Objectives: To estimate and project the economic impact of HIV/AIDS on the supply of education in Zambia.
Design: An analysis of the financial implications of HIV/AIDS for the Ministry of Education (MoE) and donors funding education in Zambia.
Methods: A mathematical model was used to project the number of primary school teachers and their HIV status under current plans for teacher training and recruitment. Cost data were compiled from the MoE, the Teacher Education Department, teacher training colleges and the donor consortium BESSIP (Basic Education Sub-Sector Investment Programme). Multivariate sensitivity analyses were performed.
Results: The impact of HIV/AIDS on the supply of primary education imply costs to the MoE and BESSIP estimated at US$1.3–3.1 million in 1999, and projected at $10.6–41.3 million over the period 1999–2010. These costs include salaries paid to teachers absent as a result of HIV-associated illness (71%), additional training of teachers to cope with AIDS-related attrition (22%) and funeral costs contractually met by the MoE (7%). They do not include the additional costs of an active care and prevention response by the MoE, or the burden of ensuring enrolment of AIDS orphans. The annual cost of HIV/AIDS is a relatively small fraction of the overall MoE budget (2.5% in 1999) but has substantial implications for resource allocation to some functions. Expenditure on teacher training will need to increase by 26% if Education for All targets are to be met in the face of AIDS.
Conclusions: HIV/AIDS has significant implications for resource allocation in the education sector in Zambia.
From the aDepartment of Infectious Disease Epidemiology and the Partnership for Child Development, Faculty of Medicine, Imperial College London, bUNAIDS, Bamako, Mali, the cMinistry of Education and the dUNESCO, Paris, France and the eWorld Bank, Washington, USA.
Requests for reprints to: Dr N. C. Grassly, Department of Infectious Disease Epidemiology, Faculty of Medicine, Imperial College of Science, Technology and Medicine, St Marys Campus, Norfolk Place, London W2 1PG, UK.
Received: 18 March 2002; revised: 16 August 2002; accepted: 29 November 2002.
AIDS is now the leading cause of death in sub-Saharan Africa. The magnitude of its impact in this region has shifted perceptions of the AIDS pandemic from being purely an issue for the health sector to a crisis of cross-sectional importance . This impact has been felt particularly in the education sector, where HIV/AIDS is affecting both the providers of education, including teachers, and those seeking education and their families. In southern and eastern Africa, many countries are losing teachers to AIDS or facing HIV-associated teacher absenteeism on a scale that has a severe impact on the ability to supply education and to reach Education for All targets [2,3]. At the same time, demand for education is affected, since many children are facing HIV-related illness or death of parents or siblings. This can compromise their ability to attend school and their performance, as they remain home to care for sick family members or find work to compensate for lost family revenues [4,5]. Tragically, it is precisely the education sector that can respond directly and effectively to the HIV pandemic, by providing both education to promote sexual behaviour change and the long-term perspectives and plans that are needed to motivate this behaviour change.
In Zambia, the magnitude of the impact of HIV/AIDS on the education sector began to be appreciated in the late 1990s, when the numbers of primary school teacher deaths reported by the Ministry of Education (MoE) began to rise steeply from two per day in 1996 to more than four per day in 1998 . This represented an annual loss of 4% of all registered primary teachers, or two thirds of the annual output of newly trained teachers from all primary level training institutions in 1998. At the same time, school authorities, parent–teacher associations and communities in Zambia reported a serious loss of teaching time resulting from prolonged illness or erratic attendance . The illness and death of teachers from HIV/AIDS has financial implications for the MoE even in the absence of an active response to the HIV epidemic. To capture these financial implications, an economic impact analysis was carried out focusing on the impact of HIV/AIDS on the supply of trained primary school teachers. The MoE's perspective was taken for costs accrued from lost productivity of teachers absent as a result of HIV illness, loss of teachers from AIDS mortality and funeral costs. These costs reflect a passive response of the Ministry where dead teachers are replaced but do not include costs that would additionally be incurred by an active planning and prevention response.
A mathematical model to measure the impact of HIV/AIDS on the population of primary and secondary school teachers has already been developed as part of the Ed-SIDA initiative . This compartmental model captures the dynamics of the teacher population in terms of recruitment, retirement, leaving before retirement, HIV infection and death using difference equations based on yearly time steps (Appendix). Relevant education statistics on teacher numbers, their recruitment and rates of leaving before retirement age were entered into the model by education planners from the MoE and Teachers Education Department in June 2001. In this study, because only primary school teachers working in government schools were considered, private and community school data are not included in these analyses. Approximately 92% of all enrolled primary school children were enrolled in government primary schools in 1996 .
Demographic and epidemiological requirements of the model were based on UN Population Division and UNAIDS/US Census Bureau sentinel surveillance site data, respectively, (available on the internet at esa.un. org/unpp/and http://www.census.gov/ipc/www/hivaidsd. html). Since the early 1990s, the prevalence of HIV among adults in Zambia has remained constant at close to 20%. It was assumed that this prevalence will remain the same over the next decade in the absence of effective interventions. The pattern of HIV incidence by age and sex was assumed to follow that observed in cohort studies in Uganda and Tanzania, where women became infected at a younger age than men (Basia Zaba, personal communication). A sensitivity analysis explored different assumptions about the future prevalence of HIV by varying the relative incidence of infections among teachers. Mortality from causes other than AIDS was based on a relational life table using a Brass African standard , while the rate of progression from infection to death from AIDS was based on data from cohort studies in sub-Saharan Africa .
Financial data were collected from the MoE headquarters, teacher training colleges and the Teacher Education Department in August 2001. Different sources of cost data were used to allow independent verification of unit costs. The average of the recurrent costs to train a teacher was used to estimate the current marginal cost of training an additional teacher. The capacity of teacher training colleges to train additional teachers was unclear and, therefore, the marginal cost was assumed to remain constant over the increments in teacher training considered.
The perspective of the economic impact analysis was defined such that the cost of HIV/AIDS for the MoE and the Basic Education Sub-Sector Investment Programme (BESSIP: the Ministry's formal programme for coordinating the financial and technical assistance of a large consortium of donors) were considered. These organizations are the major contributors to the provision of primary level education in Zambia. Costs include both direct expenditure in response to HIV/AIDS in terms of replacement of dead teachers and funeral costs, and expenditure that has either reduced or zero return because of HIV illness and absenteeism of teachers (there is no formal system of supply teaching to cover absent teachers). All costs are presented in year 2001 US dollars and discounted into the future to reflect time preferences at a rate of 5%. It is assumed that unit costs do not change in real terms over the period analysed. While salary and funeral costs are currently fixed by agreements between government and teacher unions, the cost of training a teacher may increase if HIV/AIDS compromises the efficiency of teacher training colleges. This impact is difficult to estimate and is not accounted for in the present analysis, resulting in potential underestimation of the cost of AIDS mortality. Expenditure on in-service training was excluded from the analysis since the new School Programme of In-service for the Term (SPRINT) is in the process of being scaled-up to all zones and associated costs remain unclear. Details of the cost model used to estimate the economic impact of HIV/AIDS are given in the Appendix.
A multivariate sensitivity analysis was performed to present best and worst case scenarios for the cost of HIV/AIDS to the MoE/BESSIP with respect to different discount rates (0–10%), HIV incidence among teachers (1.2–4.8% per year, equivalent to half to double that of other adults for 1999–2010) and cost estimates (for teacher training).
Results and discussion
The number of teachers at the primary level in 1999 in Zambia was 37 117, approximately 10% of whom were officially ‘untrained'. These untrained teachers were included in this analysis since they incur salary costs and must be replaced (with trained teachers) if they die from AIDS. If teachers are assumed to have the same risk of becoming infected with HIV as any other adult, then of these teachers 8114 were estimated to be HIV positive. This corresponded to a slightly higher prevalence of HIV (21.9%) than that seen in the adult population aged 15–49 (19.7%), since no teachers are in the age range 15–20 years, for which HIV prevalence tends to be lower than at older ages. The estimated number of primary school teachers in active service who died from AIDS in 1999 was 840, equivalent to 46% of all teachers trained that year. If the relative risk of a teacher becoming infected with HIV is half or double that of an average adult, then the estimated number of HIV-positive teachers in 1999 varied between 5900 and 11 063, and the number of deaths in 1999 between 597 and 1180.
Enrolment of pupils in primary school has been declining, despite increases in the size of the school-aged population, because of a lack of provision of educational opportunities (net attendance rate was approximately 69% in 1996, falling to 66.2% in 1999 and 65.6% in 2000 [8,11]). Because of these falling enrolment levels, and because of undesirable heavy reliance on the use of untrained teachers, a reformed Zambia Teacher Education Course (ZATEC) was implemented. This aims to double annual primary level teacher training and recruitment by replacing the 2-year teacher training course with a 1-year course followed by an in-school year. The reformed ZATEC was piloted in three of the ten primary teacher training colleges in 1996 and expanded to all colleges by 2001. Based on the assumption that the reformed ZATEC is operational from 2002, and using 1999 as a baseline, the number of teachers in 2010 is projected to be 49 995 in the presence of HIV/AIDS, compared with an expected 59 550 if there were no AIDS mortality of teachers.
The impact of HIV/AIDS on the teaching population, as estimated by the Ed-SIDA model, allows the identification of three sources of cost faced by the MoE/BESSIP: (i) additional training of teachers to cope with AIDS-related attrition, (ii) salaries paid to teachers absent with HIV-associated illness, and (iii) funeral costs contractually met by the MoE (Table 1).
Putting together the unit costs given in Table 1 and the results from the Ed-SIDA impact model, the economic impact of HIV/AIDS on the education system in Zambia can be estimated (see Appendix for details). Table 2 presents the expected cost to the MoE/BESSIP of HIV/AIDS over 1999, and over the period 1999–2010, if additional teachers were to be recruited to replace those lost, but no other activities were initiated in response to HIV (a ‘passive’ response). A range of costs is presented in parentheses for each estimate to reflect the best and worst case scenarios examined in the multivariate sensitivity analysis.
The bulk of the economic impact of HIV/AIDS on the MoE/BESSIP over the next decade is likely to result from teacher absenteeism (71%), with most of the remaining impact caused by the loss of trained teachers (22%). Teacher absenteeism with HIV-related illness is projected to result in the loss of 12 450 teacher-years or close to 20 million teacher-hours over the next decade. The total estimated cost of HIV/AIDS in 1999 represents only 2.5% (1.7–3.8) of the total authorized MoE budget for that year ($80 755 000), the majority of this budget paying teacher and administration salaries. However, the impact of HIV/AIDS on teacher training represents 26% (12–51) of the budget spent on primary level teacher training ($2 037 000). This impact is significant, since teacher training is one of the key responses if the national target to provide education for all is to be achieved. Further, it should be stressed that only half the cost of training a teacher is met by the MoE/BESSIP, the remaining being met by student fees charged by teacher training colleges and other college-based fund raising. If the perspective of this analysis were to be widened to include costs incurred as a consequence of HIV/AIDS by teachers themselves, the cost of AIDS would rise steeply (and would include medical costs incurred by HIV-positive teachers).
Most (> 50%) of the difference between the best and worst case scenarios presented in Table 2 for the costs over 1999–2010 of HIV/AIDS is caused by different assumptions about the incidence of HIV among teachers. Change in the discount rate contributes a further 25–43% of the differences in these cost estimates.
There are additional costs of HIV/AIDS relating to the provision of education that will need to be met by the MoE/BESSIP but which are harder to estimate. These include loss of productivity of HIV-positive teachers as a result of psychological trauma and stigmatization, teacher absenteeism as a result of HIV illness or death in their families and teacher absenteeism for attendance of funerals of colleagues and friends.
An active response to the HIV epidemic in Zambian schools, rather than the passive response considered here, will of course require further expenditure. This response might include voluntary counselling and testing for teachers, provision of antiretroviral drugs to HIV-positive teachers, curriculum changes to include education about HIV/AIDS in the classroom, peer education programmes among school pupils, and provision of support for orphans from AIDS and other causes. The financing, planning and implementation of these responses would need to include not just the MoE but also the Ministry of Health, AIDS prevention organizations and civil society. The costs of these activities will be significant, and their estimation is beyond the scope of this paper. However, their benefits will be large and wide ranging. For example, halving the incidence of HIV among teachers would save the MoE and BESSIP alone around $7 million over the next decade. If the impact on teachers themselves, their quality of life, morale, the quality of education and social capital relating to basic education were to be considered, the common sense of an active response is inescapable.
The Ed-SIDA/AIDS model of the impact of HIV/AIDS on education systems was developed under the leadership of the Partnership for Child Development with the support of the World Bank. We wish to thank Michael Kelly for reading and commenting on a draft of this paper, Bruce Jones for facilitating this study, and two anonymous reviewers for detailed suggestions for revision.
Sponsorship: KD received salary support from the Wellcome Trust. Development of the Ed-SIDA/AIDS model was supported by the World Bank.
1.Garnett GP, Grassly NC, Gregson S. AIDS: The makings of a development disaster? J Int Dev
2.Badcock-Walters P. The Impact of HIV/AIDS on Education in KwaZulu Natal.
Durban: KZNDEC Provincial Education Development Unit; 2001.
3.Kelly MJ. The encounter between HIV/AIDS and Education.
Harare: UNESCO Sub-regional Office for Southern Africa; 2000.
4.Nampanya-Serpell N. Social and economic risk factors for HIV/AIDS-affected families in Zambia. AIDS and Economics Symposium
. Durban, 7–8 July, 2000. Available at http://www.iaen.org
5.Ainsworth M, Beegle K, Koda G. The Impact of Adult Mortality on Primary School Enrollment in Northwestern Tanzania.
[Background paper prepared for the African Development Forum]
Addis Ababa, December 2000.
6.Milimo JT. Factors Affecting School Attendance. A Qualitative Approach.
Lusaka: Ministry of Education; 1998.
7.Bundy D, Kagia R, Doka A, Drake L, Maier C. A Window of Hope: Education and the Global Epidemic of HIV/AIDS.
Washington, DC: World Bank; 2002.
8.Mwikisa CN, Lungwangwa G. Education Indicators, Costs and Factors Associated with Primary School Effectiveness in Zambia.
Lusaka: Ministry of Education; 1998.
9.Brass W. On the scale of mortality.
In: Biological Aspects of Demography
. Edited by Brass W. London: Taylor Francis; 1971:69–110.
10.Morgan D, Whitworth JAG. The natural history of HIV-1 infection in Africa. Nat Med
11.Ministry of Education. BESSIP: 2000 Programme Performance Indicators.
Lusaka, Zambia: Ministry of Education; 2001.
12.Kikumbih SN, Isingo R, Boerma JT. Consequences of adult HIV infection for outpatient morbidity and treatment costs: a prospective study in a factory clinic in Tanzania. Health Policy Plan
13.Wagner HU, Kamali A, Nunn AJ, Kengeya-Kayondo JF, Mulder DW. General and HIV-1-associated morbidity in a rural Ugandan community. AIDS
A mathematical model of the dynamics of teacher numbers and the impact of HIV/AIDS
The dynamics of the teacher population can be given by the following difference equations:
where Xg,a(t), Yg,a(t) and Ng,a(t) are the number of HIV negative, positive and total number of teachers respectively for gender g and age a at time t; Rg,a(t) is the number of teachers recruited and Cg,a(t) the number of teachers leaving the profession respectively at time t for gender g and age a; ig,a(t) is the mean annual HIV incidence rate over the time t to t + 1 per HIV-negative adult of gender g and age a; q is the ratio of HIV incidence in teachers to that in the general population; μg,a is the mortality rate of teachers of age a and gender g not caused by AIDS; and αg,a is the rate of AIDS mortality by age and gender.
Cost model relating morbidity and mortality of teachers to MoE/BESSIP costs
Cost of training additional teachers to cope with AIDS-related attrition
We assume additional teachers are trained to cope with AIDS mortality, such that the current projection for the number of teachers in 2010 in the theoretical absence of AIDS is still met given the impact of AIDS. The burden of training additional teachers is assumed to be equally distributed over the period 1999–2010.
Cost of teacher absenteeism owing to HIV-related illness
Based on previous estimates of 12–14 illness episodes experienced by HIV-positive teachers before terminal illness in Zambia , median survival time of 10 years with 9–10 months living with AIDS , and a mean duration of illness of 2 weeks’ absence per illness episode, an HIV-positive teacher will, on average, be absent 1.3 months per year because of HIV illness or AIDS. These levels of morbidity are in broad agreement with cohort studies from neighbouring countries [12,13]. Teachers absent for up to 3 months from illness are officially entitled to receive full salary, after which they receive half salary until the end of the sixth month. However, for compassionate reasons, it is usual for a teacher to continue receiving full salary until she/he dies or resigns. There is no formal system of supply teaching to cover for absent teachers. The cost of teacher absenteeism from HIV is, therefore, the salary costs of teachers who are not teaching owing to HIV-related illness.
Funeral costs of AIDS deaths