Report No. 22044-SW Swaziland Selected Development Impact of HIV/AIDS April 10, 2001 Macroeconomic Technical Group Africa Region Document of the World Dank CURRENCY EQUIVALENT (April 10, 2001) Currency Unit = Emalangeni (E) Emalangeni I = US$ 0.1242 US$ 1.00 = E 8.05 (Emalangeni) ABBREVIATIONS AND ACRONYMS AIDS Acquired Immune Deficiency Syndrome ANC Antenatal Clinics CBO Community Based Organizations CIDA Canadian International Development Agency DMC Direct Manpower Cost FLAS Family and Life Association in Swaziland CMTC Crisis Management and Technical Committee EU European Union GDP Gross Domestic Product HIV Human Iminunodeficiency Syndrome IEC Information, Education and Communication MGH Mbabane Government Hospital MOH Ministry of Health NGO Non-Government Organizations RP Reduced Productivity SHAPE School HIV/AIDS Partnership SNAP Swaziland National AIDS/STD Program SSA Sub-Saharan Africa SSS Sentinel Surveillance System STD Sexually Transmitted Disease TB Tuberculosis UN United Nations UNAIDS United Nations AIDS Agency UNICEF United Nations International Children Emergency Fund USAID United States Agency for International Development WHO World Health Organization Vice President Callisto E. Madavo Country Director Fayez Omar Sector Manager Philippe Le Houerou Task Team Leader James Sackey SWAZILAND: The Development Impact of HIV/AIDS - Selected Issues and Options Table of Contents Page No. Executive Summary ....................... ....................................... v Chapter I Status of HIV/AIDS in Swaziland ....................................... I A. The Scope and Character of the Epidemic .......... .............. I B. Factors Exacerbating HIV/AIDS Spread in Swaziland ......... 6 C. Conclusions ................................................ 8 Chapter 11 Demographic, Human Resource Development and Macroeconomic Impact of HIV/AIDS ....................... . 9 A. The Demographic Impact ........................................... 9 B. Effect on Labor and Human Resource Development ............ 14 C. Macroeconomic Impact ............................................. 17 D. Conclusions . ................................................ 19 Chapter III Impact of HIV/AIDS on the Public Service ........................... 21 A. Structure of the Public Service in Swaziland ......... ............ 21 B. Estimating the Impact of HIV/AIDS ............................... 22 C. Implications for the Public Sector ................................... 25 D. Effect on the Public Service Pension Fund . ...................... 26 E. Conclusions ................................................ 27 Chapter IV Policy Options ................................................. 29 A. Ongoing Response to HIV/AIDS in Swaziland ........ .......... 29 B. Proposals to Strengthen Ongoing Initiatives ......... ............. 31 C. Conclusion ...................... .......................... 38 Bibliography ............................................................................... 40 Annexes I Modeling the Impact of HIV/AIDS in Swaziland .42 II Assumptions Used in Costing Selected HIV/AIDS Programs 53 List of Tables 1.1: HIV Surveillance Data by Site .. 2 1.2: HIV Prevalence in Selected Sample of STD/TB Patients. 3 1.3: Reported Non-regular Sexual Partnership, 1991 . . 7 2.1: Summary of Selected Demographic Impact of AIDS . .10 2.2: Swaziland - Mortality Impact of AIDS . .12 2.3: Under 5 Mortality Rates: with and without AIDS per 1000 population ..12 2.4: Swaziland -Estimates of Life Expectancy . .13 2.5: Employment by Citizenship and Skill Level (1992-98) . .15 2.6: Swaziland - Indicators of Macroeconomic Impact of HIV/AIDS (in Percentage) ..19 3.1: Public Service Employment (Excluding Teachers) by Age and Gender: 1999/2000 ..22 3.2: HIV/AIDS and the Public Service (Excluding Teachers) . .24 iii 3.3: Staffing Cost of HIV/AIDS to the Public Sector ...................... 25 3.4: Swaziland Pension Fund ................................................ 26 4.1: Strategies and Options ................................................ 32 4.2: Summary of Estimated Costs of Selected Policy Interventions .... 38 List of Figures 1.1: HIV Positive Rates for ANC Sample for 1992-98 ........ ........... 3 1.2: Estimated Prevalence Rates by Age Cohorts and Gender for STD Patients: 1998 ................................................ 4 1.3: HIV Prevalence by Age for Inpatients: 1998 .......... ............... 5 1.4 New and Cumulative AIDS Cases in Swaziland ......... ............ 5 1.5: Age/Sex Composition of AIDS Cases During 1987-99 ............. 6 1.6: HIV Prevalence by Marital Status Among Hospital Inpatients .... 7 2.1: Population Size With and Without AIDS ............ .................. I 1 2.2: Annual Growth Rate of Population With and Without AIDS ...... 13 2.3: The Size of the Labor Force - With and Without AIDS ............ 14 2.5: Skill Level Distribution (in Percentage) ............. ................. 15 2.6: HIV/AIDS Impact on Skilled/Unskilled Labor Employed (1992- 2005) ................................................ 16 2.7: Growth Rate of GDP (With and Without AIDS) ................ 18 3.1: Public service Staffing & Mortality by Age (Percentage Share) .... 23 3.2 Estimated HIV Prevalence by Age & Gender by 2010 ............... 24 4.1: Hospital Bed Days Needed for AIDS Patients .......... .............. 36 List of Boxes 4.1: Government of Swaziland's Policies for HIV/AIDS and STD 30 This report was prepared by James Sackey and Tejaswi Raparla (AFTM I). The report team received substantial support from the authorities in Swaziland, especially the staff of the Central Statistics Office who participated in the demographic and economic modeling exercise. Bala Rajaratnam (Consultant) contributed to the projections for the impact of HIV/AIDS on public service staffing. Valuable comments were received from colleagues at AFTMI, members of Swaziland Country Team, the peer- reviewers, Messers. Mead Over (DECRG) and Rene Bonnel (AFRHV). Secretarial assistance was provided by Felicidad Santos. The report was presented at a meeting in Swaziland on March 2, 2001, organized by the Swaziland National AIDS/STD Program (SNAP) and chaired by Ms. M.D. Kakudze, PS, Ministry of Finance. The comments of the participants are duly acknowledged. iv EXECUTIVE SUMMARY Introduction The economy of Swaziland registered an impressive growth in the 1980s, which was driven primarily by robust private sector investment. Real GDP growth averaged about 6 percent during the second half of the 1980s, although this tapered off to an average of about 3 percent in the 1990s. The economic gains led to substantial improvements in social indicators, with Swaziland's life expectancy estimated at 60 years, compared to 51 years for Sub-Saharan Africa (SSA); infant mortality (per 1,000 live births) at 65, compared to 91 for SSA; and gross primary enrollment (percentage of school-age population) at 118, compared to 77 in SSA. But like other African countries, the gains over the past 30 years in human resource development are likely to be wiped away by HIV/AIDS, which is emerging as a major health and development concern in Swaziland. The UNAIDS estimates that about 18.5 percent of adults (15 - 49 years), that is an estimated 81,000 persons out of Swaziland's population of just under I million, are living with HIV (excluding those with AIDS) in 1999. In addition, it is estimated that about 4,729 people died of AIDS by end-December 1999. The objective of this report is to provide an overview of selected likely development impact of HIV/AIDS for Swaziland. The purpose is to engage in a dialogue with Government, relevant stakeholders and the donor community on the appropriate actions to pursue in support of the Government's recently developed strategy on the epidemic. The review at this stage does not provide a detailed costing of the impact of HIV/AIDS on various sectors of the economy because of limitations of data. Detailed costing and policy design alternatives should be the focus of subsequent analysis. This report was initiated as an exercise to assist policy makers in Swaziland in their effort to incorporate HIV/AIDS into the planning process on a regular basis. As such, it is directed at officials at the ministries of finance and development planning. It employs conventional demographic and economic models to analyze selected development impacts of HIV/AIDS on the economy, thereby providing an illustration of how these impacts can be incorporated in the regular planning processes (including annual budgeting) in the finance and development ministries of Government. It points out the need for monitoring the progression of the epidemic through further research and improvements in existing instruments. Selected Development Impact This section deals with selected development impact of HIV/AIDS on the economy of Swaziland, focusing on the likely demographic effect of the epidemic and its implications for the overall availability of labor in the economy, the demand for and supply of manpower in the public sector, and the potential costs to the public sector of its response to deal with these effects. The objective is to outline the key factors that could contribute to reducing both private and public savings with likely negative impact on domestic investment and growth. Demographic Impact: The reduction in population due to AIDS, unlike programs for population control, is unusually damaging to the economy in two fronts. First, while planned parenthood and population programs support the increase of social capital, AIDS death does essentially the opposite. It reduces the size of the economically active population. Second, AIDS mortality tends to impose a "shock" on the household's economic structure, since the death of an v economically active individual could force changes in size, composition and socio-economic status of the household and in the use of time devoted to building human capital. In the presence of HIV/AIDS, long term planning could suffer when social contracts fail, i.e. in the sharing of work. The estimates suggest that the long-term demographic impact of AIDS is likely to be significant. The population in Swaziland is projected at 1.6 million by 2015, about 426,000 or 41 percent lower than it would have been in the absence of AIDS. The results indicate annual AIDS deaths increasing from 1,470 in 1991 to over 21,000 in 2015. Closer examination of the data reveals that by the 2005, the number of AIDS deaths may exceed the number of estimated adult deaths in the absence of AIDS. Majority of the AIDS deaths are expected to fall on the 15-49 years age group, the most sexually active and in the prime of their productive years. Along with the decline of population growth, life expectancy in Swaziland is projected to decline on account of AIDS. Life expectancy at birth measures the average number of years that a newborn child would live if mortality remained constant through out his/her lifetime. As a result of the increasing mortality due to AIDS, life expectancy has already stagnated in Swaziland and the trend is likely to continue through 2015. Life expectancy is estimated at 38.3 years for 2001, instead of 59.7 years in the absence of AIDS, a loss of almost 21.4 years over the past decade. By 2015, the difference in life expectancy, with and without AIDS, is projected to reach a staggering 35.5 years. Implications for Human Resource Development: Apart from directly reducing life expectancy, and through it the size of the overall labor force over time, AIDS also affects the dynamics of skill accumulation in the labor market. AIDS tends to kill prime-age adults, many of who are skilled and at the peak of their economic productivity. Preliminary analysis of the impact of HIV/AIDS on the labor force, and the likely effect on public service staffing by skill level was undertaken. The outcome of the analysis should be viewed as a framework for assessing relative rather than the absolute sectoral and skill shortages that would result from the AIDS epidemic. They suggest that AIDS deaths among the highly skilled in the economy could rise from an estimated 45 persons in 1992 to about 281 by 2005. The biggest impact is on unskilled labor in which the total number of AIDS related deaths could increase from about 177 persons in 1992 to about 914 by 2005. The increase in AIDS related death could be fueled by new infections. The total numbers of infections in the formal sector is estimated to increase from 7,480 in the private sector in 1992 to about 14,037 by 2005. Impact on Public Service Staffing: Two main implications may be envisaged for the public service sector. First, there will be reduced productivity resulting from higher absenteeism and morbidity due to HIV; and second, the cost of running the public service is likely to increase as a result of the higher demand for pensions due to early retirement and death and the cost for replacement and training of new staff. Productivity losses resulting from absenteeism and morbidity is estimated at E 17 million in FY1999/2000, constituting about 1.7 percent of total wage bill. If it is assumed that real wages do not increase through 2015 and the civil service to total population ratio is kept constant, the cost of absenteeism and morbidity associated HIV/AIDS could reach about 2.0 percent of total wage bill by 2010. Estimating the direct cost of AIDS deaths on the civil service sector is more complicated. The direct cost of staff death may be decomposed into three components: pension or gratuity payments to spouse and dependents, hiring costs, and retraining costs. The pension rule in the public sector in Swaziland identifies different categories of staff according to the type of contract, vi number of years in service and the level of emoluments. For the purposes of analysis, we assume that all officers who have AIDS die while in service and have successfully completed the period of probation. It is thus estimated that gratuity/pension payments could rise from about E 9.10 million in 2000 to E 26.11 million by 2015 at constant 1999/2000 prices when the impact of AIDS deaths is factored in. Thus, the direct cost of AIDS deaths in the public service could increase from about 0.3 percent of the total wage bill in 2000 to about 3 percent by 2015. In sum, both the productivity loss and the direct cost of HIV/AIDS could amount to close to 5 percent of the total annual wage bill. Macroeconomic Impact: The impact of HIV/AIDS on macroeconomic fundamentals is much more complex than with the foregoing. From the macroeconomic perspective, HIV/AIDS is likely to affect the savings/investment relations. Expenditures for mitigating the impact of HIV/AIDS at both the household and public sector levels are likely to reduce the amount of capital (both public and private) available for more productive investment; thus in the absence of external investment flows, the higher the proportion of care financed from savings, larger the reduction in growth resulting from the epidemic. The report uses a growth model extended to incorporate the increase in morbidity and mortality resulting from HIV/AIDS. The model incorporates, among other parameters, labor productivity losses and AIDS costs met from reduced savings. Using projections of AIDS deaths, it is estimated that AIDS in Swaziland would reduce the average real GDP growth rate during the period 1986 - 2015 from 3.2 percent without AIDS to 2.2 percent with AIDS. This implies that the economy will grow 1.1 percentage points smaller (or 34 percent smaller in absolute real GDP) by 2015 because of the epidemic. This constitutes a projected income loss of about one percent of per capita income per year for 2000-15 in part because of population decline. The model assumes that the baseline growth rate will be positive on the basis of past experience and the possibility of other shocks have not been incorporated. The estimates are thus conservative in view of ongoing structural changes in Swaziland that could either dampen or boost growth performance. Although the macroeconomic effects of HIV/AIDS do not appear devastating, the impact is not uniformly felt across households. At the household level, HIV/AIDS morbidity and death exacerbates poverty and social inequality. Lower income households will be less able than others to cope with the medical expenses and other impacts, including loss of income. The loss of social capital and the resilience level of the household are two key areas requiring policy focus. Policy Options Developing a policy response to the epidemic in Swaziland requires cognizance of ongoing public, donor and private activities in this area. National response to HIV/AIDS was initiated shortly after the first AIDS case was reported to the Ministry of Health (MOH) in 1987, a year after the first HIV infected person was identified. In collaboration with the World Health Organization (WHO), MOH established the Swaziland National AIDS/STD Program (SNAP) and drafted an emergency plan of action, which resulted in the formation of a number of advisory bodies and the development of new initiatives. The Government's objectives are to: * Maintain a sustained political commitment at all levels for HIV/AIDS prevention and control. * Expand the national response to the HIV/AIDS epidemic by strengthening and maintaining the multi-sectoral approach. * Improve co-ordination of HI V/AIDS prevention and control activities at all levels. vii * Ensure that the general public has access to appropriate information, education and communication (IEC) programs on HIV/AIDS and STD. * Increase the capacity of women, youth and other vulnerable or disadvantaged groups (e.g. disabled persons, sex workers, street children, etc.) to protect themselves against HIV/AIDS and other STD. * Ensure that HIV testing is used to maximize prevention and care. * Provide comprehensive health care and social support for people living with HIV/AIDS and their families. * Safeguard the human rights of people living with HIV/AIDS. * Promote HIV/AIDS related research and surveillance activities. In view of the above, and building upon ongoing initiatives to deal with the impact of HIV/AIDS in Swaziland, the report proposes that Government's focus should be directed at: (a) reducing the transmission of HIV; (b) prolonging life and reducing AIDS morbidity; and (c) mitigating the negative impact of AIDS on the economy, especially by initiating programs for skills replacement. The report outlines the key areas of action on prevention and mitigation and provides some cost estimates for several components/programs. The options defined in the study are supposed to be merely illustrative of the potential range of possible directions. They are not intended to be comprehensive. Instead they illustrate a methodology of incorporating HIV/AIDS action policy and costing into a planning framework. Recognizing that both the proposed options and the costing are for illustrative purposes, the preliminary estimates suggest that the various individual elements of prevention could each amount to less than I percent of GDP per year. In terms of mitigation, orphan care is estimated at about an average of I percent of GDP per year. The largest single cost element is hospital care, which is estimated at about 4 percent of GDP per year on average during 2000-2015. The latter points to the need for developing alternative programs for handling AIDS and terminal care. Although the illustrative programs may underestimate the likely cost of prevention/mitigation activities, they point to the likely magnitudes and suggest that such programs can be accommodated by existing resources of Government. viii Chapter I Status of HIV/AIDS in Swaziland The economy of Swaziland registered an impressive economic growth in the 1980s, which was driven primarily by robust private sector investment. Real GDP growth averaged about 6 percent during the second half of the 1980s, although this tapered off to an average of about 3 percent in the 1990s. The economic gains led to substantial improvements in social indicators, with Swaziland's life expectancy estimated at 60 years, compared to 51 years for Sub-Saharan Africa (SSA); infant mortality (per 1,000 live births) at 65, compared to 91 for SSA; and gross primary enrollment (percentage of school-age population) at 118, compared to 77 in SSA. But like other African countries, the gains over the past 30 years in human resource development are likely to be wiped away by HIV/AIDS, which is emerging as a major health and development concern in Swaziland. Swaziland falls in a very vulnerable position because of the peculiar dualism of the society that blends traditional marital with Christian monogamous norms and a small population of about one million. Weakening economic factors and increasing poverty levels may also have contribute in raising the prospects for a high WV/AIDS incidence and the likelihood for a major development disaster attributable to the epidemic. This chapter deals with the status of the epidemic, discusses some of the factors that could contribute to the high prevalence rates, and highlights the key concerns in dealing with the epidemic. It serves as the prelude to the impact analysis of HIV/AIDS in Swaziland in subsequent chapters by highlighting weaknesses in existing data and the extent of the epidemic in the country. A. The Scope and Character of the Epidemic HIV/AIDS is largely sexually transmitted and fatal.' After short initial symptoms, the virus has no visible effects of the infected person during the asymptomatic period, which can be as short as two years or as long as twenty, depending on general health status. In all but a very small proportion of cases, HIV destroys the immune system. This leaves the infected person vulnerable to other infectious diseases, which are typically fatal within six to twenty-four months. As a consequence, Reid (1993) characterized the HIV/AIDS epidemic as going through four phases following infection: • The first phase is felt as opportunistic infections occur, psychological trauma is experienced and there is increasing illness and death. * The second phase is characterized by the change in demographic and societal structure, with many children and elderly without support. • At the third phase, the depletion of labor is felt, especially in the productive and social service sectors. * The fourth phase is associated with widespread destitution, social and economic disintegration and the possibility of social and political unrest. On the basis of existing data, Swaziland appears to be in the first phase of the epidemic. The UNAIDS estimates that about 18.5 percent of adults (15 - 49 years); that is, an estimated 81,000 persons are living with HIV (including all people with HIV infection, whether or not they 1 Like other sexually transmitted diseases (STDs), HIV is difficult to transmit except by sex or other direct contact with the bodily fluid of an infected person. The major modes of transmission are sexual intercourse, reuse of contaminated syringes by injecting drug users, infection via birth or nursing from mother to child, reuse of needles in medical settings, and transfusion of contaminated blood or blood products. I have developed symptoms of AIDS) in Swaziland in 1998. Undocumented information suggests increasing mortality at younger age. HIV Prevalence: Data from antenatal clinics (ANC), which constitute part of the sentinel surveillance system (SSS), collected during 1992-98, indicate HIV prevalence rates for pregnant women ranging from 2.0 percent in Lubombo Region in 1992 to 34.8 percent in 1998 in Manzini Region (Table 1.1). 2 The wide variation in prevalence rates over time is partly a reflection of the timing of the first HIV incidence at different locations in the country and partly the state of a developing monitoring system. Table 1.1: HIV Surveillance Data by Site (Prevalence rates in Percentage) 3 Area 1992 1994 1996 1998 Outside Major Urban Areas Lubombo 2.0 16.7 26.5 31.5 Manzini 4.1 15.6 27.7 34.8 Shiselweni 4.2 16.8 23.9 29.6 Major Urban Areas Hhohho 4.3 15.5 26.3 30.3 Average 3.9 16.3 26.0 31.6 Source: Swaziland National AIDS/STD Program (SNAP), 1994, 1996 and 1998. While the sentinel surveillance system has been fully reported on a two year cycle,, national level averages are also available for 1993 and 1995. The outcome of the surveys points to a number of troubling observations: * The sharp increase in the prevalence rates over the period 1992-98 is a cause of major concern and may reflect the recent inception of the crisis in Southern Africa. Although the 1993 national estimate may be an overestimation, subsequent surveys point to an extraordinary sharp rise in the infection rate (Figure 1.1). Given the normal trend of epidemics of this nature, the situation is likely to get worse in the absence of adequate intervention. * There appears very little regional variation in prevalence rates (Table 1.1). This may be a reflection of the small size of Swaziland in terms of geographical area and population, which makes it possible for workers to live in the countryside but work in the urban areas. Blood sample from STD/TB wards provides further information on the pervasiveness of HIV in Swaziland (Table 1.2). Like the outcome for ANC, there is substantial convergence in the prevalence rates across regions. They also indicate substantially high prevalence rate in recent years. Since reporting and testing instruments were likely limited at the early stages of HIV in 2 Such surveys have been conducted at 24 sentinel sites in the four administrative regions in Swaziland during 1992 - 1998. The four administrative regions are generally evenly distributed in terms of population: Hhohho in the north with 25.3% of the population; Manzini in the west with 28.3 %; Shiselweni in the south with 22.5%; and Lubombo in the east with 22.9%. The sites were chosen to capture regional differences. Lubombo, Manzini and Shiselweni represent rural areas, while Hhohho region represents urban areas. The subjects were eligible to enter the sample only if they were at their first visit for pregnancy or first visit during the survey period. 3It is worth noting that, in general, ANC data are biased as a result of the method of data collection (focused mainly on pregnant women) and thus their extrapolation to the general population could be problematic. First, attendance at ANC does not include those who use private facilities (typically the higher and medium income groups) and those with limited access to such facilities (poor rural residents). The former could have high-risk behavior because of mobility, while the latter are more likely to have limited access to information (such as the role of condoms in limiting the spread of HIV), barrier methods, and supplies and be prone to high rate of infection. Second, ANC attendees are sexually active group and it cannot be assumed that all women of child bearing age are sexually active nor are those outside the child bearing age not sexually active. 2 Swaziland, data up to 1994 are likely to be flawed, resulting in difficulty in making useful inference about trends. Consequently, the focus of discussion will be on the outcome for the most recent survey. The 1998 STD information indicates that HIV prevalence rate among the selected patients ranged from 41.7 percent in Manzini Region to 49.8 percent inHhohho Region. The high prevalence rates do not seem to be influenced by location factors and may differ most probably on account of sampling errors. In general, the figures suggest that HIV prevalence rates are higher for STD patients than of the general population. Figure 1.1: HIV Positive Rates for ANC Sample for 1992-98* 30 2 @ - y .... . -n,,,....... ' w .S t .......... . .-:.,°. .......... S.X..S-...... ._ . - 25 ~~~~~~r i V." '' +.' \': 20 -= 15r,,,v, K. 4,;? a;,~- q':4Y9¢¢a'zw .U.,; 4 7 '. -*-National 0 _ 1992 1993 1994 1995 1996 1997 1998 T 1997 estimate is an interpolation. Table 1.2: HIV Prevalence in a Selected Sample of STD Patients Area 1992 1994 1996 1998 Outside Major Urban Areas Lubombo 12.1 28.7 41.1 49.5 Manzini 15.5 33.3 36.1 41.7 Shiselweni 9.9 26.0 29.8 48.6 Major Urban Areas Hhohho 10.1 26.0 38.2 49.8 Average 36.6 47.6 Source: SNAP (1994, 1996, 1998). While the sampling of the STD/TB patients is not necessarily random, review of information on age, sex and marital status for the HIV positive STD patients highlights a number of concerns: * There is no significant difference in HIV infection for both male and female STD patients. The rate was 47.2 percent for males and 47.9 percent for females. The regional prevalent rates also exhibited limited variations in gender terms. * Of the STD patients that tested HIV positive, the prevalence rates for females was higher than those for their male counterparts for the critical age range 15-29 years (Figure 1.2). Given data limitations, the gender aspect of the HIV infection process is a theme that needs a special focus (see below) since understanding the role played by gender factors in the spread of the epidemic is critical in devising preventive programs. The likely high level of infection among married women in rural areas is also an observation requiring further investigation. 3 Figure 1.2: Estimated Prevalence Rates by Age Cohorts and Gender for STD Patients:1998* 80. wh l m 50, __ g-XX_ 40, _ 3Males 30 . _ . _ _ _ | | _ g ll *Females o !I I 15-19 20-24 25-29 30-34 35-39 40-44 45+ Avg *Note: Sample sizes for age group 0-9 years and 10-14 years are too small to permit meaningful inference. Source: SNAP (1998). To supplement the data obtained from the SSS, SNAP conducted a survey, which investigated the HIVprevalence rates among patients in selected hospitals in 19984 The results complemented past estimates, which suggest that there is a relatively high HIV prevalence among ANC and STD patients in Swaziland. The estimated HIV prevalence in the four hospitals surveyed ranged from 45.7 percent in Mbabane Government Hospital (MGH) to as high as 53.7 percent at the Good Shepherd Hospital (GSH), yielding a weighted average of 49.5 percent. Based on data on gender, age, and marital status, the following conclusions could be derived: * There is a significant difference in the HIV prevalence between males and females: it is estimated at 52.8 percent for females and 45.6 percent for males. This highlights the higher vulnerability of females. * The age profile of the HIV positive inpatients indicated that the incidence is higher in the 20 - 39 years age cohort. More than 60 percent of inpatients aged 20-39 years were tested HIV positive (Figure 1.3). * The effect of marital status on HIV prevalence is not clear because of limitations of the sample. However, despite the limitations of the data, for the three key groups (monogamous, polygamous and single) with adequate sample size, there was no significant difference in 6 prevalence rates. It ranged from 42.2 percent for polygamous to 53.1 percent for those claiming to be single. 4 Four hospitals were selected, one from each region. They were Mbabane Government Hospital (Hhohho Region), RFM Hospital (Manzini), Good Shepherd Hospital (Lubornbo), and Hlatikhulu (Shiselweni). The four hospitals accounted for more than 40 percent of all in-patients or 65 percent of all non-matemity inpatients in 1997. The survey was conducted during a 30-days period in June/July 1998, targeting 1,600 in-patients (for a 95 percent confident interval). It involved all patients admitted to non-maternity wards for the first time in any of the four hospitals. The ELISA method was used for blood sample testing of the eligible 1379 patients out of the 1600 patients targeted. 5 Incidentally, there is no data to allow for gender comparison as in the case of STD patients. 6 The results may be misleading because of inadequate classification in responses. For example, those responding as single could be single and divorced, separated, cohabiting and widowed, which were also identified as separate classification.. 4 Figure 1.3: HIV Prevalence by Age for Inpatients: 1998 70.A 50. Wr 40 11ZZ 4 30. 4 1 l 0-9 '10-19 20-29 30-39 40-49 50-59 60+ NK Note: NK = Not classified. Source: SNAP (1998). AIDS Statistics: The Swaziland National AIDS/STD Program (SNAP) of the Ministry of Health and Social welfare maintains statistics on full-blown AIDS cases from hospitals and health institutions in the country. The cumulative number of reported AIDS cases since 1987 by end- December 1999 was 4,729 (Figure 1.4). This number is likely to be an underestimation of the actual number of AIDS cases for the following reasons: (a) inadequate diagnosis, especially because of the relationship between opportunistic diseases like TB and AIDS; (b) failure on the part of some AIDS patients to seek hospital services, especially in remote rural areas with no health facilities; and (c) poor maintenance of diagnostic AIDS data at health units, especially at the onset of the epidemic. Figure 1.4: New and Cumulative AIDS cases in Swaziland 2000 1500 1000. -- New -U--- Cumulative 500 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 Source: SNAP Despite the shortcomings of the reported AIDS statistics, the pattern of AIDS cases reflects the exponential growth characteristic of such epidemic. It is also reflective of the high rate of HIV infection in sexually active adults in Swaziland. The pattern of AIDS cases is also a cause for concern (Figure 1.5): * The reported AIDS cases are primarily clustered in the 20 - 39 years age groups, which account for 52 percent of all cases reported during 1987-1999. With the rise of new 5 infections increasingly concentrated in this age group, HIV/AIDS epidemic is likely to have substantial impact on the demographics of Swaziland during the next 5 to I Oyears. * Consistent with the rate of progression from HIV infection to full blown AIDS being faster in women than men, female AIDS cases are substantially much higher than male cases in the 20 - 39 years age group while lower in other age groups. Cumulatively, 52 percent of all reported cases was female. * Four percent of reported AIDS cases since 1987 was attributed to children under 4 years of age, although there was no evidence that the trend is increasing. Mother to child (MTC) infection could become a significant mode of transmission in the absence of policy response. Figure 1.5: Age/Sex Composition of AIDS Cases during 1987-99 900 800 8 Fem = = 700 1 300 l l_ l_ lE_l _*Female 0 B. Factors Exacerbating HIV/AIDS Spread in Swaziland Identifying risk factors in Swaziland is extremely complex because of the coexistence of traditional marital practices and western Christian norms. Nevertheless, it is estimated that 86 percent of all AIDS cases are transmitted by heterosexual contacts, and 14 percent through prenatal process (vertical transmission during pregnancy, birth or breastfeeding)7 This is collaborated by estimate of HIV prevalence by marital status among inpatients in the 1998 Hospital based survey (Figure 1.6). While it is difficult to interpret the data on marital status, if the estimates for the divorced, separated and cohabiting are discounted on account of limited sample, the HIV prevalence rates by marital status are uniformly similar. The risk factor could thus be behavioral. 7ULNAIDS: Swaziland - Epidemiological Fact Sheet on HIV/AIDS and Sexually Transmitted Diseases, Geneva: WHO, 1999. 6 Figure 1.6: HIV Prevalence by Marital Status among Hospital Inpatients 80* 6°0'- 60 .1 " 50 =Er SMU, 30 ' 20 am dm U m OM__l ii 10 a o - ,a w , ; _ _ E Z o0 ow 4S6 C 0 0 =0m 40m In this respect, two key hypotheses could be subjects for further investigation: * Women (especially, young women) may be identified as particularly vulnerable because of biological and cultural predisposition that places them at a greater risk of transmission, social disempowerment, and the inability to largely make decisions about sexual issues. * Since sex outside primary relationship is acceptable as almost inevitable in separated families, the likelihood of HIV spreading under these circumstances is very high. The situation is further exacerbated by the rise of intemal migration resulting from urbanization. Urbanization is also associated with increasing unemployment, as employment opportunities in Swaziland have become limited in recent years. The Gender Factors: The review of the reported AIDS cases indicated that the number of female cases could be much higher than those of men in the critical 20 - 39 age group (within the child-bearing age). Furthermore, most of the estimated HIV infections in the 15 - 19 years olds were female. Apart from possible biological reasons, two factors in Swaziland could help explain the disproportionate risk of young women in acquiring H[V infection: * The early onset of sexual activity by women: The review of sentinel surveillance data indicated that a larger proportion of female, relative to their male counterparts, in the 15 - 19 years category was HIV positive among the STD patients. This is both a reflection of sexual activity at an early age and the higher propensity for women to be infected' * The role played by multiple partnership in both monogamous and polygamous relationship and the inability on the part of women to influence the sexual practices of theirpartners.:For many women, the main risk factor for HIV is the problem of multiple and frequent changes in their partners partnership. In a survey conducted in 1991, about 36 percent of the men, 14-19 years, claimed of having at least one sex partner other than a regular partner in the last 12 months. A comparable proportion for the female respondents for the 14-19 years group was 6 percent (Table 1.3). Table 1.3: Reported Non-regular Sexual Partnership, 1991 (In Percentage) Age Group (Years) 14-19 20-24 25-39 40+ 14-99 Male 35.7 22.2 20.0 12.6 19.2 Female 5.9 3.7 8.2 4.6 6.1 Source: KABP Study NAP: KABP/Behavioral Studies - GPA, 1992 (Quoted by UNAIDS, 1999). 8 The 1998 Age-specific HIV Prevalence Rates indicate eleven pregnant females aged 10-14, suggesting early sexual activity and a possibility of child abuse. 7 On the basis of the above, two policy recommendations emerge: * The high rate of HIV infection among young people, especially women under 20 years of age may require a strong educational program targeted at the onset of sexual activity at the elementary level. * For many women, a major risk may be the multiplicity and frequent change in sexual partnership on the part of their male partners, highlighting the need for a campaign supporting responsible sexual behavior, including the use of condoms, targeted at the adult male. D. Conclusions The existing data on HIV prevalence and AIDS incidence no doubt point to an urgent need for action. Nevertheless, interpreting the data highlights their deficiencies. Based on the survey of existing information, a number of conclusions may be derived which may serve as the basis for improvement: First, the estimated HIV prevalence rate of 18.5 percent for adults (15 - 49 years) for 1997 by UNAIDS may be considered an underestimate of the real incidence of the epidemic in Swaziland. Similarly, the reported AIDS cases of 4729 by end-December, 1999 may be substantially below the actual incidence of AIDS on account of inadequacy in the reporting procedures, limitations of the diagnostic facilities and failure of the health care system to be accessible to all, especially those in the rural areas. Second, analysis of the characteristics of HIV and AIDS cases from various sources reveals that the epidemic is taking an increasing toll on young people (under 25 years), especially young women. The implications of this factor could be devastating for the labor force and the economic development of Swaziland. Finally, the contribution of gender factors in the high HIV/AIDS prevalence in Swaziland is cause for concern. High mobility related to the small size of the country, economic hardship and limited employment opportunities, and urbanization have all led to survival strategies that can lead to increased practice of unsafe sex, with women being most vulnerable. Consequently, the factors facilitating the spread of WHV in the highly homogenous society like Swaziland require an enhanced surveillance and preventative programs to address the specific needs of a mobile population. The above comments suggest the need for more and thorough research to serve as the basis for focused policy simulation. But despite the weaknesses and limitations surrounding existing data on prevalence and incidence of HIV/AIDS in Swaziland, it is important to underscore that all available data on Swaziland (the ANC surveillance, HIV in STD patients, Hospital surveys, reported AIDS deaths, etc.) buttress the case for concern. The epidemic seems to be spreading rapidly, at least over the past few years. Accordingly, AIDS is expected to have a major detrimental impact on the population dynamics in Swaziland, and its impact might turn out to be even worse than expected, if effective measures to prevent its continued spread are not undertaken. The analysis of the likely impact of HIV/AIDS on the demography and economy of Swaziland is the subject of the next chapter. 8 Chapter II Demographic, Human Resource Development and Macroeconomic Impact of HIV/AIDS The main and obvious impact of HIV/AIDS, like all health-related epidemics, is its likely effect on the demography and human resource development of a country. But unlike other diseases or conditions, HIV/AIDS mainly occurs in the sexually active population, which is also the economically active age group and it is fatal.l It is this characteristic that makes AIDS of great concern to economists and planners because it has the potential to reduce the human resources available for production as well as affecting their productivity. Furthermore, under certain circumstances, the resultant demographic changes could also affect savings and investment relations in various sectors of the economy, which could lead to a reduced economic growth. The reduction in population due to AIDS, unlike programs for population control, is unusually damaging to the economy in two fronts. First, while planned parenthood and population programs support the increase of social capital, AIDS death does essentially the opposite. It reduces the size of the economically active population. Second, AIDS mortality tends to impose a "shock" to the household's economic structure since the death of an economically active individual could force changes in size, composition and socio-economic status of the household and in the use of time devoted to building human capital. In the presence of HIV/AIDS, long term planning could suffer when social contracts fail, i.e. sharing of work. This chapter focuses on the likely demographic effect of the epidemic and its implications for the overall availability of labor in the economy. The first section discusses the likely demographic outcome on account of HIV/AIDS and uses the results in the second section to evaluate the likely effect on human resource development in Swaziland. The conclusions complement the findings of a study by the Ministry of Education in Swaziland that in the absence of timely intervention, the education sector would be unable to meet its development goals (MOE, 1999). A. The Demographic Impact The demographic impact analysis is based on data derived from the 1991 population census, 2using the USAID's DemProj (Demographic Projections) model. The DemProj forms part of a window-based projection programs referred to as Spectrum,3 developed by The Futures Group Intemational with funding from USAID.4 The model employs a two stage estimation framework: first, it estimates the annual incidence of AIDS on the basis of recent estimates of HIV prevalence; and second, by making assumptions about the probability of progress from HIV infection to AIDS and from AIDS to death, various demographic indicators are derived. Using the 1991 and 1996 demographic data and assumptions discussed in detail in Annex 1, the following ' Mother to Child (MTC) transmission could become a significant mode of infection (four percent of reported AIDS cases in 1987-99 was attributed to children under 4 years in Swaziland) in the absence of an urgent policy intervention. 2 Although the latest update of the population census was conducted in 1996, because the first AIDS case was reported in 1985, the 1991 data being the closest comprehensive demographic inforrnation are used as the base. 3The Spectrum Models were developed by the Policy Project, a United States Agency for International Development (USAID) - funded project implemented by the Futures Group International. Two sub-routines, DemProj and the AIDS Impact Model (AIM), were used for the projections discussed in this section. 4 The DemProj is used in preference to other demographic models, such as: the Edelston, Logistic Curve, Epimodel, the Wilkie Actuarial model, IwgAIDS model and the Actuarial Society of Southern Africa (ASSA) model because it is a simplified blend of the other models. 9 section assesses the impact of MV/AIDS by focusing on the demographic variables such as total population size, additional deaths due to AIDS, crude death rate, life expectancy at birth and infant mortality. The impact of HIV/AIDS on the demography of Swaziland is assessed by comparing the projections that make allowance for the impact of AIDS with estimates and projections that hypothetically exclude AIDS. Two scenarios are provided, the With-AIDS scenario, which is the basis for discussion, and a hypothetical No-AIDS scenario, which is used for the purposes of comparison (Table 2.1).5 Table 2.1: Summary of Selected Demographic Impact of AIDS Indicator Staff Estimate US Bureau of Census c/ 1998 2010 1998 2010 Population Growth Rate (%) With-Aids 2.1 0.1 2.0 1.7 No-Aids 3.1 2.5 3.2 3.1 Crude Death Rate (%) With-Aids 17.8 24.2 21.4 22.6 No-Aids 10.9 7.8 10.1 7.5 Infant Mortality Rate a/ With-Aids 88.7 76.3 103.4 85.3 No-Aids 80.1 61.3 83.8 58.6 Child Mortality Rate b/ With-Aids 140.3 134.8 168.1 152.2 No-Aids 114.1 83.5 114.4 77.5 Life Expectancy at Birth (years) With-Aids 43.0 32.7 38.5 37.1 No-Aids 58.5 63.5 58.1 63.2 a/ Infant (under I year) death per 1,000 live birth. b/ Death of 0-4 years (under 5 year-mortality) per 1,000 live birth. c/ U.S. Bureau of Census, World Population Profile, 1998. Population Size: Figure 2.1 presents the projected population size from 1991 to 2015 taking into account the demographic impact of AIDS as well as the hypothetically projected population excluding the impact of AIDS.6 The absolute difference between the projected population, with and without AIDS, indicates the cumulative impact of AIDS. The population in Swaziland is estimated at 1.58 million by 2015, about 462,000 fewer or 41 percent lower than it would have been in the absence of AIDS7. This effect is higher compared to the estimates by the UN (1998) which shows that by 2015, the population of Botswana and Namibia for example, is expected to be 20 percent lower than it would have been in the absence of AIDS. The projections thus indicate that AIDS is likely to have a very serious relative effect on the population size over the long term, as it is expected that given current parameters, population could decline very rapidly during the projected period.8 The reasons for the expected decline in population growth are the projected rapid rise in new AIDS cases, the limited time from full blown AIDS to death, 5 These results are compared to those derived by the US Bureau of Census. 6 1991 was used as the base year because the population census was undertaken in that year, thus providing adequate demographic information for purposes of analysis at a time when there was not significant impact of HIV/AIDS. 7 Comparable estimates for Swaziland by the US Census Bureau (March, 1999), indicate a similar lower population by 2010, with a population of 2.445 million compared to 2.745 in the current study. 8 The projections do not assume major behavioral changes, which is likely to alter the trend in the long-term population growth. 10 and the relative low fertility associated with AIDS (Gregson and Zaba, 1998). Figure 2.1: Swaziland - Population Size With and Without AIDS Total Population Size 1,600,000 1,500,000 ' 1,400,000 1,300,000 --__ 1,200,000 .. No AIDS 1,100,000,.''' ', 900,000 ., ' 800,000 Mortality Impact of AIDS: The number of deaths from 1991 through 2015 attributable to HIV/AIDS is presented in Table 2.2. Also shown is the projected number of deaths for adults (15-49), infant mortality (infant deaths per 1,000 live births) and under 5 year mortality (deaths of 0-4 year olds per 1,000 live births). The results indicate annual AIDS deaths increasing from 1,470 in 1991 to over 21,000 by 2015. A closer examination of the data reveals that by 2005, the number of AIDS deaths could exceed the number of adult (15 - 49 years) deaths in the absence of AIDS. Majority of the AIDS deaths is expected to fall on the 15-49 years age group, the most sexually active and in the prime of their productive years. Without AIDS, the annual number of adult deaths is estimated to go up from under 1,975 in 1991 to about 2,207 in 2015; with AIDS, this may increase to 20,475 in 2015, an annual rise of 5.1 percent. AIDS related deaths, on the other hand, may increase from 1,470 in 1991 to 21,730 by 2015 and exceeds normal deaths by about 19,523. The likely devastating effect of this phenomenon on the labor market is discussed below. 11 Table 2.2: Swaziland - Mortality Impact of AIDS 1991 1996 2001 2006 2011 2015 AIDS Deaths 1,470 3,849 9,689 14,641 18,309 21,730 Cum. AIDS Deaths 1,470 13,315 49,359 114,619 198,598 280,366 Adult Deaths a/ WAIDS 1,975 4,683 10,409 14,465 17,534 20,475 NAIDS 1,975 2,030 2,090 2,130 2,185 2,207 Infants b/ WAIDS 96.0 90.8 85.4 80.1 75.4 71.8 NAIDS 91.8 83.3 75.4 67.4 59.9 54.0 Under 5 years c/ WAIDS 167.6 141.3 138.7 136.3 135.0 134.8 NAIDS 133.9 119.4 106.3 933 81.3 72.2 Crude Death Rate WAIDS 14.4 16.2 20.3 22.8 24.7 27.0 NAIDS 12.6 11.5 10.0 87 7.7 7.0 Crude Birth Rate WAIDS 44.2 40.0 36.2 32.5 29.9 28.5 ___ __ NAIDS 44.2 41.8 39.0 35.6 33.0 31.5 Notes: a/ Adults defined as 15 - 49 years. b/ Infant mortality rate (infant deaths per 1000 live births). c! Under 5 mortality rate (deaths to 0-4 years per 1000 live births) WAIDS With-AIDS NAIDS No-AIDS. Crude Death Rate and Infant Mortality: The impact of AIDS on the crude death rate is usually severe in countries with high HIV prevalence. The crude death rate for Swaziland is projected to decline from 12.6 per 1,000 in 1991 to 7.0 in 2015 in the absence of AIDS, whereas with AIDS the crude death rate is projected to reach 27.0 in 2015. That is, by 2015, the crude death rate is projected at approximately 176 percent higher than it would have been in the absence of AIDS. Because of AIDS, the crude death rate is estimated to be 14.7 per 1000 in 1991-1995 and is projected to rise to 21.3 in 2000-2005 and a further 23.3 deaths per 1,000 in 2005-2010. AIDS is thus expected to account for about twice the increase in the crude death rate of Swaziland between 2005-15. Table 2.3: Projected Under 5 Mortality Rates: with and without AIDS per 1000 population 1991 1996 2001 2006 2011 2015 With-AIDS 167.6 141.3 138.7 136.3 135.0 134.8 1 No-AIDS 133.9 119.4 106.3 93.3 81.3 72.2 A contributory factor to the expected high crude death rate is the expected increase in infant mortality. It is estimated that approximately 20 to 30 percent9 of the children bom to HIV- positive women are likely to acquire the infection from their mothers. During the period under review, the infant mortality rate is projected to rise much faster than without AIDS. A key reason for this is the assumption that much of the expected decline in the mortality rate for the under 5 years attained over the past decade through public health care programs may not be realized. Table 2.3 presents the mortality rate for the under 5 years in Swaziland, taking into account the impact of AIDS and in the absence of it. A slight decline expected during 1991-92 is negated by the rapid rise in subsequent years as a result of AIDS. 9 It is estimated that mother-to-child (MTC) transmission in Zimbabwe was 20 percent during pregnancy. While 30 percent of the transmission took place during delivery and another 30 percent occurred during breast-feeding (Source: Herald from Zimbabwe July 20, 2000). 12 Life Expectancy at Birth: Along with the decline of population growth, life expectancy in Swaziland (which is a basic measure of human welfare) is expected to decline on account of AIDS. Life expectancy at birth measures the average number of years that a newborn child would live if mortality remained constant throughout his/her lifetime. As a result of the increasing mortality due to AIDS, life expectancy has already stagnated in Swaziland and the trend is likely to continue through 2015. Life expectancy is estimated at 38.3 years for 2001, instead of 59.7 years in the absence of AIDS, a loss of almost 21.4 years over the past decade. By 2015, the difference in life expectancy, with and without AIDS, is projected to reach a staggering 35.5 years. This is much worse than the estimates by the US Census Bureau (1999) which suggest that by 2025, life expectancy in Swaziland would be 53 (using a lower prevalence rate than this study). Table 2.4: Swaziland - Estimates of Life Expectancy 1991 1996 2001 2006 2011 2015 With AIDS 50.5 46.4 38.3 34.6 32.1 30.0 No AIDS 55.6 57.7 59.7 61.8 63.9 65.5 Population Growth: In view of the above, it is expected that the rate of population growth will decline mainly because of the increase in mortality brought about by the HIV/AIDS epidemic. An increased use of condoms to prevent the spread of HV could contribute to the decline but this has not been factored into the analysis. In the absence of AIDS, it is projected that Swaziland's population would have declined from the 3.2 percent growth rate in 1991 to 3.1 percent in 1997 (the census year), with further decline to about 2.5 percent by 2015 resulting from sociological changes in the family structurel°. With AIDS, the decline is likely to be staggering. Estimated at 2.7 percent per annum in the early 1990s, the growth rate is likely to be 0.1 percent by 2015. Much of the dismal outcome is derived from a higher projected AIDS deaths and a declining crude birth rate, which is projected to decline from 44.2 per 1,000 in 1991 to 28.5 in 2015, on account of AIDS related complications. Figure 2.2: Annual Rate of Population Growth With and Without AIDS Annual Population Growth Rate 3.1% ---- -------- X * .6 % No ...... . . 5 i ---