MADAGASCAR PUBLIC EXPENDITURE REVIEW 2014 HEALTH SECTOR BACKGROUND PAPER Version: September 17, 2015 TABLE OF CONTENTS ACKNOWLEDGEMENTS ................................................................................................................................................. V SECTION A. PRESENTATION OF THE HEALTH SYSTEM AND HEALTH NEEDS .................................................................6 1. Health Sector Objectives and Organization..................................................................................................6 Figure 1. . Health Facility and Hospital Pyramid Referral System .................................................................7 2. Recent trends in health outcomes ................................................................................................................8 3. The State of Health Service Delivery: Indicators and Equity .........................................................................9 1.. Figure 2: Routine Data, Immunization Coverage for DPT, Polio and Measles for children under one, 2008- 2012 10 SECTION B. TOTAL PUBLIC EXPENDITURES: SIZE AND OVERALL PERFORMANCE .......................................................15 B.1. HEALTH FINANCING ...............................................................................................................................................15 1. Size and composition of total health financing ..........................................................................................15 Figure 3. Trends in Total Health Expenditures in GDP 1995-2012, International Comparisons .....................15 2. Total public financing sources and their evolution since 2009 ...................................................................16 Figure 5. Comparing Public Education and Health Financing over 2009-2013 ..............................................17 B.2. GOVERNMENT HEALTH EXPENDITURES ......................................................................................................................18 1. General government expenditures allocated to health: international comparisons..................................18 Figure 6. Trends in Total Government Health Expenditures (Public Financing), 1995-2012 ..........................18 2. Executed expenditures by the Ministry of Health, recent trends ...............................................................19 Figure 7. Executed Expenditures of the Ministry of Health, 2006-2013............................................................19 Figure 8. Share of MoH in Government-Executed Expenditures, 2006-2013 ...................................................20 B.3. BUDGET EXECUTION..............................................................................................................................................20 1. Execution rates ...........................................................................................................................................20 Execution of the health budget ...........................................................................................................................23 2. Deconcentration of MoH expenditure ........................................................................................................24 Figure 9. Deconcentration of MoH Current Non-wage Expenditures, 2006-2013............................................25 SECTION C. ELEMENTS TO ASSESS THE EFFICIENCY OF MOH EXPENDITURES ...........................................................30 C1. ASSESSING TECHNICAL EFFICIENCY THROUGH AN ANALYSIS OF INPUT SHARES ....................................................................30 Figure 11. Decomposition of MoH Expenditures into Broad Input Categories, 2006-2013 .............................33 C2. ELEMENTS TO ASSESS ALLOCATIVE EFFICIENCY: EXPENDITURE ANALYSIS BY FUNCTION ............................33 1. Data and level of analysis ..........................................................................................................................34 2. Functional allocation of Total Health Expenditures (NHA) .........................................................................35 3. Functional allocation of public health expenditures (NHA 2010) ...............................................................37 4. Functional allocation of Ministry of Health expenditures: wage expenditures .........................................38 Figure 12. Distribution of Salaries by Facilities and Administrative Levels, 2013 ...........................................39 Figure 14. Allocation of Salaries by Level of Care and Support Activities by Type of Facility, 2013 ...............41 Figure 15. Allocation of Salaries by Level of Care, 2013 ..................................................................................41 5. Allocation of salaries by type of personnel .................................................................................................41 Figure 17. Estimated Change in Total Wage Expenditures by Category of Personnel, 2006-2013 ..................42 6. Functional Allocation of Ministry of Health Expenditures: non-wage expenditures ..................................43 Figure 18. Allocation of Non-Wage Recurrent Health Expenditures by Budget Program, 2010-2013 .............44 Figure 19. Allocation of Health Expenditures by Program —Investment Expenditures, 2010-2013 .................44 Figure 20. Classification of Non-Wage Expenditures by Program (recurrent+investment), 2010 and 2013 ...45 i Figure 21. Allocation of Non-Wage Recurrent Health Expenditures by Level of Care, 2006-2013 ..................46 Figure 22. Distribution of Non-Wage Expenditures by Program, 2008 and 2013 ............................................47 Figure 23. Allocation of Non-Wage Recurrent Health Expenditures by Type of Activity, 2006-2013 ..............48 Figure 24. Allocation of MoH Non-Wage Expenditures (incl. Investment Expenditure) by Type of Activity, 2006-2013 ...........................................................................................................................................................48 Figure 25. Evolution of MoH Expenditures Targeted to Specific Diseases, 2006-2013 .....................................49 C3. ELEMENTS OF A PRODUCTIVITY ANALYSIS ...................................................................................................................51 Figure 26. Unit Costs by Type of Facility, 2013 ................................................................................................51 Figure 27. Expenditure Shares vs Utilization Shares by Type of Facility, 2013 ...............................................52 SECTION D. DISTRIBUTIONAL ANALYSIS OF PUBLIC HEALTH EXPENDITURE .............................................55 D1. DISTRIBUTION OF MOH EXPENDITURES BY REGION AND TYPE OF RESIDENCE ...........................................................55 1. Regional allocations vs population shares ...............................................................................................55 Figure 28. Current MOH Health Expenditure Per Capita by Province in Relation to Population ..................55 Figure 30. MoH Distribution of Wage Expenditure across Regions, 2013 .......................................................57 2. Relationship between current expenditures and poverty by region ........................................................57 Figure 31. MoH Recurrent Expenditure Per Capita by Region and Poverty Rate, 2006-2010 .........................58 Figure 32. MoH Recurrent Expenditure Per Capita on Primary Care by Region, 2013 ...................................60 3. Rural/urban differences .............................................................................................................................61 D2. OUT-OF-POCKET HOUSEHOLD EXPENDITURES: DO THEY IMPACT REGIONAL AND INCOME INEQUALITIES? ..................62 1. Distribution of OOP expenditure by income level and region ...................................................................62 Figure 34. Household Out-of-Pocket Expenditure, 2005, 2010 and 2012 .........................................................63 2. Relationship between OOP spending and MoH expenditures using regional data ....................................65 3. Analysis of “catastrophic” OOP expenditure ............................................................................................66 D3. DISTRIBUTION OF MOH EXPENDITURES BY SOCIOECONOMIC CATEGORY OF USER: BENEFIT INCIDENT ANALYSIS .........68 1. Marginal benefit of government expenditure by type of facility .................................................................69 Figure 40. Lorenz Curves by Quintile for Utilization of Public Health Facilities, 2005-2010..........................69 Figure 41. Lorenz Curves by quintile for Utilization of Public Health Facilities, 2005-2012 ..........................70 2. Benefit incidence of MoH expenditure .......................................................................................................70 Figure 43. Estimated Distribution of Benefits from MoH Non-Wage Expenditures ..........................................72 Figure 44. Estimated Distribution of Benefits from MoH Expenditures ............................................................73 Figure 45. Estimated Distribution of Benefits: Simulations of Four Scenarios .................................................74 D4. ASSESSMENT OF GENDER EQUALITY ......................................................................................................................74 Figure 46. Utilization of Health Facilities by Gender, 2005 and 2010 .............................................................75 REFERENCES......................................................................................................................................................... 78 ANNEX 1. TABLE SUPPLEMENT ............................................................................................................................ 79 ANNEX 2. FIGURE SUPPLEMENT ........................................................................................................................... 83 LIST OF TABLES TABLE 3. PUBLIC HEALTH EXPENDITURES: EXTERNAL AND INTERNAL FINANCING, 2009-2013 .................................16 TABLE 4. PUBLIC HEALTH EXPENDITURES: EXTERNAL AND INTERNAL FINANCING, 2009-2013 .................................17 FINAL AND INITIAL BUDGET APPROPRIATIONS IN HEALTH AND OVERALL, 2006-2013 ...............................................22 TABLE 4. EXECUTION OF THE MOH BUDGET VERSUS EXECUTION OF THE OVERALL BUDGET, 2006-2013 .................23 TABLE 5. MOH EXECUTION RATES FOR NON-WAGE RECURRENT EXPENDITURES, BY PROGRAM, 2010-2013 ...........24 TABLE 6. MOH EXPENDITURES BY BROAD INPUT CATEGORIES (BUDGET CLASSIFICATIONS) , 2006-2013 ..................31 TABLE 7. INPUT SHARES ACROSS BUDGET CATEGORIES, SHARE OF INTERNALLY FINANCED MOH EXPENDITURES, 2006-2013 ...........................................................................................................................................................32 TABLE 8. ALLOCATION OF TOTAL HEALTH EXPENDITURES ACROSS FUNCTIONS (ALL SOURCES) ...............................36 ii TABLE 9. SHARE OF HEALTH EXPENDITURE ON INPATIENT CARE, INTERNATIONAL COMPARISON .............................36 TABLE 10. ALLOCATION OF CURRENT EXPENDITURES OF THE PUBLIC ADMINISTRATION SYSTEM .............................38 TABLE 10. SHARES OF THE RECURRENT NON-WAGE BUDGET BY LEVEL OF CARE, 2006-2013 ...................................46 TABLE 11. ALLOCATION OF NON-WAGE MOH EXPENDITURES (INCLUDING PIP) BY LEVEL OF CARE, 2006-2013 .....46 Annex Tables TABLE A1. INTERNATIONAL COMPARISONS OF PUBLIC HEALTH EXPENDITURES (INCLUDING EXTERNAL FUNDS) ...79 TABLE A2. EVOLUTION OF BUDGET ALLOCATIONS BY ADMINISTRATIVE LEVELS REPORTED BY MOH ..................80 TABLE A3. LIST OF CATEGORIES USED IN THE ALTERNATIVE CLASSIFICATION OF MOH EXPENDITURES................80 TABLE A4. TRACKING OF SPECIFIC ALLOCATION OF INTEREST IN NON-WAGE RECURRENT EXPENDITURE ............81 TABLE A5. PIP EXPENDITURES IN SPECIFIC CATEGORIES, 2008-2013 .....................................................................81 TABLE A6. ANNUAL MOH SALARIES BY PROVINCE AS A RATIO TO SIGFP REMUNERATIONS BY PROVINCE...........82 LIST OF FIGURES FIGURE 1. HEALTH FACILITY AND HOSPITAL PYRAMID REFERRAL SYSTEM ................................................................ 7 FIGURE 3. TRENDS IN TOTAL HEALTH EXPENDITURES IN GDP 1995-2012, INTERNATIONAL COMPARISONS ............. 15 FIGURE 5. COMPARING PUBLIC EDUCATION AND HEALTH FINANCING OVER 2009-2013 ............................................ 17 FIGURE 6. TRENDS IN TOTAL GOVERNMENT HEALTH EXPENDITURES (PUBLIC FINANCING), 1995-2012.................... 18 FIGURE 7. EXECUTED EXPENDITURES OF THE MINISTRY OF HEALTH, 2006-2013 ....................................................... 19 FIGURE 8. SHARE OF MOH IN GOVERNMENT-EXECUTED EXPENDITURES, 2006-2013 ................................................ 20 FIGURE 9. DECONCENTRATION OF MOH CURRENT NON-WAGE EXPENDITURES, 2006-2013 ...................................... 25 FIGURE 11. DECOMPOSITION OF MOH EXPENDITURES INTO BROAD INPUT CATEGORIES, 2006-2013 ......................... 33 FIGURE 12. DISTRIBUTION OF SALARIES BY FACILITIES AND ADMINISTRATIVE LEVELS, 2013 ................................... 39 FIGURE 14. ALLOCATION OF SALARIES BY LEVEL OF CARE AND SUPPORT ACTIVITIES BY TYPE OF FACILITY, 2013 .. 41 FIGURE 15. ALLOCATION OF SALARIES BY LEVEL OF CARE, 2013............................................................................... 41 FIGURE 17. ESTIMATED CHANGE IN TOTAL WAGE EXPENDITURES BY CATEGORY OF PERSONNEL, 2006-2013 .......... 42 FIGURE 18. ALLOCATION OF NON-WAGE RECURRENT HEALTH EXPENDITURES BY BUDGET PROGRAM, 2010-2013.. 44 FIGURE 19. ALLOCATION OF HEALTH EXPENDITURES BY PROGRAM—INVESTMENT EXPENDITURES, 2010-2013 ...... 44 FIGURE 20. CLASSIFICATION OF NON-WAGE EXPENDITURES BY PROGRAM (RECURRENT+INVESTMENT), 2010 AND 2013 .................................................................................................................................................................... 45 FIGURE 21. ALLOCATION OF NON-WAGE RECURRENT HEALTH EXPENDITURES BY LEVEL OF CARE, 2006-2013 ....... 46 FIGURE 22. DISTRIBUTION OF NON-WAGE EXPENDITURES BY PROGRAM, 2008 AND 2013 ......................................... 47 FIGURE 23. ALLOCATION OF NON-WAGE RECURRENT HEALTH EXPENDITURES BY TYPE OF ACTIVITY, 2006-2013 .. 48 FIGURE 24. ALLOCATION OF MOH NON-WAGE EXPENDITURES (INCL. INVESTMENT EXPENDITURE) BY TYPE OF ACTIVITY, 2006-2013 .......................................................................................................................................... 48 FIGURE 25. EVOLUTION OF MOH EXPENDITURES TARGETED TO SPECIFIC DISEASES, 2006-2013 ............................... 49 FIGURE 26. UNIT COSTS BY TYPE OF FACILITY, 2013 .................................................................................................. 51 FIGURE 27. EXPENDITURE SHARES VS UTILIZATION SHARES BY TYPE OF FACILITY, 2013.......................................... 52 Annex Figures FIGURE A.1. MOH RECURRENT EXPENDITURE PER CAPITA BY REGION VS PER-CAPITA CONSUMPTION, 2006-2010 .........................83 iii LIST OF BOXES BOX 1. METHODOLOGICAL ISSUES: ACCOUNTING FOR EXTERNAL FUNDING ..........................................................................16 BOX 2. BUDGET RECTIFICATION ...................................................................................................................................21 BOX 3. MEASURING THE DECONCENTRATION OF HEALTH EXPENDITURE USING SIGFP ...........................................................24 BOX 4. SUBNATIONAL FINANCING BOTTLENECKS ON VACCINATION .....................................................................................25 BOX 5. ANALYZING INPUT SHARES IN HEALTH .................................................................................................................30 BOX 6. ALLOCATIVE EFFICIENCY ANALYSIS USING FUNCTIONAL ALLOCATIONS OF HEALTH EXPENDITURES ...................................33 BOX 7. CONTRIBUTION OF DIFFERENT DATA SOURCES TO UNDERSTANDING HEALTH EXPENDITURE BY FUNCTION IN MADAGASCAR 34 BOX 8. FINANCING VACCINATION IN MADAGASCAR .........................................................................................................50 iv ACKNOWLEDGEMENTS xxx v Madagascar PER – Health |Section A SECTION A. PRESENTATION OF THE HEALTH SYSTEM AND HEALTH NEEDS 1. Health Sector Objectives and Organization 1. Since the crisis, the health sector has suffered from a lack of strategic leadership. From 2009 to 2014, there were four Ministers of Health appointed, The objectives of the National Health Strategy, which ended in 2011, was informally extended with no interim strategy put in place. This resulted in a general loss of direction in the sector and fragmentation of coordination and funding among partners. In the first year after the elections and the placement of the new Government, the Ministry of Health went through a period of transition. In March 2014, the Prime Minister was also appointed the Minister of Health. In that past year, some key developments have taken place including the launch of the development of the new health sector strategy and the revitalization of the International Health Partnership1 in Madagascar. 2. As of March 2015, a new Minster of Health and Secretary General of the MOH have been appointed with the key objectives of re-instilling strategic direction in the sector. The new Health Sector Strategy (2015-2019) is awaiting final Government validation. There are six strategic axes2 and the estimated budget needed when prioritizing maternal and child health interventions is estimated to be US$1.4 billion over the next five years. The Ministry has also explicitly committed to a the development of a Universal Coverage Strategy by the end calendar year 2015 with the first mission already having taken place. There are important challenges in the overall budget envelope and the coordination of financing (much of it being external financing) which should be addressed as they are key drivers to the success of the implementation these strategies. 3. The public health sector is organized in a pyramid structure, with four tiers of access to health services (Figure 1). The system is organized around 112 health districts, which correspond to administrative units referred to as Fivondronana, each representing approximately 100,000 inhabitants. Health services can be accessed at four different levels: basic health centers (Centre de Santé de Base: CSB) I and II; district referral hospitals (Centre Hospitalier de référence de District: CHRD) without surgery and with surgery; regional referral hospitals (Centre Hospitalier de référence Régionale: CHRR); and university hospitals (Centres Hospitaliers Universitaires: CHU) including specialized centers. Each health district typically contains 10 to 25 primary care centers and a hospital. The districts are divided into service areas for 1 Prior to the crisis, in December 2008, the IHP+ Compact had been signed by the Government and 22 development partners agreeing to overarching principles of coordination as a first step to a pooled financing approach. This was unable to move forward due to the start of the political crisis in January 2009. 2 a) improving the geographical and financial access of the population to high-quality health interventions; b) stimulating demand and use of services; c) equitable coverage and quality of health infrastructure; c) participation, coordinated and efficient government, TFP, Civil Society, the Regional and Local Authorities (CTD) and the community in the implementation and financing of health interventions; d) decentralization / devolution of the health system; e) Improving the Health Information System, which requires the acquisition of reliable health data, prompt and available to all stakeholders, reverse information, analysis of indicators for monitoring and evaluation of results for the purpose of appropriate decision-making; f) the implementation of high impact interventions to accelerate the reduction of maternal and infant mortality, reduce the prevalence of major communicable diseases (HIV / AIDS, malaria and tuberculosis), those of neglected diseases and non-communicable diseases; 6 Madagascar PER – Health |Section A community health centers (CSB 1 and CSB 2). CSB1s are managed only by paramedical staff whereas CSB2s are managed by a doctor and paramedical staff. In 2012, there were 3,074 functional CSBs and 150 CHRDs, including approximately 90 with surgical capacity (categorized as CHRD with surgery). Figure 1. Health Facility and Hospital Pyramid Referral System CHU 6 Provincial level • Specialized medical or surgical cases hospitals CHR • Complicated surgery cases 16 Regional level hospitals 150 Central District Hospitals • Complicated medical cases and surgery • Only 90 of the 150 district level health facilities 60 CHD1 and 90 CHD2 have surgical equipment. CSB I and CSB II • Simple medical cases and prevention 3074 Commune level health facilities Source: Annuaire des Statistiques du Secteur Santé, 2012. 4. The bottom 40% of the population access services at the CSB1 and CSBII levels (the population at th the 4 quintile and under in Madagascar are considered poor given the high poverty rates in the country). That said, almost 70 percent do not seek care when ill due to low quality of services and out-of pocket costs on medical consumables and services, long geographic distances to health facilities, and drug stock-outs. Higher level (tertiary care) facilities provide more primary care interventions than any other interventions and they are only frequented by the richest quintile of the population. Interestingly, most interventions delivered at that level are primary care interventions which points to an imbalance in the type of facility and the types of interventions delivered there. 5. Over the last decade, there has been increasing reliance on the private sector for health service delivery (Table 1). Private health facilities fall into two categories: not-for-profit, managed by faith-based groups or NGOs; and the for-profit health clinics, managed by private individuals. All not-for- profit primary health centers are required to adhere to Ministry of Health (MoH) norms and regulations, and must integrate their work programs into the district health planning.3 Between 2001 and 2010, the number of CSB2’s in the private sector doubled and the number of private hospitals (CHD-2) almost tripled. 3 Sharp, Maryanne; Kruse, Ioana. 2011. Health, Nutrition, and Population in Madagascar 2000-09. World Bank. 7 Madagascar PER – Health |Section A The private sector has been slowly gaining ground particularly in urban areas - with a consultation rates among private physicians in these areas increasing from 16 % in 2005 to almost 20% in 2010.4 6. In 2003, the National Health Accounts (NHA) estimated that public health providers accounted for 67 percent of all providers in terms of total health expenditures, versus 28 percent for private providers (5 percent were unidentified). The 2007 and 2010 NHAs do not separate providers into public and private, but the number of financing agents indicates that after a relative increase in utilization of the public sector between 2003 and 2007, the trend reversed in 2010. In addition, according to the 2010 NHA, nearly 58 percent of health expenditures can be attributed to the private financing5. With regards to out-of-pocket expenditure (OOP) for health, 2013 NHA-lite data indicate that OOP is an 80% share of private financing and 30% of total health expenditure. Table 1. Share of Private and Public Sector in Health, 2003-2010 2003 2007 2010 Public sector 55 62 40 Private sector 40 36 58 Rest of the World 5 2 2 Note: 2010 NHA figures are based on Current Health Expenditures (HE) instead of Total Health Expenditures (THE); Capital expenditures represented 7 percent of THE in 2010 and can be mostly attributed to the public sector. Source : Adapted from Madagascar NHA reports 2003, 2007 and 2010. 7. Risk-pooling mechanisms remain largely underdeveloped with the cost of most medical consumables borne by patients through cost recovery. Though general revenue financing provides an opportunity for implicit risk pooling and redistribution of resources, government resources are inadequate to meet all the needs. And while risk pooling mechanisms provide an opportunity for financial protection and more equitable distribution of resources, these mechanisms remain scarce in Madagascar. Furthermore, private/voluntary health insurance is limited, as only a minority of the population is willing and able to afford unsubsidized voluntary insurance given the small formal employment base. However, international experience with community-based health insurance, which has existed in Madagascar for more than ten years, suggests that such schemes can form part of a transition to a more universal health care coverage system. But there are also shortcomings with community-based health insurance due to limited incomes of community members and voluntary membership, which reduces the size of the risk pool. 2. Recent trends in health outcomes 8. Since the start of the political crisis in 2009, key social indicators declined dramatically and Madagascar is now unlikely to meet any of the health MDGs. Between 2002 to 2008, Madagascar made considerable progress on the social MDGs, and it seemed likely that the fourth MDG, on under-five 4 UNICEF 2013 Sitan. 5 However, it should be noted that it is not clear whether this increase is directly related to provision of health services through the private sector or whether households are financing more out-of-pocket expenses in the public sector. 8 Madagascar PER – Health |Section A mortality, would be achieved—child health improved, and under-five mortality rates declined from 163 per 1,000 live births in 1997 to 72 per 1,000 live birth in 2008/096. Madagascar had also started to tackle some persistent challenges, such as improving maternal health and reducing stunting among children caused by chronic malnutrition. Since 2009, however, some key health outcomes have declined sharply; and since 2012, the rate of acute malnutrition in increases in some of the most food insecure regions of country has risen by more than 50 percent. The prevalence of chronic malnutrition among children under five is one of the highest in the world—53 percent are stunted7 (short for their age) and 5.8 percent are wasted (too thin for their height).8 Maternal mortality ratios also have remained relatively high and stagnant over the last ten years: from 469 per 100,000 live births in 2003 to 478 per 100,000 live births in 2012. In 2010, pregnant women and children under five bore almost 40 percent of the total disease burden in the country. 9. Madagascar’s epidemiological profile remains comparable to many low-income countries with a high communicable disease burden, including neglected tropical diseases (NTDs), with the burden of disease falling disproportionately on the poor. About 0.49 percent of all TB cases are Multidrug Resistant TB (MDR-TB). Malaria is not as widespread as in most Sub-Saharan countries and its incidence has declined over the past few years as a result of prevention activities, but there have been spikes in the number of cases in recent years. Over the past decade, non-communicable diseases are increasing in the population, resulting in a dual burden of disease which will tax an already fragile health system. 10. There are persistently high total fertility (TFR) and population growth rates (4.6 births per woman and 2.8 percent growth respectively) with significant variations by location and income quintiles. For example, the TFR for rural areas is almost double that of the capital (5.2 versus 2.7 while that for the poorest quintile is 2.5 times that of the richest. The percentage of adolescents having given birth is almost 4 times higher for the poorest quintile than for the richest. Not surprisingly, the contraceptive prevalence rate (modern methods) for the richest quintile is double that of the poorest and the unmet need for contraception is 41 percent higher for the poorest quintile than the richest. While there has been a significant increase in the utilization of family planning (from 18 percent in 2003/04 to 47 percent in 2012), the significant unmet need and the inequitable availability of family planning services – as well as the non- health related implications of continued high population growth highlighted – makes this a concerning issue. 3. The State of Health Service Delivery: Indicators and Equity 11. Coverage of essential health services, especially key maternal and child health services, is very low and has further deteriorated since the crisis. Access to quality prenatal and antenatal care is a persistent and increasingly serious challenge, with only 38 percent of births taking place in a health center, and of those, only 44 percent attended by skilled personnel.9 This is lower than the average in the 6 The 2012 MDG Survey showed the child mortality rate to be 67 per 1000 live births. With the confidence intervals noted in the survey, this is about the same rate as 2008/2009. 7 Long-Term Anthropometric Study (2012). This is particularly concerning in that stunting compromises human capital and has long-term negative impacts on productivity and resilience due to irreparable cognitive and physical deficits. 8 UNICEF 2012 9 MDG Survey 2012/2013. 9 Madagascar PER – Health |Section A developing world, where about 58 percent of all deliveries are attended by skilled health providers and more than 50 percent of births taken place in health centers. Population policies aimed at reducing fertility by encouraging family planning brought about a 15 percent increase in the use of modern methods of contraception prior to 2009, but the contraceptive prevalence rate is still very low at 33 percent. 12. Immunization coverage has also substantially decreased since the crisis. Immunization coverage is one of the main predictors of the infant mortality rate and also it can be used as a proxy indicator for the availability of primary health care in a country. According to the Demographic Household Survey (DHS) 2008/2009, complete immunization coverage for children 12 to 23 months old in 2008 was 62 percent. Comparable data from the MDG survey in 2012 indicates a decrease in complete immunization coverage to 51.1 percent in just four years. Furthermore, this downward trend is also confirmed by other country-specific data sources. The regional MICS (2012)10 indicates a decrease in total immunization coverage to about 33.4 percent in some of the poorest areas of the country. Routine data collected from health facilities also confirms the drop with one of the greatest decreases seen in coverage of Bacillus Calmette–Guérin (BCG vaccine for tuberculosis) over the last five years (Figure 2). 1. Figure 2: Routine Data, Immunization Coverage for DPT, Polio and Measles for children under one, 2008-2012 75.4 Measles 63.3 72.5 DTC3 64.8 73.9 Polio3 64.4 83.8 BCG 61 50 55 60 65 70 75 80 85 90 2008 2009 2010 2011 2012 Percent Source: Health statistical yearbooks, 2008-2012. 13. Health Service Delivery is highly inequitable in Madagascar. In 2010, pregnant women and children under five bore almost 40 percent of the total disease burden in the country. Looking at specific outcomes such as child mortality, the bottom 40% of the population, bears most of the burden (Figure 3). Similarly, key service delivery indicators such as skilled birth attendance are 40% lower among the poorest two quintiles as compared to the richest quintile further highlighting the equity issues in accessing and utilizing care (Figure 4). Figure 3&4: Infant/Child Mortality and Percentage of Live Births Attended by Skilled Health Personnel by Quintile 10 The Madagascar MICS4, funded by the World Bank and UNICEF, was carried in 2012 by the National Institute of Statistics in the southern part of the country. The survey focused on four southern regions in Madagascar Androy, Anosy, Atsimo Andrefana, and Atsimo Atsinanana with a representative sample of 2,897 households. The MICS collected data on household members, housing characteristics, information on women between the age of 15-49 years and children under five. 10 Madagascar PER – Health |Section A 14. These trends are further confirmed by an equity analysis using results from the regional 2012 MICS survey which covered five of the poorest regions in Madagascar shows that that maternal and child health interventions such as antenatal care and immunization are reaching mostly higher quintiles of the population. Table 2. Coverage of select maternal and child health interventions according to income quintile. q1 q5 Maternal and Child Health Interventions q2 q3 q4 Total (Poorest) (Richest) Full immunization 23.2 24.7 33.8 37.9 50.9 33.4 Treatment of Diarrhea 31.2 28 27 43.1 39.7 33.2 Medical treatment of Acute Respiratory Infection 63.8 53.1 56.8 63.6 78.5 61.6 Mosquito net use by children 61.4 56.2 62.2 66.4 68.5 62.4 Skilled antenatal care 63.3 68.4 76.7 81.1 91.2 74.6 Skilled antenatal care (4+ visits) 34.5 37.9 40.6 48.8 63.7 43.3 Skilled birth attendance 10.6 15.2 25.7 39 71 28.5 Mosquito net use by pregnant women 55 68.5 68.3 64.2 68.5 64.9 15. Inequitable health service delivery has two critical dimensions in the context of Madagascar: a) affordability and b) accessibility. a)Affordability: Poor and unequal health outcomes find their root in disparities in terms of household income. Out-of pocket costs for service delivery have risen as greater numbers of households are falling deeper into poverty resulting in a population that is more vulnerable and a greater risk of falling and staying in poverty by paying for health services. The 2012 MDG Survey found that financial barriers and distance are key constraints in accessing quality health care. With the exception of the cost of prescriptions, primary care services are delivered free-of-charge at facility level, yet in practice, the use of health services also entails high out-of-pocket costs such as for supplies, medical consumables, and transportation. 11 Madagascar PER – Health |Section A In an effort to address some of the issues around financial barriers in accessing health services, a Health Equity Fund was created by the Government to provide drugs free-of charge to the poorest. Under the cost-recovery mechanism, called “FANOME”, at the health center level, a small percentage of funds (from the sale of drugs) is placed in an equity fund designed to provide free access to medicines for the most vulnerable population without adding burden to the health budget. By design, its solvency is directly tied to the population’s utili zation rate of health centers. There are large variations in the financial sustainability of equity funds, even within a district, and the overall coverage is very low. However, financial sustainability is not the biggest challenge to the effectiveness of the equity fund; in fact, funds are underutilized because targeting of eligible individuals is very difficult and linked to cultural nuances around being targeted as a poor person. b) Accessibility: Geographic barriers exist and are likely getting worse. Numerous communities are seasonally isolated for months at a time, leaving entire populations – not only the poor – with little access to health centers. Even those isolated communities with a health center suffer during the rainy season, since referrals to hospitals are impossible, replenishment of drugs is slower, and supervisory visits are virtually non-existent. Prior to the crisis, the WHO reported that only 60% of the Malagasy population had access to health facilities. By 2013, approximately 856 primary health care facilities (CSB1s), most accessed by the poor, had closed down due to the impacts of the crisis. In addition, nearly 78 percent do not have the ability to transport patients to hospitals for further treatment. The issue challenge geographic access is particularly concerning given the links to maternal mortality and morbidity outcomes. 16. Human Resource Allocation and Organization is also not equitable. According to the NHA (2010), there are approximately 2 doctors per 10,000 people in Madagascar. There are three times more doctors in urban areas than in rural areas. In the public sector, the highest concentration of doctors is in urban centers in Antananarivo province. On average, a public health center in urban area has twelve employees out of which three are doctors and an additional five are medically trained staff. On the other hand, public rural centers are staffed on average with only two people of which there may or may not be a doctor. Efforts to recruit general practitioners and nurses to work in rural areas has had limited success and there has been a reduction of 50 percent in the number of midwives between 2007 and 20011. In addition, the ratio of nurses/midwives compared with physicians is very low (range between 1.9-2.0 between 2007 and 2011). Data from the MOH indicate approximately 640 rural health facilities have closed since 2007 due to a lack of personnel. Of the CSBs that are functional, over 30% are not compliant with staff requirements as set by the Ministry. An added issue is that nearly 50 percent of public health sector staff is over 50 years old and will retire in less than ten years. Current health sector human resource policies do not address this future constraint to service delivery. 17. The quality of health service delivery is low, especially in rural parts of the country. While there are many services and system components that suffer from poor quality, health service delivery at primary care levels can be quantified around a few key indicators and these reflect major health system issues in Madagascar: a) Provider performance in compliance with diagnostic procedures is low: Observations of consultations found very few medical professionals, 15 percent, followed the correct protocol of consultations and recorded all the basic information relating to the children under five. This was also true for antenatal care consultations where only 18 percent of medical professionals followed consultation protocol. b) Supervision and monitoring functions, especially at lower levels, are weak: there is large variation in supervision across urban/rural and geographic regions with urban centers better 12 Madagascar PER – Health |Section A supervised than rural centers. Level 2 district hospitals are supervised regularly (96 percent), while basic health facilities are the least supervised (63 percent). c) Limited availability of key supply side inputs: More than 66 percent of health facilities reported at least one essential medicine was not available at the time of the health facility survey. The duration for essential medical stock-out is as high as ninety days. Key Findings Since the start of the political and economic crisis in 2009, progress made on key health indicators has stagnated or is being reversed with Madagascar falling off track to achieve the MDGs. The prevalence of chronic malnutrition among children under five is one of the highest in the world. Maternal mortality ratios also have remained relatively high and stagnant over the last ten years and the country. Contextual weaknesses  Madagascar’s epidemiological profile remains comparable to many low-income countries with a high communicable disease burden. Almost 30 percent of all deaths in Madagascar are still attributable to preventable and infectious and parasitic diseases The system is plagued by inequitable health service delivery. The two critical dimensions are:  Affordability: As greater numbers of people have fallen into poverty, there have been two likely implications on the population: i) the poor are more vulnerable and have a greater risk of falling and staying in poverty by paying for health services and ii) less of the population is seeking health services due to an inability to pay.  Accessibility: Numerous communities are seasonally isolated for months at a time, leaving entire populations – not only the poor – with little access to health centers. In addition there are major inequities in HRH distribution with the greatest negative impact on the poor who access first level primary care facilities. An added issue is that nearly 50 percent of public health sector staff is over 50 years old and will retire in less than ten years. Current health sector human resource policies do not address this future constraint to service delivery. Structural weaknesses  The quality of health service delivery is low, especially in rural parts of the country. Critical challenges include: (a) weak provider compliance with diagnostic procedures; (b) weak supervision and monitoring functions; and (c) lack of availability of key supply-side inputs.  High-out of pocket costs and scarcity of risk pooling mechanisms make it difficult for the poor to access care Policy implications There are several short and medium recommendations that Madagascar should consider implementing as a matter of priority: I. Promote equitable access to health services with a focus on delivering an essential package of high, impact maternal and child health and nutrition interventions in rural areas through tailored strategies that remove key barriers to access and stimulate demand and utilization of services.  Address financial barriers to access - Remove out-of pocket costs for services at facility level - Strengthen risk-pooling and safety-net mechanisms such as the Health Equity Fund, fee exemption schemes for services and medicines and community health insurance. 13 Madagascar PER – Health |Section A  Address geographic barriers to access - Finance existing outreach activities and approaches especially in rural areas. This includes training and deploying community health workers and expanding initiatives like Stratagie Avancee which bring services into communities. Focusing at this level can promote resilience in service delivery. - Prioritize and invest in functionality of first level rural health facilities. II. Improve the quality of health services  Ensure availability of essential commodities and drugs at the primary level, as well as investment in upkeep and maintenance of health facilities, especially in rural areas.  Invest in supervision and monitoring at lower levels. This includes training and capacity building for better management.  Strengthen relevant plans to inform priority actions for improving quality to complement the National Health Sector Strategy: - Update and implement the National Human Resources Development Plan (short, medium and longer term actions) - Develop and implement standardized “Norms and Standards” for all types of health facilities at all levels - Develop and implement a National Quality Plan. 14 Madagascar PER – Health |Section B Section B. Total Public Expenditures: Size and Overall PerformanceB.1. Health Financing 1. Size and composition of total health financing11 18. Madagascar spends less on health than three quarters of the SSA LIC countries. Since 1995, the percentage of Total Health Expenditure (THE) in GDP has remained around 4-5 percent with a slight downward trend in recent years. On average, Madagascar spent, 4.3 percent of its GDP on health between 2009 and 2012, compared to 4.8 percent in the four years preceding the crisis (2005-2008).12 This was the reverse of the trend observed in other SSA countries. (Figure 3). In real terms THE per capita expenditure has not changed since 1995. Looking at the period between 1995 and 1999 THE per capita was US$21. In the period between 2010 and 2014, Madagascar’s THE per capita was US$20. This is comparable with other LICs in SSA but far below the regional average of US$8313. Figure 3. Trends in Total Health Expenditures in GDP 1995-2012, International Comparisons 10 8 6 4 2 0 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 Madagascar Average SSA Average LICs Average HIC Source: Data from WHO/GHED. 11 This section rests on data from the WHOs Global Health Expenditure Data (GHED). Although WHO’s GHED is based on NHA exercises, there are some notable discrepancies between the two sets of data that could not be explained but are likely due to differences in the WHO methodology to create internationally comparable data. The GHED numbers have the advantage of being comparable across countries and require government clearance before publication. 12 The national NHA reports present a different picture with THE at 5.6 % of GDP in 2010, up from 4.2% in 2007 and from 3.2% in 2003. The raw NHA data that was used for the different NHA exercises could not be obtained so the source of the discrepancy with the WHO data could not be identified. 13 Source: 15 Madagascar PER – Health |Section B 2. Total public financing sources and their evolution since 2009 Box 1. Methodological Issues: Accounting for External Funding Since 2009, the AMP, managed by the GoM (Primature) records actual disbursements from all bilateral and multilateral agencies, decentralized cooperation entities from France, and the bulk of international NGOs, including direct funding by foundations. Although data prior to 2009 are available for government accounts (MoF/SIGFP), they could not be used to identify the shares of internal and external financing of public expenditure because significant differences were found between on-budget disbursements of foreign aid directed to health and externally financed expenditures in the SIGFP for 2009-2013. In fact, hardly any on-budget foreign aid was included in executed expenditure, as revealed by the following ratios of disbursed foreign aid marked “on-budget” in the AMP to externally funded expenditures in SIGFP: 2009 2010 2011 2012 2013(a) Relative to budget appropriations(b) 0.72 0.16 0.33 0.20 0.14 Relative to executed expenditures 0.13 0.01 0.09 0.03 0.02 (a) The 2013 data from the AMP include disbursements entered on the platform prior to April 30, 2014. (b) Final and initial budget appropriations were equal for the PIP in the SIGFP for health. 19. Between 2009 and 2013, total public expenditures in health increased due to large off-budget external support. There are significant constraints in assessing the total amounts of public financing to health, owning to the poor accounting of external funding in the Integrated Public Finance Management System (SIGFP) (Box 1). For the purpose of this PER, the total funding directed to health, including domestic, on-budget and off-budget foreign aid, was reconstituted using alternate government sources.14 The analysis shows that health funding continued to increase after the crisis, by about 16 percent a year between 2009 and 2013. The sector received exceptionally high amounts of public funding in 2010, owing to large off-budget external support. Directly resulting in public financing to health going from 2.8 to 3.7 percent of GDP between 2009 and 2013, with a peak at 4.6 percent in 2010 (Table 2). Table 3. Public Health Expenditures: External and Internal Financing, 2009-2013 In billion of constant 2013 Ar. 2009 2010 2011 2012 2013(a) Total Public Expenditures 585 1,007 734 807 876 Percent of GDP 2.8 4.6 3.3 3.5 3.7 (a) The 2013 data from the Aid Management Platform (AMP) includes disbursements entered prior to April 30, 2014. (b) Expenditures of the MoH marked as RPI in the SIGFP. They are mostly financed by general revenue, but budget support from foreign sources is typically included in RPI and cannot be identified separately. (c) Based on actual disbursements entered in the Aid Management Platform (AMP) from international multilateral and bilateral partners as well as major NGOs. 14 Internal resources (RPI) were evaluated using data from SIGFP, while external institutional resources were assessed using the AMP. In 2006, the MFB introduced SIGFP, a modern computerized integrated financial management system to process budget execution and accounting operations across all institutions of the government. 16 Madagascar PER – Health |Section B 20. In fact, between 2009 and 2013, 80 percent of public funding to the health sector was financed through external funds (Table 3). On the basis of the data available on the Aid Management Platform (AMP), it is estimated that the share of external funding in public financing to health varied from 88 percent in 2010 to 83 percent in 2013. During this period, external funds amounted to US$164 million on average (Ar 410 billion). Table 4. Public Health Expenditures: External and Internal Financing, 2009-2013 In billion of constant 2013 Ar. 2009 2010 2011 2012 2013(a) MoH. Exp: Internal funding(b) 21% 12% 20% 17% 17% Foreign aid: On-Budget(c) 45% 49% 56% 44% 49% Foreign aid: Off-Budget(c) 35% 39% 24% 39% 34% Total Public Expenditures 100% 100% 100% 100% 100% Percent of GDP 2.8 4.6 3.3 3.5 3.7 Percent funded by aid 79 88 80 83 83 (a) The 2013 data from the AMP includes disbursements entered on the platform prior to April 30, 2014. (b) MoH expenditures marked as RPI (Ressources Propres Internes)in the SIGFP are mostly financed by general revenue, but budget support from foreign sources is typically included in RPI and cannot be identified separately. (c) Based on actual disbursements entered in the Aid Management Platform (www.amp-madagascar.gov.mg) from international multilateral and bilateral partners as well as major NGOs. 21. The extremely low share of domestic funding to the sector poses serious concerns in terms of sustainability, ownership and efficiency of existing resources. Over the last five years, domestic funding of health increased in amount, but its share in total public funds to the sector remained at around 20 percent. This is low compared with other countries, and certainly very low compared with other sectors in Madagascar. For instance, the share of domestic funding in education varied between 75 and 80 percent over the same period (Figure 5). The over-reliance on external funding is a serious concern for the sustainability of funding to the sector, especially given the volatility of aid in fragile contexts. It also potentially raises issues in terms of alignment, harmonization and overall efficiency, given the high volume of external aid provided off-budget and the absence of alignment around a new health strategy that would facilitate the alignment and harmonization of funding. Figure 5. Comparing Public Education and Health Financing over 2009-2013 Total Amounts of Pubilc Funding Share of Domestic Funding in Total 1,200 100% Public Funds 1,000 80% 800 60% Education Education 600 Health Health 40% 400 20% 200 0% 0 2009 2010 2011 2012 2013 2009 2010 2011 2012 2013 Source: Calculated from AMP disbursements and MFB/SIGFP data. 17 Madagascar PER – Health |Section B B.2. Government Health Expenditures 1. General government expenditures allocated to health: international comparisons15 22. Total Government Health Expenditure to the health sector as a share of GDP decreased sharply after 2007, but the share of government budget allocated to health remained broadly in line with regional averages. GoM spent 2.5 percent of GDP and 12.8 percent of its executed budget on health in 2012, which is roughly on par with averages of SSA and low-income countries (LICs). From 1995 to 2007, the country consistently devoted a larger than average share of its budget to health, and stayed close to the SSA average in terms of share of GDP. The period after 2007, however, has been marked by a sharp decrease in health funding as a percentage of GDP, reflecting lower overall government expenditures, and resulting in health expenditures relative to GDP falling to below the SSA (Figure 6). Figure 6. Trends in Total Government Health Expenditures (Public Financing), 1995-2012 Source: Data from WHO/GHED. 15 Internationally comparable data produced by the WHO give the relative magnitudes of general government expenditures on health both in percentage of GDP and in percentage of the budget for most countries since 1995. The figures cover all health expenditures that go through the public administration, including expenditures managed outside of the MoH. This is the only source of data on general government health expenditure. Indeed, data obtained from the MoF for health did not include other health expenditures than those in the Health “mission”. However, it is unclear if externally funded expenditures t hat end up not going through the SIGFP are included in WHO’s G HED numbers and from the reconstitution performed for this PER, it would appear that it is not the case, leading to underestimating total funding to health in the WHO data. 18 Madagascar PER – Health |Section B 2. Executed expenditures by the Ministry of Health, recent trends16 23. Public expenditures for health decreased by 31 percent over 2007-2013, due mainly to a drop in on-budget external funding channeled to MoH. Total expenditure by the MoH (externally and internally funded) decreased from 229 billion Ar in 2006 to 158 billion Ar in 2013.17,18 This decrease was entirely due to the contraction of investment financing from external aid. Even though internally financed health expenditures, in real terms, nearly doubled over the period, this was not enough to compensate for the 95 percent drop in externally financed health expenditure (Figure 7). Although, according to Government accounting, more than half of MoH expenditures had been financed by external grants and loans in 2006 (before SIGFP was in place), externally financed expenditures dropped to less than 5 percent in 2013.19 Figure 7. Executed Expenditures of the Ministry of Health, 2006-2013 250 229 204 MoH expenditure 202 195 200 MoH exp. financed by RPI Billions of 2013 Ar. 156 158 153 146 151 150 137 122 121 119 100 87 50 0 2006 2008 2009 2010 2011 2012 2013 Note: Financing from internal resources (RPI) includes budget support. All current expenditures are financed by RPI. All externally financed expenditures are in the investment budget. Source: Data from MoF/SIGFP. 24. Overall, when looking only at domestic spending, the share of MoH spending in total government spending has remained the same since 2006. The GoM devoted 6.25 percent of internally financed expenditures to health in 2013, and this allocation had remained relatively stable, in the 6-8 percent 16 The raw data was provided by the MFB; it is extracted from the SIGFP for 2009-2013 and reconstituted in the same format for 2006 and 2008 (when the SIGFP was not operational). Throughout the document, executed expenditures are taken at the mandatement level, which corresponds to the formal clearance before going for payment to the treasury. Although they cannot be compared to other countries, these figures have the advantage of relying on primary data and provide a time consistent series up to and including 2013. 17 Given the changes in responsibilities of the ministry over time, the figures only include expenses in the programs related to health (i.e., the social protection and population components of the ministry in 2008 and 2009 are excluded). This will be the case throughout the report when looking at MoH expenditures. 18 The figures calculated from government accounts excluding external funding are based on expenditures financed by RPI. They exclude external grants and loans (most of investment expenditure) but includes budget support (budget support can be identified in 2006 but was added to RPI for comparability with later years). 19 The contraction of external funding depicted here probably reflects the overall increase in off-budget external support, but it may also be the result in inaccuracies in the way externally financed expenditures go (or do not go) through the SIGFP. Therefore, caution is required when interpreted the apparent drop in MoH- executed expenditures. 19 Madagascar PER – Health |Section B range, since 2006 (Figure 8). This result is robust across the different budget categories, except for the share of MoH in civil servant wages, which increased from 8.5 to 10 percent of the civil servant wage bill between 2006 and 2010. The stagnating share of health in Government expenditures could be a signal of sluggish political commitment to health, but since we have been considering executed expenditures rather than budget appropriations, it could also be the result of weak execution of the budget. Figure 8. Share of MoH in Government-Executed Expenditures, 2006-2013 12.00 10.00 8.00 6.00 % 4.00 MoH expenditures % of the general executed budget MOH expenditure % of the general executed budget, excluding external grants and loans 2.00 MOH non-wage recurrent exp % of non-wage recurrent gov. exp. MOH expenditure on regular salaries % of executed wage budget 0.00 2006 2008 2009 2010 2011 2012 2013 Note: All current expenditures are financed by internal resources (RPI), which may include budget support. All externally financed expenditures are in the investment budget. Source: Data from MoF/SIGFP. B.3. Budget Execution 1. Execution rates 25. Some factors need to be taken into consideration when interpreting execution rates. Normally, comparing budget appropriations to executed expenditures can be a way to assess planning and governance capacity at the line ministry level, as well as the general quality of governance in budget management. For Madagascar, several issues need to be considered before interpreting execution rates. First, executed expenditures recorded in the SIGFP may not fully account for all realized expenditures. This could lead to the over- or underestimation of execution rates, depending on the performance of expenditures not recorded (or recorded differently) in the SIGFP. Second, various blockages in the budget exist that may prevent ministries from using budgeted funds that would otherwise be committed. These blockages should not necessarily be attributed to a lack of capacity in terms of budget execution, but rather to the need to improve overall management of the budget, in particular cash flows. 26. There is a lack of clarity regarding on the inclusion/exclusion of foreign aid managed by the Government. Differences between budget appropriations and expenditures could, in large part, be attributed to the way externally financed investment expenditures are recorded in the government budget. Disbursements of foreign aid marked on-budget in the Aid Management Platform are much higher than of externally financed health expenditures in SIGFP (Table 3). This can lead to an underestimation of execution rates, as some expenditures related to externally funded projects may not be included in the 20 Madagascar PER – Health |Section B government accounts at the stage of execution but are still included in budget appropriations.20 Documentation setting out the rules regarding inclusion/exclusion of foreign aid managed by the Government would be necessary to allow for better tracking of investments and to analyze execution rates more precisely. This is particularly true of investment expenditures, which, for the most part, are financed by foreign aid and therefore not fully included in the SIGFP. Therefore, it is important to separate the current and investment budgets in the analysis of budget execution. Table 5. Under-Accounting of Foreign Aid in Government Accounts, 2009-2013 2009 2010 2011 2012 2013 AMP Foreign aid disbursements marked on-budget in billion 2013 Ar. 98.5 147.0 146.0 60.8 27.5 Of which, share included in SIGFP 0.3 6.1 8.3 20.4 10.2 Data sources: Primature (AMP), MFB (SIGFP). 27. Rules related to the execution of the budget may prevent full execution, especially for non- wage expenditure; most notably the case for regulation rates imposed by MoF. Operational budgets financed by internal resources are subject to quarterly regulation rates—a maximum percentage of appropriations that can be committed by the end of each quarter—the fourth quarter rate normally being 100 percent. Civil servant salaries are subject to a linear regulation, i.e., 25, 50, 75 and 100 percent, but the quarterly percentages for other expenses in the operational budget, including indemnities, are subject to a non-linear schedule that can be changed during the year. For 2013 these rates were: 8.8, 52.1, 67.2, and 100 percent. However, at the end of September 2013, the fourth quarter regulation rate was changed to 53 percent (Arrêté 29109/2013). The fluctuations around regulation rates are largely out of the control of the MOH. This can be due to issues like other government priorities taking precedence or a reduction in the liquidity of the state’s revenue. Box 2. Budget Rectification Budget allocations within broad categories are generally modified mid-year to accommodate requested transfers of credits. A new Finance Law is not necessary if these changes follow certain criteria. Beginning every May, program organizers and coordinators can request modifications to the SIGFP to move credits. These requests need to be approved by the Ministry of Finance and Budget and published by decree before they are effective in the SIGFP. Rules differ depending on whether the transfer is (a) across programs in the same ministry; (b) across programs of different ministries; or (c) within programs. The procedure is simplified when changes are within programs, although the changes still needs to be published by decree. It is from the Law that, for operational expenses, whatever the nature of the modification requested, the total amount in each economic category (indemnities, goods and services, transfers) cannot be changed. Regulations and procedures concerning modifications are delineated in 20 Another issues identified in the course of this PER is related to the payment of Value Added Taxes (VAT) which are paid by some externally funded projects and subsequently reimbursed by the Government. It is not clear exactly where these provisions appear in the budget, but they seem to be included in the SIGFP procedure and provisions, and need to be made as part of the budget process to pay for these liabilities. Due to lack of predictability regarding the total amount of these expenditures, there would appear to be some blockages. Unfortunately, the data obtained for this PER is not sufficient to determine the magnitude of the problem. 21 Madagascar PER – Health |Section B MFB circulaires. Any change across categories requires a rectification of the Finance Law. (Restrictions, however, do not apply to externally financed credits, in particular for the investment budget.) In years when the Finance Law is not rectified (has happened in all years except 2008 and 2010), there may be differences between total appropriations initially granted to a line ministry and final appropriations, but there should be no difference by broad category (transfers, goods and services, indemnities) for operational expenditures. Based on past budget data—especially the data from 2010—it is clear that broad allocation changes are often made to initial budget appropriations. The table below gives the ratio of modified to initial budget appropriations for health and for the non-financial general budget. The differences could be due to the reduction in budgetary aid, especially since such financing appears as internal financing in the budget, as mentioned above, and cannot be identified separately. Because of this issue, caution needs to be used in interpreting execution rates. If international budget support is included as domestic financing and may be increased or reduced after the rectification of the budget, final appropriations can no longer be used as benchmarks to calculate execution rates, and the health sector cannot be identified as a net loser or a net gainer in the budget rectification process overall. Taking the above into consideration, the health sector was a net beneficiary in 2006 (but only due to additional external funding), 2009 and 2013 and a net loser in 2011 and 2012. The magnitude of these gains/losses was small overall, except in 2013, when it reached seven percent of the internally financed budget. Final and Initial Budget Appropriations in Health and Overall, 2006-2013 2006 2008 2009 2010 2011 2012 2013 Ministry of Health 1.12 0.98 1.01 0.87 0.99 0.98 1.05 Current operations: regular wages 1.09 0.96 1.01 1.01 0.98 0.97 1.07 Other current operations 0.84 1.01 1.01 0.71 1.00 1.00 1.06 Internally financed investment program 1.16 0.92 1.00 0.30 1.01 1.00 1.00 Externally financed investment program 1.27 1.00 1.00 1.00 1.00 1.00 1.00 Internally financed health budget 1.00 0.97 1.01 0.82 0.99 0.98 1.07 General non-financial budget 1.03 1.06 1.00 0.80 1.00 1.00 1.00 Current operations: regular wages 1.00 1.00 1.00 1.00 1.00 1.00 1.00 Other current operations 0.97 1.12 1.00 0.73 1.01 1.01 1.01 Internally financed investment program 0.84 1.17 1.00 0.99 0.97 0.94 0.95 Externally financed investment program 1.19 1.00 1.00 0.61 1.00 1.00 1.00 Source: Data from MFB/SIGFP. 28. On the other hand, budget reallocation and rectification may lead to overestimating the capacity of the Ministry to execute according to plans (Box 2). The observed effective fungibility between allocations and expenses across different sub-administrations and across programs is likely to increase execution rates for total expenditures, but significantly reduce the ability to track expenditures by function or at the program level. While it may be difficult to appreciate the magnitude of the problem and its impact on allocation and execution, these issues need to be taken into account when evaluating resource allocations by function at the executed level. 22 Madagascar PER – Health |Section B Execution of the health budget 29. The budget execution rates of the MoH remain low for non-wage expenditures (Table 4).21 Execution for the internally financed part of the MoH budget was lower than for the overall internally funded government budget in the last two years (2011-2013). In particular, non-wage current operations performed poorly. This was likely due to the reduction in the final regulation rate, which limited spending to half of what was allocated (Box 3), indicating that the restriction was binding for the MoH. In fact, the MoH requested more in September 2013 but could only obtain a final rate of 53.54 percent. Given the caveats raised above about the accounting of foreign aid in government accounts, it is not possible with the information obtained to interpret execution rates for the externally financed investment program. Table 4. Execution of the MoH Budget versus Execution of the Overall Budget, 2006-2013 Executed expenditures/Final appropriations 2006 2008 2009 2010 2011 2012 2013 Ministry of Health 0.71 0.56 0.48 0.65 0.63 0.69 0.68 Current operations: regular wages 0.89 0.93 0.81 0.93 0.94 0.93 0.91 Other current operations 0.94 0.97 0.72 0.97 0.77 0.92 0.69 Internally financed Investment program 0.90 1.00 0.51 0.92 1.00 1.03 0.89 Externally financed Investment program 0.49 0.13 0.17 0.06 0.27 0.13 0.12 Total Internally financed 0.91 0.96 0.70 0.94 0.90 0.93 0.87 General non-financial budget 0.61 0.60 0.49 0.74 0.67 0.80 0.81 Current operations: regular wages 0.95 0.88 0.79 0.92 0.95 0.93 0.93 Other current operations 0.94 0.61 0.68 0.94 0.84 0.97 0.93 Internally financed Investment program 0.74 0.93 0.53 0.97 0.59 1.34 1.48 Externally financed Investment program 0.24 0.26 0.12 0.12 0.15 0.10 0.04 Total Internally financed 0.89 0.76 0.68 0.94 0.84 0.98 0.98 Note: Executed expenditures = dépenses mandatées; Final appropriations = credits modifiés. Source: Data from MFB/SIGFP. 30. Execution rates for non-wage recurrent expenditure have not been equal across MoH programs, especially in 2011 and 2013 (Table 6). Execution for medical supplies and medicines and for the maternal and child health program has been consistently above 90 percent. There is no clear pattern across years for the other programs, which had execution rates as low as 53 and 45 percent in primary health services and specific disease programs, likely indicating that these two areas suffered most from the final cut in the regulation rate. 21 Budget execution at MoH is marked by the fact that the health sector is heavily financed by external funds. 23 Madagascar PER – Health |Section B Table 5. MoH Execution Rates for Non-Wage Recurrent Expenditures, by Program, 2010-2013 Program 2010 2011 2012 2013 Health, administration/coordination 0.95 0.67 0.95 0.62 Primary health services 0.99 0.85 0.85 0.53 Hospital services 0.98 0.78 0.89 0.76 Medical supplies and medicines 1.00 0.95 1.00 0.95 Specific diseases – public health 0.96 0.84 1.00 0.45 Maternal and child health 0.96 0.99 0.95 0.91 Note: Execution rates are calculated relative to final budget appropriations. Prior years are not included in the comparison because of administrative changes in the program structure. Source: Data from MFB/SIGFP. 2. Deconcentration of MoH expenditure 31. The level at which expenditures are administered can impact the overall performance of budget execution. Importantly, deconcentration of expenditures can improve the efficiency of budget execution, allowing a closer connection between administrative and operational units. But, increased deconcentration requires additional capacity at the lower levels, as well as strengthened communication between various actors along the expenditure chain. Box 3 provides the background necessary to understand how deconcentration can be measured in government accounts. Box 3. Measuring the Deconcentration of Health Expenditure Using SIGFP After independence in 1960, Madagascar’s national health system was highly centralized, following the French administrative model. The Constitution laid the groundwork for decentralization by outlining a local service delivery structure, after which the MoH and other sector ministries began to shift limited decision-making power towards lower levels of government. This was not a true decentralization, but rather a deconcentration of spending, given that these lower-level structures had little discretion in the allocation and management of resources. Nonetheless, in an effort to improve public resource management and strengthen public service delivery, the Government introduced 22 regions (doing away with 6 provinces) and progressively integrated the administration’s deconcentrated technical services under the authority of the regional Chiefs, with the objective of harmonizing all sector activities in each region through integrated regional development plans. In 2008, for the first time, resources were allocated to the regions, making them responsible for the implementation of a small part of the investment budget. The health delivery system is aligned with the country’s administrative structure and includes central, regional and district levels. Each level has defined roles and responsibilities. The central level provides strategic direction, defines policies in the sector and oversees national coordination of sector activities. The Regional Departments of Health coordinate implementation of national health policy in the region, and provide technical assistance to the districts. The District Health Authorities provide health services through the district hospitals and health centers. Some information can be inferred about deconcentration of public expenditures by looking at the level at which they are mandated in government accounts (SIGFP). The following levels of administration relevant to the sector are coded in SIGFP: the central level (Ministry), the 6 ex-provinces, the 22 regions, the 113 districts, and the public health facilities when they directly manage funds. Regular salaries are paid at the level of the 6 provincial general treasuries (plus the central level for MoH personnel), and investment expenditures are 96-99 percent centralized. This leaves non-wages expenditures to examine, representing only 20 percent of total MoH expenditures. Source: Sharp, Maryanne; Kruse, Ioana. 2011. Health, Nutrition, and Population in Madagascar 2000-09. World Bank. https://openknowledge.worldbank.org/handle/10986/5957. 24 Madagascar PER – Health |Section B 32. The central level managed between 50 and 70 percent of current non-wage expenditures during the 2006-2013 period, with no clear trend toward de-concentration except at the level of some operational units. Figure 9 represents the relative weight of different levels of administration in mandating expenses.22 The figure clearly shows an increase in the share of expenses managed at the operational level. If we exclude the 2006 data (the lower volume of transfers to public entities in 2006 indicate that transfers to hospitals may have been included in a different account), the figure does not show a clear trend toward decentralization, with the share of central level expenditure decreasing between 2008 and 2011 and then increasing again to reach 57 percent in 2013, a level slightly higher than in 2011. The last two years also show a reduction in the share of non-wage current expenditures managed at the district level. In addition, the share of expenses that could potentially be deconcentrated has gone down sharply, from over 20 percent to 13 percent in 2013. Figure 9. Deconcentration of MoH Current Non-wage Expenditures, 2006-2013 100% 0.35 90% Operational Units (including block 0.30 80% transfers to hospitals) 70% 0.25 District (includes management of CSB 60% and CHD1) 0.20 50% Intermediary (province/region) 40% 0.15 Central excluding transfers to hospitals 30% 0.10 20% 0.05 share of total MOH expenditures 10% included (secondary axis) 0% 0.00 2006 2008 2009 2010 2011 2012 2013 Source: Data from MFB/SIGFP. 33. But, increased deconcentration requires additional capacity at the lower levels, as well as strengthened communication between various actors along the expenditure chain as illustrated by the results of a recent rapid assessment on the flow of immunization resources to districts health centers (Box 4). Box 4. Subnational Financing Bottlenecks on Vaccination In an effort to better understand the financial flows and utilization of funds for immunization from central to decentralized levels, the Ministry of Health led a rapid assessment exercise in 2014. The aim of the study, implemented by UNICEF and the Institut Pasteur de Madagascar, was to diagnose the barriers and bottlenecks to the timely flow, appropriate use, and reliable tracking of financial resources for immunization from the national to all subnational levels. The main findings and recommendations are summarized in this text box. Study objectives and coverage This rapid assessment focuses on four main areas of potential bottlenecks for financial flows: 22 Operational Units (SOAs) labeled as CH2, CHRR and CHU are considered as management at the health facility level and assigned to the operational unit level. SOAs labeled CH1 and CSB are included in the district level. 25 Madagascar PER – Health |Section B (i) the approach seeks to generate information on delays in receipt of Routine Immunization (RI) and health funds at each sub-national administrative level, relative to the start of the fiscal year (ii) the assessment of fund diversion by the transferring administrative level is examined, as well as other factors that reduce the actual amount of funds received by the subnational level compared to the amount of funds expected by that level, according to its approved health and immunization budgets (iii) each administrative level is examined to assess the scope and causes for any reallocations of funds planned for RI spending, (iv) problems with systems and capacities for accurate tracking of the receipt and use of immunization and health funds from donors and domestic sources are assessed. This study selected 10 out of 22 regions in Madagascar. In each region, one ‘strong’ EPI performer district and one ‘weak’ EPI performer district were selected, and within each of the districts four health centers (two ‘strong’ and two ‘weak’ performers) were selected using DTP3 coverage rates. Given that numerous health centers lack copies of reports and files, and subsequently, are missing data on the funds allocated to them, only 42 health centers out of 80 planned were considered for the assessment of financial flows. Findings Main results show that the regional and district health teams manage most of the public funding for health that is allocated to their respective levels. The district is responsible for distributing funding equitably among health centers. The district considers the type of health center (HC1 or HC2), total catchment population, geographic accessibility and health center needs in their allocation process. The regional health team receives funds for operational costs directly into its bank account. The district health team receives funds for the districts operational costs and the health center operations in their jurisdiction. Health centers do not have financial management responsibilities. Most disbursements to CSBs (and even to some districts) are made in cash, as there are no government-approved banks nearby that are authorized to accept and disburse funds from government accounts. This creates substantial problems with tracking use of funds, and hinders the ability of managers to ensure compliance with approved health plans and budgets. The study was able to identify the major sources of immunization funding at the national, regional and district level. However, financing details at the health center level were difficult to assess due to lack of data, archives or copies of reports to cite sources. Overall, Madagascar is heavily dependent on partners and donors for immunization funds with 92% of Expanded Program of Immunization (EPI) coming from technical and financial partners. At the district level, a little more than half (54.73%) of the total funds received by the district are used for immunization activities of which 24% comes from the state and 61.9% from technical and financial partners. Additionally, for a good number of districts, the amount received is greater than the amount planned (budgeted) and requested from partners outside of the activities of the Annual work plan (AWP) (46.8%). Delays in allocation of funds compared to the schedules of activities were identified and are mostly due to the multiplicity and cumbersome procedures of donors and complexity of banking procedures. Delays had an impact on utilization rates, which were lower due to delays in credit allocation. The utilization rate is about 35.1% at the level of the regional directorate of health and 41.9% at the level of the district. The second and third quarters are the only periods during which public funds can be used and vaccination activities funded by the government are implemented during these quarters. Thus, a large part of the public funds are prevented from being utilized in full and on time. Furthermore, delays caused some implementation bottlenecks at the district and health center level. At the district level, the date on which information for reports is sent can cause delays in the date on which funds are received. For health centers, the date funds are received overlapped with the planned date for implementation of activities causing a delay in activities. The results also indicated that the number of monitoring and evaluation supervisory visits and financial management checks vary largely with regional health directorates, districts or health centers, and in some cases are 26 Madagascar PER – Health |Section B not sufficient to ensure good management of funds. These oversight activities are generally integrated with other objectives of a supervisor’s visit, and may benefit from more explicit and systematic attention. Recommendations The findings were by the MoH, development partners and health managers from all administrative levels in late 2014. Main results of the review were: 1. Investment in maintaining health system records and archives is important for improving the availability and quality of data on financial flows for immunization. This would include making the health centre manager accountable for record-keeping. Community surveys can also contribute to validating data on immunization financing. 2. Budgetary reviews at every level are necessary for contrasting rates of expenditures between cost centres. In some instances, budgetary controls (spending limits) e.g. in the first two trimesters may be warranted. 3. Investment in bottom-up planning would strengthen the approximation of budgets with planned activities. Methods for ensuring compliance to activities and timelines of the annual work plans would also limit the number of off-budget activities. 4. Coordination between the central government and donors and development partners is essential. The aim is to better guarantee timely funding for the implementation of activities contained in the annual work plan of districts and CSBs. This analysis is being used by the government and its partners to identify practical approaches to help prioritize and overcome major financial bottlenecks that constrain achieving equitable and universal immunization coverage. Source: Case study on immunization expenditures in Madagascar, Thomas O’Connell, UNICEF, 2015 27 Madagascar PER – Health |Section B Key Findings Madagascar’s heath sector is not adequately funded and has a very constrained budget envelope, The public health sector is also largely externally financed with domestic financing very low and unstable. Level of spending for health  Madagascar spends now less on health than three quarters of the SSA countries. Since, the percentage of Total Health Expenditure (THE) in GDP has been around 4-5 percent with a downward trend in the period 2009-2012 compared to 2005-2008.  In real terms THE per capita expenditure has not changed since 1995. Looking at the period between 1995 and 1999 THE per capita was US$21. In the period between 2010 and 2014, Madagascar’s THE per capita was US$20.  Between 2009 and 2013, 80 percent of public funding to the health sector was financed through external funds. However, except for a large off-budget external investment in 2010, overall external financing drastically decreased between 2009 and 2012.  The extremely low share of domestic funding to the sector (20%) is low compared with other countries, and certainly very low compared with other sectors in Madagascar. This poses serious concerns for sustainability, ownership and efficiency of existing resources. Budget Execution  There is lack of clarity between budget appropriations and expenditures due to the different ways in which externally financed investment expenditures are recorded in the government budget under the SIGFP and the Management Aid Platform.  Expenditure by the Ministry of Health decreased by 31 percent over 2007-2013, due the contraction of investment financing from external resources, channeled through MoH.  Rules related to the execution of the budget may prevent full execution, especially for non-wage expenditure; this is notably the case for regulation rates imposed by MoF. The final yearly cuts in regulation rates have negatively impacted the execution rates of some programs more than others, namely primary health services and specific disease programs  Despite some inconsistent improvement over the last four years, the budget execution rates of the MoH remain low for non-wage expenditures.  When looking only at internally financed spending, the share of MoH spending in total government spending has stagnated at about 6%-8% since 2006. This result is robust across the different budget categories, except for the share of MoH in civil servant wages, which increased from 8.5 to 10 percent of the civil servant wage bill between 2006 and 2010. Lack of de-concentration of resources  Budget execution is highly centralized. Between 2006-2013, the central level managed between 50 and 70 percent of current non-wage expenditures with no clear trend toward de-concentration despite a tiered management and service delivery system down to primary care level.  Since 2011, there has been a reduction in the share of non-wage current expenditures managed at the district level. In addition, the share of expenses that could potentially be de-concentrated has gone down sharply, from over 20 percent to 13 percent in 2013. Overall efficiency of public spending 28 Madagascar PER – Health |Section B  The extremely low share of domestic funding to the sector poses serious concerns in terms of sustainability, ownership and efficiency especially given the volatility of aid in fragile contexts. It also raises issues in terms of alignment, harmonization and overall efficiency, given the high volume of external aid provided off-budget in the absence of current overarching National Health Strategy. Policy implications In a context of stagnation and reversal of progress across several key health outcome indicators, additional public spending on expanding access and utilization of quality health interventions are needed. Spending more  The Government needs to better prioritize the health sector in its overall Government budget by increasing public financing to the overall sector. These additional resources can come from a combination of increased resources in the total budget (given the current low share of health) and increased external financing from public and/or private sources. Executing better  Existing budgeting tools needs to be strengthened particularly the SIGFP especially with regards to including more comprehensive data on external aid, better tracking of investments and precise analysis of trends in execution rates across different programs.  De-concentration of resources to lower levels of management and service delivery should be considered; with more autonomy on execution of at least some of the non-wage budget at district and primary care levels.  Execution of regulation rates should be revisited with a specific focus on having a more equal impact across programs consistent with a prioritized budget execution strategy by the MOH.  The impending validation of the New Health Sector Strategy should be seen as a critical opportunity by Government to better harmonize financing to the sector under one national plan.  More harmonized and dynamic budgeting mechanisms should be put in place including participatory budgeting with all stakeholders and alignment of budget planning processes with calendar of the Ministry of Finance and Budget. 29 Madagascar PER – Health |Section C SECTION C. ELEMENTS TO ASSESS THE EFFICIENCY OF MOH EXPENDITURES C1. Assessing Technical Efficiency through an Analysis of Input Shares Box 5. Analyzing Input Shares in Health The production of health requires a combination of inputs (labor, capital, materials and supplies) which are characterized by a high degree of interdependence. That is, medical staff need medicines, supplies, equipment, health facilities, and training to “produce” good health; medicines cannot be administered properly without supervision, and so on. The degree of substitutability is small in the health sector relative to most other sectors. When too much is spent on one input (particularly wages), other inputs are crowded out. This imbalance impedes efficiency, reduces the quality of care, and threatens the sustainability of the system. This is why changes in the relative shares of public expenditures going to each input should be regularly assessed. Although the perfect input mix likely varies by country because of differing input prices, geographical conditions, and even cultural traditions, implicit normative benchmarks have emerged from a systematic review of PERs published between 2002 and 2012 (Gaudin and Yazbeck, 2013). The main findings show that than 50-60 percent spent on labor was considered too much; less than 5 percent on capital and maintenance was considered too little; and less than 30 or 40 percent on materials (including pharmaceuticals) was considered inadequate to ensure the normal productivity of the other inputs. 34. Expenditures on wages have been increasing, while other operational expenditures and internally financed investments have decreased (Box 5). The evolution of expenditures, in constant monetary terms, shows a clear decreasing trend for expenditures on all types of inputs except labor, which has been increasing in both in real terms and as a share of total expenditures.. This is also true in terms of shares, whether or not externally financed expenditures are included (Table 6). The share of regular wages went from 33 to 78 percent (between 2006 and 2013) of the overall budget (including both internal and external financing), and from 50 to over 80 percent during the same period excluding external financing. Other recurrent expenditures, which already received a relatively small share of the budget in 2006, at 22 percent, fell to 15 percent or less of the budget in 2013.23 This indicates that Madagascar has clearly moved to an unbalanced situation that is critical in terms of both efficiency and sustainability, especially considering the fact that the MoH wage bill has increased in terms of shares and in real terms. 23 Externally financed investment expenditures are included in the table but not highlighted, considering the caveats mentioned in section B1. 30 Madagascar PER – Health |Section C Table 6. MoH Expenditures by Broad Input Categories (budget classifications), 2006-2013 2006 2008 2009 2010 2011 2012 2013 Amounts in billion constant Ar. 2013 Regular salaries 75.8 89.1 86.7 96.3 111.9 112.3 123.3 Other recurrent expenditure 50.7 71.2 55.8 45.3 40.2 28.7 23.5 Capital expenditure - internal financing 25.6 24.9 37.9 6.6 14.4 6.0 4.5 Capital expenditure - external financing 76.5 23.2 33.1 4.6 37.1 9.0 7.0 In percent of MoH executed expenditure Regular salaries 33.2 42.7 40.6 63.0 55.0 72.0 77.9 Other recurrent expenditure 22.2 34.2 26.2 29.6 19.7 18.4 14.8 Capital expenditure - internal financing 11.2 12.0 17.7 4.3 7.1 3.8 2.8 Capital expenditure - external financing 33.4 11.1 15.5 3.0 18.2 5.7 4.4 In percent of MoH internally financed expenditure Regular salaries 49.8 48.1 48.1 65.0 67.2 76.4 81.5 Other recurrent expenditure 33.3 38.4 30.9 30.6 24.1 19.5 15.5 Internally financed capital expenditures 16.8 13.5 21.0 4.5 8.6 4.1 3.0 Regular salaries/share of recurrent budget 59.9 55.6 60.8 68.0 73.6 79.6 84.0 Data source: MFB/ SIGFP. 35. Regular salary expenditures in Madagascar have reached levels that are much higher than those generally observed in less-developed countries. While internationally comparable data are not readily available, a few points of comparison could be obtained using secondary data sources. Only two other SSA countries could be identified in past PERs with labor shares above 60 percent: Zanzibar (2003 PER) with 80-85 percent, and Ghana with about 70 percent (2009 PER)24; most other SSA countries had labor shares around 50 percent or less. Using a larger and more systematic set of countries, other studies have provided some point of comparison with labor shares in other income groups and regions. Vujicic et al. (2009) found the average share of wage in government health spending around 40 percent in Africa for the 2000-2004 period, and the average for high-income countries around 45 percent. These comparisons, however, are based on data prior to 2006. In a more recent PER, a similar trend to Madagascar was identified in a current study of Zimbabwe, where the share of wages in MoH expenditure reached 80 percent in 2013.25 36. As a point of comparison, in contrast to the Education sector, non-regular salaries are not a significant part of the non-wage recurrent budget. While it is not possible to identify all labor costs in the non- wage recurrent budget, a breakdown of expenditures by account code in SIGFP can help to gauge the magnitude of these costs. Such a breakdown can also provide important information on the structure of MoH expenditures in more detailed categories of inputs (Table7). Using further disaggregation of the budget to identify expenditures by 24 Ghana was also cited as a case of labor costs crowding out other inputs in Working in Health, World Bank, 2009. Including additional duty hour allowance, the authors calculated that the wage bill added 90 percent to expenditures as donor funding contributed solely to non-wage expenditure. 25 The Zimbabwe study, however, included all labor expenditures, including in recurrent expenditures and transfers. 31 Madagascar PER – Health |Section C input across broad categories increases the share of labor by only 3-4 percent. Even if some portion of transfers to hospitals goes to pay temporary staff, this would not make a large difference to assessing total labor costs, given that transfers to hospitals do not exceed 5 percent of the budget.26 Table 7. Input Shares across Budget Categories, Share of Internally Financed MoH Expenditures, 2006- 2013 2006 2008 2009 2010 2011 2012 2013 Labor and assimilated costs 53.4 50.5 50.6 68.5 70.6 80.2 85.2 Regular wages 44.6 42.5 42.9 57.6 60.4 71.7 73.5 Other wages 0.5 0.3 0.3 0.2 0.1 0.0 0.1 Social charges (regular staff) 5.3 5.7 5.4 7.8 7.1 5.1 8.4 Medical student grants and provisional. Salaries 3.0 1.9 2.1 2.9 2.9 3.4 3.2 Goods and services 17.8 20.6 18.7 18.7 14.7 10.5 6.7 Specific medical G&S 6.1 5.3 5.2 5.0 3.6 2.6 2.2 Fuel and transportation costs 4.3 5.1 5.0 5.4 4.4 3.4 2.3 General running costs 2.0 3.6 3.2 2.9 2.2 2.0 0.9 Maintenance and utility costs 2.6 5.5 4.4 4.1 3.5 1.7 0.7 Other G&S 2.9 1.1 0.9 1.3 0.9 0.9 0.5 Transfers and subsidies 2.4 6.3 5.3 4.4 3.9 4.0 5.0 Transfers to public entities (hospitals) 2.1 5.0 4.3 3.9 3.1 3.2 2.7 Transfers to pay staff medical costs 0.2 0.5 0.4 0.4 0.6 0.5 0.3 Other transfers 0.1 0.8 0.6 0.2 0.2 0.3 2.0 Capital costs 26.3 22.6 25.4 8.4 10.9 5.3 3.2 tangible assets 23.9 19.4 19.5 7.4 6.2 3.5 1.7 intangible assets and other 2.4 3.1 5.9 1.0 4.7 1.8 1.5 Internally financed MoH expenditures 100.0 100.0 100.0 100.0 100.0 100.0 100.0 Source: Data from MFB/SIGFP. 37. Expenditures on goods and services related to the provision of health care make up a very small share of the budget, and over the years this share has decreased considerably, with the cost of most medical consumables borne by the patient through cost recovery. Trend data also indicate that maintenance and utility costs decreased from 5.5 percent of the budget in 2008 to less than 1 percent in 2013. Fuel and transportation costs also declined, from more than 5 percent of the budget in 2008 to 2.3 percent in 2013. The raw data also indicate reductions in absolute amounts (by 50 percent in 2013 for maintenance and utility costs, and by 25 percent for transportation costs), indicating a true reduction in the level of activity. Altogether, combining input types across the health budget, it is clear that since 2010, labor expenditures have crowded out expenditures on goods and services and investments managed by the MoH (Figure 11). This is apparent looking at the decline in key service delivery indicators during the same time period. For example, the prescription satisfaction rate, a key indicator for drug availability, declined from 69 percent in 2008 to 58 percent in 2010/11 at the facility level and the utilization of basic health centers and prenatal consultations decreased by 20 percent from 2008 to 2011. 26 It is not possible to identify input shares for expenditures paid out of the health facilities own revenues, if any. 32 Madagascar PER – Health |Section C Figure 11. Decomposition of MoH Expenditures into Broad Input Categories, 2006-2013 100% 80% Labor 60% Non allocated transfers 40% Goods and services 20% capital 0% 2006 2008 2009 2010 2011 2012 2013 Note: Based on executed expenditures (dépenses mandates). Labor and capital include expenditures from the recurrent non- wage budget (category 3 expenditures). Data source: MFB/SIGFP (data reconstituted for 2006 and 2008). 38. The large bias towards salaries in domestic funding is somewhat compensated by substantial inflows of external aid targeting other aspects of the system, in particular goods and services. However, investment financed by external aid has dropped considerably, making it difficult to sustain improvement in the quality and quantity of health services delivered. 39. In conclusion, although Madagascar was found to do well overall in term of outcomes relative to expenditures before 2008, the current imbalance in the use of inputs shows that the country is not on track to keep its advantage, given that the trend since 2009 is not efficient and is unsustainable. C2. Elements to Assess Allocative Efficiency: Expenditure Analysis by Function Box 6. Allocative Efficiency Analysis Using Functional Allocations of Health Expenditures Allocative efficiency analysis generally examines the types and combinations of goods and services produced in relation to demand. The term is loosely interpreted here in the sense that consumer preferences and demand are not directly measured. Instead, the analysis rests on a general appreciation of needs and on well-known characteristics of the different types of health provision (functions) in terms of their public good (public health activities) and best value-for-money (primary care and prevention) characteristics. By affecting parameters such as accessibility and quality, the functional distribution of health expenditures is not just a response to demand or to need; it also influences the types of services that will be effectively used by individuals. There is some degree of substitutability between the different kinds of care dispensed at the primary, secondary, and tertiary levels or in different types of facilities, in terms of reaching desired outcomes (i.e., lower mortality and morbidity). Expenditures on prevention, for example, complement current expenditures on curative care contemporaneously but substitute for future expenditures on curative care, and are therefore recommended from a sustainability perspective. Finally, while efficiency and equity often involve a trade-off in other sectors, the two goals tend to be complementary in the health sector. In particular, directing more resources to primary health care and prevention is usually recommended as both efficiency and equity enhancing. In addition to looking at functions in terms of health services and levels, analysis of the distribution of expenditures by health priority is useful to assess allocative efficiency in terms of responding to needs. 33 Madagascar PER – Health |Section C The question is whether efforts are placed where they are most needed and where they will have the most effect. The need to prioritize some health programs against others in the name of allocative efficiency depends on specific conditions in the country, based on an evaluation of the situation and evolution of major health outcomes, utilization indicators, and demographic changes. 1. Data and level of analysis 40. The ability to determine whether resources are allocated efficiently (Box 6) depends greatly on the ability to classify expenditures into categories that can be matched to levels of care, health priorities, and other specifically identified needs such as those of vulnerable populations. The expenditure analysis presented below is guided by the feasibility of sorting expenditures by these types of functions using the data available. First, a breakdown of Total Health Expenditures is presented based on National Health Account (NHA) data, to give a picture of the overall situation. There is some degree of comparability over time and with other countries. This also gives an idea of the evolution of costs that are borne mostly by households.27 41. Then, the bulk of the analysis is devoted to government health expenditures measured using: (a) public expenditures from NHA 2010—-noting that these are 2.3 times larger than total MoH expenditures in SIGFP (explained below); (b) wage expenditures of MoH based on a compilation of various government data sources and author’ calculations; and (c) non-wage recurrent and investment expenditures of the MoH, based on SIGFP data. Specificities about the three types of data sources and the type of analysis they allow are outlined in Box 7. Box 7. Contribution of Different Data Sources to Understanding Health Expenditure by Function in Madagascar Different sources of data provide different opportunities to classify expenditures by function in Madagascar. This box lays out the type of analysis that can be done with each source and for different types of expenditures. Depending on the definition of health expenditure used, the analysis yields different elements of information that complement each other, to provide an overall picture of expenditure allocation. The sources presented below are ordered from most comprehensive to most narrow in terms of the range of expenditures included. National Health Account: This source encompasses the broadest range of health expenditures. It relies on expenditure information using external survey questionnaires (and may therefore differ from information obtained from government accounts). NHA methodology organizes the data into financing schemes and financing agents; total health expenditures include all financing schemes, public and private. The presentation of the data in matrix form allows identification of the subset of expenditures using central public administration as a financing scheme (HF). Expenditures under the public financing scheme exclude social insurance but include all public funds, including off-budget externally financed expenditure. For Madagascar, this total is 494 billion Ar, compared to on-budget expenditure of 217 billion Ar in SIGFP (see below). In the NHA central administration HF category: 46 percent of funds go through local governments, 29 percent through NGOs, 21 percent through the central administration, and 4 percent through international organizations. MoH expenditures. A direct analysis of MoH expenditure has the potential to reveal trends that cannot be assessed using NHA data. In particular, one can separate wage and non-wage expenditures and identify the significance of health care provision by levels of care. Looking at wage and non-wage recurrent expenditures is important for several reasons: (a) the data on non-wage 27 In particular, it is important to look at total expenditures instead of MoH expenditures when analyzing the evolution of pharmaceutical costs; these costs have been mostly borne by household in Madagascar since the system of cost recovery was implemented. 34 Madagascar PER – Health |Section C expenditure are from a single source (SIGFP) and present a much higher degree of reliability than do wage allocations based on approximations from multiple non-official sources; (b) different types of functional classifications are possible with wage and non-wage expenditures; (c) non-wage recurrent expenditures have a higher degree of flexibility, as they allow a better appreciation of short-term changes in priorities/political commitment, and can capture short-term variations in level of activity; and (d) expenditures on salaries largely dominate in Madagascar (close to 80 percent of expenditures in 2013); therefore, analysing recurrent expenditures as a whole overshadows important information that can be obtained by looking solely at non- wage recurrent expenditure. MoH wage expenditures. SIGFP can be used to evaluate the magnitude and evolution of total salaries and identify the share of salaries going to the central administration. No further analysis by function can be done using SIGFP, given that regular salaries and social charges (contributions paid by employers) are all included in the general administration program of the MoH. A reconstruction of salary shares by type of facility and type of personnel was done for 2013, using MoH data on average salaries by category of personnel combined with number and category of employees by type of facility. Approximations of salaries for 2006-2013 were obtained using a combination of MoH and MFB information on salary increases and grade changes over time. No information was available on changes in number of personnel over time. Instead, the salary estimates by function were corrected using differences with SIGFP totals each year—a method that does not take into account the relative changes between functions over time. A full description of the methodology and challenges encountered in gathering and reconciling the data is given in Annex x 3. Given the size of the effort needed to perform such estimations and the limited availability and reliability of non-SIGFP data, it is important to stress the need to change government accounting practices so that salaries can be identified by function (program, type of facility) in government accounts. Non-wage expenditures from SIGFP. SIGFP data on non-wage expenditures can be used to classify expenditures by type of health services (administrative, curative, preventive) and by levels of care (primary, secondary, tertiary). The exercise requires an analysis beyond classifications by program provided in the budget, to allow for some comparability with the pre-crisis years and to widen the range of possible categories. Only very partial information could be obtained in terms of internally financed expenditures going to different vertical health programs. Maternal and Child Health (MCH) expenses can be identified through a specific budget program (Suivie et Developpement de la Mere et de l’Enfant -SDME). The data were further analyzed using the names of administrative units to identify expenditures directed to other health priorities, but the data were not sufficient to provide a clear picture of these expenditures, especially given the importance of external funding for vertical programs and the lack of accounting of realized expenditures for externally funded expenses. A separate analysis of the immunization program was done by UNICEF this type of analysis would need to be done for other programs (immunization, malaria, MCH) to see whether the prioritization is consistent with the evolution of a broader range of indicators. 2. Functional allocation of Total Health Expenditures (NHA) 42. When considering the totality of health expenditures, both public and private, the results of the 2003, 2007 and 2010 NHA exercises reveal that Madagascar does not exhibit the common SSA pattern of over- spending on in-patient care and under-spending on preventive and public care. In fact, the share of hospital care decreased while the share of spending on prevention and public health programs increased over the period 2003- 2010 (Table 8). In Madagascar, however, these low levels of spending on curative care are likely a signal of system failure, in the sense that the majority of the population may just not be seeking care. This is verified by the 2010 EPM findings results, which showed that close to 70 percent of people in Madagascar did not seek care when ill. 35 Madagascar PER – Health |Section C Table 8. Allocation of Total Health Expenditures across Functions (all sources) 2003 2007 2010 Hospital care 7 9.2 6.1 Ambulatory care 28 7.7 25.9 Pharmaceuticals 20 19.1 16.1 Prevention and public health programs 28 24.6 32.9 Health administration 10 13.9 5.3 Note: Shares are based on total expenditures, including for 2010. The residual share is for investment, which is not separated by function. Source: Adapted from NHA reports 2003, 2007 and 2010. 43. In comparison to SSA and other comparable low-income countries, Madagascar devotes a much lower share of total health expenditures to inpatient care. Although international comparisons are difficult at this level of disaggregation, some data are available, based on country NHAs in the WHO GHED, regarding the share of inpatient care in total health expenditure (Table 9). Expenditure on inpatient care is the most expensive type of care and has the potential to crowd out preventive care, resulting in higher future curative costs. This does not appear to be happening in Madagascar, where the share of expenditures going to inpatient care is less than 7 percent. This is less than a third of SSA and LIC averages, and well below any individual countries in the comparison group (based on data availability and proximity). Again, this is likely a reflection of low health-seeking behavior across all quintiles. Table 9. Share of Health Expenditure on Inpatient Care, International Comparison Average Years available GDP/c in USD Madagascar 6.6 2003, 2007 447 Mozambique 18.2 2004-2006 565 Zambia 25.9 1995-2000, 2005 1469 Mauritius 29.5 2002 8119 Kenya 34.6 1995-2001 943 DRC 41.5 2008 262 Tanzania 50.8 1995-2000, 2006, 2010 609 Sub-Saharan Africa 23.9 1995-2008 (unbalanced panel) Low Income Countries 24.4 1995-2008 ‘’ Low and Middle Income 30.4 1995-2012 ‘’ High Income Countries 34.2 1995-2012 ‘’ Source: Data from WHO/GHED, based on NHAs. 44. Pharmaceutical costs have also remained stable, at less than 20 percent of total expenditures, and these costs even decreased in 2013. From a general sustainability perspective, given that pharmaceutical costs have had a tendency to increase in the rest of the world, this result could be seen as encouraging. But in the case of Madagascar, given that the recent health facility survey indicates a decrease in utilization as a result of a reduction in health seeking behavior, this could be concerning. The country’s FANOME system includes a cost recovery component for essential drugs but the system is currently decapitalized in many parts of the country. In areas where financing is available through donor funding, the FANOME system has been recapitalized and a fee exemption 36 Madagascar PER – Health |Section C scheme for a set of targeted services and medicines MCH services has been put in place, utilization rates have gone up drastically. 3. Functional allocation of public health expenditures (NHA 2010) 45. Knowledge of the functional allocation of total health expenditures (THE) is helpful in identifying whether there are specific gaps in the system as a whole. However, optimal prioritization varies between the public and private sectors. Higher expenditures on curative care are expected from the private sector, while the public sector needs to focus on preventive health services, which create the largest positive externalities for the population as a whole. 46. The 2010 NHA28 results indicate a distribution of public expenditures that strongly prioritizes activities with high public good characteristics, as would be expected of the public system. NHA is the only source of data that can provide an overview of all public health expenditures, including externally financed off- budget expenditures. According to these data, two thirds of public health expenditures are devoted to preventive activities, most of which are targeted to specific diseases (Table 10).29 There are some limits to these NHA results; for example, it is not clear from the report what were included as preventive health services.30 Further, since expenditure on curative care cannot be broken down by level of care, it is likely that a good number of activities carried out by primary health care facilities and some administrative costs were included as part of preventative activities. 28 Using the matrices created for the NHA 2010, current public health expenditures can be broken down by specific health services /health programs and to some extent by type of health facility (following the System of Health Accounts OECD - SHA 2011). 29 Unfortunately, it is not possible to recreate this breakdown for previous years, so trends cannot be analyzed. 30 The raw data for NHA were not available due to personnel changes combined with accidental losses of electronic files, so the data are limited to tables and matrices published in the NHA reports. 37 Madagascar PER – Health |Section C Table 10. Allocation of Current Expenditures of the Public Administration System Percent of public HE By health program/ service Prevention and fight against disease (public health) 46.60 Information, education and counseling programs 13.51 Curative services in hospitals (inpatient) 12.34 Curative outpatient services 8.13 Governance and administration 7.55 Vaccination programs 5.31 Pharmaceuticals 4.25 health monitoring programs 1.54 Early disease detection program 0.77 disaster preparedness and response 0.00 By type of facility/main activity Preventive health services 64.37 Outpatient facilities 14.80 General hospitals 13.15 Health administration 7.44 Pharmacies 0.23 Source: Calculated from NHA 2010 matrices (appendix to 2010 NHA report). 4. Functional allocation of Ministry of Health expenditures: wage expenditures 47. As noted above, given current accounting practices in Madagascar, classifications of salaries into functions rely on rough estimates. However, with salaries making up 78 percent of executed expenditures of the MoH (85 percent of domestically financed expenditures), salaries need to be considered in order to get a good idea of the level of effort by type of health service and, in particular, by level of care. Details of the methodology used to construct salary shares and raw results are presented in Annex 3. 38 Madagascar PER – Health |Section C a) Allocation of salaries by health facilities Figure 12. Distribution of Salaries by Facilities and Administrative Levels, 2013 Central Administration CHU: DRSP: Regional Administration 20% Central SDSP: District Services CHRR Administration CSB2: Health centers (include 1 doctor) 4% 29% CHRD CSB1: Basic Health center (no doctor) 7% CHRD: Primary hospitals (District) DRSP CHRR: Secondary hospitals (Region) CSB1 SDSP 5% 4% CSB2 8% CHU: Secondary/tertiary hospitals (ex-Province/Capital) 23% Data sources: MoH, MFB. 48. Even though 50 percent of the population seeking care go to primary health facilities (CSB1 and CSB2), these facilities only absorb only 27 percent of wages (Figure 12).31 At the second level of primary health care facilities (CSB2), staff absorb 23 percent of salaries while they are visited by 40-50 percent of the population seeking care. There are 1610 functional CSB2s (2013) in the country and they are typically staffed by one doctor, one midwife and a small number of support personnel. In contrast, basic primary health care centers (CSB1) are normally staffed by one nurse or midwife and one support staff. There are 875 functional CSB1s (2013), which represent less than 4 percent of the wage bill and are visited by about 10 percent of people who are sick. Personnel data by region and type of facility (2013) show that CSB1s are typically understaffed. Only 5 regions had on average more than 1 staff per functional CSB1, with the highest average in Haute-Matsiatra at 1.43 staff per CSB1. 49. Total salaries at service delivery levels have remained low since 2006, while central and regional administrative salaries increased exponentially in the same period. Despite some data limitations, the graph of salaries by type of facility reveal some important trends of shares in the MoH wage bill for CHUs, and the central administration vs. other levels of care. (Figure 12).32 33 31 Enquêtes Permanentes/Périodiques auprès des Ménages (EPM) Household Surveys 2005 and 2010. 32 Changes in staffing per facility type over time could not be obtained, and the evolution of salaries presented here cannot capture the full variation over time. The only category for which the data on evolution of salaries is not limited is the central administration. Salaries going to other facilities can only evolve independently because of differences in salary increases over time and by category of personnel. An estimate of salaries, taking account of changes in the number of facilities at the primary care level, was tested but did not improve results when compared to totals in SIGFP. The graph in Figure 13 showing the evolution of salaries since 2006 is virtually unchanged. 33 The decrease in Central Administration shown for 2009 could be due to a decrease in total salaries recorded in SIGFP. However, there were some issues in the accounting of salaries for 2009, so the decrease could be due to low data quality in 2009. 39 Madagascar PER – Health |Section C Figure 13. Estimated Trends in Salaries by Type of Health Facility/Administrative Unit, 2006-2013 35 CENTRAL MINISTRY (from SIGFP) 30 Regional Admin Billions of constant 2013 Ar. 25 District Services 20 CSB1 15 CSB2 10 CHRD CHRR 5 CHU 0 2006 2008 2009 2010 2011 2012 2013 Source: Based on data from MFB/SIGFP and from Ministry of Health/Human Resources Department. b) Allocation of salaries by level of care 50. In Madagascar, similar to other low-income countries, a great deal of primary health care is provided outside of primary health care facilities, especially in tertiary hospitals. However, this care is not measured because of lack of data, so expenses incurred at hospital end up being counted as secondary or tertiary care.34 Our estimates indicate that 60 percent of activities in university hospitals (CHUs) are at the primary care level, in the sense that they are dispensed by primary care practitioners.35 In administrative functions, about 40 percent of salaries (50 percent at the district level) go to PHC/PH personnel.36 The result is also interesting for district hospitals, which were included in the graph above as primary care hospitals37; in fact, 66 percent of salaries in district hospitals go to PHC/PH, which is not much different from the share found at the CHU level (Figure 14). 34 Glick and Razakamanantsoa (2002) report that a substantial amount of primary care is dispensed at the hospital level for individuals who have access to hospitals, based on evidence in EPM, although they do not quantify the size of primary care provision. 35 Each type of occupation was assigned to a level. General practitioners, dentists, midwifes, public health specialists, nutrition agents and health workers were automatically assigned to the primary care/public health category. Specialized doctors and specialized paramedical staff were assigned to secondary/tertiary care. Non-specialized paramedical (nurses) and radiologists were distributed according to the share of public health care staff relative to specialized staff in the facility. 36 Doctors working in administrative functions who do not dispense or supervise health care are recorded as administrative personnel. 37 The personnel data do not distinguish between category 1 and category 2 district hospitals. Category 2 district hospital offer surgery and may be closer in type of health care provision to regional hospitals that provide primarily secondary care. 40 Madagascar PER – Health |Section C Figure 14. Allocation of Salaries by Level of Care and Support Activities by Type of Facility, 2013 100% 80% ADMIN/SUPPORT 60% PHARM/LAB 40% SECONDARY/TERTIARY 20% PHC&PH 0% CSB 1 CSB 2 CHRD CHRR CHU SDSP DRSP CENTRAL Note: Type of health facility described in Figure xx above. Primary health care and Public Health (PHC&PH) includes all general practitioners, dentists, midwives, nutrition agents, and health workers. General nursing staff are assigned according to the proportion of other primary care personnel in the facility where they are employed. Data source: Authors’ calculations from data on occupation and quantity of personnel by facility type, MoH/Human Resources Department. 51. Administration costs are relatively larger in regional hospitals (24 percent) than in other health care units, including other hospitals (less than 20 percent in CHU and CHD). In CSB1 and CSB2, it is clear that the medical/paramedical staff take care of administrative tasks. 52. Overall, based on a combination of information on type of personnel and type of facility, about 62 percent of salaries go to primary and public health care, a figure that resembles what was found in NHA looking at the totality of public expenditures. Secondary/tertiary care activities absorb less than 6 percent of the MoH wage bill while administrative staff absorb about one third (Figure 15). Figure 15. Allocation of Salaries by Level of Care, 2013 Admin and support Pharmacy/ 28.9% Labs 3.8% Primary and Public Health Secondary/ 61.7% Tertiary care 5.6% Source: Data from MFB data/SIGFP and from MoH/Human Resources Department. 5. Allocation of salaries by type of personnel 53. Basic health centers (CSB1) are almost exclusively run by paramedical staff (nurses, midwives and lab technicians) with the highest concentration of medical staff salaries in tertiary care facilities (CHUs). Classification by type of personnel (medical, paramedical, administrative) allowed an analysis over time based 41 Madagascar PER – Health |Section C on actual changes in number of staff.38 The ratio of paramedical salaries to medical staff salaries is significantly lower in CHUs than in other health facilities. Salaries of paramedical staff, make up about 40 percent of the wage bill in all other health facilities except the basic health centers (Figure 16). Figure 16. Share of Salaries by Type of Personnel and Facility, 2013 100% 80% 60% Paramedical 40% Medical 20% Admin 0% Source: Data from MFB (average salary increases) and from MoH/Human Resources Department. 54. Looking at personnel, the highest increase in total wage expenditure is from medical staff (doctors). Figure 17 traces the evolution of salaries by type of personnel, taking account of overall changes in the number of personnel by type as well as the evolution of average salaries.39 Figure 17. Estimated Change in Total Wage Expenditures by Category of Personnel, 2006-2013 50 100% Nbillions of constant 2013 Admin/Support 40 80% 30 60% Medical Ar. 20 40% Paramedical 10 20% % of the total in SIGFP 0 0% (secondary axis) 2006 2007 2008 2009 2010 2011 2012 2013 Source: Data from MFB (average salary increases) and from MoH/Human Resources Department. 38 The number of personnel by type (medical, paramedical, and administrative) was available from 2000 to 2012. Some adjustments were done based on facility level data for 2012 and 2013 to match the total numbers (the numbers broken down by facility type and region underestimated the total number of staff). Average compensation estimates by category were used to reconstruct total salaries. See Annex 3 for details. 39 The data on number of personnel cover 2000-2012. The number of personnel recorded by DRH in 2012 and 2013 was used to estimate the percentage change for 2013. 42 Madagascar PER – Health |Section C 6. Functional Allocation of Ministry of Health Expenditures: non-wage expenditures a) Functional classification by budget program, 2010-2013 55. Program budgeting has the great advantage of facilitating the analysis of expenditures by function, when programs are chosen to reflect the principal health function by level of care and/or type of services. Expenditures are separated into: a) Administration and Coordination; b) Basic Health Services; Hospital Health Services; c) Supply of Medicine and Other Medical Products; d) Disease Prevention and Control; and e) Maternal and Child Health.40,41 Basic health services are fully assimilated into primary health care, while hospital services include all in-patient services and are mostly directed to secondary/tertiary health care.42 Data for 2009 are not included to ensure full comparability, with regard to program structure43 and considering that total non-wage expenditure was cut in half between 2009 and 2010. Recurrent non-wage expenditure 56. Administration and coordination takes the largest share of recurrent non-wage expenditures, more than PHC, MCH, and public health combined (Figure 18). Using the information available in the SIGFP shows that administrative expenses constitute about 40 percent of non-wage recurrent expenditure, (with a decrease in 43 in 2010 to 36 percent in 2013). The share going to hospital services stayed relatively constant at around 25 percent. The share going specifically to primary health care services was significantly reduced from 19 to 11 percent, although maternal and child health expenditures compensated for the decrease. Grouped together, non-wage expenditure on primary health care, maternal and child health and public health (which is labelled fight against specific diseases and includes some primary care), increased from 28 to 32 percent of recurrent expenditure between 2010 and 2013. 40 Note pour la traduction: Administration & coordination; Services de santé de base; Services de santé hospitaliers; Fourniture de médicaments, consommables et autres produits; Lutte contre les maladies; and Survie et développement de la mère et de l'enfant (SDME). 41 The MCH and Family Planning programs were new programs in 2010. Family planning was merged with the MCH program in 2012 and 2013. 42 If the analysis presented above regarding share of salaries can be extended to non-wage expenditures, about half of hospital services would be directed to primary care services. 43 Budget programs in 2009 did not include either the MCH program or family planning. 43 Madagascar PER – Health |Section C Figure 18. Allocation of Non-Wage Recurrent Health Expenditures by Budget Program, 2010-2013 100% Administration & 90% coordination 80% Hospital services 70% 60% Medical supplies and medicines 50% Public health 40% 30% Maternal and Child 20% Health/FP 10% Primary health care 0% services 2010 2011 2012 2013 Note: MCH and FP expenditures were not identified separately in the 2009 budget. Source: Data from MFB/SIGFP. Investment expenditure 57. Virtually all investment expenditures go to public health (fight against specific diseases) and primary care/MCH activities (Figure 19). This is based on the expenditures included in the SIGFP only. A large part of investment expenditures end up being spent off budget or outside of the MoH accounts and are not accounted for in this analysis. Figure 19. Allocation of Health Expenditures by Program—Investment Expenditures, 2010-2013 Administration & coordination 100% 80% Hospital services 60% Medical supplies and medicines 40% Public health 20% Maternal and Child Health/FP 0% Primary health care services 2010 2011 2012 2013 Source: Data from MFB/SIGFP. 58. Despite large yearly fluctuations, the overall picture on functional allocation of non-wage expenditures looks no different in 2010 than it did in 2013 (Figure 20). Combining non-wage recurrent and investment expenditure, PHC/MCH/PH expenditures went from a low of 40 percent in 2010 to 53 percent in 2013, with a peak of 68 percent in 2011. The same pattern is revealed when looking only at internally financed expenditures: from 35 percent in 2010 to 41 percent in 2013, with a peak of 47 percent in 2011. 44 Madagascar PER – Health |Section C Figure 20. Classification of Non-Wage Expenditures by Program (recurrent+investment), 2010 and 2013 Source: Data from MFB/SIGFP. b) Additional information using classifications by administrative unit 59. It is not possible to compare the results presented above to the pre-crisis period because of changes in the administrative structure and the delineation of programs, as well as changes in the way that some expenditures were classified in the budget after 2009. Program delineations also limit the type of classifications possible. To circumvent these issues, expenditures in the SIGFP were classified using the name of the Operational Activity Unit (SOA), the account code (type of expense), and current program assignments to identify the type of service, level of care, and other specific areas of focus.44 Table A3 in Annex 1 lists the classifications that were created. This alternative classification offers the advantage of being completely independent of other differences in budget accounting. Further, it is stable over time, so it can be used to confirm data quality and to compare pre- and post- crisis periods. In addition, this alternative classification provides further disaggregation by type of activity, which is useful for tracking specific efforts, including in terms of governance, given the weight of administration in current expenditure. 60. Non-wage recurrent health expenditure by health service/level of care are close to those reported by budget program for the combined PHC, MCH and Public Health functions. Results differ somewhat for the other three functions, with hospital care (secondary and tertiary care) clearly going up since 2010. 61. Relative to 2006, the weight of administration decreased in non-wage expenditures decreased, while the shares going to hospital services and to high-impact activities increased. Hospital care gained 10 percentage points, high-impact activities (PHC MCH, and Public Health) together gained 8 points, while the share of 44 About 1,945 different SOAs were identified by combining the data for 2006 and 2008-2013, about half of which were similar SOAs for different districts or regions. SOAs were manually assigned to the different categories of table x.x3 in view of its denomination, and when possible matching where expenditure of similar SOAs were classified in the latest budget programs. Additional identification was performed within some administrative SOAs using PCOP account information. In particular, hospital transfers were re-coded as secondary and tertiary health services/curative, and pharmacy and medical consumables were removed from administration when applicable. 45 Madagascar PER – Health |Section C administration decreased by 20 points. Undifferentiated services, which consist mostly of pharmacy and laboratory costs, remained stable (Figure 21 and Table 10). Figure 21. Allocation of Non-Wage Recurrent Health Expenditures by Level of Care, 2006-2013 100% Secondary and Tertiary Health care 80% Health, undifferentiated by level (mostly pharm/lab) Primary and Secondary Health Care 60% Maternal & Child Health (inc. reproduction/FP) 40% Primary Health Care (PHC) 20% Public Health General Administration 0% 2006 2008 2009 2010 2011 2012 2013 Source: Data from MFB/SIGFP. Table 10. Shares of the Recurrent Non-wage Budget by Level of Care, 2006-2013 2006 2008 2009 2010 2011 2012 2013 Primary care, MCH, and public health 24.1 28.0 24.3 28.2 31.8 26.7 32.3 Secondary and tertiary care 14.1 19.6 23.4 20.2 21.4 22.0 24.0 Undifferentiated by level (mostly pharm/lab) 15.8 13.2 16.1 15.9 12.1 15.4 14.5 General administration 46.0 39.2 31.1 31.3 30.0 32.0 26.3 Source: Data from MFB/SIGFP. 62. Including investment, the evolution of non-wage expenditures is much less stable if we look at yearly changes since 2008.45 Maternal and child health activities received a boost in 2011, when they made up 28 percent of non-wage expenditures at the MoH. Excluding wages, the combination of PH, PHC, and MCH activities reached 70 percent in 2011 and since then have remained above 50 percent of the total. (Table 11). Table 11. Allocation of Non-Wage MoH Expenditures (including PIP) by Level of Care, 2006-2013 2006 2008 2009 2010 2011 2012 2013 Primary care, MCH, and Public Health 17.3 46.5 62.5 44.4 70.3 51.2 55.0 Secondary and tertiary care 4.8 11.9 15.0 17.7 10.8 16.8 17.4 Undifferentiated by level 65.4 8.1 8.2 12.9 5.4 10.2 9.9 General Administration 12.5 33.4 14.3 25.1 13.5 21.8 17.7 Source: Data from MFB/SIGFP. 45 Expenditures on building of facilities and pre-operating costs (65 percent of non-wage expenditures in 2006) could not be identified by type of facilities and therefore could not be assigned by level so little can be said about functional allocation for 2006. 46 Madagascar PER – Health |Section C 63. A comparison of MoH non-wage expenditures (including foreign-financed investment expenditure) in 2008 and 2013 is shown in Figure 22. It is important to note that that the size of the financing envelope was greatly reduced from 115 billion Ar in 2008 to only 35 billion Ar. in 2013. This new picture reveals a larger share, albeit of a much smaller pie, given to primary health care and public health, while expenditures of administrative units went down and the share going to hospital services went up. Figure 22. Distribution of Non-Wage Expenditures by Program, 2008 and 2013 Source: Data from MFB/SIGFP. 64. The classification by function/type of activity confirms the steady reduction in administration costs, mostly to the benefit of curative activities until 2009 and to the benefit of preventive health after 2009 (figures 23 and 24).46 The weight of preventive activities increased significantly (from 7 to 21 percent) from 2009 to 2013, while the weight of curative activities increased until 2009 (from 16 to 41 percent) and remained stable or decreased slightly thereafter, to reach 34 percent in 2013. 46 The definition of administration is different in this classification compared to above, explaining the higher weight. For example, expenditures of District health services were classified as primary health by levels but as administrative(as opposed to curative or preventive) by function. 47 Madagascar PER – Health |Section C Figure 23. Allocation of Non-Wage Recurrent Health Expenditures by Type of Activity, 2006-2013 100% Mantenance, logistics 80% Training 60% Pharmacy/laboratories 40% Preventive 20% Curative Administration 0% 2006 2008 2009 2010 2011 2012 2013 Source: Data from MFB/SIGFP.Data source: MFB/SIGFP. 65. As expected, more variability is found when including investment expenditures. When investment is included, preventive activities take a larger share, reaching above 40 percent in 2011 and back to a more representative 27 percent in 2013. Curative activities make up 17-25 percent depending on the year. Figure 24. Allocation of MoH Non-Wage Expenditures (incl. Investment Expenditure) by Type of Activity, 2006-2013 100% Training 80% Pharmacy/laboratories 60% Preventive 40% Curative 20% Buildings, equipment, maintenance, etc. 0% Administration 2006 2008 2009 2010 2011 2012 2013 Source: Data from MFB/SIGFP. c) A partial look at specific allocations (health priorities, target groups, governance)47 66. Communicable diseases (CDs) and MCH dominate targeted expenditures identifiable by program. Excluding the investment budget, expenditure on CDs clearly dominate every year, followed by MCH (Figure 25). Looking at investment, CDs and MCH still make up the bulk of targeted expenditure, although there is much variation over the years between both programs. There has been some increase in non-communicable diseases (NCD) program funding, indicating that are starting to gain importance in the investment budget. 47 The classification by SOA can be used to partially identify expenditures targeted to specific priorities. Amounts targeted to specific priorities in the recurrent budget (all internally financed) and the investment budget (most externally financed) are reported in Annex 1. 48 Madagascar PER – Health |Section C Figure 25. Evolution of MoH Expenditures Targeted to Specific Diseases, 2006-2013 Targeted recurrent expenditures (a) 8 Billion 2013 Ar. 6 Communicable Diseases 4 MCH/Reproductive health Neglected Tropical Diseases 2 Non Communicable diseases 0 2006 2008 2009 2010 2011 2012 2013 Targeted Investment expenditure (b) 30 25 Billions 2013 Ar. 20 Communicable diseases 15 Non Communicable diseases 10 Neglected Tropical Diseases 5 MCH/Reproductive Health 0 2006 2008 2009 2010 2011 2012 2013 (a) Recurrent expenditure are entirely financed with internal resources (RPI) which include budget support. (b) Investments included here are limited to those included in SIGFP. No financing could be identified in the Investment category using this classification by type of disease in 2006. This could be due to changes in accounting by SOA. Source: Data from MFB/SIGFP. d) An attempt to disaggregate external aid by functional category 67. As highlighted above, external aid is an important component of total public spending on health. While the lack of systematic recording of external support to the sector makes it difficult to analyze the composition of external aid, there is substantial evidence pointing to large and continuous support from external partners in providing health inputs, such as vaccines and other health materials, over recent years. In particular, support from GAVI and the malaria funds have ensured the provision of essential inputs to the system. However, this support, while largely in alignment with specific disease control strategies, is generally channeled outside of the MoH. Box 7 summarizes the findings of a recent case study by UNICEF on financing immunization. As noted above, this type of analysis needs to be carried out across external financing for health given its large share of the total budget for health. 49 Madagascar PER – Health |Section C Box 8. Financing Vaccination in Madagascar The effective financing of the immunization program relies mainly on predicable and timely funding from external development partners such as GAVI Alliance grants, UNICEF, WHO, World Bank, with additional support (technical and resources) provided by other partners such as the development agencies of the UK, USA, French and EU. The political crisis in 2009 and subsequent macroeconomic constriction. Since the 2009 crisis, authorities have implemented tight budgetary policies to preserve macroeconomic stability which has severely constrained the available fiscal space of the government; and in turn, significantly impacted social sector spending, including on immunization services and . This has been accompanied by a steep fall in vaccination rates: complete immunization coverage for children 12 to 23 months old dropped from 62 percent in 2008 (DHS 2008/2009) to 51.1 percent in 2012 (ENSOMD 2012/2013).Routine vaccination expenditure represents 0.11% of GDP. According to data in the annual progress reports, in 2010 there was a total health expenditure on vaccinations of about US$11.5 million. Out this amount, the Government was responsible for 17%, about $2.2 million. Between 2010 and 2013, the Government share of financing dropped from 17% to 7%, representing a drop of 10 percentage points in 4 years. Figure 1: Proportional breakdown of vaccination funding Breakdown of vaccination funding Government GAVI Other Technical and Financial Partners 30% 16% 28% 17% 52% 72% 61% 76% 17% 12% 11% 7% 2010 2011 2012 2013 Over the last four years, the Government share of costs for the vaccination program is low and is declining with most expenditures being used for the purchase of vaccines, fuel, and payment of salaries. Currently, over 93% of program resources are provided by technical and financial vaccination partners. Since 1997, the country’s National Expanded Program on Immunization (EPI) has had a budget line for the purchase of routine vaccines. But with the introduction of new vaccines, the funds allocated fall short of the full costs of procuring routine and new vaccines. In the context of slow economic recovery current projections do not predict sufficient domestic revenues in the short term for the Government to meet co-financing obligations of vaccine procurement or to finance expansion of immunization coverage to achieve objectives of universal and equitable access to save lives and minimize illness from vaccine preventable diseases. With some traditional partners indicating a progressive reduction in financial support to immunization, the financing gap for immunization can only be filled over the next few years if increasing domestic resources are combined with additional financing from existing and new partners. The Government also noted that a draft law on the sustainable financing of vaccination is under preparation, which would permit the Government to earmark resources for vaccination, thereby securing funding for specific requirements of vaccine procurement and supply chain management at all levels. The main recommendations from the immunization financing review include: - Capacity building of teams responsible for vaccination at all levels - Improved monitoring of EPI expenditures at all levels - Strengthened reporting, management and feedback on expenditures at all levels - Finalization of the draft law on the sustainable funding of vaccination - Strengthened ministry leadership in coordinating interventions - Strengthened advocacy around partnership and financing for immunization with all partners Source: Case study on immunization expenditures in Madagascar, UNICEF, 2015 50 Madagascar PER – Health |Section C C3. Elements of a Productivity Analysis 68. The impact of MoH expenditures within the country can be assessed in terms of the number of direct beneficiaries. The classifications constructed for the functional analysis can be used to shed some light on whether government expenditures go to services that are most used. The unit costs presented below are based on 2013 salary expenditures.48 Annual utilization (number of people who visit a given facility over the year) is calculated based on the 2010 EPM.49 69. As expected, unit costs increase exponentially as we move up the levels of care. Relative to level 2 health centers (CSB2) that are the most frequently visited, CSB1s (level 1 health centers) cost 30 percent less per visit, primary hospitals (CHRDs) cost 30 percent more, secondary regional hospitals (CHRR) cost almost three times more, and tertiary hospitals (CHU), including main hospitals at the province level, cost about 30 times more (Figure 26). Figure 26. Unit Costs by Type of Facility, 2013 Unit costs in thousand Ar. 95.69 2.55 3.45 4.61 9.46 CSB1 CSB2 CHRD CHRR CHU Methodology note: Salaries for 2013 based on category and number of staff per facility, and average salary per category (MoH). Utilization numbers population weighted and annualized from 2010 EPM. Source: Data from MoH 2013, MFB, EPM 2010. 70. Expenditures roughly follow utilization patterns up to the CHRR level (Figure 27). Unit costs do not reveal whether differences are due mostly due to low utilization or high costs, so it is useful to present the components separately. It is clear that low utilization drives the high unit cost at the CHU level. This result need to be put in perspective, considering what was found in terms of the provision of primary care in CHU hospitals. It may make sense to deploy some of the primary care staff in CHUs to facilities that cater to the largest numbers. 48 Including other expenditures would not change the interpretation of these results. Similarly, looking at non-wage expenditure separately using SIGFP data would not be informative, given that expenditures cannot be attributed to the different facilities. 49 The calculation was done using the survey weights and multiplying the results for the last 2 weeks by 24 to get annual estimates. 51 Madagascar PER – Health |Section C Figure 27. Expenditure Shares vs Utilization Shares by Type of Facility, 2013 Expenditure % Utilization % 64% 40% 33% 15% 12% 15% 7% 7% 4% 2% CSB1 CSB2 CHRD CHRR CHU Source: Data from MoH 2013, MoF, EPM 2010. Key Findings Expenditures on labor have been increasing both in real term and in share of total expenditures, while other operational expenditures and internally financed investments have decreased. This indicates that Madagascar has clearly moved to an unbalanced situation that is concerning in terms of both efficiency and sustainability in delivering a sufficient amount of quality health services to the population. Economic analysis of public spending  Regular salary expenditures in Madagascar have reached levels (84% of domestic financing in 2013) that are much higher than those generally observed in other low-income countries.  Expenditures on goods and services related to the provision of health care make up a very small share of the budget, with the cost of most medical consumables borne by the patient through cost recovery.  Since 2010, labor expenditures have crowded out expenditures on goods/services and investments managed by the MoH.  The large bias towards salaries in domestic funding, which is already very restricted (approx. 20% of public financing for health) is somewhat compensated by substantial inflows of external aid targeting other aspects of the system, in particular goods and services. However, investment financed by external aid has dropped considerably, making it very difficult to sustain improvement in the quality and quantity of health services delivered.  Although Madagascar was found to do well overall in term of outcomes relative to expenditures before 2008, the current imbalance in the use of inputs shows that the country will likely not keep this advantage, given that the trend since 2009 is not technically efficient and is unsustainable. Functional Allocation of Total Health Expenditures (Public and Private) Madagascar does not exhibit the common SSA pattern of over-spending on in-patient care and under-spending on preventive and public care: 52 Madagascar PER – Health |Section C o The share of hospital care decreased while the share of spending on prevention and public health programs increased over the period 2003 -2010. o The share of expenditures going to inpatient care is less than 7 percent, which is less than a third of LIC averages. o Pharmaceutical costs have also remained stable, at less than 20 percent of total expenditures, and these costs even decreased in 2013. These indicators could be a signal of system failure, in the sense that the majority of the population may just not be seeking care. The 2010 Household Survey indicates that close to 70 percent of people in Madagascar did not seek care when ill. Functional allocation of public health expenditures  The 2010 NHA results indicate a distribution of public expenditures that strongly prioritizes activities with high public good characteristics, as would be expected of the public system.  Total salaries at service delivery levels have remained low and constant since 2006, while central and regional administrative salaries increased exponentially in the same period.  Primary health care facilities absorb only 27 percent of wages, while 50 percent of the population seeking care go to these facilities.  In Madagascar, similar to other low-income countries, a great deal of primary health care is provided outside of primary health care facilities, especially in tertiary hospitals. Recurrent non-wage expenditure Administration and coordination takes the largest share of recurrent non-wage expenditures, more than PHC, MCH, and public health combined. External aid analysis by functional category  Lack of systematic recording of external support to the sector makes it difficult to analyze the composition of external aid but there is substantial evidence pointing to large and continuous support from external partners in providing health inputs, such as vaccines and other health materials, over recent years. However, this support is generally channeled outside of the MoH.  A recent case study on financing vaccination found that routine vaccination expenditure represents only 0.11% of GDP. According to data in the annual progress reports, in 2010 there was a total health expenditure on vaccinations of about US$11.5 million. Out this amount, the Government was responsible for 17%, about $2.2 million. Between 2010 and 2013, the Government share of financing dropped from 17% to 7%, representing a drop of 10 percentage points in 4 years. There was also a fall in nominal terms. With the dramatic fall in immunization rates over the same period, this downward trend is concerning.  Given the high share of external aid in the total financing of the health sector, the absence of an updated national strategy and of as well as fully functioning coordination and alignment mechanisms could be impeding the realization of important synergies across sources of financing. Policy implications Over time, there is a need to redirect spending on activities and interventions that improve the delivery of quality health services Better use of resources to improve service delivery of quality HNP interventions  In the current budget envelope, there is an urgent need for the Government to address the wage vs. non-wage expenditures to improve the efficiency and strengthen the sustainability health service delivery over time. Any incremental increases on available budgets should be directed to operational budget and investments 53 Madagascar PER – Health |Section C  One of the key areas that should be addressed is the exponential expansion of central and regional administrative salaries in the last decade when service delivery salaries have not changed in the same time period.  The Government aim to improve capacity and service delivery at the primary which will require a redistribution of the wage bill to ensure that it is in line with the level at which services are being utilized.  The distribution of non-wage expenditures should be re-balanced to support the delivery of critical public health programs.  Consider the use of output-based approaches compared to the current largely input-based funding methodologies More informed budgeting  The Government could consider institutionalizing NHA exercises every two years. . The tool should be adapted to Madagascar’s specific system and needs.  More robust analysis of external aid financing is urgently needed to have a more exact analysis of the budget. 54 Madagascar PER – Health | Section D SECTION D. DISTRIBUTIONAL ANALYSIS OF PUBLIC HEALTH EXPENDITURE D1. Distribution of MoH Expenditures by Region and Type of Residence 71. While it is not surprising that private expenditures per capita are higher in richer regions, public expenditures should be expected compensate the difference in spending. Looking at the totality of health expenditure, however, per capita total current health expenditures are found to be negatively, though weakly, correlated with poverty rates (corr=-0.39, sig=0.07).50 This section shows that MoH expenditures are in fact more regressively distributed across regions than total expenditures (excluding investment). It also shows that national figures hide significant heterogeneity among regions. The analysis first looks at the per capita distribution of MoH expenditures by province and region, then analyzes the relationship between per capita expenditures and regional poverty levels. Expenditures directed to primary health care are examined separately. The last subsection briefly assesses rural/urban differences. 1. Regional allocations vs population shares 72. Looking at changes in the distribution over time, Antananarivo alone has been driving the increase in current MOH health expenditures. The allocation of these expenditures by province appears to roughly follow population ratios. This can be seen in the significant differences between some provinces, with Antananarivo benefiting from twice the amount of per capita government spending relative to Toliara and Fianrantsoa in the south (Figure 28).51 Figure 28. Current MOH Health Expenditure Per Capita by Province in Relation to Population 35 7 Millions per capita, 2013 (1000 of Ar.) 30 Average 2009-2013 6 Billion constant 2013 Ar. 25 Population 5 6 20 4 5 5 4 15 3 3 3 10 2 5 1 0 0 Source: Expenditure data from MFB, population data from INSTAT Madagascar. 50 Regional disaggregation of Current Health Expenditures (CHE) based on the NHA 2010 report and poverty rates based on EPM 2010. 51 About one third of the MoH current expenditures are allocated to the central level (up to a peak at 42 percent in 2008), the rest being allocated across provinces. Unfortunately, the analysis cannot include investment because only a very small part is identified outside of the central level in SIGFP. 55 Madagascar PER – Health | Section D 73. By contrast, at the regional level, salary shares (Table 12) show inequality with respect to population shares, (Figure 30). There is a clear disadvantage to the region of Vatovavy-Vatovivany. Analamanga accounts for about 15 percent of the population but for 20 percent of all salary expenditure, not including central ministry personnel. All other regions receive less than their estimated population shares, as depicted in Figure 30. Regions that receive significantly less than their population shares are Androy, Atsimo Atsinana, Sofia, Vatovavy Fitovinany, and Vakinankaratra. An important caveat to this analysis relative to population sizes is that Madagascar has not had a census for many years, so estimated per capita distributions may not accurately reflect the situation. Table 12. Share of Salary by Region, excluding Central Administration Personnel, 2006-2013 Expenditure shares, salary only % of total est. Region 2006 2007 2008 2009 2010 2011 2012 2013 pop., 2013 ANALAMANGA 20.8 18.3 19.6 19.1 19.9 19.6 19.8 15.3 VAKINANKARATRA 4.3 4.7 4.6 4.7 4.8 4.8 4.4 8.3 VATOVAVY-FITOVINANY 3.2 3.2 2.9 2.9 2.7 2.6 2.4 6.5 ATSIMO-ANDREFANA 4.6 5.4 5.3 5.0 4.6 4.5 4.3 6.0 ATSINANANA 5.7 5.5 5.2 5.3 5.1 5.2 5.4 5.8 SOFIA 2.7 3.3 3.3 2.8 3.0 2.6 2.5 5.7 HAUTE MATSIATRA 5.3 5.2 4.4 4.4 4.2 4.2 4.0 5.5 ANALANJIROFO 3.3 2.6 3.0 2.7 2.6 2.7 2.9 4.7 ALAOTRA-MANGORO 3.4 3.8 3.7 3.7 3.7 3.6 3.5 4.7 SAVA 3.0 2.8 3.0 2.8 2.8 2.9 2.7 4.5 ATSIMO-ATSINANANA 1.5 2.6 1.8 1.9 1.8 1.7 1.5 4.1 BOENI 3.7 4.8 4.3 3.8 3.7 3.4 3.5 3.7 ANDROY 1.6 1.4 1.5 1.5 1.4 1.4 1.3 3.4 ITASY 2.0 2.0 2.0 2.0 2.3 2.2 2.0 3.4 AMORON'I MANIA 2.4 2.8 2.2 2.2 2.3 2.2 2.1 3.3 DIANA 4.4 4.0 4.1 3.2 3.1 3.1 2.8 3.2 ANOSY 2.1 2.1 2.0 1.7 1.6 1.6 1.5 3.1 MENABE 2.7 2.4 2.4 2.1 2.0 1.7 1.7 2.7 BONGOLAVA 1.0 1.1 1.1 1.2 1.3 1.3 1.2 2.1 IHOROMBE 0.8 0.9 0.7 0.9 0.8 0.9 0.9 1.4 BETSIBOKA 0.9 0.9 1.0 0.8 0.9 0.8 0.7 1.3 MELAKY 0.9 1.0 1.1 1.1 1.1 1.0 0.8 1.3 Source: MFB, Direction de la Solde. 56 Madagascar PER – Health | Section D Figure 30. MoH Distribution of Wage Expenditure across Regions, 2013 Note: 45-degree line represents that the share spent on salary expenditure is proportional to the share of population Source: Data from MFB, Direction de la Solde. 2. Relationship between current expenditures and poverty by region 74. The following considers all non-investment expenditures, combining SIGFP and salary data.52 Poverty levels are calculated using the periodic Household Surveys for 2005 and 2010. To increase data quality, and in particular to smooth issues related to changes in personnel, averages of 2006-2008 current expenditures are compared to averages for 2009-2013.53 75. Regions with higher poverty rates received less funding on average. 54,55 The total recurrent per capita expenditure in health, ordering regions by their 2005 poverty rate in shown in Figure 31. There are two important caveats to this interpretation: a) per capita expenditures do not include transfers, in particular the transfers to tertiary hospitals, which represent about 15 percent of total non-wage recurrent expenditure and are not distributed by region in the government accounts. Given the location of these hospitals, 52 Salaries make up between 69 and 93 percent of the total, depending on the region and the period considered (the relative share of salaries is higher in the second period). 53 Non-wage current expenditures from SIGFP are averaged over 2006 and 2008 for the first period and 2009 to 2013 for the second period. Salaries are averaged over 2006 to 2008 for the first period and 2010-2013 for the second. 54 The per capita consumption figures do not show the same smoothing out of regions as do poverty rates: per capita consumption dropped more or less evenly across regions. 55 Ordering countries by decreasing per capita consumption (in PPP) instead of by poverty rates also shows the strongest differences when comparing the richest and the poorest regions (Figure A in Annex 1). 57 Madagascar PER – Health | Section D including these transfers in the analysis would likely make the unequal distribution more apparent, and b) these data do not include foreign aid. Given that foreign aid has been specifically directed to the poorer regions, it could be the case that some domestic funding in these regions has been displaced. 76. With the exception of Analamanga, the regions that received the highest share of MoH expenditure also experienced the highest cuts. Two regions with notable changes were Melaky and South Atsinanana, where poverty significantly increased while per capita expenditure of the MoH decreased by about one half. Overall, all regions were affected by the decrease in both the range and variance per capita of MoH expenditures. Figure 31. MoH Recurrent Expenditure Per Capita by Region and Poverty Rate, 2006-2010 12.0 100 % Thousand of 2013 Ariary 90 10.0 80 8.0 70 60 6.0 50 Average 2006-2008 40 Average 2009-2013 4.0 30 20 Poverty Ratio 2005 2.0 10 Poverty Ratio 2010 0.0 0 VAKINANK… VATOVAVY… ANALAMA… HAUTE… ATSIMO-… ATSIMO-… ALAOTRA-… AMORON'I… ANALANJIR… BONGOLAVA MELAKY SAVA ANDROY BOENI ITASY BETSIBOKA ANOSY IHOROMBE SOFIA DIANA MENABE ATSINANANA Sources: Data from MFB, INSTAT (population), EPM (poverty). 77. Expenditures are strongly negatively correlated to poverty ratios and strongly positively correlated to per capita consumption, which is a proxy indicator for income. The relationship becomes slightly more regressive in the later period when using consumption per capita but not using poverty ratios (Table 13). Table 13. Correlation Between MoH Current Expenditure and Regional Poverty Poverty ratio Per capital consumption 2006-2008 -0.72 0.71 2009-2013 -0.71 0.74 Note: All coefficients are significant at α<0.001. Source: Data from MFB, INSTAT (population), EPM (poverty). 58 Madagascar PER – Health | Section D 78. The regional distribution presented above is necessarily affected by the geographical location of tertiary care facilities. Contrary to primary care activities, there is an important trade-off between efficiency and regional equity for tertiary care facilities because of large economies of scale in activities that requires expensive capital inputs. In order to get at the notion of equity within the constraints of feasibility, one needs to concentrate on the provision of primary health care. 79. The analysis below looks at a snapshot of the situation (rather than its evolution, as above), to compare the 2013 primary care current expenses per capita to 2010 poverty rates by region. 56 To obtain total current expenses for primary care, non-investment expenses recorded in the basic health provision program of SIGFP were added to salary expenses in primary health care centers, primary hospitals, and district public health services.57 The finding is that salaries largely dominate, with non-salary expenses accounting for just 3 percent of the total, on average, and no more than 8 percent for a given region. A second type of analysis was done using salaries only, based on our estimates of primary care provision in all facilities (not just PHC facilities). 80. Even when restricting expenditures to the primary health care level, regional distribution is strongly regressive. Regions with lower poverty rates receive more per capita, while regions with higher poverty rates receive less (Figure 32). Although there are wide differences in per capita MoH expenditures in the middle range of poverty rates, differences are marked at the two extremes. The same pattern obtains when using per capita consumption instead of poverty rates, with higher MoH expenditures per capita in richer regions (Annex 2, Figure A.1). Correlation coefficients, although slightly lower than before, still show strong regressivity whether they are calculated against poverty ratios or consumption per capita (Table 14). 56 Although salaries by function were estimated down to 2006, the estimation could not take account of changes in the number of personnel by type of facility for 2006-2012. The 2013 estimates have the least margin of error. 57 The salary breakdowns by facilities group district hospitals of category 1 and 2 (CHD1 and CHD2) in a single category (CHDR) while the budget programs for 2013 include CHD1 expenses in basic care and CHD2 expenses in hospital care. The ratios of specialized to non-specialized staff salaries in CHRD by regions are used to estimate the salaries going to CHD2s, which excluded from the total presented here. 59 Madagascar PER – Health | Section D Figure 32. MoH Recurrent Expenditure Per Capita on Primary Care by Region, 2013 10 100 % Thousand of 2013 Ariary 8 80 6 60 4 40 2 20 0 0 Current exp/c on basic health care services (based on type of facility) exp/c on primary care based on personnel type (salaries only) Poverty rate (2010 EPM) - Secondary axis Data sources: Authors calculations based on MFB/SIGFP, INSTAT (population), EPM (poverty), and MoH/DRH. Table 14. Correlation Between MoH Expenditures on PHC and Regional Poverty Per capita Poverty ratio consumption Current expenditure in basic health care facilities and district services -0.63 0.59 Salary expenditures on PHC/PH (based on type of personnel) -0.71 0.74 Note: All coefficient significant at α<0.01. 81. Given the importance of primary care dispensed in tertiary hospitals, as discussed in section C, the argument for excluding secondary and tertiary care facilities when assessing regional equity needs to be qualified. Although it is not desirable to increase the number of tertiary care facilities due to economies of scale, primary care staff in these facilities could be deployed to different facilities at no cost in terms of efficiency. Moving forward, there needs to be a more rational allocation of primary care staff. 82. In summary, whether restricting expenditures to primary health care facilities, primary health care personnel, or including all current expenditures, expenditures of the MoH are characterized by a strongly regressive regional distribution. It is very likely, especially considering that the distribution of total health expenditures is less regressive than MoH expenditures, that externally financed health expenditures not going through government accounts are progressively distributed. Nevertheless, it is clear that efforts need to be made on the part of the Government to reach poorer regions. In particular, given that the MoH covers most expenditures on medical personnel salaries, efforts to better deploy medical staff to poorer regions need to be made. Indeed, given the large weight of salaries in MoH expenditures, these results are consistent with an unequal distribution of primary health care personnel across regions. 60 Madagascar PER – Health | Section D 3. Rural/urban differences 83. Given the differences in poverty ratios and consumption per capita between urban and rural areas (Table 15), it is important to look more closely at rural and semi-rural disadvantaged areas. In order to do the analysis by type of residence, we used the official delineation of communes into rural and urban categories according to the 2011 Decree.58 Communes were classified into four levels of urbanization so as to minimize the “border” effects and avoid over-identifying communes as rural.59 Although salaries cannot be identified by type of residence, the share of non-wage recurrent expenditure that can be identified in the data is sufficient to be indicative of the level of activity.60 Table 15. Poverty and Consumption by Type of Residence, 2005 and 2010 Poverty rate Per capita consumption % 2013 Ar, capital city 2005 2010 2005 2010 Capital city 34.4 30.1 266014 680428 Large urban centers 43.4 31.6 230112 662635 Secondary urban centers 63.4 62.1 167412 426743 Rural areas 73.5 82.2 137847 283781 Source: Household Surveys 2005 and 2010. 84. Analysis of expenditure shares by type of residence show that less five percent goes to rural communes. Considering that approximately two-thirds of the population live in rural areas, this represents a highly unequal distribution of expenditure shares. On average 36 percent going to large urban centers, 46 percent to smaller urban areas and 13 percent to the semi-rural or peri-urban areas (Figure 33 and Table 16). 58 Décret-n°2011-0042-portant-classement-des-Communes-en-Communes-urbaines-ou-en-Communes-rurales (MFB). 59 Out of a total of 1,549 classified communes, 9 were identified as large urban centers (including the capital city, which is classified separately), 63 as urban (category 2 urban), 104 communes as semi-rural or peri-urban (category 1 rural), and 1,373 as rural (category 2 rural). 60 The government expenditure data cannot be directly divided into rural and urban locations but it identifies activity units by commune code. 61 Madagascar PER – Health | Section D Figure 33. Rural/Urban Shares of Non-Wage Recurrent Expenditures Directly Assigned to Communes 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% Million 2013 Ar. 0% 2006 2008 2009 2010 2011 2012 2013 Rural 674 1003 1023 875 880 527 288 Rural/Urban 1896 3321 3275 2491 2298 1405 815 Urban 7199 12685 11059 8830 8179 5116 2757 Large urban centers 6081 8374 6929 7048 6543 3631 2744 Note: Population shares are based on standard definition used in the World Development Indicators, which may or may not include the semi-rural or peri-urban areas indicated here as rural/urban. Source: Data from MFB/SIGFP. Table 16. Share of Expenditures to Rural Areas versus Share of the Population 2006 2008 2010 2011 2012 2013 share of exp to rural* 4% 4% 5% 5% 5% 5% share of exp to rural-extended * 16% 17% 19% 17% 18% 18% Approx share of rural population 71% 69% 69% 68% 67% 67% *Note: Rural only includes the most rural communes (category 2). Peri-urban communes (category1) are included in the rural-extended category. Data source: WDI, MFB/SIGFP. D2. Out-of-Pocket Household Expenditures: Do They Impact Regional and Income Inequalities? 85. Looking at the size and evolution of household expenditures in a context of increasing poverty, it is important to gauge the need for the Government to relieve some of the financial burden on poorer households. The results of this section need to be interpreted in the context of changes in utilization, in particular whether people tend to seek care when they are sick. The analysis below is based on data from the 2005 and 2010 Household Surveys and the UN Household Survey on progress towards the Millennium Development Goals in 2012. 1. Distribution of OOP expenditure by income level and region 86. Households in Madagascar spent less than 1 percent of their budget on health in 2005, just above 1 percent in 2010, and 1.4 percent in 2012. The distribution of consumption by quintile is given in 62 Madagascar PER – Health | Section D Figure 34. In interpreting the data, it is important to note that the poverty level is situated around the average level of consumption in the fourth quintile (and closer to the high end of the fourth quintile in 2010). In 2005, the distribution of health expenditure as a share of the household budget was slightly increasing with consumption, except in the middle range. In 2010, however, the distribution changed very much to the disadvantage of the poorest quintile. In absolute terms, average household expenditure per person increased 22 percent in the poorest quintile, 19 percent in the top quintile, 9 percent in the fourth, and 5 percent in the second, while it decreased 5 percent in the third quintile. In 2012, however, the increase for the poorest quintile slowed to 15 percent, while the average household expenditures increased 18 and 37 percent for the second and third quintiles and over 50 percent for the richest quintile. Figure 34. Household Out-of-Pocket Expenditure, 2005, 2010 and 2012 1.8 20 % of budget 1.6 people under poverty level 18 2005 Thousnad constant 2013 Ar. % of household budget 1.4 16 % of budget, 1.2 14 2010 12 % of budget, 1 10 2012 0.8 8 0.6 Avg exp. per 6 person, 2005 0.4 4 Avg Exp. per 0.2 2 person, 2010 0 0 Avg Exp. per Lowest Second Third Fouth Richest person, 2012 quintile quintile quintile quintile Source: Data from household surveys 2005, 2010 and ENSOMD 2012. 87. Average per capita OOP across regions increases as regional poverty decreases, which is most noticeable in the richest three regions (Diana, Boeny, Analamanga). Figure 35). The correlation between OOP spending per capita and poverty rates is negative, statistically significant, and increasing from 2005 to 2010 (from 0.44 to 0.68). There are, however, large difference between regions as well as unequal gains and losses between 2005 and 2010. 63 Madagascar PER – Health | Section D Figure 35. Per Capita OOP Health Expenditure by Region and Poverty Level, 2005 and 2010 12.0 100 10.0 80 8.0 60 6.0 4.0 40 2.0 20 0.0 0 Average HE per person, 2005 Average HE per person, 2010 Poverty Rate 2005 (Axis 2) Poverty Rate 2010 Note: Expenditure per capita adjusted to reflect prices in the capital. Source: Data from household surveys 2005, 2010 and ENSOMD 2012. 88. Four of the six poorest regions experienced larger than average OOP. Some regions in particular stand out. Androy, which is the second poorest region, had one of the higher OOP rates in the country in 2005, and despite a decrease in 2010, it still has higher than average OOP per capita. In 2010, expenditure per capita in Sofia reached the same level as in Analamanga, with the lowest poverty rate. Menabe had an OOP rate well below expected, given its relative poverty rate in 2005, but fully lost its advantage in 2010 after its OOP per capita rate more than doubled. 89. No statistically significant correlation was found between the average OOP burden (share of OOP in the household budget) and poverty rates. As Figure 36 shows, there are large variations between regions, and Androy as an important outlier, with a OOP burden much higher than all other regions and the highest poverty rate in 2010 and in 2012 (94.5 and 96.7 percent). 64 Madagascar PER – Health | Section D Figure 36. Burden of OOP Spending by Region and Poverty Level, 2005, 2010 and 2012 3.5 120 3 100 % of household budget 2.5 80 2 60 1.5 40 1 0.5 20 0 0 Boeny Melaky SAVA Anosy Sofia Haute Matsiatra Androy Atsinanana Alaotra Mangoro Bongolava Itasy Atsimo Andrefana DIANA Ihorombe Menabe Betsiboka Analamanga Amoron’i Mania Analanjirofo Vakinankaratra Atsimo Atsinanana Vatovavy Fitovinany HE/budget, 2005 HE/budget, 2010 HE/budget, 2012 Poverty Rate 2005 (Axis 2) Poverty Rate 2010 Poverty Rate 2012 Source: Data from Household Surveys 2005, 2010 and ENSOMD 2012. 2. Relationship between OOP spending and MoH expenditures using regional data 90. The relationship between MOH recurrent expenditures and OOP per capita is positive overall (Figure 37). This is mainly due to differences in the richest three and poorest three regions. The correlation coefficients between average OOP per person and per capita recurrent expenditures of the MoH by region are positive and increase when comparing 2005 and 2010 (0.42 to 0.59). When looking at average OOP expenditure as a share of the household budget, however, the relationship is no longer statistically significant. 65 Madagascar PER – Health | Section D Figure 37. Average OOP Per Capita vs MoH Expenditure by Region, 2005 and 2010 12.00 12.0 10.00 10.0 8.00 8.0 6.00 6.0 4.00 4.0 2.00 2.0 0.00 0.0 SAVA Melaky Sofia Anosy Boeny Androy Haute Matsiatra Itasy Bongolava Atsinanana Betsiboka Atsimo Andrefana DIANA Ihorombe Menabe Atsimo Atsinanana Amoron’i Mania Analanjirofo Alaotra Mangoro Analamanga Vakinankaratra Vatovavy Fitovinany OOP/person, 2005 (thousand 2013 Ar.) OOP/person, 2010 (thousand 2013 Ar.) Average of 2006-2008 recurrent MOH expenditure per capita (secondary axis, thousand 2013 Ar.) Note: The expenditure data are averaged over 2006 to 2008 for salaries by region. Other current expenditures do not include 2007 data. Values are not deflated to reflect prices in the capital. Source: Data from Household Surveys, MFB, INSTAT Madagascar (population). 3. Analysis of “catastrophic” OOP expenditure 91. The scarcity of risk pooling mechanisms in Madagascar, combined with a cost recovery system that makes public health care expensive for the poorer quintiles, leaves poor households especially vulnerable to catastrophic health expenditures. Health expenditures are considered catastrophic if they force the individual or household to engage in behavior that has long-term negative effects on their economic well-being. This could involve selling capital assets, heavy borrowing to pay for medical care, lost wages during an illness, or reducing the consumption of subsistence goods. Following Xu and al. 66 Madagascar PER – Health | Section D (2003) and the approach used by the WHO, health expenditure is defined as catastrophic if payments for health care exceed 40 percent of the household’s non-subsistence expenditure or its capacity to pay.61 62 92. Few households are subject to catastrophic expenditure, but prevalence increased in all quintiles between 2005 and 2010, and in the middle class in 2012. The proportion of households with catastrophic health expenditure is low in Madagascar, affecting less than 2.5 percent of household overall, with significantly higher prevalence in the highest consumption quintile (up to 4 percent in 2010) and very low prevalence in the lowest quintiles (Figure 38).63 The results for 2010 show increasing overall proportions of households with catastrophic OOP spending, but the prevalence is slightly decreasing in 2012, except for households in the middle quintile, with the most noticeable increases in the fourth quintile. Figure 38. Incidence of Catastrophic OOP Expenditure by Wealth Quintile Catastrophic = OOP health share greater than or equal to 40% of non- subsitence expenditure 6.0 4.0 2005 2.0 2010 0.0 2012 Poorest Q2 Q3 Q4 Richest (Q1) (Q5) Source: Data from Household Surveys 2005, 2010 and ENSOMD 2012. 93. Very low percentages of households with catastrophic expenditures in the poorest quintiles are usually indicative of “system failure”; i.e., the poor just do not seek care. In Madagascar, the proportion of people not seeking care when ill is fairly high, an average of 70 percent in 2012, across all 61 Following the work of Xu and al. (2003), this capacity is defined as the household consumption expenditure available after basic needs have been met. Given that the poorer the household the higher the share of consumption devoted to food, the subsistence expenditure is defined as the average of food expenditure of households whose food share was in the 45th to 55th percentile. Capacity to pay of the ith household is defined as: CTPi = EXPi – SE45-55i where SE45-55i corresponds to the average food expenditures of households in the 45 th to 55th percentiles and adjusted for the size of the ith household. In cases when total expenditure was less than this basic subsistence level, the own household’s observed food expenditure was used instead. 62 The World Bank defines catastrophic expenditures as spending 10 percent or more of total expenditure at household level on health care costs. (Pradhan and Prescott 2002; Ranson 2002; Wagstaff and van Doorslaer 2003) 63 A recent analysis of catastrophic spending in Zimbabwe (PER) returned similar results, with catastrophic health expenditures varying from 0.3 percent in the poorest quintile to 1.9 percent in the richest quintile (although the subsistence income was calculated including some basic clothing and shelter) Catastrophic expenditure can, however, be very high in some countries where people actually use the health system. In Ukraine, for example, 25 percent of households in the two lowest quintiles faced catastrophic health spending, but the prevalence decreased in highest wealth quintiles. 67 Madagascar PER – Health | Section D the quintiles. More than 55 percent of the richest households and 65 percent of the poorest do not seek health care. Low utilization (rather than high cost) is therefore a plausible explanation for the low incidence of catastrophic expenditure. When looking at people who do not seek care for financial reasons, only the poorest quintile clearly stands out in 2010, at 25 percent, while the range in the other quintiles is 10 (richest quintile) to 15 (second and third quintile) (Figure 39). Figure 39. Percentage of People Who Did Not Seek Care when Sick for Financial Reasons, 2005 and 2010 2005 2010 2012 25.3 24.1 20.7 21.9 19.2 19.0 14.6 15.1 14.4 11.5 11.6 11.2 8.3 10.0 10.7 Poorest quintile Quintile 2 Quintile 3 Quintile 4 Richest quintile Source: Data from households surveys 2005, 2010 and ENSOMD 2012. 94. Catastrophic expenditures may cause 4.5 percent of people in the richest quintile and 3 percent in the fourth quintile to fall below the poverty line. The results above need to be put in perspective. Nearly 80 percent of the population in Madagascar lives under the poverty line. Most of the people in the fourth quintile are poor, and the minimum consumption level in the richest quintile is only 9 percent above the poverty line. Therefore, households at the bottom of the fifth and top of the fourth quintile are most in danger of falling into poverty and may experience negative long-term financial impacts from catastrophic expenditures (given that the poor have little to lose in terms of assets). In addition, OOP health expenditures are also impoverishing because they can cause those that are already poor to call deeper into poverty. D3. Distribution of MoH Expenditures by Socioeconomic Category of User: Benefit Incident Analysis 95. A Benefit Incidence Analysis (BIA) can be carried out using the simple utilization (or usage) approach, assuming that all individuals using the service receive the same benefits.64 The analysis is based on (a) decomposing users of public health facilities by income level, and (b) decomposing public health expenditures by type of health facility.65 For Madagascar, the first part of this process relies on good quality 64 This is the approach used in Glick and Razakamanantsoa (2002) in their BIA for health and education using 1990 data for Madagascar. 65 The usage approach assumes that all individuals using the service receive the same benefits. It also assumes that the quality of health services does not depend on average income levels in the area. Although these assumptions are not realistic, the approach has the advantage of being less demanding in terms of data quantity and quality. The unit cost approach to BIA is not be feasible for health given the structure of expenditure accounting. 68 Madagascar PER – Health | Section D data from the periodic Household Survey, which provides information on the marginal benefits of government expenditure and how each segment of the population would benefit from additional spending directed to different types of facilities. The second part of the BIA depends principally on the ability to decompose public expenditures by type of facility to match the type of facilities identified in the Household Survey. As noted in section C2, data are not readily available to break down salaries by type of facility, but a reconstruction can be done using multiple data sources and approximations using average salaries by type of personnel (Annex 3). 1. Marginal benefit of government expenditure by type of facility 96. Additional spending on CSB1s (basic health centers) is pro-poor. The distribution of benefits is consistent on the primary level. The distribution of users by quintile of consumption for different types of facilities in 2005, 2010 and 2012, using Lorenz curves, are shown in figures 40 and 41. Spending on CSB2s benefited all quintiles approximately equally in 2010 (equality appears to have slightly improved compared to 2005, likely due to the transformation of CSB1 into CSB2). Inequality becomes apparent at the CHD level and clearly increases as we move up into levels of care toward CHUs, which are clearly utilized by the rich. No clear conclusion can be drawn about changes in inequality across 2005, 2010 and 2012, especially at the hospital level, but this could be due to the fact that results are based on much lower sample sizes in 2005.66 Consistent with the catastrophic expenditure analysis above, the poorest households in the lowest two quintiles rarely use secondary and tertiary hospitals. Figure 40. Lorenz Curves by Quintile for Utilization of Public Health Facilities, 2005-2010 2005 to 2010 (2005 superimposed in dotted lines) 100 90 80 CSB1 70 CSB2 60 50 CHRR 40 CHD 30 CHU 20 10 45 Degree line 0 0 1 2 3 4 5 Source: Data from household surveys 2005 and 2010. 66 Results regarding hospital utilization rest on very small unweighted sample sizes. The Household Survey question on visits to health facilities was asked of respondents who had had a health problem in the previous two weeks, which is a very short time period to assess hospital usage. A total of only 13 respondents went to a CHU, 30 to a CHRR, and 193 to a CHD in 2005; and a total of 38, 107 and 423 in 2010. These numbers were 58, 117 and 316 in the Development Goals National Monitoring Survey (ENSOMD) of 2012. 69 Madagascar PER – Health | Section D Figure 41. Lorenz Curves by quintile for Utilization of Public Health Facilities, 2005-2012 2005 to 2012 (2005 superimposed in dotted lines) 100 90 80 CSB1 70 CSB2 60 50 CHD 40 CHRR 30 CHU 20 10 45 Degree line 0 0 1 2 3 4 5 Source: Data from Household Survey 2005 and ENSOMD 2012. 2. Benefit incidence of MoH expenditure 97. On average between 2010 and 2013, the MoH spent about 20 billion 2013 Ar. on non-wage expenditures to basic health and hospital services (based on program budgets), and 67 billion on salaries to personnel working in health facilities (excluding all administrative units).67 It is important to understand how much of this money benefited the poor. The goal of the BIA is to estimate benefits across consumption quintiles in terms of utilization. Given differences in classifications and the fact that wage expenditures are so large relative to non-wage expenditures (7 times larger in 2013), the analysis is done in two parts: a) allocating non-wage expenditures by quintile and b) allocating wage expenditures. 98. The BIA for MOH non-wage expenditures reveals that benefits are found to be regressive with the average benefits going to individuals in the richest quintile two to four times higher than those going to individuals in the poorest two quintiles (Table 17). The two poorest quintiles were found to benefit the least in most cases. The BIA rests on two groups: users of primary health care facilities and users of hospital facilities.68 Whether we consider CHR users as benefiting from basic health services (method 1) or hospital services (method 2), and whether we look at 2013 or at average expenditures over the last four years, benefits are found to be regressive. Results from 2013 are more regressive when users of CHRs are assumed to benefit from basic health care expenditures rather than hospital-related expenditures (method 1), given that basic health care expenditures were lower than hospital expenditures in 2013 (0.8:1 ratio compared to 1.7:1 on average since 2010). In addition, the total benefit was substantially lower in 2013. Figures 42 and 43 illustrate the regressive nature of MoH non-wage spending. The data 67 In 2013, the numbers were 11 billion for non-wage expenditures and 73 billion for wages. 68 Programs in the 2010-2013 budgets group all expenditures for basic health centers (CSB1 and CSB2) and for level 1district hospital into basic health care services, while the Household Survey separates CSB1 and CSB2 but does not distinguish between district hospitals with and without surgery (CHD1 and CHD2). 70 Madagascar PER – Health | Section D indicates that 40-60 percent of these expenditures benefited people living under the poverty line (fourth quintile and below). Table 17. Benefits from Non-Wage Current Expenditures of the MoH by Consumption Quintile Billions of 2013 Ar. 2010-2013 2013 (a) Method Method 1 Method 2 Method 1 Method 2 Lowest quintile 3.2 2.5 1.2 1.7 Second quintile 3.0 2.5 1.1 1.6 Third quintile 3.8 3.6 2.0 2.1 Fourth quintile 4.3 4.1 2.3 2.4 Richest 6.0 7.6 4.5 3.4 Richest to poorest 1.9 3.1 3.9 2.0 Richest to second quintile 2.0 3.0 4.0 2.1 Note: Method 1 calculates benefits across quintiles assuming that CHRs provide all basic care services (preferred method). Method 2 assumes that CHRs provide only hospital services. Source: Data from household surveys, MFB/SIGFP. Figure 42. Estimated Distribution of Benefits from MoH Non-Wage Expenditures Note: Current expenditures are restricted to basic care and hospital services budget programs in SIGFP (mandated). Utilization by quintile is based on 2010 Household Survey. Method 1 calculates benefits across quintiles, assuming that CHRDs provide all basic care services. Method 2 assumes that CHRDs provide all hospital services. Source: Data from MFB/SIGFP and Household Survey 2010. 71 Madagascar PER – Health | Section D Figure 43. Estimated Distribution of Benefits from MoH Non-Wage Expenditures 2013 30.0 Billion 2013 Ar. 20.0 29.6 10.0 8.2 8.9 13.5 13.2 0.0 Lowest Second Third quintile Fouth Richest quintile quintile quintile Source: Data from Household Survey (utilization), MoH//Human Resources Department, MFB/SIGFP. 99. Benefits from MoH wage expenditures on personnel in health facilities (excluding all personnel in administrative units) show that the richest quintile benefits 3.6 times more than the poorest quintile, and at least twice as much as households in any other quintile. 100. Distribution of benefits from all MoH expenditures directed to primary health and hospital care is pro-rich with the richest quintile benefiting from 40 percent of total expenditures. Focusing on the distribution of expenditures rather than on actual amounts, benefits can be presented in a Lorenz- curve type graph that accumulates benefits across quintiles, the 45 degree line being the line of perfect equality (Figure 44). Although the distribution of wages by facility drives the result, non-wage expenditures also benefit the rich disproportionately. 69 69 It would have been interesting to conduct this analysis over two time periods. Unfortunately, the lack of budget programs prior to 2009 does not allow for reproduction of the analysis of non-wage expenditures. Although wage expenditures could be traced back to 2006, the estimation was based on strong assumptions that limit the extent to which the allocation of personnel across facilities could change, and this would greatly limit the interpretation. This exercise, however, should be repeated in the future using these results as benchmarks. 72 Madagascar PER – Health | Section D Figure 44. Estimated Distribution of Benefits from MoH Expenditures 2013 100% 90% Cumulative benefits 80% 45 degree line 70% 60% 50% Wage expenditure 40% 30% 20% non-wage, method 10% 1 0% non-wage, method 2 Total Cumulative population by quintiles of consumption Note: The distribution is based on benefits from utilization by type of facility and expenditures directed to primary and hospital care (for non-wage expenditure). Method 1 calculates the benefits across quintiles, assuming that CHRs provide basic care services. Method 2 assumes that CHRs provide all hospital services. The total take the average of method 1 and 2. Source: Data from Household Survey (utilization), MoH/Human Resources Department, MFB/SIGFP. 101. CHUs absorb about one third of all MoH expenditures directed to health facilities but are very sparingly used by the poor, while CSB1s are the only facilities that are primarily used by the poor. In fact, all other facilities are used by the richest quintile in a greater proportion than their population share. The distribution of benefits depends greatly on the utilization rates and size of expenditures at the CHU and CSB1 levels. A few data manipulations reveal that changing the distribution of benefits to be more equitable is not an easy task. Transferring expenditures from CHUs to CSB1s could reduce the inequality, although, given the low utilization of health facilities by the poor in Madagascar, it is not feasible to generate a progressive distribution scenario. Several scenarios were examined and the resulting distribution of benefits compared to the 2013 benchmark. Scenario 1 – Reallocate general practitioners: Deploy two general practitioners per district from the CHUs to the CSB1s (775 general doctors were employed in CHUs in 2013). Scenario 2 – Increase the number of CSB1s: Add 300 new primary care facilities and staff them with one doctor and one support person (this scenario assumes that the utilization profile stays as for existing CSB1). Scenario 3 – Reallocate CHU staff time to CSB1s: Dispatch one third of the CHU staff each year to provide health care in CSB1 facilities. Scenario 4 – Relocate CSB2 facilities to reach more poor households, and increase staff: Strategically relocate half of the CSB2 facilities so they are closer to poorer segments of the 73 Madagascar PER – Health | Section D population (equivalent to allocating half of CSB2 expenditures to CSB1s), and increase the CSB2 staff so there is at least one general practitioner per facility (currently 1,190 doctors for 1,620 facilities) by deploying the equivalent number of CHU general practitioners to CSB2. 102. The system clearly needs to improve its ability to reach the poor using some of the resources currently used to staff tertiary facilities. Figure 45 presents each scenario, given 2013 wage expenditures and the same utilization profile as above (except for scenario 4). It is obvious that marginal changes to expenditure allocation would not make a large difference in inequality. Scenarios 1 and 2, which are strongly equity enhancing in a marginal sense, barely change the overall picture. Scenarios 3 and 4 do make a difference, reducing inequality by more than half, but are unrealistic to implement without large-scale structural reforms and economic reforms. A reduction of the share of salaries in MoH expenditures would also create more opportunities for change. Figure 45. Estimated Distribution of Benefits: Simulations of Four Scenarios 2013 100% 90% 80% Cumulative benefits 70% 60% 45 degree 50% line Benchmark 40% 30% Scenario 1 20% 10% Scenario 2 0% 0 Lowest Second Third Fouth Richest quintile quintile quintile quintile Cumulative population by quintiles of consumption Source: Data from Household Survey 2010 and MFB/SIGFP. D4. Assessment of Gender Equality 103. Overall, there is no discrimination against girls and women in health care. Gender is another dimension of inequality that has been shown to have important implications for growth relates to the ability to provide care to girls and women. On average, females visit all levels of health facilities in greater numbers than men (Figure 46). The CHU result is highest but cannot be compared to the other facilities because it is based on a much smaller sample. 74 Madagascar PER – Health | Section D Figure 46. Utilization of Health Facilities by Gender, 2005 and 2010 percent female using the service-50 Above 0 = Female Advantage 2005 2010 8.0 9.1 9.2 12.2 5.0 4.5 3.7 3.0 CSB1 CSB2 CHD CHRR CHU Note: Values for 2005 CHRR and CHU use are not reported because the samples are too small to make meaningful inferences (<30 total visits). Source: Data from Household Surveys 2005 and 2010. 104. The higher overall proportion of females using health centers is explained by the larger proportion of female who were sick (7-8 percentage points more than males in both surveys). Out of those who were sick in the two weeks period to the surveys, the same proportion of males and females sought care, which excludes the reverse possibility of a male bias against seeking care (Table 18). Table 18. Gender Differences in Health-Seeking Behavior when Sick, 2005 and 2010 2005 2010 Percentages from weighted sample Female Male Female Male Sick in last two weeks 54.1% 45.9% 53.8% 46.2% Sick who did not seek care 60.1% 60.1% 67.6% 67.5% Sick who did not seek care for financial reasons 14.4% 11.0% 16.1% 14.1% Source: Household Surveys 2005 and 2010 (EPM). 105. Overall averages gender differences across consumption quintiles are not significant. Looking at primary health care facilities for which number of visits are sufficiently large by quintile, the only female disadvantage is in the second quintile for primary hospitals, but the difference is not large enough to be significant (Figure 47). 75 Madagascar PER – Health | Section D Figure 47. Gender Equality in Utilization of Primary Health facilities by Quintile, 2005 and 2010 1.00 0.80 0.60 0.40 CSB1 0.20 0.00 CSB2 -0.20 CHD -0.40 1 2 3 4 5 Consumption quintiles Source: Data from Household Surveys 2005 and 2010. 106. Given these results, and in particular the fact that there is no significant differences in female advantage by type of facility, an analysis of the distribution of benefits of MoH expenditure would not yield additional information than what is presented here. Key Findings Overall, the distribution of public spending is highly inequitable with per capita total health expenditures negatively correlated with poverty rates. This has significant implications for the overall health of the population especially in the current context of Madagascar where over 80% of the population is living in absolute poverty. Distribution of public spending  Looking at changes in the distribution over time, Antananarivo alone has been driving the increase in current expenditures. Expenditures in other provinces have been relatively stable.  Salary shares have remained stable over time but show some inequality with respect to population shares. Most receive less than their estimated population shares.  Expenditures are strongly negatively correlated to poverty ratios.  Whether restricting expenditures to primary health care facilities, primary health care personnel, or including all current expenditures, expenditures of the MoH are characterized by a strongly regressive regional distribution. Regions with lower poverty rates receive more per capita, while regions with higher poverty rates receive less.  With regards to non-wage expenditures, average benefits going to individuals in the richest quintile are two to four times higher than those going to individuals in the poorest two quintiles with the two poorest quintiles were found to benefit the least in most cases.  Likewise, benefits from MoH wage expenditures on personnel in health facilities (excluding all personnel in administrative units) are also clearly regressively distributed. The richest quintile benefits 3.6 times more than the poorest quintile, and at least twice as much as households in any other quintile.  Non-wage recurrent expenditure shares by type of residence have remained relatively stable with only 13 percent to semi-rural or peri-urban areas, and less than 5 percent to the rural communes. Considering that approximately two-thirds of the population live in rural areas, this represents a highly unequal distribution of expenditure shares. 76 Madagascar PER – Health | Section D Household spending and utilization by quintile  The scarcity of prepayment mechanisms in Madagascar, combined with a cost recovery system that makes public health care expensive for the poor.  Few households are subject to catastrophic expenditure in Madagascar, but prevalence increased in all quintiles between 2005 and 2010, and in the middle class in 2012.  Very low percentages of households with catastrophic expenditures in the poorest quintiles are usually indicative of “system failure”; i.e., the poor just do not seek care. Low utilization (rather than high cost) is therefore a plausible explanation for the low incidence of catastrophic expenditure.  Inequality becomes apparent at the district hospital level (CHD) and clearly increases moving into upper levels of care toward regional hospital level (CHU), which are clearly utilized by the rich. Consistent with the catastrophic expenditure analysis, the poorest households in the lowest two quintiles rarely use secondary and tertiary hospitals. Policy implications Given the strong inverse relationship between poverty and good health, the need for publicly provided health care is greater among the poorer populations. Better equity in the distribution of spending in the health sector will need to take into consideration better resource allocation and targeting of the population, improvement in access to health care especially in rural areas and reducing the financial burden on households. More equitable distribution of resources In the current context, Madagascar needs to urgently agree on and implement pro-poor strategies to ensure better equity of health expenditure and health services amongst the population. This includes:  Redistribution of health expenditure according to geographic distribution of the population to also benefit the poor  Better allocation of existing human resources to be more equitable to the poor. Consider a diagnostic of the current human resource system.  Additional spending on first line health facilities (CSB1s), which are utilized more by the poor and maintaining spending on second line health facilities (CSB2s), which benefits all quintiles approximately equally. In addition, consider reallocation of CSB2 so they are more accessible to households.  Reallocate regional hospital staff time to CSB1s and CSB2s. In this context, an updated census, poverty map and a Demographic Health Survey are needed to help inform decisions on resource distribution. Tailor interventions for the poorest quintiles  This analysis reiterates the need to implement interventions to improve health seeking behavior tailored specifically to the poor with mechanisms such as vouchers, conditional/non-conditional cash transfers, exemption schemes, non-monetary incentives (e.g. safe delivery kits) and support to the expanding the reach of health practitioners and health workers into the community.  Redesigning existing mechanisms to be more effective such as the Equity Fund and social health insurance schemes 77 REFERENCES Baldacci, E., Gupta. S. Clements, B., and Cui, Q. (2008), “Social Spending, Human Capital and Growth in Developing Countries,” World Development. Vol. 36(8), pp. 1317-1341 Efficiency of Public Spending in Developing Countries: An Efficiency Frontier Approach: Santiago Herrera & Gaobo Pang – Policy Research Working Paper Series 3645, The World Bank. http://siteresources.worldbank.org/INTQFA/Resources/EfficiencyofPublicSpendinginDevelopingCount ries_MAY05.pdf Gaudin, S. and Yazbeck (2013). Health Sector Policy Challenges in Low and Middle Income Countries: Learning from Public Expenditure Reviews. Background paper for the health and economy program. Mimeo Glick, P. and M. Razakamanantsoa (2002) ‘The Distribution of Social Services in Madagascar, 1993–99’, Working Paper No. 128, Cornell University Food and Nutrition Policy Program http://www.cfnpp.cornell.edu/images/wp128.pdf or http://www.instat.mg/pdf/iloinstat_7.pdf Glick, P. and M. Razakamanantsoa (2005) . The Distribution of Education and Health Services in Madagascar over the 1990s: Increasing Progressivity in an Era of Low Growth. Jl of African Economies, 15(3):pp. 399-433 (published version of the 2002 paper but with less specific information) Glick P., R. Saha , and S. D. Younger (2004). Integrating Gender into Benefit Incidence and Demand Analysis. Working Paper of the Food and Nutrition Policy Program, Cornell University. http://www.cfnpp.cornell.edu/images/wp167.pdf O’Donnell, O, E van Doorslaer, A Wagstaff, M Lindelöw, (2007) Analyzing Health Equity using Household Survey Data: a Guide to Techniques and their Implementation, World Development Institute, World Bank, Washington DC, 2007. www.worldbank.org/analyzinghealthequity (ISBN: 0- 8213-6933-4) Xu, K., D.E. Evans, K. Kawabate, R. Zeramdini, J. Klavus, and C.J.L. Murray (2003), “Household Catastrophic Health Expenditure: a multicountry analysis”, Lancet 362: 111-17 World Bank. (2006) Djibouti - Public Expenditure Review (PER) - making public finances work for growth and poverty reduction. World Bank Report #34624 Hernandez Patricia, Sigrid Dräger, David B. Evans, Tessa Tan-Torres Edejer and Mario R. Dal Poz (2006) Measuring expenditure for the health workforce: evidence and challenges. Background paper prepared for The world health report 2006 - working together for health. http://www.who.int/hrh/documents/measuring_expenditure.pdf Vujicic, Marco, Kelechi Ohiri, and Susan Sparkes (2009) Working in Health: Financing and Managing the Public Sector. The World Bank, Directions in Development, Human development series. http://go.worldbank.org/PU86PVIEU0 [PDF version on WHO website: http://www.who.int/workforcealliance/knowledge/publications/partner/workinginhealth_vujicic_worldb ank_2009.pdf) 78 Madagascar PER – Health |Annexes ANNEXES ANNEX 1. TABLE SUPPLEMENT Table A1. International Comparisons of Public Health Expenditures (including external funds) Country/Comparison Group Percent of GDP 2000-2012 2005-2008 2009-2012 Rwanda 4.18 4.69 6.04 Zambia 3.67 3.64 3.85 Mozambique 3.60 3.67 2.93 Tanzania 2.52 3.19 2.99 DRC 2.48 2.05 3.25 Madagascar 2.40 2.52 2.31 Mauritius 2.12 1.96 2.41 Comoros 1.98 2.65 1.77 Kenya 1.87 1.81 1.82 Sub-Saharan Africa (developing only) 2.66 2.72 3.02 Low income 2.31 2.42 2.67 Low & middle income 3.30 3.30 3.60 High income 5.26 5.21 5.73 % of General Government Expenditure 2000-2012 2005-2008 2009-2012 Rwanda 17.7 20.2 13.2 Zambia 14.7 14.9 7.5 Madagascar 13.0 12.3 14.5 Mozambique 12.9 14.2 6.2 Tanzania 11.3 13.7 9.8 Mauritius 9.4 8.8 8.9 DRC 9.1 9.3 23.0 Comoros 9.0 11.8 11.1 Kenya 7.5 7.1 16.5 Sub-Saharan Africa (developing only) 10.2 10.6 10.8 Low income 10.0 10.4 10.4 Low & middle income 10.7 10.7 11.1 High income 13.3 13.5 13.5 Data source: WHO (GHED) 79 Madagascar PER – Health |Annexes Table A2. Evolution of Budget Allocations by Administrative Levels Reported by MoH Reproduced from PDSS 2007-2101,MoH,2007. http://www.internationalhealthpartnership.net/fileadmin/uploads/ihp/Documents/Country_Pages/Madagas car/MadagascarPDSS_25mars_2007.pdf Table A3. List of Categories Used in the Alternative Classification of MoH Expenditures Classifications by level of care or specific program Classifications by type of activity General administration Curative Primary Health Care (PHC) Preventive Primary and Secondary Health Care Curative and preventive Secondary and Tertiary health care Pharmacy/laboratories Public Health Administration Maternal & Child Health (inc. reproduction/FP) Maintenance and logistics Health: non differentiated by level Training Buildings and equipment (PIP only) Additional classification by focus of activity Type of admin (general, decentralized, planning and evaluation, finance, Information technology, etc.) Procurement type activities Type of facility Type of disease (CD, NCD) Type of public health care activity Type of primary health care Other specific focus (vulnerable people, env. health, nutrition, health promotion, outreach, etc.) Transfers to hospitals identified separately Note: Salaries were included in a specific SOA in the administrative function and identified separately. Social protection and population type SOAs were also identified as such but are not included above. 80 Madagascar PER – Health |Annexes Table A4. Tracking of Specific Allocation of Interest in Non-Wage Recurrent Expenditure In Million of constant 2013 Ar. 2006 2008 2009 2010 2011 2012 2013 Health priority/specific program Communicable Diseases 2970 6600 349 2852 5490 3560 6099 MCH/Reproductive health - all 1745 1911 37 1017 4893 960 925 of which reproductive health 716 115 0 60 54 180 82 Neglected Tropical Diseases 0 0 0 2112 1074 315 641 Non-Communicable Diseases 102 316 113 105 99 150 94 Environmental health 0 0 35 33 30 18 10 Governance and quality Partnerships with private sector 139 311 209 1832 2921 1259 757 Statistics and data management 301 980 591 240 463 129 90 Quality control activities 78 77 43 38 36 75 42 Large items Transfers to hospitals 3055 7848 7016 5416 4873 4502 4114 Paramedical school 2520 2525 2198 3238 2909 2593 2120 Note: The choice of categories is determined and limited by what could be identified in the accounts rather than what we would like to measure relative to specific needs. In addition, the amounts obtained are not exhaustive of all expenditures in the specific program (except for the MCH program), because expenditures could be included in some other general administrative units and therefore not identified by specific target. Some expenditure were identified as directed to specific target groups (vulnerable people in particular) based on SOA created in 2012 but the amount were small. Data source: MFB/SIGFP. Table A5. PIP Expenditures in Specific Categories, 2008-2013 In Million 2013 Ar. 2008 2009 2010 2011 2012 2013 Communicable diseases 21052 107 7294 3910 775 24446 Non Communicable diseases 0 3491 3905 800 2642 8882 Neglected Tropical Diseases 0 0 0 1 82 77 Note: These amounts are limited to investment expenditure included in SIGFP. The SOA used in this classification did not exist in 2006. Data source: MFB/SIGFP. 81 Madagascar PER – Health |Annexes Table A6. Annual MoH Salaries by Province as a Ratio to SIGFP Remunerations by Province 2006 2008 2009 2010 2011 2012 2013 Total salaries in wage data/ total remunerations in SIGFP 1.11 1.35 n/a 0.86 0.89 0.91 0.89 CENTRAL 1.09 1.12 0.83 0.88 0.89 0.86 ANTANANARIVO 1.17 1.44 0.83 0.86 0.85 0.85 ANTSIRANANA 1.06 1.54 0.94 0.97 0.97 0.89 FIANARANTSOA 1.09 1.40 0.84 0.87 0.99 1.00 MAHAJANGA 0.96 1.56 0.86 0.85 0.87 0.89 TOAMASINA 1.21 1.32 0.86 0.95 0.96 0.94 TOLIARA 1.08 1.34 0.99 0.93 0.98 1.00 Total salaries in wage data/ total salaries and charges in SIGFP 0.99 1.19 n/a 0.75 0.79 0.85 0.80 CENTRAL 0.99 1.02 0.75 0.79 0.83 0.76 ANTANANARIVO 1.02 1.29 0.74 0.77 0.81 0.75 ANTSIRANANA 1.04 1.29 0.84 0.86 0.91 0.79 FIANARANTSOA 0.98 1.24 0.73 0.75 0.86 0.89 MAHAJANGA 0.84 1.38 0.70 0.75 0.81 0.78 TOAMASINA 1.07 1.05 0.75 0.83 0.87 0.88 TOLIARA* 0.97 1.22 0.84 0.93 0.98 1.00 *Note: no social charges were entered in the SIGFP for Toliara in 2011-2013. The allocation by province in the salary data are done using the region rather than the section code which had salaries of the same regions spread into several provinces; in the end the difference in method did not produce noticeable differences as the difference in allocations cancelled out. Data sources: MFB, SIGFF and Service de la Solde. 82 Madagascar PER – Health |Annexes ANNEX 2. FIGURE SUPPLEMENT Figure A.1. MoH Recurrent Expenditure Per Capita by Region vs Per Capita Consumption, 2006- 2010 14.0 1100 Thousand 2013 Ar 12.0 900 10.0 700 8.0 500 6.0 300 4.0 2.0 100 0.0 -100 MELAKY BONGOLAVA SAVA ANALANJIROFO ANDROY BOENI BETSIBOKA SOFIA ATSIMO-ANDREFANA ITASY IHOROMBE DIANA MENABE ANOSY ATSINANANA ANALAMANGA VAKINANKARATRA ALAOTRA-MANGORO HAUTE MATSIATRA ATSIMO-ATSINANANA AMORON'I MANIA VATOVAVY-FITOVINANY Average 2006-2008 Average 2009-2013 p.c. consumption, 2005 p.c. consumption, 2010 Data sources: MoF, INSTAT (population), EPM (poverty). 83 Madagascar PER – Health |Annexes Figure A.2. MoH Recurrent Expenditure Per Capita on PHC by Region vs Per Capita Consumption (2003) 9 700 8 600 7 Thousand of 2013 Ariary 500 Thousand 2013 Ar. 6 5 400 4 300 3 200 2 100 1 0 0 Current exp/c on basic health care services (based on type of facility) exp/c on primary care based on personnel type (salaries only) Consumption/capita (2010 EPM) Secondary axis Note: Per capita consumption in PPP from EPM 2010. Data sources: MFB, MINSANP, INSTAT, EPM. 84