Health Expenditure and Finance in South Africa¶
Di McIntyre; Gerald Bloom; Jane Doherty; Prem Brijlal (1995). Health Expenditure and Finance in South Africa. National Health Expenditure Review. Health Systems Trust and the World Bank.
| Type | report |
| Scope | Wider apartheid literature |
| Method | policy analysis |
| Period studied | 1983–1995 |
| Regions | South Africa, former homelands (TBVC and self-governing territories), nine new provinces, magisterial districts |
| Themes | inequality poverty unemployment, macro fiscal finance debt, transition post apartheid, bantustans removals influx, demography urbanisation, business and state, methods data sources |
| Citation key | mcintyre1995 |
Argument¶
The Health Expenditure Review maps total health financing and spending in 1992/93, just before and during the democratic transition. It argues that South Africa spends a high share of GDP (8.5%) on health, yet resources are grossly maldistributed between the private and public sectors, between hospital and primary care, and between rich and poor districts. This distribution is a legacy of apartheid's racially fragmented administration. The report shows that the public sector's room to fund extension of primary care is tightly constrained by slow growth, government deficits and debt service. It estimates an additional R1.5-2.5 billion a year is needed for primary care, and sets out options for financing it through hospital efficiency, user fees, subsidy reform and donor money. It deliberately offers no policy prescriptions.
Evidence and method¶
ReHMIS facility-level survey of every public health facility (expenditure, staff, beds, workload); commissioned technical reports (private sector review, research, training, capital projects, donor funding, hospital expenditure trends); Registrar of Medical Schemes data; SALDRU/PSLSD 1993/94 household survey; census and DBSA population data; SARB and CSS national accounts; magisterial districts ranked in quintiles by 1985 per capita income; cost studies of clinics and hospitals.
Place in the debates¶
- Public versus private roles in financing health (social health insurance, tax subsidies to medical schemes) (see Living standards, health and social policy). Neutral but critical: documents rapid medical scheme cost escalation and regulatory gaps, treats the tax subsidy and civil-servant contributions as problems, and flags social health insurance as an option needing cost control.
- Reallocation from hospitals to primary care under fiscal constraint (see Living standards, health and social policy). Argues the extension of primary care must be financed largely by savings and efficiency in tertiary and academic hospitals, plus gap funding from government and donors.
- Role of user fees in public health financing (see Living standards, health and social policy). Sees limited scope to raise fees on the poor without harming access, but greater potential from private and insured patients in specialist hospitals.
Findings¶
-
Total health spending was high by international standards, but the poor receive vastly inferior care. (p. 11)
South Africa spent over 30 billion rands on health services in 1992/93. This amounted to 8.5 percent of its Gross Domestic Product (GDP).
-
Private health spending accounted for most expenditure while serving a small minority of the population. (p. 11)
only 17 percent of the population are members of a medical scheme and only 23 percent use private sector health services on a regular basis
-
Public health resources are strongly skewed toward the richest districts, which are mainly White. (p. 37)
Only 26.8 percent of the public sector health budget was spent directly on the poorest 40 percent of magisterial districts, which contain approximately 49 percent of the population
-
Per-capita public health spending rose to more than three times the poorest-quintile level in the richest districts. (p. 93)
Public sector health expenditure is lower in the poorer districts, averaging 122 rands in Q1 and 175 rands in Q2, compared with 437 rands in Q5
-
Medical aid members used health services costing about 15 times as much per person as public services in the poorest districts. (p. 40)
The ratio of spending on health services for members of medical aid schemes to the average spending by the public health services in the poorest 75 districts was almost 15:1.
-
Public health spending was hospital-centred, with tertiary and academic hospitals taking the largest share. (p. 64)
approximately 76 percent of the total was spent on acute care hospitals, 5 percent on chronic hospitals, 11 percent on non-hospital primary health care and 8 percent on other services
-
Real public health spending grew only slightly per capita in the decade to 1992/93, while inflation dominated nominal growth. (p. 61)
Recurrent expenditure on public health services increased at an annual rate of 18.3 percent a year between 1983/84 and 1992/93, from R2.5 billion to R11.1 billion
-
Weak growth and rising deficits constrain fiscal room for health spending. (p. 28)
the budget deficit was 27.4 billion rands in 1992/93, or 7.8 percent of GDP
-
Medical scheme contributions grew far faster than salaries, doubling their share of pay in a decade. (p. 52)
Medical scheme contributions were equivalent to 7.1 percent of average formal sector salaries in 1982. Ten years later, in 1992, they amounted to 15.2 percent of average salaries.
-
Extending basic primary care will cost an additional R1.5-2.5 billion annually. (p. 98)
recurrent spending on non-hospital primary level services will have to increase by at least 1.5 billion rands, and possibly by as much as 2.5 billion rands
-
Poverty and inequality were extreme; the report cites high Gini estimates and a very high income share for the richest decile. (p. 21)
51 percent of annual income goes to the richest 10 percent of households while under 4 percent goes to the poorest 40 percent
-
Per capita GDP stagnated in the 1980s and formal employment fell after 1986. (p. 21)
Since the early 1980s its GDP increased by less than 1.5 percent a year while its population grew by 2.5 percent
Evidence ledger¶
Sources used¶
| Source | Kind | Period | How used |
|---|---|---|---|
| ReHMIS (Regional Health Management Information System) national facility survey, 1992/93 | author survey or fieldwork | 1992/93 | Principal source for public-sector expenditure, beds, personnel, utilisation and fee revenue by facility, level of care, province and magisterial district. |
| Valentine and McIntyre (1994) private sector health expenditure review (HER Technical Paper No. 9) | author survey or fieldwork | 1992/93 | Provides private-sector financing, medical scheme, insurance, industry and out-of-pocket expenditure estimates and population coverage. |
| Registrar of Medical Schemes (1993) annual report, year ending 31 December 1993 | official statistical series | 1982-1992 | Medical scheme membership by race, expenditure by service, contributions per member and number of dispensing doctors. |
| South African Reserve Bank estimates of private consumption expenditure on medical goods and services (1991, 1994) | national accounts macro | 1983/84-1992/93 | Routine private health spending series compared with Valentine and McIntyre to show underestimation by 70 percent. |
| Central Statistical Service (1993c) Population Census 1991: Geographic distribution of the Population | population census | 1991 | Urban/rural population distribution and magisterial-district population base. |
| Development Bank of Southern Africa (1994) South Africa's nine provinces: a human development profile; census-adjusted population by magisterial district | secondary literature compilation | 1993-1994 | Provincial population, GDP contribution, income per capita, HDI, and SAMDC personnel counts; undercount-adjusted district populations. |
| van Wyk (1989) Personal Income of the RSA TBVC Countries by Population Group and Magisterial District, Report No. 183 | official statistical series | 1985 | Ranks magisterial districts into income quintiles, the main unit of analysis of inequality. |
| Project for Statistics on Living Standards and Development (1994) household survey 1993/94 (SALDRU), preliminary analysis | household or labour survey microdata | 1993/94 | Profile of poverty, health provider use by income quintile, reasons for not consulting, and spending of poorest quintile. |
| Bourne (1994) analysis of 1990 mortality data (from CSS death registrations) | secondary literature compilation | 1990 | Potential years of life lost by cause and share of infectious/ill-defined deaths by district quintile. |
| DNHPD (1990, 1994) Health Trends in South Africa and nutrition committee report | official statistical series | 1989-1993 | Infant mortality, maternal mortality, TB, malnutrition differentials between population groups. |
| Chetty (1994) integrated analysis of health facilities in the nine provinces (HEU Working Paper 13) | secondary literature compilation | 1992-1993 | Hospital beds (public and private), facility counts, and clinics required. |
| Rispel and Behr (1992) Health indicators: policy implications; Kaiser Foundation (1991) | secondary literature compilation | 1989/90 | Distribution of health personnel between public and private sectors. |
| Blecher and McIntyre (1994); Bunting (1994); Deloitte and Touche (1994a, 1994b); Price and Broekmann (1994) HER technical reports | commission or official report | 1991-1995 | Research and training spending, donor funding survey, planned capital projects, and provincial hospital expenditure trends. |
| Deloitte and Touche (1994b) questionnaire survey of donors | author survey or fieldwork | 1993-94 | Estimates donor health funding of at least 145 million rands. |
| Deloitte and Touche (1994a) documentation of capital projects 1993/94-1995/96 | author survey or fieldwork | 1993/94-1995/96 | Planned capital costs by facility type and IDT/CEAS clinic commitments by province. |
| McIntyre (1993) HEU Working Paper 4 and McIntyre and Owen (1994) budget commentary | secondary literature compilation | 1983/84-1994/95 | Public recurrent health expenditure trends, sources of public finance and share of government budget. |
| Central Statistical Service (1993a) South African Statistics 1993 (CPI, salaries) and CSS 1992 | official statistical series | 1982-1993 | CPI deflators and average salaries for real expenditure and contribution-to-salary ratios. |
| Fallon and da Silva (1994) World Bank, South Africa: Economic Performance and Policies; Du Toit and Falkena (1994) | secondary literature compilation | 1978-1993 | GDP, employment, government revenue/expenditure growth, inequality and poverty estimates. |
| Urban Foundation (1991) Population Trends: Demographic Projection Model | secondary literature compilation | 1991 | Population distribution by race and economic region. |
| World Bank (1993, 1994) World Development Report and Better Health in Africa; Barnum and Kutzin (1993) | international trade or finance data | 1990-1992 | International comparators for health status, spending, beds and physicians. |
| Uniform fee structure report: South Africa (1993) Final report of the committee of investigation into a uniform fees structure; McIntyre (1994a) | commission or official report | 1992/93-1994/95 | Fee schedules for public hospitals and clinics and their comparison with private tariffs (RAMS scale of benefits). |
| Steinmetz Commission Report (South Africa 1993) and Melamet Commission Report (South Africa 1994) | commission or official report | 1993-1994 | Evidence of drug shrinkage in hospitals (500 million rands) and solvency problems of medical schemes. |
| Medical Schemes Act 72 of 1967 and Medical Schemes Amendment Act 23 of 1993 | legislation gazette | 1967-1994 | Regulatory context for medical schemes and effects of reform on risk-rating and competition. |
| Unit-cost and primary-care costing studies (Lombard et al 1991; Broomberg et al 1992; van den Heever and Price 1994; Zwarenstein et al 1994; Beattie and van den Heever 1995; Price 1994 KaNgwane; Broomberg et al 1994 Ciskei) | secondary literature compilation | 1990-1995 | Cost ratios and unit costs used to estimate costs of extending primary care and to allocate homeland primary-care expenditure. |
| Kistnasamy personal communication; head-office expenditure records of former provinces/homelands; Engelhardt (1994) Hospital and Nursing Yearbook | interviews oral history | 1992/93-1994 | Defence, police and correctional services health data; validation of ReHMIS; private for-profit hospital ownership. |
New data¶
The Health Expenditure Review assembled ReHMIS, a facility-level database covering every public-sector health facility in 1992/93 (beds, staff, workload, expenditure, fees), aggregated by province and magisterial-district income quintile. It was supplemented by commissioned technical studies of private sector finance, donors, capital projects, research and training. The result is the first full health expenditure accounts of R30.15 billion for 1992/93.
Inference¶
Quantification: Descriptive tables.
Descriptive accounting of financial flows (national health accounts by source and use) and comparison across provinces, magisterial-district income quintiles, levels of care and public/private sectors for 1992/93. International benchmarking and cost projections from cited costing studies support the estimate of additional primary care spending; there is no causal or regression analysis.
Key claims¶
| Claim | Basis | Strength |
|---|---|---|
| South Africa spent about R30 billion (8.5% of GDP) on health in 1992/93, with ~58-61% from private sources and medical schemes the largest single funder. (p. 19) | Compiled from ReHMIS, Valentine and McIntyre private sector review, Blecher/McIntyre, Bunting, Deloitte donor survey and departmental data (Tables 3.5, C.1). | Partial data |
| Public resources are strongly skewed toward curative hospital care, with academic and tertiary hospitals taking 44% of recurrent spending and non-hospital primary care only 11%. (p. 47) | ReHMIS facility-level expenditure by level of care (Figure 5.3, Table 6.3). | Systematic data |
| Public health resources are inequitably distributed geographically: spending per capita in the richest districts is about 3.6 times that in the poorest, with far more doctors, nurses and inspectors. (p. xi) | ReHMIS data sorted by magisterial-district income quintiles using van Wyk 1985 income, DBSA populations (Tables 5.4-5.7, 7.5). | Systematic data |
| Members of medical schemes use health services costing about 15 times as much per person as public services in the poorest districts; the poorest 40% of districts received only 26.8% of public health spending. (p. 22) | Valentine and McIntyre contribution per beneficiary (R1,800) vs ReHMIS public spending in poorest districts (R122); assumes contributions approximate expenditure. | Partial data |
| Medical scheme costs rose faster than inflation and salaries; contributions rose from 7.1% of average salary in 1982 to 15.2% in 1992, with spending on drugs and private hospitals growing fastest, and private for-profit hospital beds nearly doubling 1988-93. (p. 34) | Registrar of Medical Schemes data deflated by CSS CPI (Table 4.8, Fig 4.1); Chetty (1994) bed data (Table 4.6). | Systematic data |
| Public recurrent health spending rose only about 3% a year in real terms (0.5% per capita) between 1983/84 and 1992/93, with real provincial hospital expenditure falling 8.4% from 1984/85 to 1990/91. (p. 44) | McIntyre (1993) and ReHMIS deflated by CPI (Table 5.1); Price and Broekmann (1994) (Table 6.8). | Partial data |
| An additional R1.5-2.5 billion a year will be needed for public primary care services to reach key preventive programmes and essential basic care. (p. 80) | Synthesis of cost estimates by van den Heever and Price, Zwarenstein et al, Beattie and van den Heever, based on ReHMIS shortfall in visits (3.66 target) and assumed unit costs of R20 per visit. | Illustrative or anecdotal |
| Health outcomes are far worse for Africans and the poor than for Whites (e.g. IMR 54 vs 7; life expectancy 8-10 years shorter), though mortality data are incomplete. (p. 16) | DNHPD (1990, 1994) and Bourne (1994) mortality analysis, with caveats on unregistered deaths. | Partial data |
The author's caveats¶
The authors note that most health data exclude the homelands, up to 40% of African deaths are unregistered and cause-of-death data are poor. The ReHMIS data were not validated until drafting began, KwaNdebele clinics are missing, Transkei expenditure was allocated by staff numbers, and fee revenue was reported by under 80% of hospitals. Out-of-pocket, training, donor and capital data are acknowledged underestimates, and homeland primary-care spending rests on a crude formula from two studies. The ranking of districts uses 1985 income. Unit-cost ratios are uncertain, and the private sector data cover only basic items.
Gaps in the evidence¶
The argument would be stronger with district-level data on the distribution of medical scheme members, private provider use and household cash spending, which were not available then. Direct health-status data (reliable mortality/morbidity, utilisation by income) would test the link between resource inequality and health needs; vital registration was incomplete, but the 1993/94 SALDRU survey could be analysed fully. Time series of real public expenditure with a health-specific deflator and longer historical series would allow a trend analysis of apartheid-era allocations; much of this data exists in departmental budget records.
Replicability: Data described but not public.