sys- tem. ■ How to deal with adverse selection (people who take coverage only when they are sick). ■ Deal with the de�cit, which reached 33 percent of SHI/VHI revenue in 2008 and which would require addressing both (a) the revenue side (premiums are al- ready scheduled to increase to 4.5 percent [1.5 percent employee, 3 percent em- ployer] next year and to 6.0 percent [2 percent employee, 4 percent employer] after that) and (b) the expenditure side by moving from fee for service to other payment methods. ■ Make VHI more affordable, especially for the near-poor, by signing up the other 50 percent of the population. ■ Improve the quality of the health services that are delivered. ■ Apply effective information technology to VSS administration, and train staff members, especially inspectors to a higher level of competence to ensure effec- tive control. China3 A key driver for the New Cooperative Rural Medical Scheme (NRCMS) was the lack of health protection in the post–Cooperative Medical Scheme (CMS) era (1978–2003), which was a recessionary period of health protection system in China. There were prob- lems with health resource allocation, which was plentiful in urban areas but sparse in rural areas, thus resulting in poor health services and infrastructure. For both NRCMS and Medical Assistance (MA), another driver was the fact that rising health care costs exceeded income growth, resulting in poor access to health care, lower health status, and illness becoming the main reason for impoverishment. The major success factors included (a) having the highest-level political commit- ment to the programs at all levels (national, provincial, local); (b) piloting to prove the concepts but not waiting to achieve perfection in the pilots before scaling up; and (c) beginning with a small premium and bene�t level that would quickly demonstrate the value of the scheme, thereby leading to both consumer demand and further political and �nancial commitment by governments to improve the level of bene�ts. As far back as 1952, free medical services were provided for civil servants, with this coverage being extended to urban workers in 1956. A CMS for rural workers and farm- ers was set up in 1978. Starting in 2001, a Basic Medical Insurance (BMI) was set up for urban workers, and free services were stopped. In 2003, there were initial steps to imple- ment the MA scheme in rural areas, followed by the piloting of the NRCMS from 2003 to 2006. The pace of scale-up was quite rapid. For example, NRCMS enrollment increased Making Health Financing Work for Poor People in Tanzania 75 from 333 counties and 80 million enrollees in 2003 to all 2,729 counties and 830 million enrollees in the �rst quarter of 2009. The system was designed to be feasible and sustainable, ensuring the needs of ba- sic medical services for different populations. The central government sets rules and provides overall leadership, while local governments implement. There is a conscious effort to perfect the system over time, rather than waiting to start with the perfect system and, especially in the NRCMS, to start with a smaller bene�t package and to use patient demand to expand both the available �nancing and the level of government funding. About 223 million are covered through BMI, representing urban workers and de- pendents, while 830 million rural workers and farmers are covered through NRCMS. Premiums for the BMI are 8 percent (6 percent employer, 2 percent employee), while NRCMS premiums are shared between the subscriber, the national government, and the lower levels of government (province, prefecture, and district) in about a 20-40-40 ratio. The MA scheme covers about 52 million poor people with premiums paid by the national government. Funds for the BMI-workers scheme are collected through payroll contribution and submission to the tax authority. Premiums from individuals for BMI-residents are col- lected by the Social Security Department and for NRCMS are collected by the county- level NRCMS offices or local agents. The agents also provide reimbursement for services obtained outside the county, where subscribers need to pay the entire amount up front and claim reimbursement afterward. Funds from national, provincial, prefecture, and county levels come through regular �nancing mechanism. For the BMI-workers, pooling of funds is at the county or prefecture level, but each employee is also his or her own risk pool. This approach is done through a combina- tion of a pooling fund and private employee health accounts: about 70 percent of the 6 percent goes toward the social pooling funds (4.2 percent total), and the entire 2 percent employee contribution plus 30 percent of the 6 percent employer’s share (3.8 percent to- tal) goes toward the private account. The employees receive what is essentially a “debit card� which they can use at their discretion to pay some or all of the deductible or copay- ment portions they are responsible for when they receive care. Management of all funds is at the county level (Social Security Department for BMI and Health Authority for NRCMS). Pooling for NRCMS is at the county level as well, with the rationale that this approach “increases risk-resilience and supervising capacity� (Wei 2009, 5). Farmers are involved in the management of the county-level risk pools. MA is also administered at the county level. There is some discussion about pooling at the prefecture level as a result of variations in county revenue-raising capability and the problems of smaller risk pools. Under the NRCMS global budget �nancing of outpatient payment is being used in a number of counties, and hospitalization is reimbursed on a fee per bed-day. The fee is based on an investigation and analysis of inpatient expenses for different levels of hospitals in recent years, together with the effect of medical cost escalation and inflation. All contracted service providers are eligible for payment. However, services are provided primarily by public providers. Efforts have been made at improving manage- ment at the Township Health Center level by posting director positions and entering a performance-based contract with those directors, as well as appointing accountants for 76 World Bank Study each center (a center has about 30 staff members and serves a population of about 20,000 people). All of the schemes pay only a portion of the total medical expenses (30–70 percent based on the type of services), with deductibles, copayments, and ceilings on the maxi- mum amount of reimbursement. These parameters are set at the local level in accor- dance with local conditions and �nancial availability, and they are revised on a regular basis. Reimbursement rates encourage the use of local medical facilities (for example, inpatient stays at township health centers are reimbursed at a higher rate than stays at county hospitals or higher-level facilities). For people belonging to MA, the reimburse- ment rate is higher, leaving about a 10–20 percent out-of-pocket payment. Once a year, the local civil affairs agency conducts poverty-level assessments. In- come levels of US$100–150 equivalent per person per year are considered poor, and lev- els under US$100 are considered very poor. Once people are assessed as being poor, they are automatically enrolled in the NRCMS and are eligible for the MA scheme. They obtain their MA bene�ts in addition to the ones they receive under NRCMS. The issues and challenges include (a) inadequate funds, with great differences in revenue-raising capacity between regions; (b) great pressure to increase the levels of compensation for BMI, NRCMS, and MA; (c) lag in information management and lack of a uni�ed platform for recording or sharing information; and (d) difficulty in supervising medical services behavior and controlling the increase health care costs. Ghana4 The National Health Insurance Scheme (NHIS) in Ghana is one of the legacies of the John Kufuor administration. During the 2000 elections, he promised to abolish what was known as the “cash-and-carry system� of health delivery. Under that system, patients— even emergency cases—were required to pay money at every point of service delivery. People died either because they did not have the money or because friends and relatives were not around to make the required advance payment. For those who survived, the fees represented a signi�cant burden, often driving people into poverty. Under the new law, a National Health Insurance Authority (NHIA) licenses, moni- tors, and regulates the operation of health insurance schemes in Ghana. Ghana has three main categories of health insurance. The �rst and most popular category is the district mutual health insurance scheme, which is operational in every district in Ghana. Under this public and noncommercial scheme, any resident in Ghana can register. Anyone who registers in “District A� and moves to “District B� can transfer the insurance policy and still be covered in the new district. The district mutual health insurance scheme also cov- ers people considered to be indigent—that is, those who are too poor, are without a job, and lack the basic necessities of life to be able to afford insurance premiums. In addition to premiums paid by members, the district mutual health insurance schemes receive funding from central government. This central government funding is drawn from the NHIF. Every Ghanaian worker pays 2.5 percent of the social security contributions into this fund; 2.5 percentage points from the value-added tax in Ghana also goes into the fund. People sign up for the district mutual health insurance scheme at the district assembly in their district or at the offices of the scheme. Making Health Financing Work for Poor People in Tanzania 77 The second category of health insurance comprises private commercial health insur- ance schemes that are operated by approved companies, which do not receive subsidy from the NHIF and which must pay a security deposit before they start operations. The third category of health insurance is known as the private mutual health insur- ance scheme. Under this scheme, any group of people (such as members of a church or social group) can come together and start making contributions to cater for their health needs, thus providing for services that are approved by the governing council of the scheme. Private mutual health insurance schemes do not get a subsidy from the NHIF. People who register under any of the schemes are given a card that can be used to seek treatment in any hospital in the country. There are no copayments unless extra services, such as a private ward, are used. Bills are then sent to the scheme provider (district, private scheme, or mutual scheme), which then pays the hospital. The card can also be used to buy prescribed drugs at accredited pharmacies or licensed chemical shops without paying at the point of delivery. The pharmacy then contacts the scheme provider to claim reimbursement. Regardless of the form of health insurance, the bene�t package consists of at least the following: ■ Outpatient services—general and specialist consultations and reviews, general and specialist diagnostic testing including laboratory investigation, X-rays, ul- trasound scanning, medicines on the NHIS medicines list, surgical operations such as hernia repair, an physiotherapy ■ Inpatient services—general and specialist services in patient care, diagnostic tests, medication-prescribed medicines on the NHIS medicines list, blood and blood products, surgical operations, inpatient physiotherapy, accommodation in the general ward, and feeding (where available) ■ Oral health—pain relief (tooth extraction, temporary incision, and drainage) and dental restoration (simple amalgam �lling, temporary dressing) ■ Maternity care—antenatal care, deliveries (normal and assisted), Caesarean sec- tion, and postnatal care ■ Emergencies—crises in health situations that demand urgent a ention such as medical, surgical, pediatric, and obstetric and gynecological emergencies, as well as road traffic accidents The health insurance schemes have the following exclusions: ■ Appliance and prostheses including optical aids, heart aids, orthopedic aids, and dentures ■ Cosmetic surgeries and aesthetic treatment ■ Anti-retroviral drugs for HIV ■ Assisted reproduction (for example, arti�cial insemination) and gynecological hormone replacement therapy ■ Echocardiography ■ Photography ■ Angiography ■ Dialysis for chronic renal (kidney) failure ■ Organ transplants 78 World Bank Study ■ All drugs that are not listed on the NHIS list ■ Heart and brain surgery other than those resulting from accidents ■ Cancer treatment other than breast and cervical ■ Mortuary services ■ Diagnosis and treatment abroad ■ Medical examinations for purposes other than treatment in accredited health facilities (for example, visa application, education, institutional, driving license) ■ VIP ward (accommodation) Recently, there have been reports of hospitals and pharmacies turning patients away, complaining that the public schemes owe them signi�cant amounts of money. Some hospitals have issued warnings that their operations could grind to a halt if the NHIS does not speed up payment of their claims. The NHIS has repeatedly assured the public and providers that it is working on the problems and, in due course, they will be resolved. In response to the increasing concerns over �scal pressures, the NHIS outlined its immediate strategic objectives to meet the challenges posed by systemic abuses and fraud and by the consequent exponential cost escalation in the NHIS. Announcing the measures, the chief executive officer identi�ed cost containment as one of the key ob- jectives in the drive for sustainability of the NHIS. Subsequent to this objective, he de- tailed various strategies designed to combat abuse and fraud responsible for the �nan- cial hemorrhage in the scheme, which, together with other challenges, had necessitated comprehensive arrangements to plug the leakages in the system, to strengthen internal controls, to inject accountability into the NHIS, and to ensure that the scheme gets value for money from service providers. Consequent key reform initiatives being implemented include the following: ■ The se ing up of a Consolidated Premium Account that would be operational from July 2010—Scheme managers will be required to deposit all premiums collected into a centralized account in either of two designated banks (Gha- na Commercial Bank and Agricultural Development Bank) with a nationwide reach and with an identi�cation code for each scheme account. The schemes have been directed to close all other accounts. This move would redress the current situation where more than 70 percent of premiums collected are not properly accounted for. ■ The establishment of a Consolidated Claims Management Centre in Accra— This pilot project will manage claims from the teaching hospitals and 10 region- al hospitals. It was operational from July 2010 and was staffed by experienced and competent claims personnel. This change would provide a greater capacity for ve ing of claims and efficient payment of service providers. If successful, the format would be rolled out in zonal claims processing centers across the country. ■ Requirement for a payment plan before the disbursement of funds—To combat irregularities in payments arrangements, scheme managers will henceforth be required to produce and forward a payment plan in advance to the authority for endorsement and before funds are released for disbursement to service pro- viders. Making Health Financing Work for Poor People in Tanzania 79 ■ Introduction of a standardized NHIS prescription form—The form will require the personal identi�cation number (PIN) of the prescriber and of the dispenser and will identify the scheme involved. The form is also designed to check sys- temic abuse and fraudulent practices in the dispensing of medicines. ■ Ways to streamline procedures for registration under the free maternal care program—Following the streamlining, pregnant women would be required to register—for free—before accessing care. The program went into effect from July 2010 and is intended to eliminate abuse of the program. ■ Withdrawal of accreditation from all private health care facilities run by full- time staff members of public health facilities—This change will eliminate con- flicts of interest and double billing of the NHIA for services to patients. ■ Introduction of capitation for outpatient department services in primary health care facilities following a pilot program that is to be carried out in a selected region and is scheduled to start by the end of 2010—Capitation would be based on research into utilization rates and costs, would be projected over a whole quarter, and would involve advance payments to service providers subject to periodic reviews and adjustments. It is expected that capitation would intro- duce efficiency and drive down costs in primary health care delivery. ■ Implementation of a unitary system of contracting with service providers—Un- der the review, contractual arrangements will involve apex bodies such as the Ghana Health Service, the NHIS, the Christian Health Association of Ghana, and other service providers. A new �nancial and operational reporting tool has been developed by the authority to streamline and standardize the electronic recording and timely transmission of accu- rate data by the schemes to the authority. This tool is expected to improve �nancial and operational discipline at the schemes and to ensure prompt processing and payment of claims to service providers. Trends in the data would also be useful for evaluation and monitoring purposes. Rwanda5 Since the introduction of user fees in public health facilities in 1996, Rwanda has pio- neered major programmatic, organizational, and health �nancing reforms aimed at im- proving the quality of care and, ultimately, the health status of the population in Rwanda. The reforms have a particular focus on the most vulnerable segments of the population. Prepaid �nancing in Rwanda was organized in the form of mutual health organiza- tions (MHOs). In 1998, there was only one initiative that could be identi�ed with a com- munity MHO and with sca ered experiences among provider-based health insurance schemes to facilitate access to care in a context of poverty. The Ministry of Health (MoH) launched pilot prepayment systems in three health districts in 1999 as an initial experi- mentation phase, which is part of developing a policy to promote MHOs in the country, and the MoH provided leadership to the overall process. MHOs were developed under the supervision of the MoH and its partners (USAID, WHO, the European Union, the Catholic relief and development organization Caritas, and other NGOs). Together, they closely involved community actors in the design and ensured that those actors would be in charge of managing MHO systems. 80 World Bank Study The scale-up was quite rapid, from 1 in 1999 to 53 in 2002, covering about 100,000 bene�ciaries. Initially, MHOs began by providing coverage for the indigent, vulnerable groups and for people living with HIV while using funds from several NGOs and a few administrative districts. The pilot experiences were evaluated between 2001 and 2003, after which Rwanda moved to implement the recommendations of the evaluations. The main recommendation was to adapt MHO organization to �t within the decentralization model that was being developed in Rwanda. As in the pilot phase, the adaptation phase received technical supervision from the MoH and subsequently from the Ministry of De- centralization and Local Affairs (MoDLA), which supervised the decentralized authori- ties and has social protection as part of its mandate. The involvement of the MoDLA and its agencies in promoting MHOs anchored them in the community and facilitated the mobilization of local authorities in the various administrative districts and district subdi- visions. This involvement also resulted in the involvement of NGOs and religious lead- ers, which raised the population’s awareness of the importance of enrolling in MHOs. The institutional arrangements for MHOs and local MHO networks were aligned with the environment of the decentralized authorities during this adaptation phase. Also, beginning in 2003, leadership at the central level was strengthened with the backing of the highest authorities in government. The MoH established a policy and strategic plan, and an MHO technical support unit to support MHO development and expansion. As part of mobilizing the local leadership in efforts to expand MHO-based health insurance, the degree of MHO promotion was made one criterion for evaluating the performance of the administrative districts. The number of MHO bene�ciaries increased from 556,000 by the end of 2003 (7 percent of the country’s population) to 3,686,000 by the end of 2005 (44 percent of the country’s population). This growth was a result of the expansion of MHOs to the entire country and of increasing the penetration of target populations. There was also an ex- pansion of the package of bene�ts covered by MHOs. Until mid-2006, most MHOs cov- ered only minor risks treated in health centers and only a limited package of major risks treated at district hospitals. In mid-2006, bene�t packages were expanded, and coverage for the indigent, vulnerable groups and persons living with HIV was institutionalized by the government and foreign partners. The bene�t packages now covered primary health care, secondary care, and tertiary care, which dramatically improved the price-quality ratio for MHO services. The quantitative expansion of MHOs grew quickly in the second half of 2006, reach- ing a coverage level of 6,283,000 bene�ciaries by the end of December 2006 (73 percent of the country’s population). By the end of August 2007, some 6,497,000 Rwandans were covered by MHOs (74 percent of the country’s population), and in 2008 it reached 85 percent. The actual number of MHOs climbed from 88 in 2003 to 226 in 2004, 354 in 2005, and 392 in 2006. In 2006, all the health centers in the country had a partner MHO under the community health center partnership. Each administrative district had a district MHO that served as a mechanism to pool the major risks for the primary MHOs. In June 2006, Rwanda established a National Guarantee Fund (FNG) and a District Solidarity Fund (FSD) to bolster �nancing mechanisms for MHO expansion in the coun- try. The FNG/FSD system strengthens equity of access to and �nancing of health insur- ance coverage in the country through two mechanisms. Making Health Financing Work for Poor People in Tanzania 81 First, the system supports the functional expansion of MHOs to harmonize the cov- erage bene�ts received by MHO bene�ciaries and the bene�ts received by the bene�cia- ries of the social insurance systems (Rwanda Health Insurance Company [Rwandaise d’Assurance Maladie, or RAMA] and Military Medical Insurance, or MMI). Second, the system supports the expansion of MHOs by providing care for the in- digent identi�ed by communities and for people with HIV/AIDS who have HIV-related opportunistic infections. The FNG is �nanced through contributions from the govern- ment, RAMA, MMI, private insurance systems, and foreign partners, including the Global Fund. The FSD is �nanced by the contributions of MHO chapters, administrative districts, transfers from the FNG, and contributions from the development partners that are involved at the district level. In addition to the MHO experience, Rwanda also led in the development of perfor- mance-based �nancing (PBF). In 2006, PBF became a major pillar of the MoH’s strategy. It was implemented to provide additional resources and incentives to health workers so as to improve efficiency and the quality of care. PBF is currently implemented at three levels: health center, hospitals, and community levels. In the national model for health centers, payments for performance are based on the quantity of outputs achieved condi- tional on the quality of services delivered. At the hospital level, performance is assessed through a peer-evaluation mechanism. Finally, the community PBF consists of decen- tralized control and decision making and of payment for a community health worker cooperative after performance has been assessed. An evaluation of the PBF approach by the World Bank concluded that it had a large and signi�cantly positive effect on institutional deliveries and preventive care visits by young children, and it improved quality of prenatal care. The authors found no effect on the number of prenatal care visits or on immunization rates. The overall effect was great- est for those services that had the highest payment rates and needed the lowest provider effort. The �nancial performance incentives that were evaluated were found to be able to improve both the use of and the quality of health services. Key lessons from the Rwandan experience and from factors contributing to the suc- cessful implementation of reforms can be formulated to provide a stronger base for fu- ture action: ■ Strong government leadership, vision, and regulatory framework at all levels will foster the short- and long-term sustainability of health sector reforms. ■ In light of the changing macro and health sector environment, the ability of gov- ernment to adapt strategies is needed to strengthen health services. ■ Independent controls and quality checks are essential for monitoring and evalu- ation of health facility performance. ■ Cultural and social factors, particularly solidarity within communities, contrib- ute to the success of several health service delivery innovations. ■ Key reforms need to include provisions for �nancial protection and other sup- port for indigent populations. ■ Government coordination of donor funding is critical to ensure that aid is used effectively and is aligned with national priorities. 82 World Bank Study Kenya6 Although the previous examples highlight reforms that have already taken place, the example of Kenya presents some of the evolving thinking on health �nancing in a system that is similar to the one in Tanzania in many respects. The health �nancing strat- egy in Kenya was developed over several years, using a process that drew heavily on the previous failure to reform the health �nancing system. In 2005, the then health min- ister prepared a plan for mandatory health insurance and was able to obtain approval by parliament for the enabling legislation. The �nance minister objected to the scheme by saying the government did not have money to fund it, and various employer and trade union groups also opposed the bill on the grounds of inadequate consultation and potential reductions in competitiveness and personal incomes. President Kibaki later refused to assent to the bill that sought to establish the scheme, arguing that the govern- ment could not afford it, and the bill expired when parliament rose for the election. In contrast, the new health �nancing strategy is the result of a broad consultation process, which included a wide variety of stakeholders and drew on both extensive analytical work and reviews of health �nancing systems in other parts of the world. While the strategy has been discussed at the National Economic and Social Council and a number of key principles endorsed at that level, there are ongoing discussions on the organization and structure of the new health �nancing system. Based on the process undertaken to date, the current version of the health �nancing strategy includes the fol- lowing elements, which are designed to be mutually supportive and reinforcing: ■ Improving efficiency, accountability, and transparency ■ Strengthening revenue collection ■ Having more effective risk pooling ■ Harnessing the informal sector �nancing potential ■ Broadening the bene�ts package ■ Strengthening provider incentives ■ Protecting the poor and vulnerable groups ■ Improving aid effectiveness ■ Ensuring sustainability The strategy identi�es a number of areas where efficiency can be improved. First, it indicates that demonstrating the most effective use of the current resources is the best approach for securing support from the government, development partners, and the general public for increased health care �nancing in the future. A central element of the drive for improved efficiency is to ensure that each component of the health system focuses on what it is best at and that it leaves other activities to other parts of the health system. For example, the strategy proposes that the MoH should act primarily as stew- ard of the health system, thereby providing overall direction and guidance and se ing the overarching policies of the system, but it should not be involved in the day-to-day management of health service providers. Having the MoH play the role of both regula- tor and service provider inevitably leads to conflicts of interest, especially in a system where various types of providers are needed to deliver necessary health services. Strong quality assurance and accreditation mechanisms are extremely important for ensuring that the services provided meet essential quality standards. Strengthening the referral system will also lead to efficiency improvements, which will include ensuring Making Health Financing Work for Poor People in Tanzania 83 (a) that each level of care provides the services that it is best equipped to provide and (b) that expensive secondary care resources are not used to provide services that can be delivered more efficiently and effectively at lower levels of care. Efficiency will also be enhanced by separating the purchaser from the provider of services, so the providers can focus on effective management of their facilities. This change will require increased autonomy for public health care providers, so that they have the ability to manage their facilities in the most cost-effective way and under the overall policy direction of the MoH. The proposed move from providing inputs to purchasing services is expected to have several important effects. First, providers will have to deliver services to patients in order to receive funding, thereby increasing the incentive to ensure that necessary high- quality services are made available. Second, the method and conditions of payment will be developed in a clear and transparent way, so that the managers of health facilities will know what level of funding to expect for a given level of activity. This approach will facilitate be er planning and management of resources. It also places the patient squarely at the center of health services delivery, because the funding will follow the patient and because providers who do not treat their patients competently will lose both patients and funds. A substantial amount of investment will be needed (a) to develop a cadre of manag- ers with the necessary skills and abilities and (b) to develop the governance and account- ability structures that will be required in a more autonomous environment. The result will be that public health facilities will be run by professional managers and overseen by management boards. Boards will be responsible for hiring the manager, se ing facility policy, and monitoring execution of that policy. There are not enough public facilities to deliver necessary health services to the peo- ple of Kenya. Thus, another key element of the health care �nancing strategy will be to make all health facilities (including FBO, NGO, and the private health providers) eligible for contracts to supply health services, provided they are accredited and willing to ac- cept the standard contract terms. Of course, in some areas, there may be more providers than necessary; in those cases, selective contracting will be undertaken. Private health providers will be encouraged to cover underserved areas through various incentive mechanisms (for example, premiums on the speci�ed tariff). The fact that many private providers are currently not fully using their capacity suggests that there is a potential for progress in those areas, if mutually agreeable terms can be established. Given the resources that are currently available, the package of health services eligi- ble for coverage would initially be limited, although it would be expanded as resources permit. For those services not included in the package, options would exist for people to take out supplementary private health insurance. Because this provision would be an important element of the overall health care �nancing and delivery system, it will be necessary to develop effective �nancial and quality oversight structures if the package is to ensure that private insurance remains sustainable. Although the amount of funding is limited, it is still very important to ensure that providers are adequately compensated for the services they deliver. In addition to pro- viders being an issue of longer-term sustainability, it is also expected that certain types of providers will not participate if they feel that the reimbursement rate is not adequate. Because all types of providers are needed to ensure coverage, this situation should be 84 World Bank Study avoided. The strategy, therefore, proposes that a Bene�ts and Tariffs Board be estab- lished to work with purchasers, providers, and other stakeholders in developing appro- priate reimbursement levels. An important aspect of this board will be ensuring trans- parency in its deliberations, so that the resulting tariffs and the rationale for determining them are clearly articulated and are available to anyone who may be interested. The other important aspect of the board’s operations will be the development of the basic bene�t package or of those services that will be covered. Although this coverage will initially be a more limited set of services, as the economy grows or the total resourc- es available for health increase, the package would be reviewed and expanded. Again, an important feature will be transparency and availability of information. In addition to stakeholder involvement, there will also be a critical role for expert advice, whether it is the effect of tariff changes or the consequences of insuring additional services. The board will need to regularly update the bene�t package to reflect both changes in medi- cal technology and updates in the available �nancing. If health care is to reach informal workers and farmers, smaller risk pools in the form of community health funds may be established, but the NHIF would essentially act as a reinsurer for those funds, covering the full range of services for the members, regardless of the claims activity. Capacity will need to be developed in actuarial analysis to determine the appropriate or minimum premium levels for such schemes, as well as for the general population of insured persons covered through formal employment or other informal group arrangements. The strategy suggests that premium levels clearly need to rise, but in the short run, this increase would have to be balanced against the public’s perceived value with respect to the services provided or covered by NHIF and the public’s ability to pay. This ap- proach means that general revenue will continue to be an important factor in the overall success of this risk-pooling structure. Initially, this change will take the form of redirect- ing the funds currently spent to �nance inputs at the (public) facility level into the NHIF, so that it can purchase services from those facilities. The transfer of funds would be used to “top up� the premiums that are paid, to pay health insurance premiums for the poor, or to purchase certain “public goods� services as speci�ed in national programs. Over time, as both the perceived value and the ability to pay grow, the premium subsidies would be reduced. The strategy proposes that the NHIF progressively penetrates the informal sector and grows membership from the current 440,000 to 7.5 million members. This growth will have the effect of increasing coverage from the current 24 percent to about 70 per- cent (leaving out the 30 percent of the population considered to be indigents). According to the National Health Accounts and the 2005/06 rate, this change has the potential of channeling into the NHIF an equivalent of K Sh 38 billion annually (through conversion of out-of-pocket expenditures into health care savings). Harnessing this informal sector potential will entail incorporating entrepreneurs, self-earning professionals, freelance workers, and members from the matatu, agriculture, and juakali sectors into the NHIF. Using a suitable model, the NHIF will seek to assess the income of members from this sector to align the actual contributions to those in the formal sector. The NHIF will pursue a pro-market strategy for increasing the participation of the informal sector that is two-pronged: communicating effectively and providing suitable incentives, particularly through a ractive products. The la er will include de�ning the Making Health Financing Work for Poor People in Tanzania 85 minimum package of care that is comprehensive and cost-effective and that includes preventive, health promotion, and outpatient care. Through the introduction of a purchaser/provider split, the opportunity exists to give clear incentives to providers for delivering services of sufficiently high quality to meet the needs of the population. However, this quality depends very much on the in- centives that are inherent in the underlying provider payment mechanisms. According- ly, the strategy will pursue payment approaches that have been proven to provide the necessary incentives and have been proven to be feasible (in terms of data availability, collection, and processing) in countries at similar levels of development, including capi- tation for primary health care and case-based reimbursement for hospitals. Protection of the poor is central to the overall strategy for �nancing health care, which means that the poor must be able to access necessary health services without either organizational or �nancial constraints. The evidence is clear that user fees at the point of service serve as a deterrent to accessing health services by the poor, so a central element of the strategy is the elimination of user fees for this group, with a view to pos- sibly eliminating user fees for the rest of the population—as and when alternative re- sources permit. However, as noted earlier, user fee revenues from the nonpoor (de�ned as either the top three or four quintiles) are substantial, so it will be important to develop coherent approaches for replacing this revenue from other sources and, indeed, �nding additional resources as �nancial barriers decline. A critical issue will be the timely and accurate identi�cation of the poor. By identi- �cation of the poor, it will be possible to ensure that, if necessary, those people can be registered with the social insurance system and that the contributions are made on their behalf through the joint efforts of the government and development partners. This ap- proach will give the poor free access to necessary medical care, while ensuring that they do not face the barrier of user fees. An important element is to make sure that health service providers do not face dis- incentives to treating the poor. Thus, it will be important to compensate providers for lost user-fee revenue when they treat poor people. The registration card that members receive would both signal to the provider that no user fee should be collected and sig- nal to the purchaser that the reimbursement should be augmented by the user fee that would otherwise have been paid. Aside from user fees, other barriers to access need to be addressed. For example, selective contracting could be used to encourage health service providers to relocate to underserved areas, thereby improving physical access. Furthermore, the introduction of a functioning referral system should ensure that care is provided at the appropriate level and that necessary care at higher levels is not impeded by the inability to pay. User fees could be increased (or kept) for patients who have not obtained a referral, further strengthening the continuum of care and guarding against the overuse of higher levels of care. The strategy indicates that full acceptance and support of the development partner community are an important determinant in how quickly the strategy can be imple- mented. Considering the involvement of development partners to date (on both the de- velopment of the health �nancing strategy and the increased use of country systems) and anticipating ongoing support in both of those areas, the role of development partners is fully integrated into the approach—from the investments needed to build capacity and 86 World Bank Study improve the physical infrastructure of facilities to the support required to ensure equity and access so that the poor and nonpoor are able to receive essential health services. This integrated approach could include a potential new access and equity fund, which would receive funding from both government and development partners, thus ensuring that everyone in Kenya is able to obtain essential health services when they are needed. This full partnership in the implementation of the health care �nancing strategy is a unique opportunity for maximizing aid effectiveness across the health sector. Colombia7 Before 1993, Colombia’s Ministry of Health (MoH) was constitutionally responsible for providing all Colombians with health care. However, inefficiency, badly targeted public subsidies, and fragmented markets resulted in only one Colombian in �ve having any protection against the �nancial risk of health shocks from serious illness. Only those with more �nancial resources could afford to join social security schemes or pay out of pocket for health care. Although policy makers believed that (a) Colombia’s public health was well targeted to the poor; (b) public health services were free to all comers, but especially the poor; and (c) the poor did not have to go to private providers for health care because public funds supported a large network of public hospitals and clinics; in fact, households contributed a signi�cant portion of health care �nancing. Cost was the most signi�cant barrier to health care before 1993. In the poorest quin- tile, only one individual in six who fell ill in 1992 sought medical care. The poor had less access to health care than the rich, paid out of pocket for public as well as private health care services, and paid proportionately more of their income for any services they received. The governmental health care delivery network did not adequately serve the poor, leaving already impoverished sick people with huge medical bills and the likeli- hood of abject poverty from which they could not recover. Before 1993, public subsidies were paid directly to hospitals rather than to users. However, the poor who could get into public hospitals incurred out-of-pocket expenses more often than did middle- and upper-income patients, who were often covered by pri- vate insurance. Rarely did the poor receive free care in public facilities. In fact, 91 percent of the poorest inpatients incurred out-of-pocket expenses, compared with 69 percent in the richest quintile. The private sector was important in both �nancing and delivering health services before the reform. According to the government’s National Household Survey in 1992, 40 percent of all health interventions and 45 percent of all hospitalizations were done in the private sector. Only 20 percent of the Colombian population had health insurance. Law 100 of 1993 �rst mandated the creation of a new national health care system with universal health insurance coverage and then reorganized �nancing and delivery. A key feature of this law was that public subsidies would go directly to individuals rather than to institutions. The reform introduced four main elements to reach the poor: ■ A proxy means testing index to target public health subsidies to the neediest (SISBEN, or Selection System of Bene�ciaries for Social Programs) Making Health Financing Work for Poor People in Tanzania 87 ■ Transformation of traditional supply-side subsidies (which �nanced the public health care network) into demand-side subsidies (which subsidize individual insurance premiums for the poor) ■ An equity fund in which revenue from payroll contributions and treasury re- sources cross-subsidize insurance premiums for the poor ■ A way of contracting for health service delivery from both the public and pri- vate sectors The new system is a universal health insurance coverage plan with two schemes: (a) a contributory scheme (RC) to cover formally employed and independent workers who contribute to the scheme so their contributions are collected by the insurer of choice, and (b) a subsidized scheme (RS) to covers the poor and indigent individuals who could not afford to make any insurance contribution. SISBEN is a general-purpose system for selecting bene�ciaries for social programs in Colombia. It has a statistically derived proxy means test index that serves as an indica- tor of household economic well-being. The variables that determine welfare include (a) availability and quality of housing and basic public services, (b) possession of durable goods, and (c) human capital endowments. The system includes norms and procedures de�ned at the central level and operated at the municipal level to gather information necessary to calculate the welfare index and to select bene�ciaries for the numerous so- cial programs. The RS scheme is one of the programs in which bene�t incidence has been the highest for those targeted with SISBEN and was bene�ting more than 11.4 million poor and vulnerable people by the end of 2002. Payroll contributions go into a national health fund with four separate accounts. The fund �nances insurance premiums for all enrolled in the RC. In the RC-RS cross- subsidization process, a share of the contributions is allocated to �nance the RS, together with treasury transfers to the territories. Those who are eligible for enrollment in the RS but are still uninsured are to rely on public hospitals for care. Every insured individual is free to choose an insurer and to consult any provider in the insurer’s network. Both schemes have a basic bene�ts package, but the RC package includes every level of care while the RS package has to be complemented with services provided by public hospitals and �nanced through supply-side subsidies. The design provides that those subsidies were to be turned into demand-side subsidies over time to achieve universal insurance coverage with the same package in both schemes. The reform of 1993 increased �nancial protection for all, but especially for the poor and rural population. As shown in �gure A.1, before 1993, 23 percent of Colombians were �nancially protected against the risk of health shocks, while a decade later, 62 percent of the population had access to health insurance. Insurance coverage among the wealthiest group increased from 60 percent in 1992 to 82 percent in 2003, while insur- ance coverage among the poorest group increased from 9 percent in 1992 to 49 percent in 2003. 88 World Bank Study Figure A.1: Insurance Coverage by Income Quintile and Year Source: Go ret, Schieber, and Waters 2008, 166. Formal insurance in Colombia reduced out-of-pocket expenditures on ambulatory care between 50 percent and 60 percent. The poor in the RS spent about 4 percent of their income on ambulatory care, but the uninsured poor spent more than 8 percent. Out-of- pocket expenditures on hospitalization among the uninsured poor absorbed more than 35 percent of their income in 2003. The poor in the RC spent a smaller proportion of their income on inpatient care than did the poor enrolled in the RS. Further, a health shock requiring hospitalization pushed 14 percent of those hospitalized and uninsured below the poverty line compared with only 4 percent of inpatients covered by the subsidized scheme. Access to preventive care was also improved by the introduction of health insur- ance. Although 65 percent of the insured saw a physician or a dentist at least once for preventive reasons and without being sick in 2003, only 35 percent of the uninsured did so. Regulations gave preference to children, single mothers, the elderly, the handi- capped, and the chronically ill for priority access to insurance enrollment in the RS. Colombia’s Demographic and Health Survey (DHS) indicated a signi�cant improve- ment in access to medical care during pregnancy, use of prenatal care, and medically as- sisted deliveries, particularly in the rural areas. DHSs between 1986 and 2000 showed a 66 percent increase in physician-assisted delivery, an 18 percent increase in institutional delivery, and a 49 percent increase in prenatal care use among rural women. Despite positive results, the Colombian social insurance scheme has been criticized for not having achieved universal coverage and for �nancing a less comprehensive bene�ts package for the poor (RS) than for the wealthy (RC). The slower-than-planned Making Health Financing Work for Poor People in Tanzania 89 transformation of supply-side subsidies (which �nance public hospitals) into demand- side subsidies (which �nance health insurance for the poor) has also come under �re for slowing down the expansion of the RS bene�ts package. Colombia still has many challenges to surmount to complete the consolidation of the social insurance scheme, not only to cover the entire population but also to improve the efficiency and quality of health care. The changeover to the new, insurance-based system has been difficult politically, administratively, and technically. Mexico8 In the 1940s, Mexico established social security institutions for formal private and public employees that collect premiums from employees and their employers, that obtain ad- ditional funding through government subsidies, and that use those resources to provide services to the covered employees. The system has worked relatively well for the half of the Mexican people who have access to it. The poorer half of the population—which tends to work in smaller and poorly regu- lated establishments, to be self-employed menial laborers, or to be unemployed—is far less fortunate. Because they are outside the formal labor market, they usually cannot access health insurance through the workplace or cannot afford it. This segment of the population was left to seek lower-quality services provided, at a fee, through health min- istry facilities or through the expensive and often unregulated private sector. The Mexican health system was also seriously unbalanced and underfunded, and government health programs were not able to address the challenges. Funding was not distributed efficiently or equitably—close to half of the population was uninsured, yet only a third of the federal funds for health went to them. There was great risk of impov- erishment caused by health spending: 2 million to 4 million households spent 30 percent of their annual disposable income on health. The government that took office in 2000 assigned a high priority to correcting this imbalance. It introduced a set of reforms that featured a new health insurance program, Seguro Popular (SP), for all those not covered by existing social security plans. This voluntary program would provide the people with subsidized insurance coverage com- parable to that available to social security bene�ciaries. The health ministries of Mexico’s 32 state governments are responsible for identi- fying, enrolling, and serving eligible SP participants. SP was designed to focus on the poorest families �rst. Premium payments are subsidized on a sliding scale by the state, and the families from the poorest 20 percent of the population do not pay. Participants are entitled to treatment at no additional cost through some 266 interventions at speci- �ed institutions (mostly state government facilities). SP participants are also eligible for 17 interventions at a network of higher-level tertiary care institutions operated by the federal and state governments. The gap between premium revenue and the program’s total cost is covered by gov- ernment subsidies. Most of the subsidies come from the federal to the state governments, with federal funding covering 83 percent of the premiums. The funding formula factors in the level of development of the state (poor states get more funds than be er-off ones), as well as the number of families enrolled. This approach gives states an incentive to im- prove the quality of services and to increase the number of participating families. State governments bear the remaining costs from their own resources. 90 World Bank Study The SP program is mostly responsible for the recent increase in federal government resources allocated to health. For example, the budget of the federal health ministry increased by about 53 percent between 2000 and 2005—largely a ributable to the pro- gram’s implementation. The SP program originally aspired to universal coverage by 2010, reaching be er-off and poor uninsured people alike, but ensuring that the disadvantaged are served �rst. The program design includes three related measures to increase the focus on the poor. First, premium rates are graduated by economic status, with exemptions for the poor. Second, mechanisms are in place to identify poor families and to determine premium subsidies or exemptions. Third, there is a system for paying implementing agencies that gives them an incentive to focus on enrolling poor families. By the end of 2009, approxi- mately 31 million of 48 million eligible individuals were covered by SP (65 percent). The government recently took out a US$1.25 billion World Bank loan to increase this cover- age to 85 percent and to achieve universal coverage by 2012. The system is based on prepayment for a package of services that will expand as demand and funding permit. This growth has already happened, and bene�ts covered by the SP have expanded signi�cantly, from 76 health interventions at inception to 266 interventions today. Families pay up to 5 percent of disposable income (de�ned as total spending after basic needs are covered). Farther down the economic ladder, the pre- mium falls and the government subsidy rises, with people in the poorest 20 percent of the population paying nothing at all. States have several options available for identifying the people who qualify for dif- ferent levels of subsidy. They can use the approach of the existing PROGRESA/Oportun- idades program, whose enrolled participants are automatically eligible for SP coverage at no cost. They can also use the approach developed by the SP program or developed by several other federal subsidy programs, or they can enroll all members of speci�ed groups without evaluating the individual members of those groups. The PROGRESA/Oportunidades and SP approaches both use a proxy means test, which identi�es the poor on the basis of information about the assets of households and the people living in them, rather than income or expenditures. Field workers score each item on the basis of guidelines developed through statistical studies of household economic status, sum the results, and compare the overall outcome with that of other members of the population. A family’s SP premium is based on where the family �ts on the economic spectrum thus identi�ed: the lower the family is, the less it pays. Federal SP support to state governments is replacing federal assistance based on inputs and is shifting to the number of people the states serve rather than the number of people they employ and the facilities they operate. This change gives states an incentive to enroll as many people as possible in SP. Because the poor do not pay premiums, they are easier to register and thus are prime candidates for early enrollment. By the end of 2005, SP had enrolled about 11.5 million people, just under 20 percent of the 65 million to 75 million Mexicans not covered by the social security system. About 40 percent of the enrollees were in the poorest quintile, compared with less than 3 per- cent in the richest quintile. SP participants in need of health services were signi�cantly more likely to obtain them than were those who did not participate. Thus, the SP pro- gram is very pro-poor, as shown in �gure A.2. Making Health Financing Work for Poor People in Tanzania 91 Figure A.2: Enrollment in Seguro Popular by Income Quintile Source: Yazbeck 2009, 251. Two principal challenges have been identi�ed for the SP program: ■ The ongoing commitment of the federal government to provide funds for this program will continue to be a challenge, especially in light of the global eco- nomic crisis, which saw the Mexican economy contract by 6.5 percent. The re- cent renewed commitment to achieving universal coverage by 2012, together with explicit borrowing from the World Bank for this purpose, signals that—for now at least—this commitment is being maintained. ■ The second challenge is how to encourage those above the poorest income quin- tile to pay the premiums required of them to participate in SP. Those people will constitute an ever-increasing proportion of the additional people who will have to be enrolled in the coming years if SP is to meet its objective of universal care. Notes 1. This section is based on �ndings from a 2009 Africa health insurance study tour by the Ethiopia delegation, Tanzania delegation, Uganda delegation, World Bank Mission Team, and Abt Associ- ates Mission Team (personal communication to author). 2. This section is based on �ndings from a 2009 Africa health insurance study tour by the Ethiopia delegation, Tanzania delegation, Uganda delegation, World Bank Mission Team, and Abt Associ- ates Mission Team (personal communication to author). 3. This section is based on �ndings from a 2009 Africa health insurance study tour by the Ethiopia delegation, Tanzania delegation, Uganda delegation, World Bank Mission Team, and Abt Associ- ates Mission Team (personal communication to author). 4. This section is based on Diop and Ba (2010). 92 World Bank Study 5. This section draws heavily from Basinga and others (2010), Diop and Ba (2010), and the Rwanda Ministry of Health and World Bank Africa Region Human Development (2009). 6. This section is based on Kenya Ministry of Public Health and Sanitation and Ministry of Medical Services (2010). 7. This section draws heavily on Yazbeck (2009, 205–14); Giedion and Uribe (2009); and Go ret, Schieber, and Waters (2008). 8. This section draws heavily on Yazbeck (2009, 241–46) and World Bank (2010). References Basinga, P., P. J. Gertler, A. Binagwaho, A. L. Soucat, J. R. Sturdy, and C. M. Vermeersch. 2010. “Paying Primary Health Care Centers for Performance in Rwanda.� Policy Research Working Paper WPS5190, World Bank, Washington, DC. Borghi, J., G. Mtei, and M. Ally. 2011. “Modelling Health Insurance Expansion in Tan- zania.� SHIELD Work Package 5 report, Ifakara Health Institute, Dar es Salaam. CNAS (Romanian National Health Insurance Fund). 2010. Situatia Privend Executia Bugetlui Fondului National Unic de Asigurari Sociale de Sanatate, pe Anii 2007–2009. Bucharest. Denmark, Ministry of Foreign Affairs. 2007. “Joint External Evaluation of the Health Sec- tor in Tanzania, 1999–2006.� Copenhagen. Diop, F. P., and A. Ba. 2010. “Mutual Health Insurance, Scaling-Up and the Expansion of Health Insurance in Africa.� Health Systems 20/20, USAID. h p://www.health- systems2020.org/content/resource/detail/2523. Estonian Health Insurance Fund. 2010. “Estonian Health Insurance Fund Annual Report 2009.� Tallinn. Georgieva, L., P. Salchev, S. Dimitrova, A. Dimova, and O. Avdeeva. 2007. “Bulgaria: Health System Review.� Health Systems in Transition 9 (1): 1–156. Giedion, U., and M. V. Uribe. 2009. “Colombia’s Universal Health Insurance System: The Results of Providing Health Insurance for All in a Middle-Income Country.� Health Affairs 28 (3): 853–63. Go ret, P., G. J. Schieber, and H. R. Waters, eds. 2008. Good Practices in Health Financing: Lessons from Reforms in Low- and Middle-Income Countries. Washington, DC: World Bank. JAHSR (Joint Annual Health Sector Review Final Report). 2010. Development Partner Statement. Ministry of Health and Social Welfare, Dar es Salaam. Jamu, N., L. Nduhiye, J. Macha, F. Kessy, J. Borghi. 2009. “A Feasibility Study for the In- troduction of TIKA in the Three Municipalities of Dar es Salaam.� Ifakara Health Institute, Dar es Salaam. Kamuzora, P., and L. Gilson. 2007. “Factors Influencing Implementation of the Commu- nity Health Fund.� Health Policy and Planning 22 (2): 95–102. Kenya Ministry of Public Health and Sanitation and Ministry of Medical Services. 2010. Accessible and Affordable Quality Healthcare Services in Kenya: Financing Options for Universal Coverage. Nairobi. Kurowski, C., K. Wyss, S. Abdulla, N. Yémadji, and A. Mills. 2003. “Human Resources for Health: Requirements and Availability in the Context of Scaling Up Priority Interventions in Low-Income Countries: Case Studies from Tanzania and Chad.� HEFP working paper 01/04. Funded by the Department for International Develop- ment (DFID), and carried out under the auspices of the London School of Hygiene and Tropical Medicine Health Economics and Financing Programme, London. 93 94 World Bank Study Ku in, J. 2000. “Towards Universal Health Care Coverage: A Goal-Oriented Framework for Policy Analysis.� HNP working paper, World Bank, Washington, DC. Kuwawenaruwa, A., J. Macha, and J. Borghi. 2010. “Willingness to Pay for Voluntary Health Insurance in Tanzania.� Ifakara Health Institute, Dar es Salaam. Lankers, C., K. Mgonela, G. Mtei, and A. Mushi. 2008. “Suggestions towards a Frame- work for the Future Development of Health Insurance in Tanzania.� Report for the Ministry of Health and Social Welfare, Tanzania, �nanced by the Tanzania- German Program to Support Health, Dar es Salaam. Makawia, S., J. Macha, M. Ally, and J. Borghi. 2010. “An Assessment of the Distribution of Health Service Bene�ts in Tanzania.� Report on SHIELD Work Package 3, Dar es Salaam. [SHIELD is funded by the European Commission (Sixth Framework Pro- gramme; Speci�c Targeted Research Project no:FP6-2004-INCO-DEV-3-032289) and by the International Development Research Centre.] McIntyre, D., V. Govender, E. Buregyeya, D. Chitama, E. Kataika, E. Kyomugisha, R. Kyomuhangi, T. Mbeeli, A. Mpofu, S. Nzenze, A. Walimbwa, and B. Chitah. 2008. “Key Issues in Equitable Health Care Financing in East and Southern Africa.� EQUINET Discussion Paper Series 66, Health Economics Unit, UCT and EQUI- NET, Harare. Mtei, G., and J. Borghi. 2010. “An Assessment of Health Care Financing Progressivity in Tanzania.� Report on SHIELD Work Package 2 , Dar es Salaam. [SHIELD is fund- ed by the European Commission (Sixth Framework Programme; Speci�c Targeted Research Project no:FP6-2004-INCO-DEV-3-032289) and by the International De- velopment Research Centre.] Mtei, G., and J.-A. Mulligan. 2007. Community Health Funds in Tanzania: A Literature Re- view. London: Consortium for Research on Equitable Health Systems (CREHS). Mtei, G., J.-A. Mulligan, M. Ally, N. Palmer, and A. Mills. 2007. “An Assessment of the Health Financing System in Tanzania.� Report on SHIELD Work Package 1, Dar es Salaam. [SHIELD is funded by the European Commission (Sixth Framework Pro- gramme; Speci�c Targeted Research Project no:FP6-2004-INCO-DEV-3-032289) and by the International Development Research Centre.] NAO (National Audit Office). 2008. A Performance Audit Report on the Management of Pri- mary Health Care: A Case Study of Health Centers—A Report of the Comptroller and Auditor General of Tanzania. Dar es Salaam: NAO. NBS (National Bureau of Statistics). 2007. Integrated Labour Force Survey (ILFS), 2006. Dar es Salaam: NBS. ———. 2009. Household Budget Survey 2007—Tanzania Mainland. Dar es Salaam: NBS. Rwanda Ministry of Health and World Bank Africa Region Human Development. 2009. Rwanda: A Country Status Report on Health and Poverty. Washington, DC: World Bank. Savedoff, W. D., and P. Go ret, eds. 2008. Governing Mandatory Health Insurance: Learning from Experience. Washington, DC: World Bank. Sendro, J. E. 2007. “Community Health Funds as Complementary Financing Option in Tanzania.� CHF Best Practices Workshop, CHF Coordination Unit, Ministry of Health and Social Welfare, Dar es Salaam. Making Health Financing Work for Poor People in Tanzania 95 Swiss TPH (Swiss Institute for Tropical Medicine and Public Health). 2010. “Situation Analysis of Health Financing in Tanzania: Debrie�ng.� Health Promotion and Sys- tem Strengthening, development partner–sponsored meeting, Dar es Salaam. TACAIDS (Tanzania Commission for AIDS). 2010. Public Expenditure Review 2007–2009, HIV and AIDS, Tanzania Mainland, Final Report, June 2010. Dar es Salaam: TACAIDS. Tanzania MoHSW (Ministry of Health and Social Welfare). 2008a. Health Sector Strategic Plan III, July 2009–June 2015. Joint Annual Health Sector Review. Dar es Salaam. ———. 2008b. Tanzania National Health Accounts (NHA) Year 2002/3 and 2005/6 with Sub- accounts for HIV and AIDS, Tuberculosis, Malaria, Reproductive and Child Health. Dar es Salaam. ———. 2008, 2009, and 2010. Public Expenditure Review. Dar es Salaam. ———. 2010. “Health Financing in Tanzania.� Technical Review Meeting as part of Joint Annual Health Sector Review, Dar es Salaam, September 6. Tanzania National Health Insurance Fund. 2005. 2003/04 Actuarial and Statistical Bulletin. Dar es Salaam. ———. 2007. 2005/06 Actuarial and Statistical Bulletin. Dar es Salaam. ———. 2008. Actuarial and Statistical Bulletin, 3rd Issue. Dar es Salaam. ———. 2009a. Action Plan for CHF Operations, 2009–2012. Dar es Salaam. ———. 2009b. Actuarial and Statistical Bulletin, 4th Issue. Dar es Salaam. ———. 2010. Annual Report 2009/10. Dar es Salaam. Tanzania NSSF (National Social Security Fund). 2010. Actuarial Valuation of the National Social Security Fund as of 30 June 2009. Dar es Salaam. TC-SWAp (Technical Commi ee of the Sector-Wide Approach). 2011. Agenda Item 8, February 9. Tanzania Ministry of Health and Social Welfare, Dar es Salaam. WDI (World Development Indicators) (database). World Bank, Washington, DC. h p:// data.worldbank.org/data-catalog/world-development-indicators. Wei, Fu. 2009. “The Establishment and Development of NRCMS.� Presentation by the Department of Rural Health Management, China Ministry of Health, to World Bank Health Financing Study Tour, Beijing, July, 2009. WHO (World Health Organization). 2004. “Reaching Universal Coverage via Social Health Insurance: Key Design Features in the Transition Period.� Discussion Pa- per No. 2, WHO, Geneva. ———. 2010. The World Health Report: Health Systems Financing: The Path to Universal Cov- erage. Geneva: WHO. World Bank. 2010. Eighth Poverty Reduction Support Credit Program Document. Washing- ton, DC: World Bank. ———. 2011a. Country Assistance Strategy for Tanzania. Washington, DC: World Bank. ———. 2011b. “Tanzania Pension Policy Note.� Washington, DC: World Bank. Yazbeck, A. 2009. A acking Inequality in the Health Sector: A Synthesis of Evidence and Tools. Washington, DC: World Bank. ECO-AUDIT Environmental Bene�ts Statement The World Bank is commi ed to preserving In 2010, the printing of endangered forests and natural resources. these books on recycled The Office of the Publisher has chosen to paper 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 * 40 feet in height and 6–8 inches in www.greenpressinitiative.org. diameter M aking Health Financing Work for Poor People in Tanzania 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. Health �nancing in Tanzania is currently highly fragmented, with many different sources of funds and programs directed at speci�c population groups. Yet many people still do not have access to health services because of �nancial barriers. This burden falls disproportionately on the poor. This study looks at the current situation with respect to health �nancing as well as experience in other countries to address health �nancing not only for the poor, but also for the general population, using a common analytical framework. The study then explores a number of policy options in the areas of revenue generation, pooling of funds, purchasing, and service provision, and also looks at the regulatory and political environment, making speci�c recommendations for consideration in each of these areas. The focus of these recommendations is on improving �nancial health protec- tion for the poor. Economic, �nancial, and service delivery implications are then examined, using several different scenarios for extending pre-paid health insurance coverage to the population. It is hoped that this study will help stimulate the discussion of options and will also help Tanzania develop a health �nancing strategy to meet its long-term needs. Making Health Financing Work for Poor People in Tanzania will be of interest to readers working in the areas of health care and public health, social protection, and social analysis and policy in Tanza- nia and in other countries aiming to improve their health �nancing systems. World Bank Studies are available individually or on standing order. The World Bank Studies series is also available online through the World Bank e-library (www.worldbank.org/elibrary). ISBN 978-0-8213-9473-1 SKU 19473