Preparing PERs for Human Development

Specific Guidelines for Health

1 2

User’s Guide

CHECKLISTS

The guidance note given here is not to be taken as a minimum list where the authors must tick every sub-box. Every PER must be selective in what it covers, with the selection of topics based on many factors—what is needed to underpin the country dialogue, what is already known and packaged elsewhere, what is manageable to do given constraints on time, data, and funding, etc. This guidance note is meant to remind the analyst of the main features that might normally be included in the social protection chapter of a PER. Omissions will often be made, but with some justification in mind. In addition to agreeing in the concept note on the planned coverage of topics, it may be useful to convey to the reader of the full report the reasons for omissions of major themes. Similarly the depth of treatment and number of programs covered in depth will need to be considered, agreed, and explained.

Note also that the guidance note is organized as a checklist rather than an outline or table of contents. While a report might be organized along these lines, there are many other outlines that could be effective. One option might be to work around the core PER questions of: Where does the money come from? Where does the money go? What does it buy? How could spending be improved? Another outline might be to present first the situation with all basic analysis, followed in a second section by a discussion of issues, and in a third section by options for reform.

NOTES

In many places along the right hand margin of the checklist the symbol Note X appears. The text of all the notes follows the checklist itself. Some are short text that explain further what is meant in the checklist. Often the notes contain references to methodological material or to sources from which international comparators may be drawn.

EXAMPLES

In many places along the right hand margin of the checklist the symbol Example X appears. The text of the examples follows the text of the notes, each starts on a new page, with running headers to help the reader navigate. The examples are excerpts of a page, table or series of pages meant to show at least one interesting case of application of the themes contained in the checklist. The examples are largely drawn from the universe of PERs done in FY00 to FY02, significantly supplemented by the body of social safety net assessments commissioned by the LAC region, and other sector work that I happened to be familiar with or was referred to me by others as part of the peer review process. The use of examples from other sector work is necessary because of the limited treatment of 3 social protection in PERs (Office of the Chief Economist, 2003). It also allows a somewhat better (though still not wholly balanced) coverage of regionsof regions.

In addition to using the varied examples that form some sort of composite “model” PER chapter, it may be useful to the task team to look at a few actual PERs, though of course no single report is exemplary in all ways. The 2002 Bulgaria and 2003 Armenia reports provide very solid overall treatment of SP in cases with a relatively generous budget, time frame and prior policy dialogue and knowledge base. The 2003 Ethiopia report is a nice example of what can be done in a very much more constrained environment.

BIBLIOGRAPHY

Short references are given in the individual notes and examples. Full references are contained in the unified bibliography. For the majority of documents contained in the bibliography, materials are available via the World Bank website and the URLs are given. 4

SPECIFIC GUIDELINES FOR HEALTH

This document provides guidance for preparing a PER for the health sector. The PER may be prepared specifically for health, or it may cover the social sectors more generally, or it may be a specific chapter in a larger PER. The guidelines that follow this introduction are divided into sections dealing with specific topics—these might correspond to subsections in the health chapter of a larger PER, or to individual chapters in a background paper on the health system. For general perspective and orientation, however, some points are worth noting at the outset.

The guidelines here are comprehensive. They assume an ideal situation where the team can deal with all the issues and have all the data and resources required—including time —for the task. This is, of course, unlikely in the majority of situations. This begs the question: What should be the “core” topics, required of any PER? But this is the wrong question to ask. The scope and depth of the analysis should be driven not by some mandatory requirements, but by the issues and characteristics important in each case.

The PER might confine itself to issues of particular interest, or of special concern in the country. Or as is frequently the case, data and resource limitations, possibly compounded by institutional and political constraints, may limit the scope of the work. It is important therefore to acknowledge at the outset the scope of the report and the issues addressed. These choices, and the limitations implied, must be justified explicitly in the report.

The PER is about finance. But finance is not an end in itself. It is a means to enable and facilitate the provision of health care in an equitable and efficient manner. This is what distinguishes a PER from National Health Accounts. Note 1

Financing for health care comprises multiple sources of funding, methods of allocating funds, and modes of paying providers. Different sources of funding health care affect the economy and society differently. And private financing and provision are often more important than public. So, to ignore the private sector would severely handicap the understanding of the sector. Although the PER should focus on public expenditure, it is important to examine the interactions between the public and private sectors, especially since one function of public expenditure is to encourage private action to improve health.

The multiplicity of sources of funding care, including donor contributions, raises the question of sustainability—the availability of funding over the intermediate and long terms. This issue is particularly acute when certain types of care depend on funding (such as from donors) that might not be sustained.

The PER is primarily concerned with public financing and public provision of care. Neither should be taken for granted. Financing and provision are either highly regulated or outright managed by the state. But the rationale for state involvement in financing and provision must be explicitly justified, and its consequences explicitly acknowledged. Public involvement in finance and provision is often defended by reference to theoretical 5 abstractions such as scale economies and market imperfections. These conditions may hold, but they must be supported by evidence rather than merely assumed.

The PER must address the following broad questions:

 How are health spending and services distributed among the population?  How efficient is the provision of health care and other related interventions?  What are the macroeconomic consequences of revenues and expenditures?  What is the impact on equity of sources and levels of revenues?

And now, the guidelines. 6

GUIDELINES FOR PREPARATION

1. INTRODUCTION

 Objectives of the PER  To examine the flow of funds within the public sector and with a public policy focus.  To examine the performance of the system in ensuring and financing the provision of care and improving welfare.  To examine some specific aspect of sector performance. Note 2 Example 1  Objectives of the health system: goals of the government and the ministry  In general and for specific target groups (such as children or expectant mothers).  For preventive and curative care.  Equity.  Cost containment.  The relevance of these goals, given resources and risks. Note 3 Examples 2 Example 3  National context  Income, income distribution, and poverty.  Growth and the macroeconomic environment.  Associated economic policies (HIPC, PRSP).

2. HEALTH STATUS AND HEALTH RISKS

 Health status indicators  Over time. Example 4  In comparison with other countries.  Incidence among population—by region, ethnic group, and so on. Note 4 Note 5 Examples 4 Example 5 Example 6  Health risks  Demographic, environmental, behavioral.  Communicable diseases such as HIV/AIDS. 7

 Non-communicable diseases such as hypertension and heart disease. Example 7  Emerging concerns—what factors will become more important in the short and medium term.

3. PUBLIC RESOURCE MOBILIZATION AND SOURCES OF FINANCE

 Public funds  In total.  By ministry and level of government.  Public funds—sources  General and earmarked revenues.  Specific mandatory and voluntary contributions.  Debt (including arrears).  Donor contributions – bilateral, multilateral, NGO. o general budget support o specific (such as project) expenditure o sustainability of donor funding  Off-budget funds  Domestic (such as national health insurance funds).  Foreign—debt, donors and NGOs.  Public revenues from private sources Example 8  Out-of-pocket. o fee-for-service o informal payments for public services  Formal co-payments and insurance premiums (such as social insurance).  Sickness funds.  Fiscal and macroeconomic impact of different sources of finance  Taxes.  Debt.  Donors.  Equity and welfare impact of different sources of public finance  Taxes.  Debt.  Donors.  Out-of-pocket. Note 6 8

Examples 9 Example 8 Example 10

4. HEALTH EXPENDITURE Note 7

 Total expenditure on health services  Public expenditure. Example 11 o in total o by ministry/agency and level of government, including local o on and off-budget expenditure o direct expenditure by donors (not through budget) o arrears  Private expenditure. o employers o households (including out-of-pocket, catastrophic)  Other.  Functional allocation of expenditures  By level and type of care. o primary o secondary o tertiary and higher o public health interventions (e.g. aerial spraying, information and education campaigns)  By output, cost center or broad DRG. Example 12  Economic allocation Example 13  Capital (investment) expenditures.  Recurrent expenditures.  Direct allocations to consumers  Vouchers.  Tax credits to employers and consumers.  Inventory of resources – manpower, inpatient beds, and the like, if available  Distribution—by region, level of service, and so on.  Public sector. o ministry of health o other ministries/agencies o district and local governments  Private sector.  Incidence of expenditure and resource allocation (such as doctors per capita), across regions, income classes, and the like. Example 14 9

Example 15  Public expenditure.  Private expenditure. o incidence of ‘catastrophic’ expenses o household use of savings, transfers, etc.  Measures of distribution and incidence. Notes 7 Note 8 Note 9 Example 13

5. HEALTH SYSTEM COVERAGE AND OUTPUTS Note 10

 Volume and mix of services delivered  Type of care (immunizations, supervised deliveries, ante-natal care, outpatient visits, inpatient days).  By whom (central ministry, districts, other agencies, private sector, donors). Example 16 Example 17  Distribution of services—by region, urban/rural, income class, ethnic group, sex.

Examples 18 Example 19 Example 20

 Who is covered by public services (number and characteristics of beneficiaries)  Insurance (general population).  Social security (define eligible groups).  Civil service coverage (nature and extent of programs).  Who is covered by private services (number and characteristics of beneficiaries)  Formal, employment-based insurance.  Other private formal insurance.  Informal insurance (such as cooperatives).  Medical savings plans.  Community financing.  Pre-payment schemes and sickness funds.  Incidence and equity of coverage Note 11 Examples 21 Example 22 Example 23 10

 Moral hazard—differences in behavior of those covered and not covered Example 18

6. HEALTH SECTOR POLICY, BUDGETING AND MANAGEMENT

 Budget management Example 33  What is the budget process?  How are levels of expenditure determined?  How is the allocation of expenditures determined?  How are the sources of revenue determined?  How are fees established?  Cost containment policies and practices.  Payment mechanisms to facilities  Direct budget allocation.  Fee-for-service.  Capitation.  DRGs.  Other.  Payment mechanisms to individual providers  Direct payment of wages.  Fee-for-service.  Capitation.  DRGs.  Other.  Institutions and stakeholders  Who are they? o government (central/local) o provider associations o consumers  What role do they play and how much influence do they have in policy formation?  Governance Example 13 Example 14  Indicators of management effectiveness. o inefficiency, incentives, and principal-agent problems o mismanagement and corruption (informal payments, missing expenditure, absenteeism)  Contracting for service provision  Public contracts for private providers.  Public contracts for public providers.  Terms and conditions of contracts (such as performance-related payments).  Training and medical education 11

 Who pays?  Who provides training? (Ministry of Health, Ministry of Education, other)  Number and specialization of students.  Public research and development investment Note 12

7. RECOMMENDATIONS

 Summarize the evidence Example 34  What are the goals of the health system?  What are the major obstacles to meeting these goals? o supply-side constraints (insufficient supplies, poor management, etc.) o demand-side constraints (poverty, high prices, travel times, etc.)  Recommendations for changes to improve policy, programs, and actions Example 35  Assess short-run and long-run costs, and other implications of recommendations (possible unintended consequences)  Establish priorities  Suggest indicators for monitoring  Counterfactual – what is likely to happen in the absence of reforms? Note 13 12

NOTES

Note 1. PERs and National Health Accounts PERs and NHAs are highly complementary. NHAs provide the data for analysis, and they can be an essential component of a PER. The purpose of NHA is the “systematic, comprehensive, and consistent monitoring of the resource flows in a country’s health system” (Poullier and others 2002). This includes all flows of resources within the health system: resource mobilization, pooling, allocation, and distribution.

NHAs are primarily an accounting tool – a sequence of identities – in which total revenue is required to equal total expenditure, which equals consumption plus investment. These accounting identities are usually presented as a series of tables that include and classify all agents and transactions within the sector, and which can be used to trace the flows of resources from source to final use. This is analogous to general national income accounting, or more precisely to input-output matrices. NHA tables are often sufficiently disaggregated to provide data for benefit incidence analysis; and they may include inventories and outputs, as well as revenue and expenditure. As part of an effort to expand and codify NHA, the OECD and WHO have devised a system of classification for health-related activities (WHO 2003), which should be used if the data permit. These define categories of services provided, financing sources, and providers.

Note 2. Goals of the PER The PER in health can follow three complementary tracks: (1) it can study the flow of funds, (2) it can focus on the institutions that make and execute public budgets, or (3) it can examine sector performance.

 The first track resembles a National Health Accounts exercise. But even this should include outputs—that is, what you get for the money – as well as the flow of funds.  The second track is now commonly referred to as a Public Expenditure and Institutional Review. This focuses more attention on the process of policy formation and implementation: that is, the stakeholders and the institutions that determine how the money is raised and spent. But it should still describe the flow of funds.  The third track is potentially broader and more ambitious, perhaps beyond the scope of the PER. A 1998 review of the impact of PERs argues that overly comprehensive reports take longer, are more expensive, and may be “less persuasive and less useful to groups interested in specific expenditure problems” (World Bank 1998:10).

The comprehensive checklist presented here does not imply that such a broad review is possible or necessary in all cases. In planning the PER, the team should review previous research, policy documents, and existing data to determine what aspects of the sector 13 deserve a more thorough examination. These guidelines should be used to help select the issues appropriate to the national context.

The 1998 review also reported that “the objectives and design of PERs have not been clearly defined to meet the needs of any well defined set of end users. …If it is not designed to meet the needs of specific users, a PER is not likely to have an impact on the behavior of client governments or external participants” (World Bank 1998:5).

Note 3. Goals of the health system The health care system is attempting to achieve many diverse and often contradictory goals: better health status, social equity, cost containment, efficiency of service delivery, and consumer satisfaction with service and care. Clearly, better health is the key goal, especially in developing economies. But the other goals are important for their overall social and economic implications (also on health, beyond medical care). The relative importance of these goals varies across countries and over time within one country.

Note 4. Health status and health risks This section should be brief, no more than a few figures or tables and the related discussion. Its purpose is to describe in summary the health status of the population—to provide a picture of the system’s performance, in terms of outcomes—over time and in relationship to other “comparable” systems. This section should portray the system’s performance along multiple dimensions, though it may be necessary or desirable to restrict the discussion to a few key characteristics. Trends and comparisons should highlight the system’s potential and priorities relative to other cases (but see note 5).

People and health systems differ in their environments, behavioral factors, and risks. An understanding of these characteristics is necessary to establish the priorities of the system and the modes of intervention. This section should provide an overview of risk factors or causes of morbidity and mortality. More important—it should provide evidence of issues that might become important in the near future—such as HIV/AIDS—that will have implications for the fiscal and budget management of the sector.

Note 5. Comparators Nations are commonly compared with other nations of the same level of GDP per capita. But other comparisons should be considered. For example, a nation can be compared with its geographical neighbors, with nations that share its political history (former Soviet Union), with political or economic allies (EU or OECD), with countries that represent ideals in the sector, or with those that share other characteristics (small island states).

Note that the choice of comparison countries is arbitrary. Much of what makes countries different, invisible to the casual observer, can be elicited only with more careful analysis. Simple comparisons with other “similar” countries are perfunctory and superficial, not explanatory. Casual comparisons cannot tell you how a country “should” or even “could” 14 be performing. Some apparently obvious comparator countries are too similar to be of much use (Belarus and Ukraine): they share the same constraints and problems, and the performance of one should not be used to judge the potential of the other. Conversely, some countries are too different (Botswana and Malawi): it is difficult to imagine that one would be like the other if only the policies were changed.

Comparisons are useful, however, to situate the country in context, to illustrate potential problems (relative to other countries), and to see whether the country is outside the range of other, similar countries. The figure gives an example of comparisons of user fees in Africa. The figure shows the enormous variety of country experience and policy, even among countries one might expect to exhibit similar characteristics. This comparison should not be used to drive policy, but to motivate further analysis. For instance, what explains the difference in user fees between Benin and Mali, or Lesotho and Swaziland?

Figure. Share of MoH budget from user fees, sub-Saharan Africa

0.20

0.15 e r

a 0.10 h S

0.05

0.00

Country

Source: World Bank, WDI.

Note 6. Public resource mobilization and sources of finance Examine all resources and revenues accruing to the public sector, including private expenditure for public services (out-of-pocket and insurance contributions). This section does not include private expenditure on private services, which are dealt with in section 5. Nor does it include allocation, expenditure, or the services purchased by the public sector. It should focus only on public funds.

The section covers both the levels and sources of funding available to the health system. Funding for health care may come from general or earmarked national government revenues, local public sources, private sources, and external sources. Different sources of revenue have different consequences for system performance, equity, and for the economy. Ideally, this would take advantage of National Health Accounts, if available. 15

Public funds may include mandatory contributions into social health insurance schemes. Often considered “off-budget,” these are not channeled through central or local state budgets. Private finance comprises funds available through voluntary insurance schemes as well as out-of-pocket expenses in the form of direct payments to providers. These may include co-insurance and co-payments in conjunction with private insurance as well as public finance. Also include discussions of deficits and arrears in funding, of particular concern in transition countries.

Deal with donor financing, which can be quite large in low-income countries—equal to or greater than domestic public finance—and is not always included in government budgets. Allocation of donor funds may or may not be coordinated with government priorities or decided in consultation with the national government. There may also be activities that are funded primarily by donors, especially if donors focus on project lending rather than general budget support. Donor involvement can have two related problems. First, the sustainability and continuity of expenditure are not under the domestic government’s control, complicating policymaking. Second, donor activities can alter the incentives facing governments and private actors. They may “crowd out” domestic priorities and activities, or crowd them in, creating potential problems in either case.

Different sources of revenue have different consequences for the macroeconomy: for example, donor funding may have Dutch-disease effects, or expenditures may be financed by domestic debt (usually arrears), or foreign debt. The consequences for equity may also differ, and this should be reflected in benefit-incidence analysis (see note 8).

Note 7. Health expenditure This section is the heart of the PER: total expenditure for health care and the inputs used to provide care, including manpower, facilities, and high-technology equipment. Although the PER should focus on public expenditure, this section should include private expenditure (for both public and private services) as well, especially since the latter is the majority of total expenditure in many countries. As with section 3, this could rely on a system of National Health Accounts.

Discuss both recurrent expenditure and investment in human and physical capital. How efficiently and equitably does the system uses available resources, and how well does it prepare for the future? The first involves the allocation of expenditure to regions, programs, and populations. The second involves funding of investments. Do not deal with management and institutional issues, covered separately in section 7.

Discuss public expenditures at all levels of government (where possible) and by all agencies that either provide or purchase health services. This includes line ministries other than the Ministry of Health (such as the military), and expenditures by regional and local governments. Local public health expenditures are not likely to be included in the budget of the central Health Ministry. Data on local expenditures may be difficult to obtain, but these activities should be acknowledged in the report. 16

Also include private expenditures, as well as expenditures of NGOs and donor agencies. In many countries, expenditures by donors and NGOs can exceed that of the domestic public sector. This has serious implications for budget discipline, as well as possible adverse consequences in terms of crowding out and “Dutch disease” shifts in incentives.

Note 8. Benefit incidence and distribution The benefit incidence may be computed for any characteristic that is distributed among the population, including (but not limited to) the allocation or consumption of health services. The analysis may be presented by income class, ethnic group, region, age, and so on, as appropriate. There are a number of different methods for computing the incidence of benefit or burden.

The most common method is to examine the benefit incidence directly. This involves classifying individuals (or households) according to a comparable welfare measure, such as total per capita expenditure, and then examining the benefit received from public services for each individual or class along the welfare rank. Examples are presented in the figures. The first shows the distribution of infant mortality among Brazilian households, ranked by income.

Figure Under-two mortality rate by income decile- Brazil, 1996 Source: World Bank, 2001. Public Expenditures for Poverty Alleviation in Northeast Brazil: Promoting

12

10

8

6

4

2

0 P oorest 2 3 4 5 6 7 8 9 Richest

Growth and Improving Services. Report No. 22425-BR

The second shows the distribution of public health expenditure to the richest and poorest quintiles in a sample of countries. As in the discussion of comparators (Note 5 supra), this figure provokes a number of questions. Why is public health expenditure targeted towards the poor in Costa Rica, but not in Ecuador? 17

Figure Percent of total public health expenditure accruing to richest and poorest quintiles

60

50

40 t n

e poorest 20%

c 30 r richest 20% e p 20

10

0 ) ) ) ) ) ) ) ) ) ) ) ) ) ) ) ) 5 0 3 3 3 0 d 5 2 9 4 3 8 4 2 6 / 9 0 9 9 9 9 n 9 9 9 9 9 9 9 9 ( ( ( ( ( ( ( ( ( ( ( ( 5 ( ( (

) r 9 r e h l a a a a a a a a a ( r i i i i i a m

o s i a c c r e n s i n n i c r n a a d e o r a i n a s e a a f u e n i R v a d g r t d I h z a n l ' u ( A u a m m e n a l n g o u i d r G I t c o G

a h g a a d s t B V A E y e n R d n T t o u I n a a o o C e B M C S K country (year)

Alternatively, one can calculate the concentration index (CI), a summary measure describing benefit incidence. The advantage of the CI is that one can also easily compute standard errors, which permit robust comparisons of statistically significant differences across classes of individuals or households. Excellent instructions on how to compute the CI, and benefit incidence analysis in general, are presented in Technical Notes 7 and 12, respectively, at http://www.worldbank.org/poverty/health/wbact/health_eq.htm.

Note 9. Net benefit incidence Analysis of the equity of health care must account for the distribution and incidence of expenditures on health services, as well as the value of services consumed. Evaluation of the incidence of health care finance requires examination of all sources of health sector funding, not only the direct payments that are made exclusively for health care. In addition to out-of-pocket payments, health insurance contributions, and earmarked health taxes, the distributional burden of all direct and indirect taxes can be examined where these make up a significant part of total health sector revenue. They can be examined in the same way as benefits (see Technical Note 16 at http://www.worldbank.org/poverty/health/wbact/health_eq.htm).

Expenditures paid and benefits received may be combined in an effort to understand net benefits incidence. In health the net benefit is the value of the good or service (transfer) after accounting for user fees and other payments. The figure shows that the net incidence of public health expenditure is more progressive than the gross incidence. Once the burden of tax payments is accounted for, the poorest receive about 16% of health expenditures, while the wealthiest actually pay more in taxes than they receive in benefits. 18

Figure Distribution of net benefits, total public health expenditure, Mongolia

Source: Taken from Filmer, Hammer, and Pritchett, 1997, Table 9. e

r 0.25 u t

i Gross benefit Taxes d n

e 0.20

p Net benefit x e Gross h 0.15

h Taxes benefit

taxes a t

i net benefit p

a 0.10 c

r e p

0.05 f o

e r

a 0.00 h

S Net benefit -0.05 Poorest 2nd 3rd 4th Richest quintile Note 10. Health system outputs Here you answer the questions, “What do you get for the money?” What goods and services does the health system produce? This should be an inventory of the volume of services provided, and the type of care: such as the number of child immunizations, and the number of outpatient visits. The level of disaggregation (clinical, administrative, and geographical) should be determined by the country context. For instance, certain diseases and related interventions (malaria) may be important in only some regions of a country.

As for expenditures, take care to include services provided by agencies outside the central Health Ministry. In many countries the military is a significant provider of health services to soldiers and their dependents. And many countries have decentralized provision, so that health services are under the control of local departments or districts. These will often be under the supervision of (and in the budget of) the Ministry of Local Government rather than the Health Ministry.

Again, even though the focus should remain on outputs of the domestic public sector, other actors provide significant volumes of services, which have an impact on the health of the population and their demand for public services. The private sector provides the majority of services in many countries. NGOs and donor agencies often directly provide services. This is increasingly the case for HIV/AIDS interventions, such as anti-retroviral therapy (ART), which are considered too expensive for national governments in client countries.

Note 11. Resource-pooling and risk-pooling Briefly describe the incidence of public and private insurance and how risks are spread among communities. This section differs from those before and after in that the unit of analysis is people rather than money. How do individuals gain access to care? What is the structure of resource-pooling and risk-pooling? Who has access to public insurance, and 19 who to private insurance? Are some services or types of coverage restricted to specific individuals or groups? Coverage and access to health care can be obtained through private purchases in the marketplace, as an entitlement, or as most common, a combination of the two.

Note 12. Health sector policy, budgeting, and management The public budgeting process, including raising public funds, is the basic mechanism to facilitate the functions listed in the previous section. It involves the pertinent decision- making structure (decentralization, hierarchy, and so on). This process is guided by and involves legislation and regulation that also guide contracts and public-private relationships. Identify stakeholders who play a role in determining and implementing policy—and who will be directly affected by any changes.

Describe the regulatory frameworks, the budgetary process, and the political economy guiding the different elements of finance as listed in the section outline. These are also manifest in contracting arrangements involving public (and possibly private) funds. Changes that may be ultimately called for to improve the performance of the system will involve changes in the budgeting process and related decisionmaking processes and regulations.

Note 13. Recommendations A common failing of PERs and other policy papers is that they often do not derive recommendations not from evidence but from some presupposed or idealized model of health care policy. These a priori recommendations include the idea that a country “should” spend X percent of its GDP on health and “should” provide more to primary care and less to tertiary care. These recommendations may actually apply, but they must be based on a solid analytical foundation. This will also ensure that the recommendations are specific and relevant, not general.

Another problem of PERs is that they provide an impossibly large set of recommendations for policy. The review of PER impact referred to earlier (World Bank 1998) reported that PERs have on average 100 recommendations. Limit yourself to a small set of suggestions and focus on those that will be most realistically implemented. You must also provide some guidance for the implementation of these suggestions. This will ensure that the recommendations remain in the realm of the possible, rather than the improbable ideal. You must, in addition, suggest indicators for monitoring the success of the policies recommended. If the recommendations are adopted, how can the government, the Bank, and civil society track the progress of implementation?

Finally, the PER must also provide some discussion of the assumptions underlying the recommendations. What circumstances or other factors must prevail in order for the recommendations to succeed? More important, what is the counterfactual? That is, what might happen if the recommendations are not implemented successfully, or if they are poorly or partly implemented? This requires acknowledging the risks involved in both the analysis and the policy process. 20 21

DATA SOURCES

 Official data  National accounts  National Health Accounts  Government budget documents o Central government consolidated accounts o Line ministries (MoH, Social Security, etc.) o State or provincial governments, if separate from consolidated government accounts  Annual reports from line ministries  Tax reports  Trade statistics  Census data and reports  WB and IMF data and documents  Country economic memoranda  Medium-term economic frameworks  PRSP / PRGF documents  HIPC assessments  Poverty assessments  Previous project / loan documents  Country-specific survey data  Public Expenditure Tracking Surveys (PETS)  Quantitative Service Delivery Surveys (QSDS)  Household surveys (e.g. LSMS)  Demographic and Health Survey (DHS) data and equivalents  Marketing / social marketing surveys  Firm (provider) / sector surveys  Other country-specific sources  Insurance industry reports and records  Providers’ (e.g. physicians’) association reports  Provider records  Other international sources  Reports from donor countries and agencies  International agencies (OECD, WHO, PAHO, etc.)  Academic studies  NGO reports 22

V. EXAMPLES

Example 1. Objectives of the health sector PER. .Error! Bookmark not defined.

Example 2. Objectives of the health system (1)...Error! Bookmark not defined. Example 3. Objectives of the health system (2)...Error! Bookmark not defined. Example 4. Health outcomes in historic and regional perspective...... Error! Bookmark not defined. Example 5. Geographic distribution of health outcomes...Error! Bookmark not defined. Example 6. Equity of health outcomes...... Error! Bookmark not defined. Example 7. Risk factors...... Error! Bookmark not defined. Example 8. Private payments to providers...... Error! Bookmark not defined. Example 9. Impact of different sources of finance on equity...Error! Bookmark not defined. Example 10. Equity implications of out-of-pocket expenditures...... Error! Bookmark not defined. Example 11. Assessment of Gaps in Expenditures and Allocation...... Error! Bookmark not defined. Example 12. Classification of expenditures by program……...Error! Bookmark not defined.

Example 13. Economic classification of expenditures (2). Error! Bookmark not defined. Example 14. Geographic distribution of health services....Error! Bookmark not defined. Example 15. Geographic distribution of resource allocation....Error! Bookmark not defined. Example 16. Organization of resources, by level of service.....Error! Bookmark not defined. Example 17. Health services delivered by other agencies..Error! Bookmark not defined. Example 18. Equity of health services (1)...... Error! Bookmark not defined. Example 19. Equity of health service (2)...... Error! Bookmark not defined. Example 20. Equity of health services (3)...... Error! Bookmark not defined. Example 21. Insurance and entitlement to care (1)...... Error! Bookmark not defined. Example 22. Insurance and entitlement to care (2)...... Error! Bookmark not defined. Example 23. Private insurance and equity...... Error! Bookmark not defined. Example 24. Entitlement, insurance and moral hazard...... Error! Bookmark not defined. Example 25. The budget process...... Error! Bookmark not defined. Example 26. Efficiency of service provision (2). .Error! Bookmark not defined. 23

Example 27. Efficiency and impact of health services...... Error! Bookmark not defined. Example 28. Recommendations, priorities and evidence...Error! Bookmark not defined. Example 29. Recommendations for reform...... Error! Bookmark not defined. Other references: Error! Bookmark not defined. Other suggested PERs...... Error! Bookmark not defined. Institutions, governance, and the general budget process...... Error! Bookmark not defined. Expenditures in the context of PRSP. Error! Bookmark not defined. Decentralization and state-level analysis...... Error! Bookmark not defined. About Public Expenditures...... Error! Bookmark not defined. National Health Accounts...... Error! Bookmark not defined. 24

Example 1. Objectives of the health sector PER

Turkey Public Expenditure and Institutional Review: Reforming Budgetary Institutions for Effective Government. Report No. 22530-TU, August 20, 2001

4. PUBLIC EXPENDITURE ON HEALTH

1. PRINCIPLES AND PERSPECTIVES

4.1 This chapter and the following chapter on education undertake a dual task: a conventional analysis of the efficiency and equity effects of public expenditure in these sectors as well as an assessment of how "policy" is determined in these sectors and whether it is meaningful in determining budget allocations. While the bulk of this chapter is devoted to the former assignment, the chapter does provide specific glimpses into the formulation and implementation of policies which is the concern of the larger report and which is further discussed in chapter 6.

4.2 Public expenditures on health should eventually be justified either on the grounds that they contribute to a more effective and efficient functioning of the health system and/or contribute to poverty alleviation and redistribution of income.1

4.3 The efficiency justification is often based on public and merit goods arguments. Common examples of public goods in health are vector2 control, sanitation, mass education, programs to provide clean water, etc. According to economic theory, private production will not be forthcoming for such goods and services, and it is socially optimal for governments to finance and possibly provide these goods. Similarly, many health goods are merit goods, and create large positive spillover effects (e.g., vaccination), produce greater social benefit than private benefit (e.g., family planning), and/or possess significant interpersonal utility values (e.g., medical services for the vulnerable segments of the population, emergency services for trauma patients, etc.). These and other market failures in health care and health financing (like those resulting from incomplete information, incomplete capital markets, etc.) mean that government intervention can raise welfare by improving the way those markets function.

4.4 A government role in health can also be justified on equity grounds. Since the poor cannot always afford health care, the government can play an important role in promoting equitable access to health care that would improve their productivity and well-being.3

4.5 While governments worldwide have played an important role in achieving improved health status of their citizens, there is nevertheless widespread evidence of misallocation of public resources, inefficiency of government expenditures, and the general inability of governments to address equity issues effectively.4 Public funds are often spent on health interventions with low relative cost-effectiveness, such as surgery for most cancers, while

1 See Hammer, "Economic Analysis for Health Projects", The World Bank Research Observer (1997). 2 Vectors are the carriers of disease such as mosquitoes for malaria. 3 Hsiao, "Abnormal Economics in the Health Sector", Health Policy 32 (1995): 125-139. 4 Nandakumar, Chawla and Khan: "Utilization of Outpatient Care in Egypt and its Implications for the Role of Government in Health Care Provision", World Development, January, 2000. 25 highly cost-effective and often most critical interventions, such as treatment of tuberculosis, remain under-funded. Government expenditures tend to be inefficient and wasteful: health workers are badly deployed and supervised, hospital beds are under- utilized, expensive equipment is purchased but poorly maintained, and much of the money spent on drugs is on brand-name pharmaceuticals when generic drugs could inexpensively substitute.5 Studies in many countries show that government spending for health often goes disproportionately to the affluent, in the form of subsidies to sophisticated public tertiary care hospitals, while the poor lack access to basic health services and receive low-quality care.6

4.6 It is against this broad conceptual backdrop that this chapter examines public expenditures in the health sector in Turkey. The chain of reasoning starts with an examination of government policies toward health, and continues through the translation of policies into budgetary allocations and expenditures of public funds, in the process evaluating the efficiency of allocations and impact on health outcomes.7 The following questions motivate the chapter:

 What are the government priorities in the health sector, either in terms of stated or implicitly established policy?  How do budgetary allocations and public expenditures in health support the strategic priorities in the sector?  How efficient and effective is the use of public funds in achieving the desired outcomes?

Example 2. Objectives of the health system (1)

Bangladesh Public Expenditure Review. Report No. 24370-BD, May 25, 2003

185. Despite this relatively strong record, there are still numerous gaps in health outcomes. Bangladesh is not likely to achieve some Millennium Development Goals. At 392 per 100,000, its maternal mortality ratio is among the highest outside of Sub-Saharan Africa, mainly because of the low level of clinic deliveries (95 percent of deliveries take place at home) and insufficient number of trained midwives. And despite increasing use of contraceptives, the total fertility rate has been stagnating at around 3.2 births per woman for the last five to six years. This is due in part to insufficient diversification of available modem contraceptives and their irregular use.

186. Another Millennium Development Goal- reduction m child malnutrition-remains a challenge. A first generation of strategic goals in reducing malnutrition has been achieved in Bangladesh. This includes the generation of a high level of policy support,

5 Berman, Peter: "Rethinking Health Care Systems: Private Health Care Provision in India" World Development, 1998. 6 Mahal, "Public Expenditures on health in India", Draft manuscript, World Bank mimeo (2000). 7 Filmer and Pritchett find a weak impact of public spending on health status in "The Impact of Public Spending on Health: Does Money Matter?" Social Science & Medicine, 1999, though other studies, like Anand and Ravallion, "Human Development in Poor Countries: On the Role of Private Incomes and Public Services", Journal of Economic Perspectives (1993) report a significant impact of public spending. 26 development of a programmatic approach, and evolution of conducive structures in the health and food policy arena. While future economic growth will provide strong impetus for continued reductions in malnutrition, it is most likely that achievement of this Millennium Development Goal will require additional interventions. For example, a recent study of poverty and child malnutrition in Bangladesh shows that there are a host of factors that need to be addressed to improve child malnutrition (Deolalikar 2002). Using data from both the Child Nutrition Survey (CNS) 2000 and the 1999-2000 Demographic and Health Survey (DHS), the study shows that delayed and early terminated breastfeeding, low family food intake, low mother's education, lack of access to health facilities, low village electrification, high food prices, and lack of access to safe water and sanitation are associated with worse nutrition outcomes for children.8 Thus efforts to improve child health outcomes should be supported through cross-sectoral interventions that include improvements in access to clean water, currently threatened by such contaminants as arsenic and pathogens. Future interventions should also build on the success of the Bangladesh Integrated Nutrition Program to expand the coverage of nutrition programs (Box 6.2).

Box 6.2 Bangladesh Integrated Nutritional Program

The Bangladesh Integrated Nutritional Program (BINP) is a large, coordinated effort launched by the Government in 1995 to address the high prevalence of child malnutrition in the country. Under the community-based nutrition component a large number of female community nutrition promoters are employed to undertake growth monitoring and promotion among children, nutritional support for pregnant and lactating women, behavior change communication about nutrition and related issues for the whole community, and a supplementary feeding for severely malnourished or growth-faltering children and pregnant and lactating women. Services are provided through 9,000 community nutrition centers donated and managed by village communities and 14 non-governmental organizations contracted by the Government to mobilize communities and deliver services.

In 1998 a World Bank team reviewing projected progress noted that sever malnutrition among children under two had declined from the baseline figure of 13 percent to 2 percent. Also, the number of low-birth-weight babies decreased by 30 percent, and there had been an improvement in weight gain among at least half of the prename women in project areas. The success of the BINP, undertaken in limited area, has demonstrated that community mobilization and community-based nutrition services delivered with the help of NGOs can bring rapid, sustainable reductions in severe malnutrition among children and deliver targeted food and micronutrient supplements to reproductive-age women suffering chronic energy and micronutrient deficiency.

Source: Poverty in Bangladesh: Building on Progress WB 2002

187. There are other lingering problems, such as low birth weight and the persistence of infectious diseases. In addition to the emerging additional burden of arsenic poisoning, there are serious challenges associated with the changing disease patterns. Currently, infectious and communicable diseases as well as non-communicable diseases are prominent causes of mortality. Injuries (road traffic accidents and drowning) are by far 8 The presence of public food distribution programs (Food for Work and Vulnerable Group Feeding) does not have large effect in reducing malnutrition on average, but has a large positive effect in reducing malnutrition among the poorest. For more detail, see the companion Poverty Assessment Report (2002). 27 the leading cause of death for 5-14 year olds and 15-44 year olds (Appendix Table A6.5). Respiratory and digestive problems take the highest and second highest toll on the 0-4 and 5-15 age cohorts, respectively. Tuberculosis is the second most common cause of death among adults ages 15-44, and cardiovascular diseases for those ages 45 and older. The contribution of injuries and non-communicable diseases to total morbidity and mortality is projected to increase, placing new challenges on the health system.

188. Besides the shortfalls in health outcomes, substantial inequalities persist in many health indicators, especially those relating to child health (Table 6.6).

Table 6.6 Bangladesh—infant and under-five morality rate and children’s nutritional status, 1996/7

Quintile Infant Under-five Children Children Children mortality mortality stunted moderately severely rate* rate* (percent) underweight underweight (percent) (percent) Poorest 96.3 141.1 50.5 60.3 28.7

Second 98.7 146.9 50.8 53.5 26.2

Middle 97.0 135.2 41.9 49.2 21.7

Fourth 88.7 122.3 34.8 41.8 13.1

Richest 56.6 76.0 23.5 28.1 5.6

All 89.6 127.8 41.3 47.6 19.8

Poor to rich 1.70 1.85 2.149 2.15 5.13 ratio

* deaths per 1,000 live births Source: Demographic and Health Survey (DHS) 1996/7

189. There are three institutional problems in the health care system, that if not addressed, will seriously inhibit the Government's ability to address the increasing health needs of the population.9 These problems pertain to (i) the slow progress in implementing the HPSP; (ii) the lack of efficiency in the public sector; and (iii) the ill-defined role of the non-public sector.

 The HPSP is facing implementation problems. The most important of these problems are (i) lack of capacity in procurement, financial management, accounting, and auditing, which leads to delay and disruption in project implementation; (ii) delay 9 There is also a growing perception of increasing corruption m health service delivery. Doctors as well as the public are very vocal about systemic corruption in procurement, the registration of clinics, the provision of medicine and supplies, and the appointment, posting, and promotion of medical professionals. Even after admission, extra payments are routine aspects of treatment, whether in government or private facilities. Another set of problems pertains to the rigidities in budget management and the duality of the health budget, leading to sub-optimal geographic allocation rules and imbalances between recurrent and development expenditures. These problems, however, are symptomatic for nearly all sectors of the economy and thus discussed in Chapter 4. 28

in the unification of health and family planning services at the district and central level, which leads to confusion over roles and responsibilities at the field level, wastage of resources, and duplication of activities; and (iii) lack of coordination and politicization of site selection for community clinics, a large and basic component of the program, which has made many facilities functionally ineffective.10 The Government recognizes these operational problems and delays. It has expressed reservations about the complexity of the program and has indicated that it finds it difficult to monitor the various components of the program.11 In order to make the program effective, the Government intends to commission a full review by the Implementation and Monitoring and Evaluation Division (IMED) of the operational strengths and weaknesses of the HPSP sector approach. Notwithstanding these problems, the Government stated its unambiguous intentions to retain the SWAPp and possibly expand it to other social sectors.

 The productivity of thana and union facilities is very low by international standards. Poor management of thana and union facilities, sub-optimal location and staffing of the facilities, and doctor absenteeism and inadequate provision of medical supplies lead to low quality and hence low attendance rates and low productivity and high unit costs relative to facilities in other large low-income countries (Appendix Table A6.6). This situation is particularly worrisome because these facilities attract the bulk of resources and represent the backbone of health service delivery. One of the reasons for low productivity is that government clinicians have weak incentives to increase their productivity in government work, since they can sell their services to private clients. In fact, the public (especially the urban non-poor) is more likely to seek medical care from government-employed doctors working in private practices than from those working in public facilities (Appendix Table A6.7).12

 Contrary to the objectives of the HPSP, there has been little progress toward efficient mobilization of private providers. Given the deep penetration of private providers in the health market and the large volume of private health spending, regulation and enforcement of quality and standards have become essential to ensure the effectiveness of private spending.

Example 3. Objectives of the health system (2)

Maldives Public Expenditure Review. Report No. 24238-MV, December 6, 2002 b. Strategic Objectives and Policy Direction

4.49 The Republic of Maldives has set out an ambitious agenda for improving the health of the population. Various documents by the Ministries of Health and Planning lay out 10 See Bangladesh Health and Population Sector Program, Aide-Memoire, Mid-Term Review, and Annual Program Review (December 2000) for a complete list of implementation problems facing the HPSP. 11 In addition, the Government is concerned about the large amount of resources devoted to consultants' wages and training financed from donor resources. 12 If, as is likely, these government-employed doctors treat private patients during official hours of work, then part of the government health subsidy intended for the public is misappropriated by them. 29 the main thrusts and strategic objectives. A complete list of priorities is identified in the Health Master Plan 1996-2005. The following thrusts summarize the policy directions:

 Equitable access to health services. The emphasis has been on infrastructure investments that decrease travel time to facilities for the population of the most distant islands.

 A focus on prevention of diseases and the promotion of healthy lifestyles. The vision statement of the Ministry focuses on empowering citizens to improve their health status by engaging in healthy lifestyles and appropriate use of preventive and curative care.

 Comprehensive reproductive health services, including maternal and child health (MCH). There is a recognition that maternal and reproductive outcomes continue to lag behind and require focused policy and investments.

 Strong communicable disease control function. Building on success.

 A new emphasis on non-communicable diseases. Adapting to the epidemiological transition.

 Improvements of health sector manpower. Responding to the need to develop technical and managerial skills currently lacking.

Example 4. Health outcomes in historic and regional perspective

Georgia Public Expenditure Review. Report No. 22913-GE, November 25, 2002

5.5 Available data indicates that Georgia's health outcomes measured in terms of life expectancy at birth, infant and under-five mortality rate and maternal mortality rate compare favorably with other CIS countries. (Table 5.1).13 However, it should be noted that Georgia's health indicators have worsened significantly over the past decade. The infant mortality rate (IMR) has increased from 9 in 1991 to 24 deaths per 1000 live births in 2000. Similarly, the maternal mortality rate (MMR) in Georgia is 51 per 100,000 live births which, while low in comparison to other developing countries, is way up from the MMR of 32.4 reported in 1993. The majority of maternal deaths in Georgia (38 percent ) percent) occur from bleeding, one of the causes of which is anemia which typically needs to be diagnosed and addressed during the prenatal stage. This is yet another indication of poor health system performance.

5.6 Similarly, incidence rates of infectious and parasitic diseases in Georgia have increased in the last decade. For example, there has been a huge increase in the number of new tuberculosis cases, from 1,531 in 1991 to 4,515 in 1999. Also, while the number of diphtheria cases has declined from 425 in 1995 to 60 cases, it is still considerably higher than 1992 (when 3 cases were reported). The situation in noncommunicable diseases has

13 Data on health outcomes is Georgia are not very reliable since disease surveillance is rather weak. 30 also worsened. Overall, circulatory system diseases cause about 71 percent of all deaths in Georgia, followed by neoplasms (11 percent). Other social illnesses, particularly cases of drug addiction, are on the rise (from 1,347 in 1984 to 4,873 in 1996).

Table 5.1 Health outcomes in Georgia and other CIS countries (2000) Source: WDI, 2001, UNICEF (for Infant and Under-5 mortality rates) Countries GNP per Life Crude Maternal Infant Children capita expectancy death rate mortality mortality mortality (US$95) (years) (per 1,00 rate (per rate (per rate (under live births) 100,000 live 1,000 live 5, per 1,000 births) births live births)

Georgia 620 73 8 51 24 29 Other FSC Countries Armenia 490 74 6 35 25 30 Azerbaijan 460 71 6 43 74 105 Kazakhstan 1,250 65 10 70 60 75 Kyrgyz Republic 300 67 8 65 53 63 Tajikistan 280 69 5 65 54 73 Uzbekistan 720 70 6 21 51 67 ECA 1,190 68 — — 20 26 Low-income 420 59 — — 77 116 countries

Example 5. Geographic distribution of health outcomes

Maldives Public Expenditure Review, Report no. 24238-MV, December 6, 2002.

Table 4.6 HNP outcomes in Malé and the atolls, 1997-8

Atolls Life Expectancy Stunting (%) Contraceptive Use Alif Dhekunu Buri 65 26 26.9 Alif Uthuru Buri 58 16 28.3 Baa Atoll 66 26 24.7 Dhaat Atoll 58 37 25.9 Faafu Atoll 54 60 12.5 Gaaf Ahf Atoll 62 70 10.8 Gaaf Dhaal Atoll 58 49 14.6 Gnaviyam Atoll 69 50 12.8 Haa Alif Atoll 58 54 16.4 Haa Dhaal Atoll 62 48 16.1 Kaafu Atoll 66 22 34.1 Laamu Atoll 59 43 24.0 Lhaviyani Atoll 58 36 26.8 31

Malé 68 0 13.6 Meemu Atoll 62 37 33.3 Noonu Atoll 56 47 24.0 Raa Atoll 52 56 19.6 Seenu Atoll 69 23 16.6 Shaiyani Atoll 53 50 24.1 Thaa Atoll 54 35 18.3 Vaavu Atoll 61 39 23.6 Source: Ministry of Planning and National Development and the United Nations Development Programme, Vulnerability and Poverty Assessment, 1998. (1999)

Table 4.7 Mortality rates: Malé and the Atolls — 2001

Malé Atolls

IMR 13 19 Under 5 mortality 17 30 Early neonatal death rate 7 9 Late neonatal death rate 4 3 Post neonatal death rate 3 8 Neonatal death rate 10 12 Source: Ministry of Health

Example 6. Equity of health outcomes

Public Expenditures for Poverty Alleviation in Northeast Brazil: Promoting Growth and Improving Services. Report No. 22425-BR, June 11, 2001.

3.4 That the health of the poor is worse than that of others should come as no surprise. In Brazil, however, the difference between rich and poor on this dimension of well-being is extraordinary and further reflects the severity of the gap between them. Figure 10 shows the relationship between income and the probability of death of children. This information is derived from the PNDS of 1996 and the indicator variable is a little unusual. It represents the fraction of children ever born to women under 40, at the time of the survey, who have died from any cause. It is not the same as the more common indicator of child mortality (the probability of dying between the ages of one and five), but allows for a much larger sample to be used in its calculation. 32

3.5 The difference between the richest and poorest decile of the wealth measure used from the survey is a factor of five. Children born to the poorest families are five times more likely to die before their mothers reach an age of 40 than are children in the top 10 percent of families. For comparison, in India the same mortality rate doubles between the poorest and the richest.

3.6 What are the poor dying from? A large proportion of children die from communicable and environmentally determined diseases. Diarrhea and acute respiratory infections are common among the poorest. Other infectious diseases are still common among the poor even though they are quite rare among the better-off. Among older people, the noncommunicable diseases are responsible for more deaths and are increasing in prevalence. Indeed, poor people suffer from the noncommunicable diseases more than do richer people (Murray, Yang, and Qiao 1992). However, the relative incidence of communicable and environmentally determined diseases between rich and poor is much higher. Therefore, a much higher proportion of money spent on communicable diseases will reach the poor than the same amount on noncommunicable diseases.

Example 7. Risk factors

Vietnam – Managing Public Resources Better – Public Expenditure Review 2000, Report no. 24469, vol.2.

6. Vietnam has experienced an epidemiological transition during the last 20 years, with a major decline in the share of communicable diseases in morbidity. While communicable diseases accounted for 50-56 percent of morbidity in 1976, the corresponding share had fallen to 27 percent by 1997 (figure E.3). Much of this decline was achieved during the last ten years. Most dramatic has been the increased incidence of accidents, injuries and 33 poisoning, whose share in morbidity increased several-fold between 1976 and 1997. The share of noncommunicable diseases also increased considerably over this period.

7. These shifts reflect the success of communicable disease-control programs, especially the expanded program of immunization, which has reduced the incidence of vaccine- preventable diseases in the country.

8. In 1997, the five leading causes of morbidity in hospitals were infectious and parasitic diseases (accounting for 23 percent of all cases), respiratory ailments (17 percent), diseases of the digestive system (12 percent), genito-urinatory diseases (5.3 percent), and diseases of the circulatory system (5.1 percent) (Ministry of Health, 1997).

9. Despite the relative decline in the incidence of infectious diseases, the latter still remain a main public health problem, particularly some new or reemerging diseases, such as tuberculosis, HIV infections, dengue fever and Japanese encephalitis. For example, the incidence of dengue fever has increased dramatically in the last 4-5 years, with the infection rate increasing almost five-fold between 1994 and 1998. Likewise, the number of tuberculosis cases has also increased substantially in recent years. In 1997, the incidence of tuberculosis was about 90 cases per 100,000 inhabitants and 2,402 deaths from tuberculosis were recorded at hospitals (Ministry of Health, 1997).

10. HIV/AIDS is a new disease whose incidence has increased rapidly in the past few years. The number of new HIV cases hasve increased ten-fold between 1993 and 1998. It is estimated that there are 88,000 persons living with HIV/AIDS in the country. By the end of the year 2000, it is estimated that this number will rise to 186,000. Even though 65 percent of the HIV infections reported so far are associated with intravenous drug use, heterosexual transmission among young people is increasingly becoming more common (UNDP, 1999b).

11. Even though implementation of malaria control measures in the past 6-7 years have been very successful and have brought down the incidence of malaria from 16 cases per 1,000 persons in 1991 to only 6 in 1998, malaria continues to remain a major public health problem in the mountainous and ethnic minority areas of the country.

12. Accidents, injuries and poisoning are a major cause of death in hospitals. In 1997, there were 475,000 cases of and 5,424 deaths from accidents, injuries and poisoning. As noted earlier, this represents a dramatic increase from the situation twenty years ago. Many of these deaths are caused by traffic accidents. This is a trend that is characteristic of many developing countries experiencing economic growth, rapid urbanization and traffic growth.

13. Finally, although the incidence of diseases of affluence, such as cancer and heart disease, has increased over time, they still account for a small proportion (less than I percent) of total morbidity or mortality in the country. Of course, these diseases are much more prevalent in urban centers, such as Hanoi and Ho Chi Minh City, but their incidence, at least as of now, is confined largely to better-off populations in the urban areas. 34

Example 8. Private payments to providers

Georgia Public Expenditure Review. Report No. 22913-GE, November 25, 2002

Private Expenditures on Health Care

5.27 Private spending consists of out-of-pocket payments in the form of official co- payments, fee-for-service payments and informal payments. Reliable estimates of private expenditures on health are not available for each year. One of the first assessments was made in 1998 based on a 1995 UNICEF household survey, which estimated private expenditure on health to be almost GEL272.7 million. Since then, several national household surveys have been conducted, especially in recent years. Results of these surveys vary, and it is difficult to say with certainty how much is spent out-of-pocket on health care. Different sources of data on out-of-pocket spending indicate different figures for out-of-pocket spending, ranging from 66 to 87 percent of total health spending. For example, a recent study by the Department of Statistics, Georgia (SDS) found that out-of- pocket payments were approximately $28 per capita or 73 percent of total health expenditures. In 2000, according to the Tbilisi Household Survey and the Quarterly Survey of Georgian Households, total out-of-pocket expenditures amounted to GEL132 million, or about 66 percent of total spending on health. Most of the out-of-pocket payments go towards the purchase of drugs (53 percent), followed by hospitalization (23 percent), and outpatient services (17 percent). In addition to formal co-payments and fee- for-service payments, informal payments are probably significant and are very much part of the system. Private health insurance in Georgia is not significant, and amounts to less than GELO.5 million per year.

Example 9. Impact of different sources of finance on equity

Public Expenditures for Poverty Alleviation in Northeast Brazil: Promoting Growth and Improving Services. Report No. 22425-BR, June 11, 2001.

3.27 Regardless of the direct effect on health status, how does the health subsidy look in terms of redistribution to the poor? This depends on who is paying.

3.28 Figure 16 shows the usage of public facilities in the northeast by consumption decile. The deciles used are based on the national distribution of consumption. Note that there are very few people in the highest deciles that use public facilities. This is because there are very few people in the northeast who are in these deciles. The usage pattern is very progressive from this perspective. To the extent that usage of public facilities can be translated into subsidies (discussed below), very little of the public subsidy goes to people who are wealthy relative to the Brazilian population, and over 50 percent of these subsidies accrues to people in the poorest 30 percent of the consumption distribution (the lowest 30 percent corresponds roughly to those in poverty). If the budget for health were 35 paid from federal sources, i.e., from taxes collected from people on average richer than those in the northeast, it would be an excellent means of redistributing income.

3.29 However, if funding were to switch to the states themselves, the distributional impact becomes less clear. Figure 17 shows the proportion of visits to facilities from people in different deciles of the consumption distribution defined within the northeast. That is, people in the lowest decile in figure 17 are the poorest 10 percent of the northeast population not the poorest 10 percent of the national population who live in the northeast. 36

3.30 If visits correspond to use of public money and if the money needs to be generated within local communities, then the pattern of use in figure 17 represents the beneficiaries of subsidies relative to the population paying for them. In this case, the distributional impact is not nearly so progressive. It is not the poorest of the poor in the northeast who use health services disproportionately, but the well-off relative to others in the region, though not relative to the country as a whole. If the people in the northeast have to support health services from their own tax revenues, these services will lose their ability to serve a redistributive function.

3.31 Since the data in figures 16 and 17 are visits and not money, it is not possible to assess the incidence of subsidies. Should it be the case that richer people can bargain for better and costlier services, figure 16 would be flatter and figure 17 would be steeper. If the government is successful in targeting its subsidies and gives better (or at least more expensive) services to the poor, then the reverse is true and using health as a redistributive device is even better than it looks here. International experience suggests things are worse than they look.

3.32 The distributional effect still depends sensitively on whether money is coming from within the region or from the federal government. The states can choose to add to federal transfers to different degrees. Ceará, for example, added about US$23 per capita to its budget on top of its federal transfer of US$43 whereas Bahia added US$16 to its allocation of US$32 (World Bank 1998). The distribution of income of beneficiaries relative to the taxpayers footing the bill can differ substantially.

3.33 Health services do not appear to make much difference in terms of mortality, and, if paid for from local sources, may not be a very good vehicle for redistribution. On the other hand, the poor don't have much choice. Figure 18 shows the use of different services by consumption group (going back to national definitions). Less than 15 percent of visits to health-care providers by the poorest 30 percent are to private facilities, leaving the large majority to rely on public services. This leaves us with a cruel dilemma. The services may not be very good and they may not go primarily to the poor, but the poor have no alternative. This is another manifestation of the extreme skewness of the consumption distribution. So little of the expense in aggregate terms helps the poor, but that small amount represents a large contribution relative to the poor's resources. 37

Example 10. Equity implications of out-of-pocket expenditures

Georgia Public Expenditure Review. Report No. 22913-GE, November 25, 2002

Out-of-Pocket Spending on Health

5.36 Private spending is a major source of funds for the health sector. In 2000, according to the Tbilisi Household Survey and the Quarterly Survey of Georgian Households, total out-of-pocket expenditures amounted to GEL132 million. Out of all types of medical services, inpatient treatment incurred highest costs (average GEL311 per visit) followed by out-patient services (average GEL70 per visit) and emergency care (average GEL44 per episode).

5.37 To assess the burden of out-of-pocket expenditure on the poor, we analyze the data from the Georgia Households Survey (fourth quarter, 2000). Of the total of 39,648 respondents, 19 percent (7,467) reported suffering from illnesses during the recent three months. Of those suffering from an illness, 25 percent required urgent medical care, and 81 percent of these sought medical treatment. The poorer quintiles reported higher incidence of illness and lower percentage of those seeking treatment when ill (Table 5.4).

5.38 Of the several reasons why many of the ill do not seek treatment, shortage of money is the most prominent, affecting over 90 percent of those who do not seek any treatment, predominantly in the poorest two quintiles. Among those who seek treatment, 57 percent report no expenditure on treatment, 33 percent report no expenditure on drugs, while 26 percent report no expenditure overall. For the sample as a whole, the average annual expenditure per individual on illness is GEL47, with the richest spending four times as much as the poor (Table 5.4).

Table 5.4 Instances of illness (not requiring hospitalization) 38

Quintile Group Percentage reporting Percentage requiring Percentage seeking illness out-patient medical care care 1 23.2 21.5 67.2 2 17.4 25.4 79.7 3 18.4 23.7 82.1 4 19.1 23.5 82.1 5 17.7 28.2 87.2 Total 18.8 24.6 80.9

5.39 As far as inpatient care is concerned, of the total of 39,648 respondents to the Georgia Households Survey, only 195 people (0.5 percent) reported being hospitalized. The richest quintile reports twice as many cases of hospitalization than the poorest quintile. Of 195 people who had undergone inpatient treatment, 38 percent (74 respondents) respondents reported no expense on inpatient treatment. Of those who incurred some expenses, about one-fifth reported monthly overall expenditure over GEL500.

Table 5.5 Total annual expenditures on out-patient care by the non-chronically ill

Quintile Total health care expenditures

Mean Standard deviation Minimum Maximum 1 20.9 41.5 0 350 2 32.2 73.0 0 690 3 35.9 79.8 0 700 4 37.2 67.6 0 610 5 86.8 496.3 0 5500

5.40 An analysis of total expenditures related to the hospitalization by income quintiles shows that the richest spend four times on inpatient care as much as the poor, and the average annual expenditure per individual is GEL376 (Table 5.6). The calculations of expenditure on inpatient care by income quintiles shows that just one incidence of hospitalization can potentially completely wipe out the annual income of a household in the poorer quintiles.

Table 5.6 Total annual out-of-pocket expenditures on inpatient care

Quintile Total expenditures on inpatient care

Mean Standard Minimum Maximum Deviation 1 145.9 275.8 0 1213 2 367.9 484.3 0 1850 3 185.3 247.0 0 1090 4 257.9 252.5 0 1000 39

5 680.0 1413.1 0 5500

5.41 In terms of the percentage of income spent on health care, households in the poorest quintile can expect to spend over 5 percent of their annual income on health care, while those in the three highest quintiles can expect to spend approximately 1 percent of their annual incomes on health care.

5.42 There is little doubt that the poor in Georgia have suffered disproportionately from both the collapse in public spending on social services as well from misallocation of available resources. As out-of-pocket expenditures have become the major source of health financing, access to basic health services for the poor has been adversely affected, and increasingly large numbers of health conditions are going untreated. Shortage of money is consistently mentioned as the most common reason for the lack of medical treatment, providing a strong basis for the inequities in the system. Expectedly, the rich also undergo inpatient care treatment in larger numbers than the poor. Those among the poor who seek medical treatment spend twice or even three times as much as the rich with respect to their income. The data show that very few episodes of illness are sufficient to completely impoverish those with low incomes who seek health care when ill.

Example 11. Assessment of Gaps in Expenditures and Allocation

Bangladesh Public Expenditure Review. Report No. 24370-BD, May 25, 2003

Health expenditures declined in real terms between 1997 and 2000, largely because of deficiencies in the procurement system.

175. Analysis carried out by the Health Economics Unit (HEU) of the MoHFW over 1993-2000 shows that, while the budget allocations to health continued to increase, since 1997/98 actual expenditures declined both in real per capita terms (see figure 6.1)14 and as a share of GDP. Total spending in 1999/2000 was only 84 percent of the original budget (Table 6.2). This expenditure under-run, which is mostly in the development budget, is largely the result of less than full utilization of donor contributions, which represent over 70 percent of all development budget resources.

14 During the last two years of the HPSP total annual health spending has declined by 0.8% in real terms, compared with an 18% increase in the previous two years (HEU, MoHFW, 2001a). 40

Table 6.2 Bangladesh—ministry of health and family welfare current and Annual Development Program expenditures, FY94-00(Tk millions at current price, etc.)

FY94 FY95 FY96 FY97 FY98 FY99 FY00 Current expenditure Expenditure 5,040 5,930 6,470 7,330 7,860 8,760 9,430 Expenditure with a share of budget allocation (percent) 116 103 94 103 101 101 97 Share of MoHFW in total current expenditure (percent) 6.6 6.2 6.0 6.2 6.0 6.0 3.4 Annual Development Program (ADP) expenditure Expenditure 5,690 7,810 8,120 10,250 11,120 9,810 10,260 Expenditure as a share of budget allocation (percent) 109 102 91 108 98 77 75 Share of MoHFW in total ADP expenditure (percent) 6.7 8.1 8.5 92 9.0 6.9 10.9 Total MoHFW budget (current and ADP) Expenditure 10,720 13,740 14,590 17,580 18,980 18,570 19,690 Expenditure as a share of budget allocation (percent) 112 102 92 106 100 87 84 Share of MoHFW in total Government expenditure (percent) 66 7.2 7.2 7.6 7.5 6.5 5.3 Total MoHFW expenditure as a share of GDP (percent) 1.12 1.27 1.21 1.36 1.34 1.15 1.10

Source: Health Economics Unit, MohFW (2001a)

176. The HEU analysis suggests that the shortfall in spending is due to two factors: (i) the general and continuing lack of understanding about procurement procedures (IDA and GOVERNMENT) due to an acute shortage of trained procurement experts; and (ii) cumbersome and time-consuming guidelines for procurement of supplies and services, particularly through the reimbursable program aid pool (RPA). A recently completed World Bank mid-term review of the HPSP identified procurement as the single most critical operational issue in the HPSP. Disbursements from IDA and pooled funds during the first two years of the implementation of HPSP were US$136 million, or about 40 percent of the original commitment (World Bank 2000d).

Example 12. Classification of expenditures by program 41

Vietnam – Managing Public Resources Better – Public Expenditure Review 2000, Report no. 24469, vol.2.

64. Greater disaggregation of Ministry of Health expenditure on (vertical) health programs is shown in Table E.4. While overall spending on disease-control programs increased by 15 percent annually in real terms over the 1993-97 period, there were some important reallocations within this spending category. The malaria control program saw a 40 percent real decline in its, while funding for EPI, tuberculosis and leprosy control programs increased sharply in real terms expenditures. (Actually, the shortfall in budget spending for malaria is compensated by funding from the World Bank financed National Health Support Project). Table E4 Ministry of Health expenditure on health programs, 1993-97* Year Malaria Iodine EPI AID TB Lepros Nutrition Other** Total deficienc S y y

Expenditure in nominal terries (million VND): 1993 44,799 27,235 10,738 1,362 84,134 1994 46,693 17,518 32,961 9,880 107,052 1995 35,277 30,419 37,103 4,770 930 108,499 1996 43,232 28,796 40,690 6,766 22,073 2,689 1,904 14,606 160,756 1997 37,642 25,087 55,618 33,897 6,685 42,828 201,757

Expenditure in real terms (millions of constant 1994 VND): 1993 50,223 30,533 12,038 1,527 94,321 1994 46,693 17,518 32,961 9,880 107,052 1995 29,719 25,627 31,258 4,019 783 91,406 1996 34,669 23,092 32,630 5,426 17,701 2,156 1,527 11,713 128,914 1997 27,780 18,514 41,046 25,016 4,934 31,607 148,898

% share of each program in total MoH expenditure on health programs 1993 53.25 32.37 12.74 1.62 100.00 1994 43.62 16.63 30.79 9.23 100.00 1995 32.51 28.04 34.20 4.40 0.86 100.00 1996 26.89 17.91 25.31 4.21 13.73 1.67 1.18 9.09 100.00 1997 18.66 12.43 27.57 16.80 3.31 21.23 100.00 Notes: * exclusive of national program expenditure spent by provinces as authorized by the center ** includes the hospital equipment upgrade program Source: Ministry of Health estimates

Example 13. Economic classification of expenditures (2)

Vietnam – Managing Public Resources Better – Public Expenditure Review 2000, Report no. 24469, vol.2.

Composition of Government Health Expenditure 42

56. Economic Composition. There are three important observations to be made on the economic composition of public spending on health. First, capital spending comprises a large share of the state health budget, although this share has declined from 28 percent in 1991 to 21 percent in 1998 (Table E.2). Nearly one-half of capital spending over the years (but rising to 82 percent in 1998) is on construction, which is again very large share in comparison to the health budgets of most developing countries.

Table E2 Distribution of public health spending by source and economic type, 1991-98 Recurrent As % of total recurrent expenditure Capital As % of total capital expenditure expenditure expenditure as % of total as % of total expenditure expenditure Year Salaries Goods Subsidies Others Purchas Capital and and and e of construction wages services transfers fixed assets State budget on health (central plus local) 1991 71.70 22.67 71.29 5.89 0.15 28.30 54.48 45.52 1992 70.73 16.34 76.79 6.34 0.53 29.27 67.70 32.30 1993 76.22 27.45 69.57 2.56 0.43 23.78 59.19 40.81 1994 77.99 32.92 62.32 4.74 0.03 22.01 53.07 46.93 1995 76.21 31.36 62.51 6.10 0.02 23.79 48.25 51.75 1996 77.51 29.22 65.75 5.02 0.00 22.49 40.38 59.62 1997 78.63 27.59 67.82 4.58 0.01 21.37 52.60 47.40 1998 79.08 29.38 59.38 4.02 7.22 20.92 17.73 82.75 Total public speaking on health (including user fees and insurance) 1991 73.34 20.88 73.56 5.43 0.14 26.66 54.48 45.52 1992 72.42 15.04 78.64 5.83 0.49 27.58 67.70 32.30 1993 78.42 24.99 72.00 2.31 0.70 21.98 59.19 40.81 1994 80.07 29.41 65.84 4.18 0.57 19.93 53.07 46.93 1995 79.49 26.33 67.99 5.04 0.64 20.51 48.25 51.75 1996 81.39 23.47 71.91 3.96 0.66 18.61 40.38 59.62 43

1997 82.31 22.19 73.63 3.62 0.56 17.69 52.60 47.40 1998 82.77 22.94 67.80 3.08 6.18 17.23 17.25 82.75 Notes: Figures in bold are percentages of total health spending. The remaining figures are expressed as percentages of total recurrent or capital expenditure. Source: Ministry of Finance data.

57. Second, only about 29 percent of the recurrent health budget is spent on salaries and wages, while 59 percent is spent on goods and services. Since health insurance reimbursements and user fees are typically not used to pay for salaries, the share of salaries and wages in total government health expenditure (i.e. state budget plus insurance plus user fees) is even smaller – 23 percent.15 In this sense, Vietnam is unlike other developing countries which spend the majority share of their recurrent health budget on salaries and wages, and have very little to spend on consumables such as medical supplies and drugs.

Example 14. Geographic distribution of health services

Georgia Public Expenditure Review. Report No. 22913-GE, November 25, 2002

Equity in Health Expenditures

5.32 A little over one-fifth of the population of Georgia lives below the national poverty line (June, 2000). About half of the poor, i.e., around 10 percent of the population, live in extreme poverty. Ill health is both a dimension of poverty and an important factor in generating poverty. A recent poverty profile of Georgia shows that approximately 60 percent of the population are at risk of falling into poverty, and the costs of seeking health care is one major factor (World Bank, Poverty Assessment, 2001).

Public Spending on Health

5.33 In the absence of data on utilization by income quintiles of publicly-provided or financed health services, we examine allocation of municipal budgets, the only public expenditure on health data available by region, and the regional distribution of poverty, to draw inferences on patterns of use.

5.34 The distribution of poverty varies significantly across regions in Georgia. Regardless of which measure of poverty is used, Adjara and Samegrelo have the lowest incidence of poverty, while Imereti - home to almost one-third of Georgia's poor - has the highest (The World Bank, 1999). An equitable distribution of public expenditures would be one in which the poorer regions have a relatively higher per capita allocation of public expenditures compared to the richer regions. However, an examination of the allocation of municipal expenditure and poverty distribution shows that the reverse is true: richer regions, i.e., those with fewer persons below the poverty line, tend to spend more public money per person compared with the poor regions (see figure below). Imereti, with the

15 User fees may be used, however, for paying bonuses to health workers. 44 largest number of poor in the country, spends the least, while Adjara, with the least number of the poor, spends the most.

5.35 Municipal allocation to health varies considerably across the years, however, and to this extent is not a robust indicator of resource use. No other data is available on resource allocation and utilization of health services that can be used to assess the impact on equity of public expenditures on health.

Example 15. Geographic distribution of resource allocation

Maldives Public Expenditure Review, Report no. 24238-MV, December 6, 2002.

4.72 Geographic distribution. The final area of expenditure analysis is to look at regional allocations of public sector spending in health. As in the case of the functional categorization, the financial information system does not allow for a direct analysis of regional allocations. Indirect methods were used to impute expenditures by atolls for the three levels of care delivery. Since the only tertiary care facility, IGMH, is located in Male, the negative equity implications of spending at that level are both obvious and inevitable. A similar problem exists when looking at the secondary care facilities as only a limited number can be supported and have to be placed in the selected atolls. An enhanced cost recovery effort at the secondary and tertiary levels, with appropriate exemption mechanisms, can, however, mitigate some of the negative equity implications.

4.73 The only level of care where there is an opportunity to compare across the atolls is primary care (the combination of island-level and atoll-level facilities and expenditures). One way of looking at the equity aspect of spending at the primary care level is to compute estimates of per capita spending by atolls and look at the variations as they relate to different atoll characteristics. The average per capita spending at the primary care level was Rufiyaa 41, but the range varied from a low of Rufiyaa 7 in one atoll to a high of Rufiyaa 142 in another atoll in 1997/98. Using an index of vulnerability developed for the Maldives: Vulnerability and Poverty Assessment 1998 (VPAS) study cited earlier, we can look at the relationship between per capita spending and 45 vulnerability. In figure 4.8, the atolls are ranked from the most to the least vulnerable and there does not appear to be a relationship between the spending and need.

4.74 A more direct measure of the need in the health sector can be constructed using health outcome measures. In figure 4.9, the atolls are ranked by infant mortality rates (from the lowest to the highest rate) and the corresponding per capita primary care expenditures are graphed. The results are similar in that there does not appear to be a relationship between the need and spending.16

16 Population concentration is obviously a key factor in affecting the allocation of primary care expenditures among the atolls. However, many of these atolls do have broadly comparable populations. 46

Example 16. Organization of resources, by level of service

Democratic and Popular Republic of Algeria, Public Expenditure Review of the Social Sector. Report No. 22591-AL, January 2002.

9. Algeria has a three-tiered vertically integrated delivery system of care - there are 2.1 beds, 1.3 physicians (40 percent specialists), and 2.9 paramedical staff per thousand population. Over 99 percent of all beds, 71 percent of all physicians, and 63 percent of all dentists practice in the public sector, but only 8 percent of pharmacists do. Overall, 43 percent of all specialist physicians and 31 percent of the generalists work in the private sector. The private sector is an increasingly important source of health care (especially at the primary health care level), but coordination between the public and the private sector has so far been limited. 47

Example 17. Health services delivered by other agencies

Maldives Public Expenditure Review. Report No. 24238-MV, December 6, 2002

4.60 It is possible that the expenditure and budget numbers used here actually underestimate the Government's commitment to the health sector in the Maldives. This is due to the fact that, in addition to the Ministry of Health, a number of the other line ministries have budget items related to health sector spending. figure 4.4 presents a flow of funds diagram covering public spending on health. While the bulk of spending on health goes through the Ministry of Health, a complete public spending picture includes:

 Ministry of Women's Affairs and Social Security (MOWASS) provides financial aid for Maldivians needing care outside the Maldives as well as to Maldivians receiving treatment in Male (primarily at IGMH). In 2000, for example, MOWASS provided financial aid of more than Rf 8.5 million to 589 individuals receiving treatment abroad and 1485 individuals receiving treatment within the country.

 The Presidential Palace. The Presidential Palace also provides aid to individuals that receive treatment abroad.

 Ministry of Defense and National Security. Some spending takes place on health services for staff and their families.

 Cost Recovery. IGMH, some regional hospitals, the Ministry of Health, the Water and Sanitation Authorities, and Department of Public Health charge for a variety of services. Revenues are not retained/recycled at facilities; they are remitted to the Treasury. 48

4.61 The flow of public funds depicted in figure 4.4 and the difficulty of finding disaggregated expenditure and budget information for all the intermediaries highlight the need for a well functioning financial information system and National Health Accounts. The inability to track sectoral expenditures from sources to intermediaries to uses makes it difficult for policy makers to ensure that resources are allocated and used in effective and efficient ways and that priority activities are receiving appropriate funding.

Example 18. Equity of health services (1)

Public Expenditures for Poverty Alleviation in Northeast Brazil: Promoting Growth and Improving Services. Report No. 22425-BR, June 11, 2001.

3.7 The questions for public policy to answer are first, how much of public money spent on health actually reaches the poor in terms of more or better services, and second, what kinds of policy are most effective in improving their health or well-being? The first question depends on the fraction of users of publicly provided health services who come from the poorer groups. The second depends on understanding the determinants of disease and ill-health. In particular, we need to identify the main determinants that are influenced by public policy. This chapter deals with the latter question first and then returns to the issue of health care.

3.8 There is a tremendous degree of complementarity between the pursuit of equity and efficiency in the health sector when policies are directed at communicable diseases. Public goods, externality-correcting policies and basic health information are all of high priority in the traditional public health agenda and are well justified in terms of standard efficiency arguments: they directly address market failures. At the same time, pursuing 49 them is particularly pro-poor in their incidence. Policies exhibiting such a degree of complementarity are not common and provide an unusual opportunity.

3.9 Several different studies investigate the relationships between public policy and health status. While there are a few differences among them, they tell a remarkably consistent story.

3.10 The first study relies on the study and tries to explain the pattern of deaths of children shown in figure 10. The variables chosen to explain this pattern are (1) a set of variables that measure household characteristics, (2) a set of determinants that are not part of the health sector but which are expected to have important health effects, especially the education of mothers, and (3) determinants that are part of traditional public health programs, including access to water and sanitation services.

3.11 By decomposing the mortality numbers into these sets of determinants, we can make an assessment of the relative importance of different policies in improving the health of the poor. By comparing the effect of these variables with the differences between rich and poor, we can attribute such differences to their constituent parts.

3.12 Appendix 4b, table A4.1 shows the equation that results from regressing the mortality variable on these hypothesized determinants. Expenditure and education play important roles in the relation as in many other studies of this type. Of particular note, however, is the effect that the source of water and sanitation facilities have on mortality.

3.13 Not only are sanitation and water supply important as determinants of health, they are also services which sharply distinguish the poor from the non-poor. Figure 11 and figure 12 show the relationships between expenditure and access to sanitation facilities and the source of drinking water. The figures paint a stark picture. In the poorest decile of the population, disproportionately in the northeast, almost 70 percent are without any sanitation facilities. Furthermore, the rate at which this proportion falls is dramatic. By the third decile, still poor by national standards, this number drops to less than 15 percent. By the middle of the consumption distribution, sanitation facilities are basically universal. 50

3.14 The quality of drinking water is almost as revealing. The survey did not actually measure contaminants in the water. It merely indicated the source. However, some of the sources are clearly good and others are clearly bad . The relatively good sources are piped into the household and bottled (which is fairly common in the highest decile). The clearly bad sources were from surface ponds, streams, etc. The remaining sources were basically wells and community pumps. These could be good or bad; since we cannot distinguish between them in the data they are not included here. On the basis of these results, one could almost define a poor person as someone who does not have piped water or sanitation services, the correlation with wealth being that close.

3.15 Combining the results of the regression equation with the evidence concerning differential access, yields figure 13. This picture takes the household profile of a poor, northeastern family with average values for education, expenditure and so forth, and simulates the effect of replacing the probability of having sanitation and safe water sources of poor people with the probabilities of the top decile (for sanitation to a 100 percent probability). This is done for water and sanitation separately and jointly. The proportion of children born to these people who subsequently die is cut in half by the 51 combined effect of better water and sanitation, particularly of the former. Figure 10 showed a gap between the mortality rates of rich and poor to be 8 percentage points - 10 percent for the poor, 2 percent for the rich. The results in figure 13 shows that 5 percentage points of these 8 are attributable to differential access to water and sanitation. This is a tremendously important concern.

Example 19. Equity of health service (2)

Vietnam – Managing Public Resources Better – Public Expenditure Review 2000, Report no. 24469, vol.2.

34. Implications for Progressivity of Public Health Spending. Public subsidies to a particular level of health facility (e.g. commune health centers) will be weakly progressive if the share of the poor among all users of that facility is greater than their share in aggregate national. consumption expenditure. In such a case, the distribution of commune health center users is more progressive than the underlying distribution of personal consumption; hence, public subsidies to commune health centers will be weakly efficient (in terms of targeting). On the other hand, public subsidies will be strongly progressive if the share of the poor among all users of that facility is greater than their population share.

35. Figure E.8 suggests that public subsidies to commune health centers and public hospitals together are weakly pro- poor, since the distribution of users of these public health facilities is more progressive than the underlying distribution of personal consumption expenditure. However, only public subsidies to commune health centers (but not to public hospitals) are strongly pro-poor, as the share of he poor among commune health center users exceeds their share in the population. 52

Example 20. Equity of health services (3)

Bangladesh Public Expenditure Review. Report No. 24370-BD, May 25, 2003

Public health expenditures play an important redistributive role. Outlays for child care and prenatal care benefit the poor the most.

181. The benefit incidence analysis of public expenditures on health was conducted in parallel with that of education expenditures.17 District-level disaggregated data from the MoHFW were used in conjunction with utilization episodes from the 2000 Household Income and Expenditure Survey to infer the per-visit health subsidy.18 The analysis shows that the share of all public health expenditures accruing to the poor is an estimated 45 percent, while their share in the population is 50 percent and their share in overall income is an estimated 26 percent (Table 6.5). Such a distribution of health expenditures reduces inequality and adds proportionately more to the welfare of the poor. In fact, the health subsidy represents 1.45 percent of the average per capita expenditure of the poor and 0.8 percent of that of the non-poor. Also, public health expenditures are more equitable than private health expenditures. Of all categories of health expenditure analyzed in this report, ESP allocations to child health reduce inequality the most. The distribution of

17 Caveats to BIA are mentioned in Chapter II. In addition, a recent finding shows that a substantial part of health expenditures is diverted from health care as absenteeism among doctors reaches about 40 percent at the thana level and nearly 75 percent at the UFWC. 18 As in the case of education, allocations for health are made from the revenue and development budgets. But unlike for education, development expenditures are used to finance salaries and operating costs. This analysis is based on recurrent and apportioned ADP expenditures. Additionally, the following assumptions were made: (i) child health expenditures benefit users of immunization services and users of curative care for child illnesses; (ii) curative care expenditures as well as expenditures on the non-ESP components benefit users of curative care in government facilities and from government doctors, (iii) maternal health expenditures benefit women who utilized prenatal or postnatal services or delivered a child in an institutional setting; and (iv) expenditure on family planning and communicable disease control benefited the population in a district as a whole. 53 these outlays is found to be strongly pro-poor, in large part because poor households tend to have more children. By contrast, allocations to limited curative care show the most unequal distribution, with the poorest 20 percent of the population having considerably less access to the subsidy than to the rest of the population.

182. The rural and urban poor command about the same proportion of overall health subsidies in their respective sectors. However, the very poor (the poorest 20 percent) of the urban population commands a smaller share of subsidies than their rural counterparts (Appendix Table A6.2). This pattern is driven by the fact that the poorest 20 percent of the urban population has considerably less access to government subsidies for curative care.

Table 6.5 Bangladesh—distribution of units of government-provided care, private health expenditures and government subsidies across income strata, 2000

By quintile By poverty status Total 1 2 3 4 5 lowest highest Poor Nonpoor Per capita expenditure 8 12 16 22 42 26 74 100

Visits to public facilities Curative visits 15 13 23 24 26 38 62 100 Child curative visits 16 17 27 21 18 46 54 100 Prenatal visits 19 20 21 20 20 50 50 100 Deliveries 10 12 14 21 42 28 72 100 Postnatal visits 3 16 18 23 30 38 62 100 Immunization visits 27 23 20 17 13 61 39 100

Private expenditures (for curative visits to government facilities Fees 9 5 13 24 49 19 81 100 Drugs 7 10 18 21 44 23 77 100 Transportation costs 5 10 13 28 45 19 81 100 Informal payments 4 5 14 8 71 14 86 100 All private medical costs 7 8 15 18 52 20 80 100

Public subsidies Family planning and control of communicative disease 18 18 19 19 24 46 54 100 Limited curative care 11 21 22 18 28 42 58 100 Maternal health* 20 13 20 18 29 44 57 100 Child health ** 23 21 19 18 18 54 46 100 All health expenditure 16 19 21 18 26 45 55 100 Total health subsidy as a percentage of per capita expenditures 1.65 1.38 1.16 0.77 0.60 1.45 0.78 1.11

* Combines expenditure on maternal health with utilization of prenatal and postnatal services and institutional deliveries. ** Combines expenditure on child health with curative visits for child illnesses and immunization visits. Source: World Bank staff estimates from 2000 HIES and MoHWF data. 54

183. An especially encouraging finding is that the HPSP targeting of the poor with prenatal care and immunization services has been successful. As shown in Figure 6.2 and Appendix Table A6.3, the utilization rate for poor women below 25 years of age, estimated at 43 percent, is higher than that for their non-poor counterparts, estimated at 37 percent. The difference in immunization rates between poor and non-poor children is negligible, with both estimated at about 90 percent (Appendix Table A6.4).

184. The HEU has carried out benefit incidence analysis based on an exit survey of 1,000 patients during August-September 2000 in four divisions of the country- Barisal and Rajshahi and later Chittagong and Sylhet-in a range of facilities at upazila level and below, sampling both outpatients and inpatients (LEU 2001a). Its analysis of attendance rates shows that primary health services are primarily used by lower-income groups. The bottom quintile accounts for more than 35 percent of visits, while the richest group accounts for only 15 percent (Figure 6.3).

Example 21. Insurance and entitlement to care (1)

Public Expenditures for Poverty Alleviation in Northeast Brazil: Promoting Growth and Improving Services. Report No. 22425-BR, June 11, 2001. 55

3.36 A further benefit of medical services to the poor also relates to insurance. Very few poor people have formal insurance (Figure 19). For these people, the prospect of free or heavily subsidized care through public facilities serves a function similar to insurance. It frees them from worry about financial vulnerability. How much this is worth can be approximated by calculating the "risk premium" associated with assurances against financial loss. Appendix 4c presents a description of this concept and its calculation. Figures 20 and 21 show the size of this benefit as a fraction of the costs associated with providing the services themselves as a function of the cost of services and of the income of the recipient. 56

3.37 Several points are noteworthy. First, the insurance value of providing free service can be a very high fraction of the cost of the services themselves. Second, the value of this benefit goes up with the cost of the service. It is more valuable to insure against expensive procedures rather than relatively cheap ones. This implies that the role of primary health centers as a source of referrals to hospital treatment may be more important than its role in providing free care. People may be willing and able to pay for routine curative care and subsidies at that level may not be worth much. Finally, the insurance benefit of the services is higher for poor people than for rich at every level of expense. Therefore, this could offer an important means of helping the poor relative to the rich.

3.38 However, there is an important caveat to this conclusion. Free hospital services for the poor, will require that access to the services be based either on income itself or on medical need. Since the determination of a person's income is often difficult, the better prospect is to assure that the referral system operate fairly. If the poor had the same probability of receiving care once sick as other people, their higher disease incidence would be enough to make the system pro-poor. In conjunction with the higher insurance benefit per unit lost to poor people than rich from expensive hospital care, the potential is very large. The reality depends on how referrals are administered.

Example 22. Insurance and entitlement to care (2)

Turkey Public Expenditure and Institutional Review: Reforming Budgetary Institutions for Effective Government. Report No. 22530-TU, August 20, 2001

Health Coverage

4.17 The population can be divided into two main groups: people who have some sort of health coverage, comprising about 86 percent of the total population, and people who do 57 not have any sort of health coverage, comprising the remaining 14 percent (Table 4.3). Those who have any sort of coverage can be further divided into six subgroups depending on the coverage scheme to which they belong.

(i) Civil servants and their dependents, who enjoy free health care, principally from MOH and university hospitals.19

(ii) Members of SSK, who use health services provided by the SSK facilities and are financed by their own contributions.

(iii) Self employed workers covered by health insurance from another social security institution, Bag-Kur (BK), which purchases health services from public sector organizations, like the MOH, SSK and University facilities.

(iv) Retired civil servants and their dependents, who are covered by Emekli Sandigi (ES), a government employees retirement fund. Financed by the contributions of active civil servants and the state, Emekli Sandigi contracts with MOH and University facilities for health services for its beneficiaries.

(v) A small group of people, mainly working for banks and insurance companies, whose expenditures are covered by private funds.

(vi) Rural and urban poor covered by the Green Card scheme.20

4.18 This leaves a big subgroup comprising almost 14 percent of Turkey's population that is not covered by any health insurance program other than the Green Card program. There is a provision under Code 3294 (Law on Incentive to Social Aid and Solidarity Fund) under which the state can pay for health expenditures of those are not insured, are unable to get a green card and cannot afford to pay for their health care, but the allocations and use of the Fund resources is far from adequate to address the healthcare needs of this population. Clearly, there is a pressing need for some insurance system that will expand coverage and extend health care to these 8.8 million people.21

Table 4.3 Health coverage of population, 1998 Coverage type Percentage of population SSK 51.5% Pension fund (ES) 13.0% BK 21.4%

19 Since 1983, civil servants are required to pay 20 percent of the prescription charges for medicines. 20 Enacted under Law 3816 of 1992, the Green Card Program provides free health care services to its beneficiaries. Managed by the Ministry of Health and financed via general taxes, the Green Card program is seen as a transitional solution that would prevail until the General Health Insurance System is brought into application. To qualify for the Green card, an individual should be a Turkish citizen, not be covered by any social security system, and have a monthly income of less than one-third of the minimum wage amount (excluding taxes and social security premiums) as determined by Code 1475. 21 At the time of writing, Bag-Kur has prepared an innovative program to cover all agricultural workers, including the landless, under its insurance scheme. Participation is voluntary, and it remains to be seen how effective the measure will be in application. 58

Private funds 0.5% Subtotal 86.4% Covered population Uncovered population 13.6% Total population 64.8 million Source: SPO

Example 23. Private insurance and equity

Vietnam – Managing Public Resources Better – Public Expenditure Review 2000, Report no. 24469, vol.2.

90. There are two main issues related to the health insurance program. First, the health insurance program primarily benefits the more affluent population in Vietnam, as enrollees in the compulsory health insurance scheme are civil servants and salaried employees of state and large private enterprises, most of whom are middle-class urban residents. This is borne out by Figure E.23, which shows that not only are health insurance coverage rates much smaller among the poorer consumption quintiles than among the richer quintiles, but that health insurance enrollees in the country are drawn disproportionately from the more affluent groups in society. For example, while only 8 percent of insurance enrollees are drawn from the poorest consumption quintile, as many as 37 percent are from the richest quintile. Thus, if the social health insurance program is to benefit the poor in Vietnam, there will need to be substantial expansion in coverage of the rural population by the voluntary insurance scheme.

91. One implication of this income bias in membership of the health insurance program is that a large portion of insurance reimbursements go toward treating diseases and ailments of affluence, such as heart disease, cancer and diabetes (Dunlop, 1999).22 Health 22 Actually, data on reimbursements by type of diseases treated are not reported routinely by the VHIA from its reimbursement records. However, the data reported by Dunlop (1999) were collected by the author 59 insurance reimbursements for cancer and heart disease treatments have lately averaged VND 1-3 million per case, while reimbursements for kidney dialysis have averaged VND 50-60 million per case (Dunlop, 1999). These diseases are not representative of the mortality or morbidity profile in the province.

Example 24. Entitlement, insurance and moral hazard

Vietnam – Managing Public Resources Better – Public Expenditure Review 2000, Report no. 24469, vol.2.

92. Second, the health insurance program has run into major financial problems in part because of uncontrolled growth of insurance reimbursements. In 1996 and 1997, many of the provincial health insurance funds were overdrawn in comparison to the paid premiums during that year. In 1996, nearly 25 percent of provinces had insurance expenditures exceeding insurance revenues, while in 1997 about 10 percent of the provinces ran a deficit (Dunlop, 1999). While the overall financial integrity of the insurance fund was not in jeopardy, such problems are likely to keep cropping up in the future. One reason for the increase in insurance expenditures is the existence of consumer and provider moral hazard.23 Table E.7 shows that, while the annual outpatient service contact rates of health insurance enrollees were not significantly different from those of the general population in 1997, the inpatient admission rate was significantly higher (by about 70 percent). In addition, both the outpatient contact rate and the inpatient admission rate have grown much more rapidly over the period 1993-97 for the insured than for The general population. Indeed, the inpatient admission rate for the general population has been more-or-less stagnant over the entire period, while that of the insured has grown by about 41 percent. The growth rate of inpatient admissions among the insured is probably unsustainable.

93. Table E.7 also indicates that the average expenditure on an inpatient admission for insurance program enrollees has been increasing much more rapidly than the average expenditure for an outpatient service contact. In 1993, the cost of an inpatient admission for an enrollee was about 9 times the cost of an outpatient service contact. By 1997, the ratio had increased to a factor of 12. The increase in inpatient expenditures probably reflects provider moral hazard, as hospitals respond to reduced public subsidies by increasing the number of billable and reimbursable services to insured patients.

94. In response to these financial difficulties, the provincial health insurance offices have been placed under the direct control of VHIA since 1999, and VHIA has tightened the guidelines regarding the control and use of funds. Several changes to the health insurance package have been implemented, including:

for Phu Tho province in 1998-99. 23 Moral hazard is the practice of excessive utilization of covered services by the insured. This can occur because (i) the insured demand more health services as the effective price of care for them is lower (consumer moral hazard), and (ii) providers deliver more services to the insured because they have a financial incentive to do so through the fee-for-service system (provider moral hazard). 60

 a 20 percent co-payment for those insured under the two schemes (with school children exempted), up to an annual maximum of six months' basic salary (about VND 860,000);  limitation on maternity benefits for only the first two children;  extension of waiting period (for benefits to commence) from one month to three months.

Table E7 Service and health expenditures of health insurance enrollees, 1993-97 Year Health Insurance program General population Health Program Enrollees enrollees insurance Average Average Ratio of Average Average Ratio of Average Average Ratio of number of number of inpatient number of number of inpatient exp. per exp. Per average exp. inpatient inpatient admissions outpatient inpatient admission outpatient inpatient Per admissions admissions to contacts admission s to contact admission outpatient per per outpatient per capita s per outpatient (000 (000 contact to enrollee enrollee contacts capita contacts VND) VND) average exp. per per capita Per inpatient enrollee admission 1993 0.48 0.07 0.14 0.84 0.09 0.11 11.67 103.06 8.83 1994 0.95 0.12 0.12 0.98 0.08 0.08 14.12 122.11 8.65 1995 0.97 0.11 0.11 1.18 0.07 0.06 17.63 169.27 9.60 1996 1.20 0.12 0.10 1.49 0.07 0.05 18.08 193.95 10.73 1997 1.36 0.12 0.09 1.52 0.07 0.05 18.56 215.71 11.62 Source: Dunlop, 1999

95. In addition, many provinces have placed their own rules to limit reimbursements. For example, Ben Tre has limited outpatient reimbursements to health facilities to 55 percent of the total facility charges, effectively raising the co-payment by individuals. The province has also placed a ceiling of VND 102,000 per case on reimbursements for inpatient care.

Example 25. The budget process

Vietnam – Managing Public Resources Better – Public Expenditure Review 2000, Report no. 24469, vol.2.

The Public Expenditure Decision-Making Process in Health

36. In Vietnam, the public health budget covers four levels of government: the central and the three levels of local governments: provincial, district and commune. Since some expenditures by communes from their own resources are not included in the State budget, comprehensive database which aggregates the four levels into a consolidated statement of sectoral expenditure is lacking. However, central, provincial and district governments account for about 90 percent of the national budget.

37. The Ministry of Health has a limited role in deciding the overall budgetary allocation to the health sector. It is the Ministry of Finance (MOF) and the Ministry of Planning and Investment (MPI) that play deciding roles in this process, albeit with some discussion 61 with the Ministry of Health. The overall level of recurrent spending on health is determined by projected growth rates of total revenue and of total recurrent expenditures (after subtracting estimated expenditures for new policies, such as increased salaries, etc.) and the share of health in total recurrent expenditures. This type of incremental budgeting pays little attention to health goals and health priorities set by the Ministry of Health.

38. The overall budgetary resources to the health sector are allocated on the basis of certain norms across two dimensions: functional (curative versus preventive) and sectoral (Ministry of Health, other sector ministries, and the provinces). These transfer norms are based either on the number of hospital beds belonging to a ministry (for curative expenditures) or on the population of the province (for both curative and preventive expenditures). These amounts are supposed to cover salaries for a centrally-determined number of health workers in the provinces, fees and drug exemptions for the poor, and other recurrent expenses. The allocation norms vary across five geographical regions, presumably to capture variations in the unit cost of delivering services in different geographical regions. In addition, Hanoi and Ho Chi Minh City have a norm that exceeds those of other cities by 50 percent. In general, the norm for curative expenditures is about 3 times as high as that for preventive expenditures, reflecting the higher unit cost of curative interventions.

39. In addition to central government-disbursed funds, the provinces, especially the richer ones, can (and do) spend on health out of their own revenues.

40. Interestingly, however, in the annual budget allocation document sent to provinces, by Ministry of Finance only an aggregate recurrent budget is provided -- not its composition across sectors.24 Thus, provinces are at liberty to make their own spending decision on each sector, including health. The provinces also can spend on health out of their own revenues.25 The financing of health expenditure out of provincial revenues is largely what causes large disparities in government health expenditure per capita across provinces, as richer provinces are able to spend more than poorer provinces.

41. Although, in principle, provinces have to inform the Ministry of Finance about how they allocate their aggregate recurrent budget, this is an ex post formality. It is not at all clear how strongly, if at all, the Ministry of Finance can influence provincial spending on health out of the aggregate budget already approved for the provinces. The Ministry of Health has an even smaller role in this process. Although current regulations require provinces to notify the Ministry of Health of their budget allocation for health in practice this is not observed . Provinces are not required to provide the Ministry of Health with information on actual health expenditures or on results achieved. As a consequence, the Ministry of Health has very little or almost no information on provincial health budgets and expenditures.

24 In fact, provinces are provided with two minimum amounts for spending on education and science and technology and a maximum spending on administration. Other than these, no sectoral breakdowns are provided. 25 A province can retain revenue that it collects in excess of its assigned target. It can choose to spend this retained revenue in any way it wishes, including on health. Likewise, provinces may have to reduce spending on health if their revenue performance fall short of the assigned target. 62

42. In addition to the budgeted amounts for health, the provinces receive funds from the various national health programs (such as the Expanded Program of Immunization, the Tuberculosis Program, etc.). Until this year (2000), specific amounts were earmarked for each program, and provinces had no authority to reallocate funds across programs. This resulted in under utilization of some program funds and excess demand for other program funds in the same province. Starting with FY 2000, a new mechanism of fund allocation has been introduced whereby provinces are given block grants for all targets of the national health program and have discretion in allocating the funds across the different programs.

43. In principle, the national health programs should serve as vehicles for providing additional funds for disadvantaged areas and addressing various disease patterns specific to each province. However a recent study by Ministry of Health experts found that the allocation of program funds is more-or-less equally distributed across provinces and does not take account of the specific disease profile of a province.26 This suggests such expenditure may not be well targeted.

44. Districts receive most of their support from the provinces, although they have some own revenue sources. Provinces have almost complete discretion in the resources they pass on to district health services. While some provinces split their allocations evenly (i.e. 50 percent for the province and 50 percent for all the districts in the province), other provinces may spend two-thirds of their allocation on nationally-determined staff salaries, which leaves little room for discretion. This results in even greater variation in per capita resource available at the district level, than that existing at the provincial level.

45. Before 1995 the communes were expected to be basically self-sufficient, on health expenditures, and these were not included in the integrated state budget. Since 1995, however, the government has taken over the payment of salaries to approved staff of communal health centers out of the state (provincial) budget. Virtually all communes supplement their small budgets through the sale of drugs. They buy drugs from the province or district pharmacy and resell it with a markup (typically between I and 10 percent). In richer communes households are sometimes assessed for contributions to defray the operational or capital expenses of the commune health center. Thus commune health centers in richer communes typically offer better-quality services than those in poorer communes. However households' willingness to make general contributions for the upkeep of the commune health center is limited and these taxes are not very popular. (See discussion of Fees and Contributions in Chapter 2 and in Annex C)

Example 26. Efficiency of service provision (2)

Georgia Public Expenditure Review. Report No. 22913-GE, November 25, 2002

Efficiency of Public Expenditures

26 Draft Report, Evaluation of Current Status and Impact of some National Programs. December 1999 - National Health Support Project, Sub-component 3. 1.2). 63

5.28 Public expenditures on health care do not necessarily meet allocative efficiency criteria. While originally designed to make the most effective use of limited public funds for health care, the BBP has grown and currently covers programs which do not meet the public or merit good criteria. Overall, the public expenditures envelope for health is small, and these small amounts are spread thinly over many programs. Moreover, prioritization between competing programs and needs is ad-hoc and often does not correspond with the criteria of ensuring the highest positive externalities. The net result is that maximum efficacy gains for the society are not realized.

5.29 Georgia inherited a health system characterized by misallocation of resources, underutilization of capacity in some areas and under-supply in others, and shortage of drugs and medical supplies. Since then, several waves of health sector reforms have focused on downsizing, restructuring and privatization of health facilities, especially of hospitals. Despite these attempts, Georgia still has a high number of specialized hospitals, and a bed-to-population ratio of 4.1:1,000 population, significantly higher than ratios in the much better-off developed countries. A bigger problem exists with the reorganization of medical personnel. The number of physicians in Georgia has hardly declined over the years and in 1999, Georgia had 21,520 physicians which translates into 3.9 physicians per 1,000 population, which is higher than the CEE and FSU averages of 2.5 and 3.7 respectively. At the same time, the number of nurses in Georgia is low, with 1.4 nurses per physician compared with 2.4 in the Central and Eastern Europe (CEE) and 2.1 in FSU.

5.30 Efficiency in public spending can also be examined by looking at technical efficiency measures such as the average length of hospital stay, staffing ratios, prescription of drugs, mix of factors of productions, etc. The average length of hospital stay (ALOS) in Georgia is 10.6 days, which compares favorably with other countries in the region, and is slightly below the ECA and low and middle income country averages of 14 and 12 days respectively.27 However, hospital occupancy rates in Georgia are very low (31 percent), indicative of both low levels of utilization as well as excessive number of hospital beds. There is also the severe problem as a result of excessive staffing, as is evident from the ratio of 1.5 physicians per occupied bed. Moreover, the majority of doctors are specialists, which contributes to increasing costs in the production and delivery of health care.28

Example 27. Efficiency and impact of health services

Public Expenditures for Poverty Alleviation in Northeast Brazil: Promoting Growth and Improving Services. Report No. 22425-BR, June 11, 2001.

27 World Development Indicators, 2001. 28 Zoidze A, Gzirishvili D, Gotsadze G. July 1999. Hospital Financing Study for Georgia. Small Applied Research 4, Bethesda, MD: Partnership for Health Reform Project, Abt Associates Inc. 64

3.19 As for public medical inputs, the results are discouraging. Figure 14b shows the distribution of coefficients on the numbers of public sector doctors and nurses as well as on the training of nurses. Here the results are striking. There are considerably more perverse effects that are significant than effects that go in the "right" direction. Sixteen coefficients are significantly negative and only seven are positive. Furthermore, of the ones that are positive, most are related to the training of nurses and not the coverage of the population of either doctors or nurses. It is possible that this reflects the importance of maintaining quality in public facilities or the contribution of relatively less well-trained professionals (nurses vs. doctors), as long as standards are met. It is also possible that this merely reflects higher earning capabilities or general educational attainment for women in the district which is reflected in higher wages and higher standards for employment. In any case, it is impossible to infer any effect of publicly supplied medical inputs on infant mortality.

3.20 Finally, a recent study of water and sanitation was done by the World Bank (1998b). This also examined census data for 1990 for four states: Minas Gerais, Pernambuco, Rio de Janeiro, and Sao Paulo. The relevant results are presented in table 4. Once again, water and sanitation appear highly significant. The results are so consistent across studies with differing data sets, researchers, and methodologies that it is hard to believe they are spurious.

Table 4 Impact of safe water and sanitation on mortality rates Mortality Infant Less than 5 years old Average morality rate 39.4 8.80

Change due to a 10 percentage rise in Urban access to piped water 0.80 0.25 65

Urban access to sewers 0.60 0.15 Source: World Bank (1998).

3.21 What are the policy implications of the results on water and sanitation? The jump from empirical findings and policy is not straightforward. Figure 11 and figure 12 directly imply a pair of policies based on distributional concerns alone. The government has in its control two instruments: the extent of the distribution system for water and sanitation via direct investments-the quantity dimension-as well as the proportion of costs that are subsidized by charging fees, which are too low-the price dimension. Since current possession of water and sanitation services are closely related to income, or, at least, the poorest do not benefit much from them at all, a progressive policy option is to cut current subsidies by raising fees. "Average" incidence of the policy of subsidization is not good. Very little of these subsidies reach the poor so there is not much of an equity benefit from the policy and they tend to distort the use of water, encourage wasteful practices, and thus compromise efficiency. On the other hand, since all of the non-poor already have sanitation facilities and good sources of water, any expansion of the system to those not currently enjoying these services will be progressive. The "marginal" incidence of expansion is very pro-poor. So, for any level of subsidy chosen, expansion is progressive and for any level of coverage (up to the point of universal coverage) raising fees is progressive as well.

3.22 This leads to an unambiguous answer to the question of whether the pursuit of better health for the poor via extension of services would be terribly expensive. With appropriate pricing schemes, extending the system to include the poor may have little or no fiscal impact at all. The World Bank study of Brazil cited above, as well as many other studies around the world, have found very high willingness-to-pay by people for good, reliable drinking water as well as sanitation systems that remove wastes from the house. The fees that people appear willing to pay are almost always higher than average costs of provision in urban areas and certainly higher than marginal costs. To the extent that very poor people in urban areas may not be willing to cover all costs, the amount of subsidy needed to reach them may be very small indeed. Neither water nor sanitation are public goods though they may have externalities justifying small subsidies. On the other hand, the private sector has not been much of an actor in the sector. Why this is so may be worthy of further study but given that it is so, this presents an opportunity for the public sector to improve efficiency and the health of the poor at the same time. It simply does not necessarily require large outlays of money.

3.23 The results of the analysis of the PNDS data show the actual, physical source of water to be the important factor. Depending on the mechanism by which better water leads to better health, the types of policies appropriate to rural areas will differ. Increasing coverage will not necessarily entail hook-ups to houses for waste disposal. Many of the policies in these cases may rely instead on public information and health education campaigns. Promoting basic hygienic practice (not something the private sector does) is a standard public health role which may have great payoffs in terms of improved health of the poor. 66

3.24 Turning to medical care, why is it so difficult to find a correlation between measures of access to publicly supplied medical care and health status? There are several possible explanations. Appendix 4b, table B4.4 presents the results of an analysis of the PPV data in which examines people's likelihood of seeking care once sick in either the public or the private sector. Several conclusions are worth noting. First, figure 15 shows that the likelihood of seeking care is closely related to the type of illness experienced. People find a way to handle relatively serious illnesses such as accidents and heart attacks, but may not bother to seek medical attention for colds or general pains. This might help explain why it is hard to find an effect of medical care on such gross measures of health status as mortality rates. When people are faced with life threatening situations, they find some way to cope: so it would only be in the less severe cases or in financial protection that we might see the benefit of medical services, not in mortality.

Example 28. Recommendations, priorities and evidence

Vietnam – Managing Public Resources Better – Public Expenditure Review 2000, Report no. 24469, vol.2.

A Menu of Reform Options

There are several major issues emanating from the above analysis

96. At about US$6 per capita per annum, government health expenditure per capita is about one-half of the US$12 recommended for low-income countries for implementation of a basic package of health services. To achieve this level of spending, Vietnam would need to spend an additional $468 million annually on health. While increased levels of user fees and expansion of health insurance could meet some of this increase (see discussion below), the bulk of the increase would have to come from the state health budget. Currently, households account for nearly four-fifths of total spending on health; an increase in public spending would also serve to reduce the financial burden of medical expenditures on households, especially those that are poor.

97. The efficiency and equity of public health spending are as important as its quantity. The current pattern of public spending on health is inefficient as public goods like preventive

NB – paragraphs 97-103 omitted from PDF

103. …persons represents less than 4 percent of the population, which implies that the scheme would cover only a small fraction of the poor in the country.

104. As soon as a program for assisting the poor with their health-care costs is in place, the government should consider raising hospital user fees; since this will allow the government to reduce its subsidies to the hospital sector. As noted earlier, reduction of subsidies to hospitals, especially at the secondary and tertiary level, will not only increase 67 the efficiency and cost-effectiveness of public health spending, but will also improve the equity of these subsidies. While user fees are onerous for some users, such as the poor and people living in remote mountainous regions, they are not unusually high for most of the rest of the Vietnamese population. (For instance, a single hospital visit constitutes only 4-5 percent of annual nonfood expenditure per capita for a person in the highest income consumption quintile). There is evidence that because user fees have not been revised since 1995 and are perceived to be too low, hospitals have increased their gross revenue from user fees by increasing the number of diagnostic tests and procedures performed per patient and by increasing the average length of hospital stays for inpatients. From a societal perspective, this is a wasteful method of raising finance.

105. Another means of raising revenues for the health sector will be through the expansion of the health insurance program, especially the voluntary scheme. While the health insurance program has expanded very rapidly in the last five years, starting from a zero base, the compulsory scheme within the program now covers over three-quarters of the target population (i.e. civil servants and salaried employees of large private and state- owned companies). Since the voluntary scheme covers only 5 percent of the target population, future increases in health insurance coverage will have to necessarily come from this population.

106. There is also a need to use public spending as a policy tool for reducing interprovincial disparities in private health-care spending and access to health care. Currently, government health expenditure reinforces--rather than compensates for-- provincial disparities in private health spending. Some inter-provincial disparity in government health expenditure is unavoidable, since a portion of provincial health spending is financed by provincial governments out of their own revenues, which tend to be strongly correlated with provincial incomes. But, in principle, it should be possible for the central government to reduce this disparity by adjusting the formula by which the health budget that falls under provincial control is allocated to the provinces by the Center. Unfortunately, this budget is allocated by the provinces on the basis of norms such as the number of hospital beds in the province (for curative expenditures) and on the population of the province (for preventive expenditures). These norms do little to redress the inherent inequality in provincial health spending per capita. It may be worthwhile for the government to reexamine the formula used by the government in allocation the state health budget across provinces. The formula should incorporate additional factors, such as the level of private health spending per capita, incidence of poverty, and the level of health infrastructure in a province, so that disadvantaged and under-served provinces can receive disproportionately larger allocations of the state health budget.

107. In the longer run (next 5-10 years), the government should rethink its roles in the health sector. It may make more sense for the government to reduce its role in the direct provision of curative, hospital-based health services, and focus on (i) provision of high- quality preventive health services that have a clear public goods character, (ii) greater financing of health expenditures, especially for the poor, and (ii) improving regulation and monitoring of the large and growing private sector in health. 68

108. Finally, the information base for analyzing the functional and other composition of government health expenditures and the incidence of public health spending is weak. The current budget classification system does not permit disaggregation of government health spending by the level of service (i.e. primary, secondary and tertiary)--something that is essential for determining the benefit incidence of public spending. While health budget data are classified into broad functional categories of preventive and curative expenditures there is not full agreement among government agencies as to what individual items of expenditure are aggregated into these broad categories. For instance, until recently, one of the national programs included under preventive expenditures involved upgrading equipment at public hospitals.

109. In addition to the weaknesses in budgetary data collection, information on health service utilization and health status of different economic groups in the population has been seriously lacking until the release last year of the VLSS 1998 data. The government has in the past emphasized the routine collection of facility-based (administrative) data on utilization of health services. While such data are important, they do not permit analysis of income (and other) differences in utilization of health services. This makes it impossible to know which groups in the population are benefiting from government programs and facilities. Regular household surveys are often the only means of obtaining such information. The forthcoming National Health Survey--the first nationally- representative multi-purpose health survey planned by the Ministry of Health in 2001- will go a long way toward meeting this data gap. However, it is essential for the Ministry of Health to establish a regular system of household surveys to monitor the impact of its programs and its spending on different socioeconomic groups.

Example 29. Recommendations for reform

Democratic and Popular Republic of Algeria, Public Expenditure Review of the Social Sector. Report No. 22591-AL, January 2002.

D. POLICY RECOMMENDATIONS

D.1 The Current Context for Reform

50. The MOHP and CNAS have developed plans for an important series of reforms that bear directly on the analysis in this study. These reforms reflect many of the themes discussed in the Health Sector Overview and Strategy Note (June 2000) and are in accordance with the recommendations in the Note. Specifically, the MOHP and CNAS are either in the process of, or plan to take measures to do the following: a) Develop a health sector master plan, to include the growing private sector. A master plan would determine the optimal size and functions of the health care delivery system over the next decade - helping to rationalize the delivery system for the future needs of the Algerian population. This exercise includes public and private physical structures, equipment, and most importantly health care personnel. The MOHP has 69

already begun the planning of a comprehensive master plan (carte sanitaire). The components of a master plan at the secteur sanitaire level could be the following:

1) Review existing norms and standards for health services.

2) Develop revised norms and standards for health services in accordance with the Government's health sector strategy.

3) Develop a model for number of beds and specialties based on demographic and epidemiologic data.

4) Develop planning criteria for physical infrastructure by type of facility.

5) Develop a model for work force requirements by type of facility and beds.

6) Develop a model for estimating recurrent cost requirements by type of facility and beds. b) Revise existing treatment norms for health facilities and personnel, as a function of a rational hierarchy of care based on the level of the facility. c) Develop and implement a contracting system between the MOHP and CNAS, so that the MOHP can submit bills by medical act and by insured patient, instead of the current lump-sum transfer. d) Revise the reimbursement schedule for private health care - this schedule has not been updated since 1987 and is considerably below market prices. e) Maintain and reinforce access to a package of essential health care services – including preventive health care, contraception, and the most cost-effective curative services. f) Conduct an analysis of the pharmaceutical sector, rationalize drug consumption and reduce the rate of pharmaceutical inflation. Analysis of Spending on Pharmaceuticals. The MOHP is spending 16 percent of its recurrent health expenditures on drugs, and the absolute level of spending on drugs has almost doubled over the past 10 years. Despite the large amount spent on pharmaceuticals, shortages of drugs are reported from all parts of the country at all levels of service delivery. As a first step to taking measures to rationalize drug consumption in Algeria, we strongly recommend an analysis of the pharmaceutical sector in the country, including spending, regulation, distribution, quality control, use of generics, and physician and consumer education. g) Carry out a National Health Accounts (NHA) study. NHA provide information on the total (public and private) resources going into the health sector, their sources, and the types of services funded. There is currently not enough information on health care financing in Algeria to make comprehensive policy recommendations. The Government sent a team to a regional seminar sponsored and organized by the World 70

Health Organization, the World Bank, and USAID in Bahrain in March 2001. It is important that the individuals involved in the study come from different stakeholders within the health sector, including the MOHP, the MOF, and the Central Statistical Office. NHA are not a one-time exercise, but rather should be institutionalized and updated every year - facilitating the rapid policy making for the health sector.

51. The World Bank strongly endorses these proposed reforms. They would make the Algerian health system both more efficient and effective. The envisioned reforms respond directly to the priority problems noted in this document. Nonetheless, implementation is likely to be a complex process, necessitating political and social consensus. A step-by- step approach to the reforms, accompanied by social dialogue, is most likely to succeed. The reforms - particularly a system of contracting - will also require the use of a number of specific approaches, including:

 Accurate health information systems.

 Improved management capacity.

 Calculation of the real unit costs of delivery for specific health services. Unit cost studies are essential in order to establish a realistic schedule of reimbursement rates under a contracting system. Practical methodologies - including Activity-Based Costing are available for this type of study.29

 A provider payment mechanism adapted to Algerian needs. Possible methods include global budgets, line-item budgets, capitation, Diagnostic-Related Groups, and fee-for- service - or some combination of these. Each has different incentives for providers in terms of costs and service provision, and each has different requirements in terms of information (see Table C4 in Statistical Annex). To establish an efficient mechanism is a challenge, even in the most sophisticated OECD countries. Nevertheless, there is clear evidence of success (5 to 20 percent savings) in systems that: (i) separate provision of services from financing; (ii) have money follow the patient concepts; and (iii) use incentive-based provider payment mechanisms that encourage efficiency, clinical effectiveness, quality, access, and consumer satisfaction.

Other references:

Other suggested general PERs

Florencia Castro-Leal, Florencia. 1999. “The Impact of Public Health Spending on Poverty and Inequality in South Africa.” World Bank, Country Department I, Africa Region., 1999.World Bank, Washington, D.C.

29 A working group in the MOHP has already conducted one such study, entitled "Système de calcul des coûts par la méthode des sections homogènes. 71

Oscar F. Picazo, Oscar F.. 2002. “Better Health Outcomes from Limited Resources: Focusing on Priority Services in Malawi.” Africa Region Human Development Working Paper. Series, April 2002 World Bank, Washington, D.C.

World Bank. 2001. “Zambia Public Expenditure Review – Public Expenditure, Growth and Poverty: A Synthesis.” Report No. 22543-ZA,. World Bank, Washington, D.C. December 2001.

Institutions, governance, and the general budget process

World Bank. 2003. “Cambodia: Enhancing Service Delivery through Improved Resource Allocation and Institutional Reform, Integrated Fiduciary Assessment and Public Expenditure Review.” Report No. 25611-KH, September 8, 2003.. World Bank, Washington D.C.

World Bank. 2003. “The Republic of Uganda Public Expenditure Review 2003: Supporting Budget Reforms at the Central and Local Government Levels.” Report 27135-UG. World Bank, Washington, D.C.

World Bank. “Poland: Towards Fiscal Stability and Growth – A Public Expenditure and Institutional Review.” Report No. 25033-POL, ND. World Bank, Washington, D.C..

World Bank. The Republic of Uganda Public Expenditure Review 2003: Supporting Budget Reforms at the Central and Local Government Levels. Report No. 27135-UG, September 2003.

Expenditures in the context of PRSP

World Bank. 2003. “United Republic of Tanzania Public Expenditure Review FY03: Managing Public Expenditures for Poverty Reduction – Report on Fiscal Developments and Public Expenditure Management Issues.” Report No. 26807- TA,. June 2003.World Bank, Washington, D.C.

Decentralization and state-level analysis

World Bank. 2003. “Decentralizing Indonesia: A Regional Public Expenditure Review, Overview Report.” Report No. 26191-IND., World Bank, Washington, D.C.June 2003.

World Bank. 2003. “Mexico State-Level Public Expenditure Review: The Case of Veracruz-Llave.” Report No. 25162-ME, October 8, 2003. World Bank, Washington, D.C.

About Public Expenditures 72

Sanjay Pradhan, Sanjay. 1996. “Evaluating Public Spending,” World Bank Discussion Papers, No. 323., World Bank, Washington, D.C1996.

Gurushri Swamy, Gurushri and Richard Ketley. 1999. “Review of Public Expenditures: Efficiency and Poverty Focus.” World Bank, Washington, D.C.Country Department I, Africa Region, 1999.

AuPhil Swanson, AuPhil and Leiv Lundethors. 2003. “Public Environmental Expenditure Reviews (PEERS): Experience and Emerging Practice.” Environment Strategy Papers No. 7, May 2003.

Dominique van de Walle, Dominique and Kimberly Nead, (eds). 1995. Public Spending and the Poor. Baltimore: Johns Hopkins University Press., 1995.

World Bank. 1998. “The Impact of Public Expenditure Reviews: An Evaluation.” Report No. 18573. World Bank, Washington, D.C.November 13, 1998.

National Health Accounts

Poullier, J.-P., P. Hernandez, and K. Kawabata. 2002. “National Health Accounts: Concepts, Data Sources, and Methodology.” Evidence and Information for Policy Cluster Discussion Paper WHO/EIP/02.47. World Health Organization, Geneva, WHO, 2002..

WHO World Health Organization ( World Health Organization WHO). 2003. Guide to Producing National Health Accounts. Geneva: WHO, WHO, 2003.

FUNSALUD / SIDALAC Fundación Mexicana para la Salud / Regional AIDS Initiative for Latin America and the Caribbean (Fundación Mexicana para la Salud / Regional AIDS Initiative for Latin America and the CaribbeanFUNSALUD / SIDALAC). 2001. Technical Handbook for Estimating the National Accounts on HIV/AIDS. México DF, FUNSALUD, 2001.