Coping with the AIDS epidemic in Tanzania : survivor assistance

FILE COPY II a A Report No.:11279 Type: (MIS) Title: COPING WITH THE AIDS EPIDEMIC Author: AINSWORTH, MARTHA Ext.: 0 Room: Dept.: JULY 1992 Coping with the AIDS pdeiC in Tanzania: Survivor Assistance Martha Ainsworth and A. A. Rwegarulira The World Bank Africa Technical Department Population, Health and Nutrition Division Technical Working Paper No. 6 July 1992 COPING WITH THE AIDS EPIDEMIC IN TANZANIA: SURVIVOR ASSISTANCE by Martha Ainsworth Africa Technical Department Population, Health and Nutrition Division The World Bank and A. A. Rwegarulira Social Welfare Department Ministry of Labor and Youths Development Republic of Tanzania July 1992 This paper was written as a background paper for the "Tanzania AIDS Assessment and Planning Study", Report no. 9825-TA. The opinions expressed are those of the authors and do not necessarily reflect the policy of the World Bank, its member countries, or the Government of Tanzania. Comments are welcome. i ACKNOVMLEDMENTS This paper was prepared as background for the "Tanzania AIDS Assessment and Planning Study", a collaborative effort between the World Bank and the World Health Organization Global Programme on AIDS for the Ministry of Health of Tanzania. The report was managed by Jeannette Murphy of the Southern Africa Human Resources Division of the World Bank. Most of the Information for this pape: was assembled during a two-week mission to Tanzania in November 1990. Such a paper could only be written with the full cooperation and support of numerous individuals in the Tanzanian Government. We particularly indebted to Mr. S. Ngallaba, Senior Demographer, and Mr. E.D. Malmu, Principal Systems Analyst, both at the Bureau of Statistics, for providing data from the 1988 Census on the basis of which estimates of the number of survivors could be made. Numerous individuals at the Ministry of Labor, Culture and Social Welfare provided valuable information on the costs of existing programs and the elements of new programs under consideration: Mr. P. Mwankyuse, Senior Welfare Officer; Mr. Gedeon Mugyabuso, Regional Social Welfare Officer, Kagera Region; and Mrs. Frida Ndone, Director, Kurasini Children's Home, Dar es Salaam. Appreciation is also extended to officials at UNICEF in Dar es Salaam, who provided advice on the costs and inputs of the feeding post and day-care elements of their Child Survival and Development Programme: Ms. Laetitia van den Assum, Director, UNICEF/Dar es Salaam; Mrs. I. Tukal, Assistant Project Officer, Child Care and Development; and Mr. P. Mushi, Project Officer, Kagera CSD Programme. Several officials of WAMATA, the counselling and assistance organization for persons with AIDS, also gave freely of their time in describing the costs and inputs into their programs: Dr. M.C. Mukoyogo, Legal Advisor, Faculty of Law, University of Dar es Salaam; Mr. Conrad Mushi, Counsellor Coordinator; and Mrs. Alice Rutahindurwa, Laboratory Technologist and Counsellor. Helpful comments from Althea Hill, Elizabeth Preble, Taryn Vian, Jeannette Murphy, Jill Armstrong, David Dunlop and participants in the World Bank's AIDS Working Group seminar series are gratefully acknowledged. Last but not least, we would like to thank Andrew Nyamete and Sara Kim for assistance in data analysis. Survor Assistance 1 July 1992 TABLE OF CONTENTS E Acknowledgeme ................................................... I I. Iltroduction .................................................... 1 H. The impact ofte AIDS epidemic on survivors .............................. 2 A. Theimpact ofA DSon mortality .................................... 2 fIeBrelatlonbetweenHVinfectionandmortaflty ........................ 2 Current e ovlshofInfection ....................................... 3 Currentlevdsofadultmortality ................................... 4 B. Whoarethesurvivorsandhowaretheyaffected? .......................... 6 C. Themagitadeofthesurvivorproblem ................................ 8 Orphansandotherchildrenatrisk .................................. 9 Widowsandvwidowers ......................................... 18 Theelderly .................. .............................. 19 Survivorhouseholds ........................................... 20 D. Thesietofsurvivorgg eups: asummary ............................... 24 I. Strategiestoassistsurvivors ......................................... 25 A. Criteria fr selecting interventions .................................... 25 Costs andbenefits ............................................ 25 Target groups of beneficiaries .....................................26 Sustan ayofp ams ....................................... 27 B. Some examples of servivor assistance program ........................... 28 Orphanges ................................................28 Directa tn es ..............................................31 Feeding posts/childc can ................................... 32 Subsidofscholofsadii m ................................34 IV. Conclusions ...................................................39 Rehre am . .. . . .. . . .. . . .. . . .. . . .. . ..42. .. .4 Survivor Assitawe iii July 1992 Annxe Eage 1. Method of calculating the number of orphans from both parents .................... 45 2. Children'shomesinTanzania, September l990 .............................. 47 3. Costcalculations,KurasiniChildren'sHome ............................... 48 4. Cost calculations, Social Welfare Office survivor assistance program in Kagera .......... 51 5. Co calculations, feeding post/child care centers ............................. 52 Tables 1. Approximate relation between HIV infection and adult mortality in a cohort of1000sexuallyactiveadults,15-49 .................................... 3 2. Proportion and number of orphans under varying assumptions ..................... 16 3. Estimates of the total number of orphans from either parent ...................... 17 4. Char e s of households with mateal orphans and with a recent adult death ........ 21 5. Multivariate analysis of the characteristics of survivor households in Kagera region ........ 22 6. Summary of some survivor groups, Kagera region, 1988 ........................ 24 7. Actual and potential number of children in existing orphanages, Kagera Region, June 1990 ... 30 8. Out-of pocket costs for schooling, Tanzania, 1988 ............................ 35 1. Adult mortality rate per 1000, age 15-49, twenty regions of Tanzania, 1988 ............ 5 2. Seroprevalence and adult mortality, Kagera region, by district, 1988 ................. 6 3. Number of maternal orphans by region, mainland Tanzania, 1988 .................. 10 4. Adult mortality and orphan rates, twenty regions of mainland Tanzania, 1988 ........... 11 5. Orphan rates by age group, six regions of Tanzania, 1988 ....................... 13 6. Orphan rates and adult mortality, Kagera region, by district, 1988 .................. 13 7. Orphan rates by age groups, six districts of Kagera region ....................... 14 8. Number of matenal orphans, six districts of Kagera region ....................... 14 9. Widowing rates and adult mortality, six regions of Tanzania, 1988 .................. 19 10 Distribution of households with an adult death by district, Kagera region, 1988 .......... 20 11. Distribution of households with orphans by district, Kagera region, 1988 .............. 23 12. Enrollment rates, orphans and non-orphans, three regions of Tanzania, 1988 ............ 36 13. Enrollment rates, orphans and non-orphans, six districts of Kagera, 1988 .............. 37 COMN WITH TH AID EPIEMIC IN TANZANIA:* . ~SURVIVOR ASSISANC I. AV LQZL*M Estimates by various UN agencies, WHO and the Tanzanian AIDS Control Programme place the total number of Tanzanians currently Infected with the human immunodeficiency virus (HW), the causative agent of AIDS, at between 400,000 and 800,000 persons (Chin, 1990; Ministry of Health, 1990). This corresponds to a seroprevalence rate of 3.5 - 7.0 percent among adults, in a total population of 24.7 million. The rapid spread of this incurable, deadly disease in Tanzania is a tragedy in human terms and also in terms of its impact on future economic development. There is no known cure for AIDS, nor is there likely to be an affordable cure in the next few decades. Prevention of future cases is unquestionably the only way to reduce the prevalence of AIDS in Tanzania; urgent efforts must be made to find the most effective and least costly ways of preventing AIDS. Unfortunately, because of the large number of persons already infected with HIV and the long latent period from infection to development of full-blown AIDS', the impact of preventive efforts will not be felt for many years. Even if all future cases of HIV infection could be prevented from this day forward, every year for the next two decades roughly 5 percent of those already infected with HIV will develop full-blown AIDS and die. Although AIDS can kill people of any age, AIDS deaths - unlike deaths from other major diseases in Africa, such as measles, malaria and tetanus - are concentrated among the sexually active adult population that is also most productive. Thus, AIDS is selectively killing the generation of adults that is breadwinner for numerous dependents - both children and elderly parents - leaving large numbers of vulnerable survivors with no means of support. The objectives of this paper are: to characterize the impact of AIDS on surviving family and community; to estimate the magnitude of the potential target groups for survivor assistance interventions - - orphans, widows, widowers, elderly and affected households; and to evaluate available information on the costs and benefits of selected interventions to assist survivors. There are many programs that could be targeted at survivors in need of assistance; most of these are already being implemented in one form or another in ongoing development projects. Ideally, one would want some measure of both the cost and the benefits of each potential program, on the basis of which alternative programs could be compared. Unfortunately, the effects of most programs have never been properly measured and their quantification in such a way as to permit comparisons across interventions is beyond the scope of this paper. The choice of programs for costing in this paper is neither comprehensive nor indicative of programs that are recommended. Rather, our objective is to demonstrate the benefits, drawbacks and issues in targeting potential programs to persons most in need. The median time between EIV infection and development of full-blown AIDS is 10 yeaus in the United States (Moss and Bacohetti, 1989), possibly a few yeas shorter in Sub-Saharan Africa (Killewo et al, 1990). Swvvor Assistance 2 July 1992 II. T I ACT OF T AIIDS EIDEMIC ON SUV R S The major modes of transmission of AIDS in Tanzania are through heterosexual Intercourse and from mother to child during pregnancy. Thus, the mortality impact of the AIDS epidemic will be concentrated among children under 5 and among sexually active adults. This section begins by summarizing the posited relationship between HIV infection and mortality, and examines the levels of adult mortality in several regions of Tanzania from the 1988 census for evidence of the impact of AIDS. The nature of the impact of the epidemic on survivors Is then discussed, followed by an attempt to estimate the size of survivor groups, based on the 1988 Census. Particular attention is paid to comparing the magnitude of the survivor population created by AIDS with the underlying prevalence of survivors not related to AIDS. A. The impact of AIDS on mortality Relaton between HIV iqfection and mortatsy The relationship between levels of HIV infection and mortality among adults is based mainly on studies from Western countries. The median time between HIV infection and development of fil blown AIDS is about 10 years2; 15-20 percent will develop AIDS and within 5 years, half within 10 years and all by 15-20 years (Moss and Bacchetti, 1989). Whether these same parameters appy to Africa is uncertain; it has been suggested that the mediaa time among adults may be shorter (Killewo et al, 1990). An approximate rule of thumb for assessing the impact of a given, constant level of HIV infection on the mortality rate of sexually active adults in Africa is as follows. Assume, first, that the underlying mortality rate In Sub-Saharan Africa for adults in the age group 15-49 is roughly 5 per thousand and, second, that in a cohort of infected aduits, about 5 percent will develop AIDS and die annually over 20 years. Then, in a group of 1000 adults 15-49 with no MV infection the mortality rate will be 5 per thousand. With an IV infection (seroprevalence) rate of 10 percent, 100 of the adults will be infected and 5 will die each year from AIDS, raising the adult mortality rate to 10 per thousand. With a seroprevalence rate of 20 percent, 10 will die each year from AIDS, raising the adult mortality rate to 15 per thousand. This rule of thumb is summarized in Table 1 below. Every 10 percent increment in a (constant) level of MV infection should roughly raise the adult mortality rate among persons 15-49 by 5 per thousand. A seroprevalence rate of 10 percent will thus double the adult mortality rate, and roughly half of all adult deaths will be due to AIDS. A seroprevalence rate of 20 percent will triple adult mortality, and two-thirds of adult deaths will be due to AIDS.' L Iemp at al (1990) have recently projected a longer median incubation period for IV among homosexual and bisexual men in San Francisco, of 11 years. s As is true for any rule of thumb, these relationships are not exact. The probability of dying of AIDS should be nt of the proSability of dying of other causes (a 5/1000 chance for each person who dies of AIDS). When this is factored in, the mortality rates in an AIDS epidemic are a fraction smaller than in Table 1. For example, under a 10 percent seroprevalence rate the adult mortality rate is 9.975/1000 instead of 10/1000. It is also worth noting that if seroprevalence rates are increasing, then the mortality rates will be less than implied by the current seropevaleace rate in the chart. The adult mortality rate for persons 15-49 in Tanzania was 5.8 per thousand in 1988, according to the Census. Every 10 percentage points' increase in seroprevalence will add roughly 5/1000 SW-vior Asstance 3 Ju 1992 TABLE 1: Approximate relation between IIV infection and adult mortalilty In a cohort of 1000 sexualv ative adults Number of annual Mortality rate Seroprevalence (%) AIDS deaths (per 1000 per year) 0 0 5 10 5 10 20 10 15 30 15 20 AIDS will have both direct and indirect effects on the mortality of children under 5 years of age. About 30-40 percent of the children born to HIV-infected mothers will be infected. The time from infection to development of full-blown AIDS and death among infants is much shorter than among adults. The increase in infant and child mortality due to AIDS infection will likely offset recent successes in lowering childhood mortality. Preble (1990a) estimates that in 1990 under-five deaths in 10 seriously- affected central African countries were 4-6 percent higher than they would have been in the absence of AIDS. By 1999, she projects that under-five deaths will be 21-43 percent higher than without AIDS. AIDS will also raise child mortality indirectly by throwing into poverty the healthy orphaned children of parents who have died from AIDS. This additional indirect child mortality caused by AIDS has not been estimated. Carmnt levels of ifetlon There has been no nationally-representative seroprevalence survey in Tanzania. The number of AIDS cases reported as of the end of 1990 (21,175) is considered a severe undercount (NACP, 1991). Estimates of current levels of HIV infection are largely extrapolations based on smaller studies of specific groups, such as bar workers, blood donors and mothers attending antenatal clinics, and range from a low of 400,000 (Chin, 1990) to a high of 800,000 (Ministry of Health, 1990). These estimates would imply a national seroprevalence rate among adults on the order of 3.5 - 7.0 percent. The rate of infection is higher in urban areas than in rurak areas. More detailed information on infection rates is available for three regions thought to be the most seriously affected: Kagera, Dar es Salaam and Mbeya. A population-based seroprevalence survey of the Kagera region conducted in 1987 found a regional rate of 6.62 percent among adults 15-54 (Killewo et al 1990).' The seroprevalence rate for adults varied considerably among districts, however, from 0.4 percent in the least-infected districts to 24.2 percent in the regional capital. Testing of blood donors in Dar es Salaam over the period 1986-1990 revealed an adult seroprevalence rate of 7.12 percent to this baseline adult mortality rate (see Table 1, column 2). 4The unweighted average rate found among adults 15-54 in the study was 9.6 percent. The rate of 6.62 percent has been calculated by weighting the results for different districts by their population size. Swvivor Assistance 4 July 1992 (Ministry of Health, 1990). A third seriously affected region is thought to be Mbeya, through which runs the major trucking route to Zambia. Sentinel seroprevalence data from pregnant women attending antenatal clinics in urban and rural areas of Mbeya suggest an adult seroprevalence rate in 1990 of 13 percent in Mbeya town and 7.6 percent in rural areas (Hemed, 1990). It is not known how representative these figures are of the entire Mbeya region, however. Levels of adult modtalty The 1988 Census of Population provides some evidence of the impact of the AIDS epidemic on adult mortality in Tanzania. The questionnaire submitted to a 15-percent subsample of the population asked, among other things, whether anyone in the household had died in the 12 months prior to the census and, if so, their age and sex. Although this information is subject to many different types of error and is clearly an underestimate of true mortality levels, the differences in the level of adult mortality among regions may be indicative of the impact of AIDS. Figure 1 presents the number of deaths per thousand adults aged 15-49 for the twenty regions of mainland Tanzania, according to the 1988 Census. The adult mortality rate (AMR) exhibits considerable variation, from 3.4 per thousand in Singida to 9.4 per thousand in Kagera. The AMR is determined by many factors, including the socioeconomic status of the population and the availability of health care. This would lead to differences in rates among regions even in the absence of the AIWS epidemic. How er, as was demonstrated above, HIV infection leads to dramatic increases in adult mortality rates. 'igera region has a markedly higher adult mortality rate than the other regions, which must be in part attributable to the AIDS epidemic. The first case of AIDS in Tanzania was diagnosed in Kagera in 1983, and by 1988 the epidemic was already widespread in several districts of Kagera. There is evidence that mortality has risen in Kagera since the 1978 census: at that time, Kagera and Arusha regions had relatively high life expectancies of 45 and 50, respectively.' However, by 1988, Kager's adult mortality rates were almost three times the rate in Arusha in Figure 1. Surprisingly, the adult mortality rate in Mbeya (6.3 per thousand) is on a pai with Iringa and below that of Morogoro and Coast, which had not been greatly affected by the AIDS epidemic in 1988. The epidemic may not have been sufficiently advanced in Mbeya in 1988 to have had much effect on the adult mortality rate. * For example, the deaths of persons who lived alone and of persons whose households disintegrated are not included in these mortality rates. Further, there is almost always under-reporting of infant and child deaths. Omissions of adult deaths in families where they occurred may be less severe, however. * In 1978, Arusha, Kagera and Morogoro had the highest life expectancies of 50, 45 and 44 years, respectively, while Irings (41 years), Mbeya (41 years) and Ruvuma (43 years) had the lowest (Sembqiwe 1983). Survivor Assistance 5 July 1992 FIGURE 1: Adult mortaity rate, 15-49, per 1000 Twenty regions, mainland Tanzania, 1988 4 ShMsa*ITM 8hi DodI 10 0UZwLh1UwMrTnid Mbe OarMar CSg Region The clearest evidence of the relation between AIDS and adult mortality in Tanzania comes from a comparison of seroprevalence and mortality rates for districts within Kagera region. In Figure 2, the seroprevalence rates for persons 15-54 from the 1987 population-based survey in Kagera (Killewo et al, 1990) are compared with the adult mortality rates in the age group 15-49 obtained from the 1988 Census. The two districts with the lowest seroprevalence rates, Biharamulo and Ngara, also have relatively low adult mortality rates of about 4.6 per thousand. In Muleba district, with a seroprevalence rate of 10.0 percent, the adult mortality rate is almost three times higher at 13.2 per thousand. The highest seroprevalece and adult mortality is found in Bukoba.' Although the seroprevalence rate is somewhat elevated in Karagwe district, it seems not to have affected the adult mortality rate as of 1988. According to Figure 2, a one percentage point increase in seroprevalence is associated with an increase of 0.91 deaths per thousand adults 15-49.8 An increase in the seroprevalence rate of 10 percentage points would be associated with an increase in the adult death rate of 9.1 per thousand. These mortality levels are higher than would be expected by the "rule of thumb" based on developed country data. This suggests that the rate of progression from HIV to AIDS may be faster in Tanzania.9 ' Bukoba Urban and Bukoba Rural districts have been combined because of tho high degree of mobility between them. The coprevalence rate for Bukoba in Figure 2 is an average of the rates of the two districts, weighted by poPulation. 8 The relationship is statistically significant at .02 (T=5.212, 3 df, R2=.90). * Based on the seroprevalence levels, our "Wule of thumb' would have predicted an adult mortality rate of 10 in Muleba and about 11.5 in BMkoba. In both instances the measured mortality rates are higher. If the census data are underestimates, then the true adult mortality levels are higher still. Survivor Assistance 6 July 1992 FIGURE 2: Aoult mortality and seroprevalence Kagera Region, by District, 1988 1480 4.6 4% 48 4.6 .0 District Adult sero rate Adult death rate Source: Over et al (forthcoming) B. Who are the survivors and how are the affected? The increase in mortality of adults in their prime productive years due to AIDS will leave increasing numbers of surviving household members with reduced means of support. Among these survivors will be orphaned children, widows or widowers and the elderly parents of adult 4DS victims. The impact of AIDS on survivors begins before the loss of a productive adult. Wh the onset of full-blown AIDS, the infected individual will become increasingly less able to perform his/her normal activities. Other household members, both adults and children, will be called upon to make up for the lost labor of the patient by working longer hours. There will be additional claims on their time for the care of the person with AIDS who, in his/her final months, may be completely bedridden. Household members will attempt to maintain their current income level by reallocating their time, working longer hours and spending less time in school. Large sums of money and time may be spent seeking medical treatment prior to the diagnosis of AIDS; following the diagnosis even larger sums may be spent in the futile search for an effective cure. Over et al (1988) have estimated these "direct costs* of medical treatment in Tanzania at between US $104-631 per symptomatic HIV-positive person. In this environment, the remaining housenold members will be left with fewer resources for their own consumption, Including food and medical care, and for investments in future productivity, including schooling and agricultural inputs. With the labor to land ratio altered, the household may have to hire additional labor, sell-off land or adopt new technologies. They may have to liquidate their assets, deplete Survivor Assistance 7 July 1992 their savings or borrow. The health and autrition status of the remaining household members will suffer and children will likely be withdrawn from school for financial reasons and because their labor is needed at home. These non-monetary impacts of adult mortality from AIDS In Tanzania have not been quantified.* With the death of the adult AIDS patient, the household permanently loses his/her labor and future Income. The value of this "indirect cost" of AIDS mortality in terms of the foregone earnings of lost members has been estimated at $2425-5093 (In 1985 dollars) per HIV positive person in Tanzania (Over et al, 1988). After the death, the remaining household members face the costs of mourning and funeral arrangements. They may lose entitlement to housing if the deceased was employed by the Government; depending on inheritance practices, the widows of AIDS victims may have difficulty gaining access to their deceased spouse's land and property. The death of the breadwinner may lead to the dissolution of the household, including the fostering out of children to relatives and the migration of widows, or the consolidation of households by remarriage. Throughout the illness and after the death of an adult, the family reorganizes its activities to minimize the impact of fatal illness and to optimize welfare under a new set of constraints imposed by AIDS. These efforts can be termed "coping" behavior and they entail dynamic costs not included in the direct and Indirect costs quantified above (Mujinja et al, 1990). The "costs of coping" will be felt not only by households who lose a productive adult, but also by the extended family and community, as they are asked to contribute to the funerals of neighbors and relatives, assist In the financing of health care, help in field work or the care of the sick and take in orphaned children. In the absence of outside support, familie that agree to care for orphaned children are implicitly reducing the per capita consumption of their households and lowering the resources available for their own children. Thus, both orphaned children and the own children in a foster home are likely to experience a reduction in welfare. The three types of costs discussed above - direct medical treatment costs, inalrect costs in terms of foregone earnings and the costs of coping - are incurred for all chronic, fatal diseases affecting productive adults. However, there are reasons to believe that the impact of AIDS illness and death on survivors may be greater than the impact of death from other causes, like traffic accidents or maternal mortality. First, since it is an infectious disease affecting sexually active adults, AIDS Is more likely than other causes of adult death to kill spouses in the same family, leaving no productive adults to care for remaining elderly and orphans. Second, in the absence.of external assistance programs, households and communities have developed traditional "coping mechanisms" to deal with adult deaths, including remarriage for widows and widowers, fostering of children to relatives, contributions from neighbors and communities for funerals and direct transfers. Many of these traditional coping mechanisms may not be available to survivors of AIDS because of fear that the survivors themselves are infected and may be contagious. Third, the clustering of AIDS in families and communities may completely exhaust any traditional coping mechanisms, even in the absence of discrimination. 1 Research is currently underway in Kagem region to measur the full nature and aagnitude of the impact of fatal adult illness, however. See Over at al (1989). Survivor Assistance 8 July 1992 C. The maMnitude of the survivor problem Effective planning to mitigate the impact of the AIDS epidemic on survivors will require some knowledge of the size and characteristics of survivor groups. There are two approaches to estimating the number of survivors: Vrotecting the number of survivors, based on numerous plausible assumptions; and actually countin survivors at a given point in time, through an exhaustive enumeration or a sample survey (Preble, 1990a). Each method has its advantages and disadvantages. Pfections are the only way to estimate future trends in the number of survivors, but they are based on many uncertain assumptions. For example, models projecting the number of maternal orphans typically require as an input estimates of the prevalence and incidence of HIV among women of childbearing age in urban and rural areas now, in the past and in the future. There are few countries in Africa for which such figures have been reliably measured, and they are the source of considerable debate among epidemiologists and demographers. However, to estimate the number of orphans, one must make additional assumptions for which even less information is available: the average time from contracting HIV infection to the onset of disease; the distribution of AIDS mortality among women of reproductive age; the number and age of children of women who die from AIDS; the underlying infant and child mortality rates in the absence of AIDS; the perinatal transmission rate; and the estimated mortality of AIDS orphans. Projections can demonstrate the degree to which estimates of future numbers of orphans and the long run trends are sensitive to underlying assumptions. Their accuracy depends critically on the accuracy of the assumptions and parameters, however. The alternative, an enameratlon, provides a more accurate estimate of survivors during the current period. There are different ways of collecting such information. A census can be conducted of the entire population of survivors, as has been done for orphans in the Rakai and Masaka districts of Uganda (Hunter, 1990) and in the Kagera region of Tanzania (Rwegarulira and Mushi, 1991). This type of operation is too costly to repeat often, but can provide accurate estimates of the current situation and serve as an input for projections. Once such baseline information has been collected, survivor estimates can be updated periodically at lower cost if some type of monitoring system is adopted. A less costly alternative to a census is enumeration of a representative sample of the population, from which estimates for the entire population can be imputed. Implementing a sample survey depends on the availability of an accurate sampling frame (such as a national census) in order to select a sample of households with known probability. In the case of Tanzania, most of the parameters necessary for projecting the number of survivors - including the past and future prevalence and incidence of HIV among women of childbearing age - are unknown. The information on survivors presented below is based on extrapolations from the 1988 Census of Population. In addition to providing a count of the total population, the Tanzanian census administered a longer questionnaire to 15 percent of households. This questionnaire included detailed information on maternal survival, the mortality of household members and social and demographic characteristics of all individuals. From the census data it is not possible to directly measure the excess mortality caused by AIDS. However, estimates of excess mortality can be made by comparing areas known to be severely affected with those not affected by the epidemic. One of the advantages of using the census data is that it provides an estimate of al survivors of adult mortality, regardless of the cause. The estimated excess number of survivors attributed to the AIDS epidemic can be compared with the number of survivors from all other causes. Survivor Assstance 9 July 1992 Orphans ard other chidren at risk The youngest survivor group is children who have lost one or both parents to AIDS. Traditionally, orphaned children are absorbed into the extended family. However, these coping mechanisms may break down as the clustering of AIDS within extended families creates a growing dependency burden. Instances of adoption by unrelated families are relatively rare in Tanzania and such arrangements may be more difficult for AIDS orphans if they are suspected of being infected. There have been two enumerations of orphans in Tanzania. The first was launched by the Department of Social Welfare in 1988-89 in two districts of Kagera, the hardest-hit region outside of Dar es Salaam (Mutembel 1988, 1989). This study counted roughly 7,000 children aged 0-16 in Bukoba and Muleba districts who had lost at least one parent to AIDS. About 30 percent of these children had lost both.parents.u Among those enumerated, 18.9 percent were under 6 years of age, 32.6 percent were between the ages of 6 and 10.and 48.5 percent were age 11-16. A second enumeration of orphans in the entire Kagera region was conducted in May 1991 by the Social Welfare Office and UNICEF. The enumeration counted 35,291 children under age 15 who had lost one or both parents (Rwegarullra and MuShi, 1991).12 Unfortunately, the first study did not achieve total coverage of Kagera region and neither study included other regions of Tanzania that have been hard hit, such as Dar es Salaam and Mbeya. It has also been suggested that some communities may have exaggerated the number of orphans in hope of obtaining assistance. Even if these numbers are accurate, it is not clear how they compare with the underlying orphan rate in the absence of AIDS. In order to put into perspective the impact of the AIDS epidemic on the population of orphans, we need estimates of the underlying orphan rates in the country in the absence of AIDS and the excess orphaning caused by the AIDS epidemic. The remainder of this section is devoted to estimating the numbers of orphans due to AIDS, based on data from the 1988 Census. The census questionnaire asked a subsample of households about the survival status of each individual's mother. From this information we have estimated the number and proportion of all children who are orphaned from their mothers (or "maternal orphans"). We then estimate the number of paternal and two-parent orphans. Finally, we estimate the excess number of orphans in Kagera region due to the AIDS epidemic. ' Materi rphn. In 1988, about 210,000 children in Mainland Tanzania under age 15 had lost their mothers - 1.96 percent of children under 15. Mwanza had the largest number of maternal orphans (20,238), followed by Kagera (18,125), Shinyanga (16,927), Iringa (13,095), Morogoro (12,073) and Tanga (11,565) (see Figure 3). In Figure 4, the regions of Mainland Tanzania are ordered from lowest to highest adult mortality rates in the year preceding the census. The maternal orphan rate for children 0-14 ranged from a low of 1.2 percent in Kilimanjaro, to more than twice that percentage in This enumeration has been continued by Sister Deborah Brycke of the Evangelical Lutheran Church of Tanzania; as of August 1990 she had enumerated 17,000 orphans over a much larger area. However, it is not known what proportion of these children were orphaned due to AIDS. 12 An additional 12,795 orphaned children 15-21 were also counted. However, this is thought to be a severe undercount of orphans in this age group. Those who had married, for example, were not included. Srvvor Assitnwwe 10 July 1992 FIGURE 3: Number of maternal orphans, 0-14 by region, mainland Tanzania, 1988 1892 1802 ai A&u aRW uT & D a o coW me unuwar a asO co.n Region Kagera (2.92 percent).33 In Figure 5, maternal orphan rates for six regions are presented by age group. As children age, so do their parents. Thus, the older is the child, the greater is the likelihood of being an orphan. While the percent of orphaned children 0-4 does not seem to have any relation to the level of adult mortality, for both of the older groups of children the positive relation between the orphan rate and adult mortality rate is statistically signiticant." Although all regions have now reported AIDS cases, seroprevalence levels in most were still relatively low in 1988 and would not be expected to have had an appreciable impact on adult mortality rates or orphan rates. The epidemic has been raging longest in Kagera, followed by Dar and Mbeya. However, since AIDS mortality trails the seroprevalence rate by a few years, adult mortality levels in Dar and Mbeya might not have been notably affected by the seroprevalence rate in 1988. Thus, for at least seventeen and possibly nineteen of the regions in Figures 3-5, the maternal orphan rates reflect levels in the absence of AIDS. The 0-14 maternal orphan rate for Kamera (2.92 percent) is roughly one percentae point hiher than the mean for the mainland (1.96 percent). or about 50 gercent higher than the rate in reion unaffected . by the epidemic in 1988. 3 The relationship between the 0-14 orphan rate and the 15-49 adult mortality rate (AMR) for these 20 egions is statistically significant at .01 (=3.26, 18 df): an increase in the AMR of 1 per 1000 is associated with an increase of 0.14 percent in the maternal orphan rate (1V = .37). 14 An increase in the AMR of 1 per 1000 is associated with an increase of 0.21 percent in the orphan rate for children 5-9 (significant at 5 percent, R2 = .67) and of 0.28 percent in the rate for children 10-14 (significant at 10 percent, R2 = .63). The reason for a lack of relation at younger ages is probably due to the fact that children who are orphaned at a very early age have higher mortality and are therefore less likely to be observed. A study of orphans in neighboring Uganda also notes the dearth of orphans in the youngest age groups (Hunter, 1990). Survvor Assstance 11 July 1992 FIGURE 4: Adult mortallty and orphan rates* Twenty regions, mainland Tanzania, 1988 10 Region 2.11 17 LJNOA MWU~A MARA TANGA TAA AOMA OOA MNRA Regon Orphan reas Adult death rate Chfide who hav lost ~hi mo#hu Survivor Assistance 12 July 1992 Within Kagera region, the maternal orphan rate ranges from a low of 2.28 percent in Biharamulo (with the lowest seroprevalence and adult mortality levels) to 3.69 percent in Bukoba (with the highest levels)(see Figure 6)." 1* The orphan rate for young children aged 0-4 has no correlation with the adult mortality rate (see Figure 7). However, the adult mortality rate is positively correlated with the orphan rate for children 5-9 (coefficient of .142) and for children 10-14 (coefficient of .113). Of the 18,125 maternal orphans in Kagera, 10,362 (57 percent) were located in the most heavily populated and highly infected districts of Bukoba and Muleba (see Figure 8). Altogether, 13,284 of the maternal orphans in Kagera (73 percent) were of school age (7-14). Comparing the average of the 0-14 maternal orphan rate for the three least-affected districts where adult mortality was not yet much affected by the epidemic (2.42 percent) with the rates for the two most seriously affected districts, Mulesbaa an orphan rate 0.71 percentage points higher than exoected and Bukoba (urban and rurall a rate 1.27 points higher than expected. This implies an excess of 892 maternal orphans in Muleba and 2.210 in Bukoba due to the AIDS epidemic. Thus. Rossibly about 30 percent of all maternal orphans in Muleba and Bukoba were due to AIDS in 1988 and, assuming few AIDS orphans in the other districts, about 17 percent of all maternal orphans in Kagera were due to AIDS.' Total ophans. The census data underestimate the total munber of orphans because children orphaned from their father were not identified. Furthermore, they provide no guidance on the number of orphans who have lost bth natural parents and who are presumably most vulnerable. Because AIDS in Tanzania is spread primarily through heterosexual intercourse, we might expect that a higher proportion of AIDS orphans than children orphaned from other causes will have lost both parents. Given the maternal orphan rate and various assumptions about the share of all orphans who have lost both parents, we have estimated the total number of children orphaned from one or both parents (see Table 2). The number and proportion of children orphaned are estimated for the twenty regions of mainland Tanzania and within Kagera region, under four scenarios. In all of the scenarios, it is assumed that the paternal orphan rate is twice the maternal orphan rate."' The scenarios differ in their 15 An increase in the AMR of 1 per thousand is associated with an increase in the 0-14 orphan rate of 0.11 percent (1V = .86); an increase in the seroprevalence rate of one percent is associated with an increase in the orphan rate of 0.09 percent (1e - .77). Both relations are statistically significant at 5 percent. " Based on estimates of mortality of women of reproductive age, Preble (1990a) estimates that in 1990 0.4 - 1.0 percent of children under 15 were orphaned from their mothers due to AIDS in 10 central African countries, including Tamania, and that by the end of this century the figure will escalate to 6-11 percent. The maternal orphan rate of 3.69 percent for children 0-14 in Bukoba district (urban and rural) in 1988, where seroprevalence in 1987 was 14.8 percent, exceeds the rates implied by UNICEF projections for 1990. However, it must be kept in mind that the rate for Bukoba includes children orphaned for reasons other than AIDS. n7 Note, however, that these two districts are considered the more prosperous within Kagera region, so the underlying maternal orphan rate in the absence of AIDS may well have been lge than the mean of the low- seroprevalence districts. 8 This assumption is based on the results of the 1969 Ugandan Census, which asked about orphaning from both parents. The ratio of paternal to maternal orphan rates for ages 0-14 was 1.92 (Republic of Uganda, 1976). The 1977 Census of Malawi found a paternal to maternal orphan ratio of 1.72 (Malawi Government, 1977). Household surveys in Cote d'Ivoire (1985) and Ghana (1987), conducted before the onset of the AIDS epidemic there found ratios of 1.5 in Cote d'Ivoire and 2.2 in Ghana. There are several reasons why we might expect more children to be fatherless than motherless. First, since children whose mothers die have very high mortality, we would expect proportionately fewer to survive to report the mother's death. Second, fathers are older than mothers on average and therefore have higher mortality. Third, in Kagera region a war was fought in 1979, presumably elevating male Sunvor Assstwance 13 July 1992 FIGURE 5: Orphan rates by age group Six reglons of Tanzania, 1988 4~~ .-...-........e--... 40, Regdon ChMre 0-4 MChMdrena MUj Chldr 1 FIGURE 6: Orphan rates and adult mortdlty Kagera Reglon, by District, 1987-88 1 -. ..-10 27 6 0 ChhimldNgren - ICilr en 5 AMb 8hdeinb10W1 Orphan rate Adult death ate mortality. Survivor Assistance 14 Jly 1992 FIGURE 7: Orphan rates by age and adult mortality Six districts of Kagera, 1988 10 0a0 .. . . . 1 District Chidren 04Chdren 5.9 ChWdrn 10-14 E Adult death rats FIGURE 8: Number of maternal orphans, 0-14 Kagera region, by district, 1988 N~o em 1eoS District Survivor Assistance 15 July 1992 assumptions about the proportion of orphans who have lost both parents. In the first scenario it is assumed that being orphaned from the mother and the father are independent events, so that the probability of being orphaned from both parents is the product of being orphaned from each. The second through fourth scenarios assume that the share of orphans who have lost both parents Is 10, 20 and 30 percent of all orphans, respectively. The fourth scenario corresponds to the share of two-parent orphans found in Kagera by Mutembet (1988, 1989). Of course, the average share of two-parent orphans, even in a population suffering from AIDS, should be lower than that exclusively among AIDS orphans. Details on the calculations behind Table 2 are in Annex 1. The effect of the AIDS epidemic is to raise the maternal orphan rate and also raise the proportion of children losing both parents. As is evident from reading across the scenarios in Table 2, for a given maternal orphan rate, an increase in the proportion of children who have lost both parents raises the rate of orphaning from both parents but lowers the orphan rate from at least one parent. Which of these scenarios Is most likely? A 1989 census of orphans in Uganda's Rakal district, adjacent to Kagera, found that an estimated 12.8 percent of all children 0-18 had lost at least one parent (Hunter, 1990). Among orphans, 23 percent had lost both parents.19 A similar census in Hoima district, which was left relatively untouched by the war and is still relatively free of AIDS, found a 0-18 orphan rate of 4.4 percent with 12 percent of orphans having lost both parents.0 The inclusion of older children in the Ugandan studies would tend to raise the orphan rates. Based on the Ugandan studies, the assumptions for scenario 2 seem most plausible for areas of Tanzania in the absence of AIDS and the assumptions for scenario 3 are most plausible for areas already heavily infected. This implies that the 0-14 orphan rate from either parent for mainland Tanzania in 1988 was 5.3 percent (see Table 3). Of the 545,000 orphans, 10.6 percent (57,637) had lost both parents. The two-parent orphan rate in the general population was thus 0.53 percent. Kae IO h 46604 orphans 0-14. for a rate of7.5 percent; 16.7 percent of the Kagera orphans (7,771) had lost both parents. The two-parent orphan rate in Kagera was 1.25 percent of children 0-14. 19 The studies in Uganda enumerated orphans but did not enumerate other children. The last complete census in Uganda was conducted in 1970: one of the 2 areas enumerated (Rakai) suffered a war in 1979, which surely altered the age structure of the population. Therefore, the orphan rate can only be estimated, based on an esdmated number of children under 18 in the population. The accuracy of this orphan rate depends on the accuracy of the estimate of the number of children in the population at large. x For comparison, in the 1977 Malawi Census, 4.5 percent of children 0-14 had lost either their mother or their father and 0.5 percent has lost both, for a total orphan rate of 5.0 percent. Thus, 10.2 percent of all orphaned children had lost both parents. The baseline orphan rates in West Africa seem higher. The Cote d'Ivoe Living Standards Survey, conducted in 1985 before the onset of the AIDS epidemic in that country, found that 6.3 percent of children had lost one parent and 0.5 percent had lost both parents. Among orphaned children 7-14, 7 percent had lost both parents. Among children age 0-14 surveyed in the 1987 Ghana Living Standards Survey, 6.7 percent had lost one parent and 0.4 percent had lost both, for a total orphan rate of 7.1 percent. 勵〞編絨〞油州h朋卹16二為今j夠吃 ’…〔〕〕！丰〕〕一‘ SwWvor Assbrance July 1992 TAM 31 Natimatgo of the total nobit of orphans- from eithat garent, mainlnfi Taggania - - --- - --- - --------------------------- -- ------- TOTAL NWWR ORPHAN 2-PARENT QB?JjW REGION/DISTRICT or ORPHANS 0-14 RATE (0) NM48ER RATE (%) ------------------------------------------------------------------- - Arusha 25,978 4.13 2,S79 .41 Coast IS,IS6 S.46 I'Sis SS Dar on Salaam 21,428 4.19 2,143 .42 Dodooka 2S,789 4.S4 2,S79 .4S Irings, 3S,714 6.31 3,571 .63 Zagora 46,604 7.51 7,771 1.25 Kigama 24,396 5.76 2,439 .58 K414manjaro 17,242 3.36 1,724 .34 Lindi 14,OOS 5.11 1,400 Sl Mara 26,S69 5.70 2,657 .57 Mboya 30,895 4.61 3,089 .46 x0rogoro 32,927 6.10 3,292 .61 Mtwara 22,252 6.18 2,225 .62 Mwanza 55,19S 6.39 5'S19 .64 Rukwa 13,596 3.99 1,359 .40 Ruvuma 18,393 5.42 1,839 .54 Shinyanga 46,165 5.18 4,616 S4 Singida 16,983 4.67 1,698 S2 Tabora 24,491 5.14 2,449 .47 Tanga 31,541 5.49 3,154 .55 20 RRGIONS 545,318 S.31 -57,637 .56 DLstricts of Ka-gerat Biharamulo 6,379 6.22 638 .62 Ngara S,716 7.44 572 .74 Karagwe 9,076 6.44 908 .64 Muleba 9,663 7.69 2,148 1.71 Bukoba 15,770 9.06 3,505 2.01 ------ --------- -------------------- ---------- Rum gcpb= din to AEDS in NU=, To estimate the excess number of orphans in Kagera due to the AIDS epidemic, we assume that in 1988 AIDS had a negligible impact on orphan rates in the lowsetoprevalence, 1OW-1n0ttality districts of Kagera (Biharamulo, Npra, Karagwe). The estimated excess number of orphans due to AMS in Muleba and Bukoba districts is the diffiam between the number of orphans estimated in Table 3 and the number of orphans there would be H the avenge maternal orphan rate were only 2.42 percent (the avenge of the other 3 districts). TbIs yields the result din W 1988. 5,648 glyhm (Qj 22.2 U=W qff all gMb= fi-orn either VAMW in Bukoba Muleba gM be anflbuted to AIDS. Sixty-five percent (3,674) of -children who had lost both parents and 10 percent (1,,974) of children who are orphaned from one parent can be aftnnted to the AMS epidemic.'" In the absence of AUDS, the number of orphans In Kagera would have been 40,956 instead This is- don of the number of chiW en counted by Mutembei (1988, 1989); however, he included older children in his study and conducted bis count up to 6 mouft after the census. By 1990, the numbers of orphnns would have grown. Survivor AssAtance 18 July 1992 of 46,604.u s . 12.1 pRe of all orpans in Kae are 91uaed to have be caused by the ADS There is no reason to believe that the number of AIDS orphans has remained stable In Kagera since 1988, however. Increasing numbers of infected persons are succumbing to AIDS - due both to past infection and an annual Incidence rate for new cases estimated at 1.4 percent (Killewo et al, 1990). In her projections of orphan rates for ten hard-hit Sub-Saharan countries, Preble (1990a) notes that although the orphan burden appears "modest" in 1990, a dramatic increase is projected by the end of the decade. Other children at risk. Our earlier projections estimated that there were 18,125 maternal orphans age 0-14 in Kagera Region in 1988. There were also 21,266 non-orphaned children age 0-14 living in households that have taken in maternal orphans. To the extent that orphans deplete the resources available to all household members, these children, too, are also indirectly affected by AIDS mortality. idow and widowers A second group of survivors potentially in need of assistance are widows and widowers. AIDS widows presumably may have lost their major means of support and have few or no Inheritance rights over the assets of their deceased husbands. Since household members often specialize in different modes of production, widows may be ill-equipped to continue the productive activities of their husbands effectively. They may be blamed by others for the death of their spoust, lose access to their children /and have difficulty remarrying. Widowers, although generally more able to hold onto their property, often have difficulty running the household, cooking, shopping and caring for their children if their marriage was monogamous (Bertozzi et al, 1990). Unlike widows and widowers from other causes, those whose spouses have died of AIDS face the additional burden of knowing that they themselves may be infected and may face an extended period of illness and short life expectancy. Since AIDS is selectively killing prime-aged adults, we might expect AIDS mortality to have an impact on widowing rates in the 20-49 age group. Figure 9 presents adult mortality rates and the percent of adults who are widows or widowers in the age group 20-49 for six regions of Tanzania. The "widowing rates" range from a low of 1.78 percent in Ruvuma to 3.6 percent in Iringa. There appears to be no correlation between adult mortality rates and widowing rates, however. Determinants of marital status is complicated by the fact that widows and widowers remarry and migrate. The more rapid the uturnover" or remarriage, the lower the widowing rate will appear, even if mortality Is high. The widowing rate in a region suffering from AIDS may be further depressed by outmigration of AIDS widows and widowers for remarriage. An additional consideration is that many widows or widowers of AIDS victims are presumably infected and may themselves die of AIDS, reducing the population of widows and widowers. Assessing the number of spouses who are survivors of the AIDS epidemic clearly requires more information than is provided by the widowing rates. In the interest of preventing the spread of 2 This is far larger than the results of the May 1991 orphan enumeration in Kagera, which counted about 35,000 orphans under 15. The largest potential sources of error in our calculations are: (a) the assumption of twice as many paternal as maternal orphans; and (b) the inability to control for migration. However, it is also possible that the 1991 enumeration was incomplete. The results were not based on a door-to-door count of orphans, but rather on a form completed by village authorities. Survvor Assisance 19 July 1992 AIDS, policymakers may actually want to support AIDS widows and widowers so that they need 10 resort to the traditional coping mechanism of remarriage. This would tend to ake the widowing rate from AIDS, but in this context would be a desirable outcome. FIGURE 9: Adult mortality and widowing rates Six regions of Tanzania, 198 .10 OS* -1- o rf .3 I-2 Region m Wdowfng rae * Autdea r7ae The elderfy One of the traditional rationales for having large families in Africa is to provide for parents' security in old age. In heavily-hit areas, however, AIDS has wiped out the entire generation of productive adults within the same family, leaving the elderly to care for their grandchildren and with no source of support.2 These "elderly orphans" often have few resources and insufficent energy to generate the income necessary to support other survivors. Yet they are often the caretakers of AIDS orphans. 4 No questions were asked on the 1988 census that would allow estimation of the number of elderly orphans. Mutembel's (1989) enumeration of Bukoba district found 1,084 persons over the age of 50 who had lost their adult children to AIDS. Fifty-nine percent were between 50-69 and the remainder were 70 or older. In one division of Muleba (Nshamba) there were 101 elderly orphans. a Bertozzi et at (1990) cite the example of a girl in one of their focus groups who *had lost her mother in 1988, her father in 1989, and her paternal uncle the week prior to the interview. Her uncle's widow is ill and all 9 cousins have moved in with their grandfather who already had his 9 youngest children (of 13) living with him." (p.10) Survivor Assistance 20 July 1992 Survivor households Households may be easier to target for assistance programs than individuals and, to the extent that adukt deaths from AIDS affect people through their Impact on the household as an economic and social unit, it may be moie appropriate to address survivor assistance to households.2 The analysis below considers two groups of survivor households of interest to public policy: households that experienced an adult death and those with maternal orphans.25 These points are again illustrated with the 1988 census data for the Kagera Region. Households with an adult death. An estimated 4,769 households in the Kagera region had suffered an adult death in the ages 15-49 during the 12 months preceding the 1988 census. Almost half of these households (45.9 percent) were located in the high seroprevalence districts of Bukoba urban and Bukoba rural (see Figure 10). For the Kagera Region as a whole, 1.77 percent of households suffered an adult death in the age range 15-49; this varied by district, from fewer than one percent of households in Ngara (0.85 percent) and Karagwe (0.93 percent), to about one percent in Biharamulo (1.11 percent), and more than two percent in high-setoprevalence Bukoba urban (2.72 percent), Bukoba rural (2.52 percent) and Muleba (2.42 percent). FIGURE 10: Distribution of Households w/an Adult Death by District, Kagera Region, 1988 Ngara Karagwe Biharamulo c(6.3a Muleba 3ukoba Rural Bukoba Urban 7 Note, however, that households are dynamic. AIDS deaths may weaken an existing household, cause it to dissolve or cause two households to merge. Assistance strategies directed to either households or individuals thus have the potential to influence household composition. 5 The analysis here will refer only to households with a maternal orphan; although we attempted to estimate earlier the total number of maternal and paternal orphans, attributing them to households is far more complex and beyond the scope of this paper. This would involve assumptions about the distribution of paternal orphans across households - for example, how many households have at least one, how many have more than one, etc. SurvIvor Assistance 21 July 1992 The first two columns of Table 4 compare the characteristics of Kageran households that did and did not experience an adult death. More than a fifth of the households that had an adult death also had maternal orphans, while only 3.6 percent of households that did not have an adult death had maternal orphans. Although these differences are large and highly significant, it is nonetheless surprising that a greater share of households with an adult death in the 15-49 age range did not leave behind maternal orphans. Possible explanations are: under-reporting of maternal orphans in instances where they are being cared for by foster mothers; higher mortality among men in the 15-49 age range than among women; or higher mortality among orphans who lose their mothers, leaving fewer surviving maternal orphans. On the other hand, if the ratio of paternal to maternal orphans is really on the order of 2:1 (as assumed in the calculations earlier), then we might expect a relatively smaller share of households to leave behind maternal as opposed to paternal orphans. Households that had an adult death were significantly more likely to live in larger dwellings and to own their nousing and less likely to rent, compared with households that did not have an adult death. Households with an adult death were more likely to be headed by a female (a finding consistent with the seemingly low share of households with a maternal orphan). The heads of households that had an adult death were on the whole less likely to have been schooled and had received less schooling than those without an adult death. The results in Table 4 are difficult to interpret because they may represent conditions that led to the adult death or conditions resulting from an adult death. Furthermore, two-way tabulations may be deceptive because of the failure to control for other confounding variables. For example, one reason why the heads of households with a death may seem less schooled is because they are more likely to be headed by women and elderly persons after the death of the main breadwinner. It is impossible to do TABLE 4: Characteristics of Households with maternal orphans and with a recent adult deatha Among Households... Percent of With an Without an With a Without a households... adult death adult death maternal orphan maternal orphan With maternal orphans 21.14 3.61 100.00 0.00 With an adult death 100.00 0.00 9.72 1.48 With 4 or more rooms 61.36 43.42 59.11 43.11 That rent housing 8.87 16.80 6.98 17.05 That own housing 87.85 80.04 90.00 79.78 With female head 39.42 31.13 28.88 b 31.37 b With unschooled head 39.98 38.43 41.79 38.33 With head w/primary 31.83 28.39 32.01 28.31 With head w/secondary 28.19 33.18 26.20 33.37 n 823 44,728 1,790 43,761 - Unless otherwise noted, differences are statistically significant at at least .0001. b Difference is significant at .02. Survyor Asstance 22 July 1992 away with the *cause vs. effect" problem with the data at hand; this would require longitudinal observation of households. The second problem can be dealt with in a multivariate analysis, however. The first column of Table 5 provides the results of an ordinary least-squares regression of whether a household had an adult death in the ages 15-49 on available characteristics from the census. The results show that households with a death had significantly older heads that were more likely to be female. Households in Bukoba rural, Bukoba urban and Muleba were significantly more likely to have had an adult death than households in Biharamulo (the omitted region). Households with an adult death also had more *own" children (those who were not maternal orphans) than households without an adult death. The schooling of the household head is not significantly related to the likelihood of an adult death. These variables explain less than I percent of the variation in the dependent variable, however. Unfortunately, variables potentially measuring the welfare impact of an adult death were not collected in the census. Finally, although not in either table, about one fifth of the households experiencing an adult death (21.9 percent) had no remaining adults in the economically productive age group 20-49. This is not much greater than in the population at large, however, in which 18.5 percent of households had no productive adult. The share of households experiencing an adult death and without a productive adult is also highest in Bukoba rural district (27.8 percent), with the highest seroprevalence rate among rural districts. TABLE 5: Multivariate analysis of characteristics of survivor households in Kagera region Dgoendent Variable (1) (2) Household with Household with an adult death a maternal orchan Explanatory variables Coefcient T-Statistio Coefficient T-Statistio Age of household head 0.00033 8.75 0.00054 9.84 Years of schooling of head -0.00007 -0.40 0.00019 0.74 Female head 0.0051 3.69 -0.0055 -2.76 Number of 'non' orphans in household 0.0015 4.84 -0.026 -6.05 Household had an adult death (15-49) .173 25.51 KaMgwe -0.0012 -0.54 -0.0042 -1.38 Bukoba Rural 0.013 6.11 0.0069 2.22 Muleba 0.012 5.37 -0.00084 -0.27 Ngara -0.0025 -1.10 -0.0038 -1.18 Bukobs Urban 0.018 8.44 -0.013 -4.31 Intercept -0.0078 -3.03 0.022 5.94 R2 .007 .019 Notes: Ordinary least squares regressions. Dependent variables are defined as follows: household with an adult death = 1 if there was a death, otherwise = 0; household with a maternal orphan = 1 if the household has one or more maternat orphans, otherwise = 0. The left-out district in the regressions is Biharamulo. Sample size is 45,551 Survivor Assistance 23 July 1992 Housld haraloM n. An estimated 11,461 households In the Kagera region, or 4.25 percent of all households, had at least one maternal orphan under 15 years of age in 1988. This is almost three times the number of households having experienced an adult death in the past 12 months, due to the fact that many children were orphaned in previous years and because a single adult death may result in fostering orphan siblings to different households. More than a third of the households with a maternal orphan were located in Bukoba rural, while more than a fifth were located in Muleba district (see Figure 11). On average, there were 1.56 maternal orphans in households with at least one maternal orphan. The proportion of households with a maternal orphan ranged from fewer than 4 percent in Biharamulo (3.88 percent), Ngara (3.65 percent) and Karagwe districts (3.49 percent) to more than 4 percent In Muleba (4.44 percent) and the combined Bukoba urban and rural (5.03 percent).? FIGURE 11: Distribution of Households with Orphans * by District, Kagera Region, 198 Ngara Karagwe Biharamulo oeer Muleba Bukoba Rural Bukoba Urban *cmenofl-14 wthnvelatewaimowrs. The second two columns in Table 4 compare the characteristics of Kageran households with and without a maternal orphan. Almost 10 percent of the households with maternal orphans had also experienced an adult death in the past 12 months, compared to only 1.5 percent of households without a maternal orphan. Many of the maternal orphans might have resulted from deaths in previous years in the same household; it may also be the case that maternal orphans have been fostered to the families of relatives. Households that have taken in a maternal orphan are likely to have more rooms, are less likely to rent and more likely to own their housing. This suggests that households taking in an orphan may be wealthier than those that do not. Households with a maternal orphan are less likely to be headed by a female but the head is also less likely to be schooled than households without a maternal orphan. In the multivariate analysis of Table 5, we see that households with maternal orphans are more likely to have had an adult death and to have an older head that is not female. Households in Bukoba rural are * Th figures for Bukoba urban and rural are 3.06 percent and 5.30 percent, respectively. Survivor Assistance 24 July 1992 significantly more likely to have taken in a maternal orphan and those in Bukoba urban less likely to have taken in a maternal orphan than households in Biharamulo. Thus, households in both Bukoba urban and Bukoba rural are more likely to have had a death, but households in Bukoba urban are far less likely to have taken in an orphan. The more *own" children in a household, the less likely It is to take In a maternal orphan. Again, however, only a small proportion of the variation is explained by these variables (2 percent). D. The size of survivor groups: a summary Table 6 summarizes the estimated size of different survivor groups in the Kagera Region in 1988, based on the analysis of this section. These are clearly underestimates of the current size of these groups, since they are three years out of date. It was also not possible to estimate the number of households with paternal orphans. As was mentioned above, a large share of these survivors are not linked to the AIDS epidemic, but are the result of underlying high adult mortality rates in the population of Tanzania. The overwhelming message of Table 6 is that, even accounting for underestimates, there were large numbers of persons affected by adult mortality in the Kagera Region alone. The resources do not exist to help such large numbers of people in one region, let alone the other nineteen regions of Tanzania.2 We have not been able to examine the welfare levels of these households and individuals to assess the consequences of adult deaths; there are without a doubt many very needy families that have not experienced a recent AIDS death and other families that have experienced AIDS or other deaths but are not in need of assistance. Using "survivor status" as a targeting criterion for assistance is likely to be very inefficient in channelling the assistance to those most in need. Additional objective criteria will be necessary to identify the neediest individuals and households among survivors. TABLE 6: Summary of some survivor groups. Kagera Reson. 1988 Survivor aM Size Ofr 1. Individuals Maternal orphans 18,125 Children orphaned from one or both parents 46,604 Children orphaned from both parents 7,771 Non-orphaned children in households with maternal orphans 21,266 2. Households Households with a recent adult death 4,769 Households with a maternal orphan 11,461 2 Recall that the total estimated number of orphans from one or both parents in 1988 came to 545,318; only Kagers evidenced relatively elevated orphan rates indicative of some impact from the AIDS epidemic. Strvivor Assistance 25 July 1992 III. STRAT ES TO ASSIST SUR The AIDS epidemic will throw more Tanzanians into poverty and augment the numbers of orphans and elderly dependents In the population." The types of programs that could be Initiated to help the survivors are familiar development programs designed to reduce poverty, such as: * Provision of community services, such as water supply, day care, schooling and medical care * Subsidies of schooling, medical care, housing and basic services * Transfers of food, clothing or cash * Technical training * Orphanages and boarding schools * Income-generating projects * Free or subsidized credit * Free or subsidized farm inputs * Legal action The strategy for survivor assistance thus becomes one of selecting among these alternative interventions and deciding to whom the interventions will be addressed. The resources available to implement these programs will be scarce, however. Rational decisions among the alternatives require an assessment and comparison of their costs and benefits. The most efficient program will be that which siomitaneously maximizes the value of program benefits for a given cost and minimizes the cost of achieving a given level of program benefits, compared with other programs. In addition to cost-benefit calculations, additional criteria may enter into the selection process, such as the distribution of benefits among beneficiaries, the number of persons who can be helped and the sustainability of programs. A. Criteria for selecting Interventions Costr and ben(fjtr Three types of information are needed to evaluate the costs and benefits of alternative programs: * The costs of the intervention * The effectiveness of the intervention (the relation between the intervention and its benefits) * The value of the benefits of the intervention a This is the major impact insofar as the welfare of the survivors is concerned. There will be many other short- and long-term macroeconomic impacts, however, such as redistribution of income, reduced availability of scarce skills that could reduce growth and productivity in the short run, labor scarcity in the production of cash crops that could reduce the volume of crop exports, etc. Skrvivor Assistance 26 July 1992 Calculating these three pieces of information is not straightforward, however. The benefits of programs to assist survivors are multi-faceted, difficult to quantify and to value. For example, a program that provides Improved water supply will Improve health and sanitation, save time in home production and indirectly raise Incomes, In the current period and for several years to come, for Individuals, households and communities. To properly evaluate these benefits for comparison with other programs, one would have to quantify all benefits for each period, attach a value to them, discount the benefits In future periods and sum all of the discounted benefits across all periods. One way of simplifying the problem is to dispense with valuing the benefits and evaluate the costs of achieving a common, specified benefit across all programs. For example, the objective of raising orphan school enrollments to 100 percent could be achieved by: raising incomes through income- g,nerating projects; subsidizing school fees; organizing day care centers that would free up children's time; or constructing boarding schools for orphans. The program selected would be the most aaL- effectiv. that is, the one that minimizes the cost of obtaining a given effect, in this case the enrollment target for orphans. Cost-effectiveness studies are often used to evaluate health programs, precisely because of the difficulty in attributing a value to health improvements (Dunlop, 1975). However, the cost-effectiveness approach is not a *shortcut" for evaluating programs with multiple effects aimed at reducing poverty. The ideal effect to compare across all programs would be a given improvement in welfare. Thus, one is still faced with the problem of quantifying different types of intermediate effects and knowing by how much they are likely to raise incomes. Further, the effectiveness of different antipoverty programs in various settings is not often known; research is necessary to establish these parameters. Tauget groups of benefdaries For any given intervention, the costs, benefits and effectiveness can be expected to vary, depending on the characteristics of the recipients and the size of the beneficiary group. For example, survivor assistance programs could be offered to AIDS orphans, all orphans regardless of the cause, orphans who have lost both parents, families that have taken in orphans, low-income families that have taken in orphans, communities that have been hard hit, and so forth. Each of these alternatives will entail different costs, achieve different levels of effectiveness and be directed at a different te although the objectives may be the same. The cost-effectiveness of programs will be enhanced to the extent that they can be targeted at the survivors who can benefit the most or who are in greatest need, with minimum "leakage'. Thus, the two key issues in defining target groups for donor assistance are: * Who are the persons or groups most in need of assistance or likely to benefit the most? * How can they be identified for program assistance? Effective targeting of interventions to the neediest groups requires ta g criteri that correctly identify as many of the intended beneficiaries as possible while excluding persons who are not supposed to benefit. The criteria should be easy and inexpensive to verify, and not easily circumvented by opportunistic behavior of potential beneficiaries. For example, if subsidies were offered to all orphans not enrolled in school, then families with orphans already enrolled would be tempted to disenroll" them to receive a subsidy for their re-enrollment. The cost-effectiveness of the program would be diminished, Suv1wr Assistance 27 July 1992 since many families who could afford to pay enrollment fees would be subsidized while the enrollment of orphans from these families would essentially be unaffected. In this example, the targeting criteria are easily verified but are also easily circumvented by the opportunistic behavior of potential beneficiaries. Assistance can be channelled directly to individuals, through households or through communities. Many households of survivors are already receiving assistance from relatives and neighbors in the form of transfers, gifts, labor assistance and taking in the children of victims. The most efficient programs will reinforce these traditional modes of assistance, not replace them. Some studies have advocated using AIDS as a targeting criterion for survivor assistance strategies - that is, targeting AIDS orphans, AIDS widows or families associated with AIDS survivors (Mutembel 1988, 1989; Rutayuga, 1990). Targeting based on AIDS raises at least two ethical issues, however. First, it may reinforce the stigmatization of survivors. Second, targeting based on AIDS will not necessarily identify the neediest people. There are many needy people in Tanzania who do not live in AIDS-afflicted households (including those in households that have taken in AIDS orphans) and some AIDS survivors come from relatively well-off families that are capable of absorbing them.* These problems highlight the need to develop other targeting criteria to identify the neediest persons In areas affected by AIDS. One approach for rural areas would be to develop programs for entire communities that are hard-hit, without distinguishing between families that have and have not suffered from AIDS. These programs could be designed to assist the entire community (e.g., by providing a community water supply) or channelled to the neediest persons according to the information of community leaders (e.g., food and clothing for the neediest families). By working though local organizations, such programs would reinforce community coping mechanisms and local responsibility for the success of interventions. The high mobility and heterogeneity of urban dwellers makes coununity- based argeting problematic in urban areas, however. Swatumai0y ofpogwras The AIDS epidemic is likely to have an impact over at least the next few decades in Tanzania. To ensure their sustainability, survivor assistance programs should be selected that can be maintained at low cost with minimal reliance on external funds, in order to reinforce the self-sufficiency of target individuals and communities. 2 Bertozzi et at (1990) report that in two hard-hit Tanzanian villages adjacent to the border with Uganda, "the average orphan was living in better conditions than the average child with living parents*, a situation attributed to the fact that many of the families first struck by the AIDS epidemic were those of elatively affluent traders. (p. 15) Survivor Assistance 28 July 1992 B. Some eaIme of svir assistance PMrogrm To illustrate these issues in the evaluation of survivor assistance programs, the remaining part of this section examines more closely several types of programs already being implemented in Tanzania or that are under consideration by the Government or NGOs: orphanages; direct relief; child care/feeding posts; and subsidization of school fees. These particular programs are discussed for illustrative purposes only; they have been selected because they are under discussion or already in existence, wot because they are thought to be more cost-effective or better targeted than other ongoing or potential programs. Indeed, despite experience implementing these programs in Tanzania, there is a remarkable dearth of information about their benefits, effectiveness or the criteria for eligibility. For each program of survivor assistance we present: rationale (the problem addressed and the expected benefits of the program); an overview of existing programs; the costs of existing or hypothetical programs; and a discussion of issues in the evaluation of each program. Orphaages BM Im. There are several reasons why the AIDS epidemic might dramatically increase the need for orphanages in Tanzania. First, it is creating unprecedented numbers of children orphaned from both parents in the severely affected zones, creating demands on the extended family that may exceed their ability to cope. Second, because of the potential stigma attached to AIDS orphans, including the fear by adoptive parents that the children themselves might be infected, even relatives may be reluctant to take them in. If the epidemic produces many children with no parents who are unadoptable and for whom there is no alternative, orphanages may have to be considered as an option. Even in the hardest-hit area, however, this does not seem to have happened yet (see below). isting ogram. As of September 1990, there were 34 orphanages in Tanzania, sheltering 1,083 children, ages 0-18 (see Annex 2). All but one of these orphanages is run by non-governmental organizations; NGO orphanages are subsidized by the Government and supervised by officers of the Ministry of Labor, Culture and Social Welfare. In most instances when a child loses one or both parents, he/she is taken in by relatives. Resort to an orphanage is quite rare. The vast majority of orphanages in Tanzania are temporary homes for children aged 0-3 who have lost their mothers in childbirth. Often the widower does not have the means or time to care for the infant. The child is left with the orphanage until he/she is old enough to return to his/her family". The Department of Social Welfare makes every attempt to find adoptive parents for children who have been abandoned or who have no family. The number of adoptions by unrelated adults is few, however. Children who are not claimed by relatives or adopted are sent at age 3 or 4 to the Kurasini National Children's Home in Dar es Salaam, where they live until age 18. 3 Bertozzi et al (1990) corroborate this practice. They report, based on their focus group interviews, that paternal orphans virtually always remain with their mothers, while maternal orphans are less likely to live with their fathers, particularly when they are very young. (p. 8) Survivor Assistance 29 July 1992 Orphanages provide complete support for their wards, including room and board, clothing, schooling, medical care and recreation. Although there is often a presumption that orphanages provide inferior care to that which can be provided by families, this is not always the case. A study of war orphans in Uganda by Nalwanga-Sebina and Sengendo (1987) found that orphans living in orphanages had a higher standard of living than those who had been taken in by relatives or foster families. In particular, the study found that children in orphanages were more likely to be attending school than orphans living elsewhere and that this was an important reason why children were sent to orphanages. Bertozzi et al (1990) visited the Ntoma orphanage run by the Evangelical Lutheran Church of Tanzania in Kagera, which they described as in "excellent condition, well equipped and maintained" and staffed by 2 nurses, 11 nurses aides and 3 caretakers, for (at that time) 13 children. Children in the Kurasini National Children's Home were housed 2-4 to a room and supplied with bedding, mosquito nets, a cupboard and clothes. School fees were paid for all children, even those enrolled at private secondary schools, and every child received an "allowance" of 80/= Tshs per month. There was a small playground and soccer field. The grounds were well kept and landscaped. The Home was managing a vegetable plot and a poultry project. fak. Estimates of the costs of supporting orphans were obtained from the Kurasini National Children's Home In Dar es Salaam. This is the only completely state-run orphanage in Tanzania, serving children 3-18. The cost of supporting one child for a year in the home is estimated at 125,310/= Tsks (US $649), assuming that the home provides 100 child-years of care annually, or 179,014/= Tshs (US$ 928) assuming that it provides 70 child-years of care (see Annex 3).1 The cost of maintaining an orphanage for older children outside of Dar es Salaam, where rents are lower, would be less. The cost to the Government of subsidizing young children at NGO-run orphanages is 29,200/= Tshs (US $151) per child per year." Discussion. Orphanages are not the preferred solution of any Tanzanian official or NGO representative contacted during numerous visits to Tanzania. A number of reasons are cited: (1) Orphanages de-socialize children; it is better psychologically and socially that they remain with relatives or adoptive families. (2) Orphans become permanent wards of the state - even as adults.34 (3) Orphanages are expensive to build and maintain. Many policymakers expressed a preference for subsidizing unrelated foster families over placing children in orphanages. Given the shortcomings of orphanages and the inability to accommodate large numbers of children due to the expense, such alternatives deserve to be considered in the event that child abandonment becomes a major outcome of " Unfortunately, the study failed to control for the fact that orphanages are all located in urban areas where schooling is more accessible. Part of the difference in enrollment rates can be attributed to the fact that orphans living with families were in rural areas. * Using a 1990 exchange rate of Tabs 193/$. " Tshs 80/= per child per day, 365 days per year. 34 After age 18, orphans from the Kurasini home live in the community but are welcomed back whenever the need arises. At times, the entire family of an adult orphan will present itself at the Home for food, shatter and assistance. Swviver Astance 30 July 1992 the AIDS epidemic. Although there have been many reports of the severe strain that AIDS orphans are exerting on the resources of adoptive families, to date the anidemic has not increased demand for nace in orhanages in Kagera. the hardest-hit region. To the contrary, the three orphanages In Kagera are operating at only 57 percent of their capacity of 120 children (see Table 7 below). Further, few of the Institutionalized orphans are thought to be orphaned from AIDS. While many children in Kagera are being taken in by their extended families, less clear is the fate of children orphaned from AIDS in Dar es Salaam. Possibly orphans in Dar es Salaam are being sent to their parents' home regions to live with relatives. Alternatively, they may remain in Dar es Salaam living on their own and contributing to a growing number of homeless delinquent children (a problem noted often in The Daily News). TABLE 7: Actual and Potential Ss of Chidre in Existing Ordames. Kanere Renton Aune 30. .190 Total jnrolled Enrollment as Orhanage . ate -Female Totat Canactt of Canacity Ntem 10 8 18 35 51 (Evangefcal Lutheran Church of Tanania) nyanahanga 16 27 37 50 74 (Union of Pentecostal Churches) Rulenge 6 7 13 35 37 (St. Sernadetta Sisters) TOTAL 32 42 68 120 57 Source: Social Welfare Office, Kagera Region. 0 Only one very young orphan (less than 6 months old) out of 16 was suffering from AIDS in the Ntoma orphanage in Kager visited by Bertozzi et at (1990). In an interview in Febwmay 1990, tbi assistant Regional Welfare Officer in Mwana could recall only one AIDS orphan ever admitted to the Nyoged orphanage, the only orphanage in Mwanza. To date, there has been only one AIDS orphan at the Kurasini Hon; the child eventually got sick and was taken by his relatives up-country. Survivor Assistance 31 July 1992 DIrect maffers aMie. This program would immediately raise the consumption levels of survivors by providing them with food, clothing and basic commodities, Improving their short-term welfare. EIt prr . Programs of direct transfers to AIDS victims are not widespread in Tanzania. Their greatest concentration is In Kagera Region, followed by Dar es Salaam. In Kagera, the key actors are the Social Welfare Office of the Ministry of Labor, Culture and Social Welfare (MLCSW) and numerous NGOs. The most active organization in Dar es Salaam is WAMATA, an NGO that specifically assists persons with AIDS and their families. The Social Welfare Office in Kagera is the only field office of the MLCSW involved in relief directed to AIDS orphans and families. The program offers food ind child care counselling to needy families with orphans who are identified by the village leadership and are verified by a Social Welfare Officer. The families of these children are visited once every two months." Between April 1 and June 30, 1990, the Social Welfare Office assisted roughly 557 children in Muleba District (from 5 divisions, 15 wards and 45 villages) and 647 children in Bukoba District (from 7 divisions, 29 wards and 61 villages, including 250 children in Bukoba town), for a total of 1,204 children assisted. The MLCSW has a proposal to the National AIDS Control Program for training and transport for the support of victims of the AIDS epidemic (see Annex 6). NGOs are also heavily involved in direct assistance to survivors in Kagera, although the numbers of survivors receiving assistance could not be determined. The Tanzanian Red Cross is providing support to needy families In 9 villages in Bukoba Rural district, In the form of sugar, maize, flour, soap, cooking oil and blankets." The Eangical Lutheran Church of Tmnia (LCT) and Parta e avec les Enfants du rs Mond are heavily involved In distribution of food and clothing and In providing shelter for AIDS orphans. The Bapis Church is supporting some 50 families to the amount of 2,000/= Tshs per month as part of their Family Support Programme (Bertozzi et al, 1990). In Dar es Salaam, WAMATA provides direct assistance to orphans, widows and persons with AIDS. Unlike assistance provided by the Social Welfare Office in Kagera, WAMATA provides cash x The Social Welfare Office in Kagera has 7 social workers, of whom 3 deliver family counselling. Five are based in Bukoba and 2 in Bibarando. Only two fo the 7 social workers are engaged in food distribution. Other services offered by the social welfare office include: family and child welfare services (placement of orphans and children born out of wedlock, inheritance disputes; custody disputes and marriage counselling); probation counselling; and rehabilitation of the disabled. Of the 30 cases handled by the office per month, it is not known how many are related to AIDS. " The Red Cross also provides, in coqjunction with this direct assistance, health education, AIDS counselling, condoms and drug kits. S Other services provided by WAMATA include: AIDS prevention and control; medical treatment and home care; blood testing; legal assistance; and counselling individually and in group sessions. As of fall 1990, WAMATA was serving 160 families affected by AIDS, 173 orphans who had lost one or both parents to AIDS and 32 persons Survivor Assistance 32 July 1992 assistance. In the fall of 1990, WAMATA was providing 72 orphaned toddlers with 2,500/= per month for food. Persons who are HIV-positive were receiving a similar amount per month in cash to deal with their needs. WAMATA has also experimented with giving AIDS widows grants of 20,000/= to start up small businesses. Of the 47 women who received these grants, however, only 20 are thought to have used the money for businesses. These women were generally those whose husbands had a business or who already had their own. The other widows used the money to pay for rent and other bills. Q0gl. The costs of ongoing programs of direct assistance in Kagera are not well documented, while the costs of the WAMATA programs are difficult to establish because it is a small organization that depends heavily on volunteers and donated buildings. Further, many of the programs sponsored by NGOs include multiple interventions from which It is difficult to assess the costs of any one. It appears that the Social Welfare Office was spending roughly 383/= Tshs (US $2) on food per child per month for the 1,204 children helped between April and June 1990 in Kagera, for a total level of 4,596/= Tshs per child per year (US $23.81) (see Annex 4). Note, however, that this amount of assistance per child reflects the amount actually delivered, which is not necessarily the optimal amount to sustain these children, and Is exclusive of overhead and transport. Some simple calculations in Annex 4 project that the administrative costs of expanding the food assistance program of the Social Welfare Office in Kagera to an additional 1,806 children would cost roughly 745/= Tshs per child per month ($3.86) or 8,940/= Tsabs per year ($46.32), assuming that the amount of food provided per child remains constant. These estimates are extremely tentative, however, as tbi total number of beneficiaries of current efforst, the level, frequency and costs of support have not been well documented. Disusio. Direct transfers of food, clothing and other basic needs helps poor families through acute crises. Alone, they are unlikely to have an impact on long-term improvements in welfare unless sustained over a long period. Clearly, the resources are not available domestically or Internationally to support a large and growing number of survivors in Tanzania indefinitely. These transfer programs should be viewed, therefore, as short-term assistance with a finite duration. They should be accompanied by programs with a longer time horizon that will foster self-reliance and long-run sustainability. Feeding posts and cMfd care center Raoal. AIDS has produced a labor shortage at the same time It has raised the dependency ratio within households. The feeding post/child care center would have two main benefits: (a) it would ensure at least one good meal per day for participating children under the age of 7; and (b) it would free up the time of adults and older children for productive activites and schooling, respectively. with AIDS. Survivor Assistance 33 July 1992 Eisting w am At present, the feeding post and day care center programs are operationally and conceptually distinct. There are 252 feeding po In 5 regions, serving 13,362 children age 0-6, most of whom are under two years of ago." The average feeding post serves 53 children. Feeding posts serve meals; very little child care is provided. The parents or older siblings of the children bring them to the feeding post once per day for a meal and escort them home. Recurrent inputs for the feeding posts, including food and the time of the staff, are largely contributed or financed by the community with finance for transport, training and supervision from external donors such as UNICEF. Da care centers are organized more formally for children aged 2-6. In addition to a meal, they provide education, developmental activities and play materials. They are supervised by the Department of Social Welfare and are maintained under Law #17 of 1971. They generally operate half a day, from 6-12 am. The child care workers are formally trained and receive a salary. There are 1,692 day care centers in all 20 regions of Tanzania, with 117,413 children enrolled, or about 70 per center. They include the day care centers organized by public institutions and parastatal organizations (such as tea and sisal plantations) for the children of workers. The Department of Social Welfare is considering expanding feeding posts to include some day careldevelopmental activities and of organizing them in villages hard-hit by the AIDS epidemic. Cga. The recurrent and capital costs of setting up a combined child care and feeding center are calculated in Annex 5. Annual costs for a post of 40 children, including staff, equipment and building, range from 133,500/= to 295,750/= Tshs (US $668 - $1,479), or 3,338/= to 7,394/= Tshs (US $17 - $37 ) per child per year. Roughly 80-90 percent of these costs attributable to feeding posts are already being borne by communities. The value of external inputs from donors is estimated to be an additional US $4 per child per year, for a total annual cost of US $21-41 per child.41 Disso. One attractive aspect of this intervention is that it has both short- and long- run time horizons - the immediate supplementation of children's diets and the freeing up of adult labor to raise incomes. Another attractive aspect is that, once launched, it relies heavily on the self-reliance and initiative of the community. This type of activity may thus be more sustainable over the long run. There is already substantial experience in Tanzania with this intervention and attempts have been made to evaluate its costs and effectiveness (in conjunction with other interventions) in reducing child malnutrition (see JNSP, 1988).2 Of course, improved child nutrition is only one of the benefits that such a program " Kagera, Morogoro, Kilimmjaro, Mtwara, and Ruvuma regions. SThere are 56 feeding posts in Kagers region, serving 3,064 children (1,567 male, 1,495 female). 41 Not included in these costs are the costs of supervision, management, training and monitoring and evaluation, provided both by external donors and by district and regional authorities. 2 The prevalence of severe malnutrition, as measured by less than 60 percent of the standard weight for ago, dropped from 5 percent to 2 percent in participating villages in Iringa. The percent of underweight children less than 80 percent weight for age dropped from 48 to 37 percent (INSP 1988, P. 32). Survvor Assistance 34 July 1992 would offer to the survivors of AIDS. A potential problem, however, is that the programs now In existence rely heavily on community resources - resources like food and organizational capacity that communities hard-hit by AIDS may not have abundantly. Hard-hit communities mvy initally require outside resources, such as externally donated food, to launch this Intervention. Subsidies of school fees and un(foms . This program is a subset of direct transfers, considered above. Its impact differs conceptually from other basic needs transfers, however, because schooling is both a consumption and an investment good. Children who receive schooling now will reap higher incomes and contribute more to society in future periods. Subsidies of school fees and uniforms for AIDS survivors will help to ensure that they are not denied access to schooling and acquisition of human capital.0 is aroam. WAMATA provides grants to school-aged children in families of AIDS patients of 5,000/= Tshs per child per year for 54 primary students and 20,000/= Tshs per child per year for 20 secondary school students in Dar es Salaam. The Social Welfare Office in Kagera has not subsidized school fees or uniforms to date, but is aware of some 200 children in need of such assistance. Partage and ELCT have been providing school uniforms and/or fees in the villages where they are active. The amount being spent by NOOs in Kagera on school fee subsidies and school uniforms is not known, nor is the criteria for eligibility for any of these programs. CAa. The annual out-of-pocket costs for schooling for primary and secondary students in 1988 are presented in Table 8, below. The out-of-pocket costs of enrolling a child in primary school are 3,700/= Tshs per year (US $19). At the secondary level, costs amount to 13,7001= to 17,000/= Tshs per year (US $71 -$88) in public day and boarding schools, respectively, and roughly 3,000/= Tshs more per year in private schools of each type. Out-of-pocket expenses are only part of the cost that families face in enrollment decisions, however. What may be even more Important, especially in the case of survivor households, is the value of the forgone labor of the child in home production during the period when he/she is in school. With the loss of the most productive adults, the value of children's time in home production is likely to increase. Problems paying school fees were cited as the most critical issue by virtually all of the focus groups interviewed by Bertozzi et at (1990). Survivor Asstance 35 July 1992 TABLE 8. Out-of-gMket cots for schootina. Tamania. 1 (Tabs) Level of schooling Public Secondary Private Secondary Type of expense Prfeary Day schoot Boarding Day schoot Boarding Fees 200/4* 700/.4 4,000/- 3,200- 71000* 3,700/a 7.500/ Uniform 2,000/a 3,000/. 3.000/a 3,000/= 3,000/a Stationery/Books 1,500/u 10,000/u 10,000/a 10.000/ 10,000/2 TOTAL 3,700/ 13,700/a 17,000/C 16,200- 20,000- 16,700/- 20,500/u Source: Vortd Bank (1990a). These figures do not include special contributions for furniture or bultdings. BIm on. The gross primary enrollment ratio in Tanzania has declined from 96 percent in 1979 to 78 percent in 1988; primary enrollments have dropped absolutely from 3.5 million to 3.2 million over the same period (World Bank, 1990a). This dramatic drop is thought to have been the result of two factors: (1) a preference for keeping children at home due to the low quality of instruction; and (2) limited ability to pay the out-of-pocket and indirect costs, in terms of the forgone labor of the children while they are away at school. This drop in enrollment has occurred at a time when the quality of schooling has also declined, largely due to the undedinancing of recurrent inputs. The public sector is extremely strapped for cash and is attempting to elicit greater participation from households and communities in the finance of schooling so as to improve its quality and availability. This situation raises many issues with respect to a school fee subsidy intervention. First, since there are many Tanzanian children not enrolled, ethical Issues are likely to arise in the decision about which children to help: there are many poor children not enrolled who are not AIDS orphans and many AIDS orphans who are already enrolled in school. Figure 12 shows the enrollment rates for maternal orphans and non-orphans in three regions - Kagera, Mbeya and Morogoro - according to the 1988 census. Although maternal orphans are less likely to enroll than other children, in all three regions a very high proportion of maternal orphans is enrolled compared to non-orphans. All children have low enrollment rates up to about age 11, and for no group of children does enrollment reach 100 percent. Kagera, with a higher orphan rate presumably due in part to ADS, seems to have no larger a gap between the enrollment of orphans and non-orphans than do the other regions. Figures 13A and 13B compare the enrollment rates for maternal orphans and other children across districts within Kagera. Recall that the districts hardest-hit by AIDS are Bukoba urban, Bukoba rural and Muleba. Maternal orphans in the two Bukoba districts are more likely to be enrolled than any group of children, orphaned or not, in Biharamulo district, which is poorer. Further, the largest gaps between the enrollment of orphans and non-orphans are also in the relatively less wealthy districts of Karagwe and Blharaimlo, where AIDS was not yet pronounced in 1988. 4 Fee paid to local governments. 4 With lunch. The cost excluding lunch is 350/= per year. SWn~r Asstoace 36 July 192 IIIGURE 12: ENROLMENT RATES, ORPHANS & NON-ORPHANS KAGERA 720 10- 0 7 9 10 i1 12 13 14 5 MOROGORO 100 go- 40 60- 0 b b iO 1i i2 1å 14 1s5 MBEYA 100 0 10 40- 1> 10- 7 8 b 1O 1 12 13 14 15 A~ E --OrgI.an/Mmhl -NU- Orphmn/Feai -ag- Non.Oph/Male ----NOpI Swrher A4ssistance 37 July 1992 FIGUE 13A: ENROLLMENT RATES, ORPHANS & NON-ORPHANS BUKOBA URBAN 100 0 -80- ~60- 10- 0 7 8 9 10 11 12 13 14 15 BUKOBA RURAL 100 -80- 70-. MULEBA 10 0- 20- 10 7 8 9 10 11 12 13 14 1.5 so- -O~d N-OVmb- o4b ~ .OpW Su3r sn . 38L992 FIGURE 133: ENROLLMENT RATES, ORPHANS & NON-ORPHANS BIHARAMUID 100 0 70- 160 400 S30- -20 - ___ _ _ _ _ _ 10 ----a S10 11 12 13 14 15 NARAE 100 0 80- 70- 60- 10 8 10 11 12 13 14 15 NGAG 100 90- 780 710 0 1 1 1 4 1 7 - Oihu/J 1- OiIm/.ai 12 13OpWa 4 i5-wp Swvwvr Assistance 39 July 1992 This raises the second issue: targeting criteria. Many orphans are already enrolled In school; targedng based on being an orphan will result In subsidizing fees for a large number of orphans who would have enrolled anyway and will be unfair to other children whose parents are alive but cannot afford the fees. Of course, a figure that distinguished between AIDS orphans and other children or between two-parent orphans and others might tell a different story - perhaps that two-parent orphans or AIDS orphans do not enroll at all. Unfortunately, this information is not available. Even so, targeting based on being an AIDS orphan might stigmatize the child and may still result In the subsidization of many children who otherwise would find the wherewithal to enroll. The third issue raised is that of cost. This intervention is potentially very costly due to the large numbers of children who could be involved. During a period of public sector austerity, any type of public policy to exempt or subidize one group of children would Inevitably result In higher costs for the remaining children. The funding for this intervention would thus most likely have to be found from private sources. In fact, the intervention would not promote self-reliance and has the same sustainability problems as other direct transfer interventions. Fourth, it is not known what the bnpact of this intervention would be on school enrollments. If AIDS orphans more than other children have large labor responsibilities at home due to the absence of adults, or if the remaining adults do not see it as a good investment, then offering school fees may not raise enrollment greatly. Quite possibly, a policy reducing the labor constraint in home production would have a greater Impact on the enrollments of orphans. Finally, if the quality of schooling has slipped substantially, the benqfits of a policy of subsidizing schooling may be low relative to other interventions to assist survivors that may raise incomes with greater certainty. The high cost, potential leakage, unknown effectiveness and low sustainability of this intervention indicate caution. More needs to be known about the characteristics of those in greatest need who would not otherwise have enrolled and the reasons why they have not enrolled. Survivor Assistance 40 July 1992 IV. CONCLUION An Important impact of the AIDS epidemic will be to create new groups of persons In poverty and new Inequalities. However, not all AIDS survivors will be poor and there are many poor Tanzanians who do not have AIDS. The first step in developing a survivor assistance strategy at the national, regional or district level is to assess how many survivors will be created, the types of problems they are likely to encounter, who among the survivors is in greatest need of assistance and how they can be Identified. This paper has marshalled existing information, primarily from the 1988 census, to establish rough orders of magnitude for the number of survivors, particularly orphans. In doing so, we have attempted to contrast the survivor problem created by AIDS with the baseline number of survivors in the population. A second objective of this paper has been to suggest criteria for evaluating survivor assistance programs in Tanzania and to review the experience to date with several ongoing programs. Programs to assist survivors are generally anti-poverty Interventions that are already widely Implemented in Tanzania and elsewhere in the absence of AIDS. The issue then becomes how to select among alternative Interventions. Three criteria were proposed: costs and benefits; the target group of beneficiaries; and sustainability. Four ongoing programs were discussed. * To date, the demand for places in orphanages is still low. If child abandonment becomes commonplace because of the AIDS epidemic, however, orphanges raay have to be considered in spite of their expense and inherent disadvantages; experimentation with other less expensive arrangements would be highly desirable before embarking on such a course. * Considering that direct transfers are currently the major form of survivor assistance offered in Tanzania, it Is surprising how ittle information Is publicly available about their costs, the number of persons being assisted, the duration of assistance, the Impact of these programs and the criteria for eligibility being used by dMMeren organizations. Direct transfers, although they satisfy acute needs, are not sustainable over the long run for the large and growing number of survivors. Such assistance should be viewed as temporary, with a finite duration, while other more sustainable interventions can be launched. * There is substantial experience with feeding posts and child care centers in Tanzania. Further, they are potentially more sustainable, as they rely heavily on local contributions. This asset may be a disadvantage in areas severely affected by AIDS, where the local population may not be able to make as substantial a contribution and where the organizational capacity may have been severely damaged. Hard-hit communities may require external assistance at the outset. SWvivwr Assistance 41 July 1992 * Subsidiation of school fees is another type of direct transfer, although it has some longer-run implications in raising the earning power of recipients. At a time when enrollments have declined proportionately and absolutely and when shortages in public finance have resulted in declining quality of instruction, it is unrealistic to believe that resources for such a policy could come from the public sector. Exempting one group from paying is likely to result in higher contributions from other groups or, failing that, even greater decline in the quality of instruction. Even if financial assistance could be found from private sources, however, such a policy is likely to be expensive in light of the large numbers of children not enrolled in school and the difficulty in establishing eligibility criteria. Issues about the ethics of selecting eligible children, the effectiveness of the intervention in raising welfare (relative to other potential Interventions) and the sustainability of such a policy were also raised. The selection of policies for review was arbitary, focusing on interventions already implemented or discussed in the context of survivor assistance. Many other policies of potential interest were not reviewed but deserve consideration. For all policies there are many unsolved operational issues, including: how to identify those in greatest need and who can be helped; how to evaluate the effectiveness of alternative interventions; and how to guarantee the sustainability of programs. Although the information on costs and beneficiaries for the reviewed interventions was insufficient, by far the largest gap in knowledge concerned their effectiveness. A first step in rationally considering the alternatives would be to conduct a comprehensive study of the costs, benefits and criteria for eligibility of existing programs being conducted by the Government and NGOs. Suevivor Assistance 42 July 1992 Bertozzi, Stefano, Maxine Ankrah and Magdalene Ngalza (1990). "Tanzania: Assistance to Survivors of the AIDS Epidemic: A Review of the Policy Options". Draft. June 16, 1990. Bulatao, R. (1990). "The Demographic Impact of AIDS In Tanzania". Draft. June 13, 1990. Chin, James (1990). "The Epidemiology and Projected Mortality of AIDS in the United Republic of Tanzania". Background paper for the AIDS Sector Report. Mimeo. Dunlop, David W. (1975). "Benefit-cost Analysis: A review of its applicability in policy analysis for delivering health services". Social Science and MedIcine 9(3): 133-139. Government of the United Republic of Tanzania and UNICEF (1988). "A Report on the Mid-Term Review of the Government of Tanzania/UNICEF Programme of Cooperation, 1987-1991". Dar es Salaam, December 1988. Hemed, Dr. (1990). "Sentinel Surveillance of HIV in Mbeya Region, Tanzania". Presentation to the Fifth International Conference on AIDS in Africa. Kinshasa, Zaire, October 12-15, 1990. (check date and title of talk) Hunter, Susan (1990). "Orphans as a Window on the AIDS Epidemic in Sub-Saharan Africa: Initial Results and Implications of a Study in Uganda". Social Science and Medicine 13(6):681-690. Killewo, J., K. Nymuryekunge, A. Sandstrom, U. Bredberg-Raden, S. Wall, F. Mhalu and G. Biberfeld (1990). "Prevalence of HIV-1 infection in the Kagera region of Tanzania* A population-based study". AM 4(11):1081-1085. Killewo, J., A. Sandstrom, U. Bredberg-Raden, K. Palsson, S. Wall, F. Mhalu, G. Biberfeld et al. (1990). "Incidence of HIV Infection in the Kagera Region, Tanzania A Population-Based Study". Presented at the Vth International Conference on AIDS In Africa, Kinshasa, Zaire, October 1990. Malawi Government (1980). Malawi Population!Census 1977. Final Report, Volume 1. Zomba: Government Printer. Ministry of Health, United Republic of Tanzania (1990). "Overview of AIDS Control Activities up to September 1990: Workplan and Budget 1st January - 31st December 1991", September. Moss, A.R. and P. Bacchetti (1989). "Natural History of HIV Infection". AI 3:55-61. Mujinja, P., 0. Koda, G. Lwlhula and I. Semali (University of Dar es Salaam) and M. Alnsworth and M. Over (The World Bank) (1990). "Coping with the Economic Impact of Fatal Adult Illness: An Analytic Framework". Mimeo. November 29, 1990. Mushi, P., UNICEF Project Officer, Kagera Child Survival and Development Programme. "Notes for the Annual Review - Kagera". Memorandum of September 28, 1990. Sarviwvr Assistance 43 July 1992 Mutembel, I.B. (1989). "AIDS Socio-Economic Impact on Adolescents: The Case of AIDS Orphans in the Lake Zone, Tanzania'. Ministry of Labour, Culture and Social Welfare. A paper prepared for presentation to the Second International Symposium on AIDS Information and Education, Yaounde, Cameroon, October. - (1988). "A Report on the Number of Orphans and Aged People Resulting from AIDS Disease in Kagera Region. Report based on a survey done in Kagera from October 11 - November 24, 1988. (Translated by Dr. Owenye, Tanzania Red Cross Society.) Nalwanga-Sebina, Abby and James Sengendo (1987). "Orphaned and Disabled Children in Lawero and Kabale Districts and in Ugandan Child Care Institutions: A comparative profile to the general child population". December. National AIDS Control Programme (1991). Sentinel Surveillance System for January 1 - December 31, 1990. May 15, 1991. Nyamuryekunge, K., J. Killewo, U. Bredberg-Raden, A. Sandstrom, F. Mhalu, G. Biberfeldt and S. Wall (1988). "HIV Infection in Kagera Region, Tanzania". Poster presented at the IVth International Conference on AIDS. Stockholm, Sweden, 1988. Over, M., M. Ainsworth, P. Mujinja, G. Koda, 0. Lwihula and I. Semall (1989). "The Economic Impact of Fatal Adult Illness from AIDS ane Other Causes in Sub-Saharan Africa: A Research Proposal". Submitted to the World Bank " esearch Committee, November 1989 and approved in January 1990. Over et al (forthcoming). Over, M., S. Bertozzi, J. Chin, B. N'Galy and K. Nyamuryekung'e (1988). "The Direct and Indirect Cost of HIV Infection in Developing Countries: The Cases of Zaire ad Tanzania" in The bal IpactfAIDS pp. 123-135. Preble, Elizabeth A. (1990a). 'Impact of HIV/AIDS on African Children", Social Science and Medicin 13(6):671-680. (1990b). "AIDS Orphans in African. Paper presented to the Meeting of Norwegian NGOs on AIDS. Oslo, Norway, December 4, 1990. Republic of Uganda (1976). t on the 12 puai ua . Volume IV, The Analytical Report January. Rwegarulira, A. A. and Peter Mushi (1991). "Strategies for Community-Based Care: A New Approach for Service Delivery and Care of Orphans". Paper presented at the Regional Officer's Workshop on the Impact of AIDS. Tanga, Tanzania. October, mimeo. Sandstrom, A., L. Kablgren, J. Killewo and G. Lwlhula (1990). "Socio-geographical spread of EV infection in Kagera region, Tanzania". Mimeo. September 21. Stevior Assiance 44 July 1992 Sembawe, I.S.L. (1983). "Mortality Differentials" In United Republic of TAnzania, 1978 Put Co Volum VI. Bureau of STatistics, Ministry of Planning and Economic Affairs, Dar es Salaam UNICEF (1990). Children and AIDS: An Imoonding Calamity. New York: United Nations. World Bank (1990). "Staff Appraisal Report: The United Republic of Tanzania: Education Planning and Rehabilitation Project". Report No. 7998-TA. April 26, 1990. (1990b). Wold D ment Rpo 1990. New York: Oxford University Press. Survbor Assistance 45 July 1992 ANNEX 1: Method of calculating the number of orphans from both parents Let p. equal the probability that a child Is orphaned from its mother; pt equal the probability that a child is orphaned from its father; and P.r equal the probability that a child is orphaned from both parents. The proportion of children who have lost at least one parent is given by: pa + pt - PMf Paternal survival data were not collected in the 1988 Tanzanian Census. Household surveys in Cote d'Ivoire (1985) and Ghana (1987) found the ratio of pf/p, for children 0-14, 1.5 and 2.2, respectively. Al four of the following scenarios for Tanzania assume that pt = 2p.4 The scenarios differ in their assumptions about P.r . The first scenario assumes that maternal and paternal deaths are Independent. That is, that P.r , P. X Pr Clearly, in a population undergoing an AIDS epidemic we would al expect maternal and paternal deaths to be independent. Scenario One thus provides a low-case estimate of orphans who have lost both parents. The remaining scenarios assume that the share of children who have lost both parents among those who have lost at least one parent is 10 percent (Scenario Two), 20 percent (Scenario Three) and 30 percent (Scenario Four). These assumptions are summarized below: AssMpoforphan _estimates Scenarlo 1 Par = p., x pt Pr = 2p,2 Scenario 2 pdI (p + pt - Par) = .10 P.r = .2727 p, OR Scenarlo 3 Pr (p, + Pt - Pr) = .20 P.r = .5455 p Scenario 4 par I (p. + pt - P.r) = .30 Par = .8182 p., A figure of 30 percent is used for a high-side projection of children who have lost both parents, based on the 1988-89 enumeration of AIDS orphans in Kagera region, which found that 30 percent had lost both parents. Since even in Kagera there are orphans independent of the AIDS epidemic with lower probabilities, this proportion is likely to be on the high side. * Soure: data tapes from the 1985 Cote d'Ivoire Living Standards Survey and the 1987-88 Ghana Living Standards Survey. Survivor Assistance 46 July 1992 These four scenarios using Identical estimates of the maternal orphan rate, result In different projections of both the total number of children who have lost any parent and the number who have lost both. For a given maternal orphan rate, as the proportion who have lost both parents rises, the total number of orphans falls. Thus, scenario 1 produces the highest estimate of orphans who have lost at least one parent, but few in the high risk group who have lost both parents. Scenario 4 produces the lowest projection of total orphans but the highest estimate of the high risk group. This tradeoff is illustrated below with data for the Kagera Region, where p. = .0292. Scenario 1 Scenario 2 Scenario 3 Scenario 4 Number who have 1,059 4,943 9,887 14,830 lost both parents Number who have 52,257 44,489 36,601 24,715 lost one parent TOTAL 53,316 49,432 44,488 39,545 As a percent of all children 0-14 8.60 7.97 7.17 6.38 Survivor Assistance 47 July 1992 ANNE[ 2: CHILDREN'S HOMES IN TANZANIA, ETMBER 1990 Name fNumb Children's Home 898 of Chl Nkoaranga Arusha 8 Dongobesh Arusha 24 Kurasini Dar es Salaam 47 Mtoni Dar es Salaam 38 Msimbazi Dar es Salaam 47 Mburahati Dar es Salaam 80 Poloni Dodoma 16 Ilembula ringa 10 M n Iringa 19 Tosamaga Iringa 27 Ntoma Kagera 25 Nyamahanga Kagera 42 Rulenga Kagera 12 Matiazo Kigoma 57 Muhange Kigoma 43 Moshi Chapel Moshi 41 Upendo Kilimanjaro 56 Igogwe Mbeya 38 Mbod Mbeya 18 Berega Morogoro 29 Mgolole Morogoro 38 Ukwama Morogoro 29 Katandata Rukwa 26 Kilangala Rukwa 24 Chipole Ruvuma 12 Mbesa Ruvuma 32 Gongoni Tabora 42 Kwamkono Tanga 60 henme Tanga 29 House of Joy Dodoma 46 Bulongwa Iringa 19 Myegez Mwanza 22 Myuji Home Dodoma 7 Kongowe Dar es Salaam TOTAL LM Swvvor Assistance 48 July-1992 ANNEX 3: COST ESTIMATES, KURASINI NATIONAL CHILDREN'S HOME The V Natinal was started around 1966. It is the only completely state- run children's home in Tanzania, housing children age 3-18 years. The Kurasini Home receives children older than three who are transferred from orphanages in the rest of the country. It also takes In "lost* children temporarily until their parents or relatives can be contacted. Mrs. Ndone reported that the typical number of children at the home is 70, and at times it has been as high as 100. Most orphans who are admitted come at age 3 and stay for 15 years, until age 18. After that, they live in the community but continue to draw on the orphanage for assistance when needed. Orphans are not the majority, however: "lost" children may spend only a few months. The home is entirely financed by the state. It raises additional funds and food through the sale of chickens from its poultry project, which produces 600 chickens/year. The food budget is also supplemented with home-grown vegetables. The value of these additional benefits to the home are not taken into consideration in the cost calculations. There are currently 52 children staying at the home (as of October 25, 1990), of whom 12 are orphans and only 14 are girls. Typically, it serves 70-100 children. The Home provides the following services to its wards: . General maintenance and care: bedding, clothes, food. . Education: the Home runs a nursery school for children aged 3-7. Older children are registered in nearby primary schools. The Home pays a UPE fee of 600/= per child per year in primary school and 1,500/= for uniforms. Secondary school fees are paid by the Commissioner for Family Welfare directly and do not enter into the home's budget. . Medical care: doctors and nurses from the nearby Temeke District Hospital visit the home twice per month gratis. The home has its own nurses and pays for medicines for sick children. . Recreation: toys, outdoor games, radio cassette music and news. . Religious instruction: provided by numerous religious authorities who visit the home. . Fleld trips: Volunteers provide transport and food for outings to the national museum, major hotels, beaches, etc. (le home does not have a functioning vehicle.) SWrvvor AssIstance 49 July 1992 Staf. The following people are employed at the home: 2 welfare officers 10 nurses 4 day care workers/nursery school teachers 3 cooks 2 sweepers 4 watchmen 2 launderers 1 welfare assistant in training 28 The monthly wage bill is 82,000/= plus 22,216/= overtime, for a total of 104,216/= Tshs, or $540. Value of housing and buildins. The orphanage consists of 5 buildings, with an estimated rental value of 80,000/= per month. Expenditures on minor maintenance and rehabilitation amount to about 100,000/= per year. There are two staff who receive free housing from the home - the director and one nurse. It is estimated that the monthly rent for these two abodes would be 3,0001= and 1,500/= respectively on the open market, for a total of 4,500/=. Survivor Assistance 50 July 1992 COST CALCULATIONS. KURASIlI CHILDREN'S HOME Reeurent costu per month B*£..Year Wages and overtime 104,216/a 1,250,592/0 Transport attomance for staff 72,800/u 873,600/- Food (budgetted) 431,333/a 5,176,000/a Maintenance (soap, toothbrushes, 1.3501000/- underwear, shoes, socks) Recreation 605,000/a Laundry 70,500/a Dormitory 685,000/a Schooling 415,000/a Attowances for chiLdren at school (80/u per chiltd per month) 3,200/- 38,400/- x 40 children Medicines 140,000/a Hiring of car/taxis 240,000/a Office supplies 60,000/a Kitchen utencils, etc. 182,900/u Postage 6.000/a Etectrifty 25,000/n 300,000/a Water 2,000/= 24,000/a Rental value of buildings & staff housing, 1,114,000/- plus building repairs TOTAL RECURRENT COSTS ' 12,530,992/a Per chtid per year, assuming 100 chitdren 125,310/a a , assuming 70 chtidren 179,014/a Dotter costs per child: $649 (100 chtd years), $928 (70 chtid years) Aniver Assstance 51 Jlly 1992 ANNEX 4: COST CALCULATIONS, SOCIAL WELFARE OFFICE SURVIVOR ASSISTANCE PROGRAM IN KAGERA The survivor assistance programs of the Ministry of Labor, Culture and Social Welfare In Kagera region have not been costed. In this annex, we attempt to estimate the value of commodities currently being distributed and the administrative overhead. T'he value oftrnfs Over the period April 1-June 30th 1990, the Kagera office distributed 16,870 kg of rice, 4,477 kg of beans, and 1,180 liters of cooking oil to the families of 1,204 children in Bukoba and Muleba districts. The total recurrent cost of this program for April-June, excluding salaries, was 1,488,142/= Tshs, of which 105,430/= Tshs was for transport expenses.0 Thus, the program is spending roughly 3831= Tabs per child per month for food (US $2), or 4,596/= Tss per child per year (US $23.81). Administrative.co Two social welfare workers are presently able to help 1,204 children in 3 months, or 400 children per month. It is estimated that adding one landrover and three additional social welfare officers to the Kagera office would allow the program to reach an additional 1,806 beneficiaries every three months, raising total assistance by 600 per month to a total of 1,000/month, assuming that the food is provided for distribution. Allowing the value of the landrover to be distributed over 10 years, this amounts to an annual overhead cost of 362/= Tshs per additional child per month. Calculations are provided below. Assuming an additional 383/= Tshs per month for food, each additional child would cost 745/= Tshs per month ($3.86), or 8,940/= Tshs per year ($46.32). 10= Cot per month Cost ner ea Landrover 1,000,000/0 (10 mittion Tabs spread over 10 years) Fuet and maintenance 80,000/- 960,000/* Driver 5,000/. 60,000/ Three social welfare officers 48,750/a 585,000/. (8 5,000/a each for saLary" per month plus 15 nights away from the office per month 8 750/= per night) TOTAL ADMINISTRATIVE COSTS 2,605,000/a 4These transport expenses cover the cost of fuel alone. The Social Welfare Office in Kagern does not have its own luy and must borrow from other govermment ministries when needed. Occasionally a lorry is hired from paast organizations. SBas salary for social welfare officers ranges from 3,345/= to 6,375/- per month, depending on training and seniority. An intermediate value of 5,000/- per month was arbitrarily chosen for these calculations. Survivor Assistance 52 July 1992 ANNEX 5: COST CALCULATIONS, FEEDING POSTS/CHILD CARE CENTERS This annex provides cost and Input Information for operating feeding post/child care center for 40 children aged 0-6. Not Included in the calculations are the start-up costs of training village leaders, conducting reconnaissance trips to mobilize villages or supervision costs, which would include transport and fuel. Annual operatingcot I Estimates of inputs and prices for the annual operating costs have been obtained from both the Department of Social Welfare of the Ministry of Labour, Culture and Social Welfare and from UNICEF. To accommodate the range of prices provided by these two sources, both a low and a high cost estimate are given. The value of replaceable equipment was distributed over the number of years it would last. The total annual operating cost for a post with 40 children ranges from 133,5001= to 295,750/= (US $668 - $1,479), or 3,338/= to 7,394/= (US $17 - $37) per child per year (see Table below).4 In similar ongoing programs (e.g., UNICEF's Child Survival and Development Programme), the food, fuel, building and staff are paid for by the community. This amounts to roughly 81-89 percent of the operating costs per child. External costs The estimates of external costs for a child care/feeding post program are based on the 1988 Evaluation Report of the Joint WHO/UNICEF Nutrition Support Programme (JNSP) in Iringp region. This program, launched in 1983, included many different interventions designed to improve child nutrition, of which child care and feeding posts were one component; a total of 168 villages in 5 districts of IMga were covered. There are two types of external costs - those financed by external donors and those that are domestically Fnanced by villages, district, regional and national governments. The external inputs from eternal donors for the child care component of the JNSP amounted to $131,500 over 5 years, or $26,300 annually (in 1987 dollars). These costs included those for vehicles, equipment, personnel, local transportation, international transportation, purchased services and other supplies.0 A survey conducted in the project area suggested that 93 percent of all villages (or 156) participated in the child care component and that, on average, 53 children were participating in each village with child care. The average cost per village per year came to US $210.50, or roughly US $4 per child per year. The evaluation document notes that virtually all of these costs were "start-up" or "expansion* costs that would not have to be incurred again. * The estimates of the operating costs in the Table have been made independently of the carefully documented cost estimates for the Joint WHO/UNICEF Nutrition Support Programme (1988), but on the items for which the JNSP provided details on local operating costs, figures are consistent A survey of 103 of the participating villages in Iring. region found, for example, that Day Care Assistants were being compensated an average of 400/- per month in 1987, including the value of payments in cash, in kind and in labor compensation (p. 69). 0 Not i are the costs of management, training or monitoring and evaluation. Survivor Assistance 53 July 1992 Estimates of the annual oerating costs of a day care/feeding center ITEM LoW etmate High esti Potsft 4,000 4,000 CupsO2 2,000 2,000 Plastic buckets" 3,000 3,000 Mats-U 2,000 2,000 Subtotal 13,000 13,000 Building" 12,500 18,750 &ar 12,000 24,000 Food and cooking fuel 96,000 240,000 Total cost, 40 children 133,500 295,750 ($668) ($1479) Cost per child 3,338 7,394 ($17) ($37) 1 Two 20-litre pots @ 10,000/= each, lasting 5 years. 3 One plastic cup @ 100/= each x 40 children; each lasts 2 years. a Two plastic buckets @ 1,500/= each, lasting one year. a44 rats 0 500/= each per year. * Mud brick house with a thatch roof for 40 children, assumed to last for 8 years. Total cost is 100,000- 150,000/r, depending on the scarcity of building materials. a Two persons to care for 40 children, receiving 5001= to 1,000/- each per month. MwTh value of food and firewood contributed by children is estimated at 200-500/- per child per week. Survivor Assistance 54 July 1992 The external Inputs from national sources for child care were valued at 10,480,700/= shillings over four years, or 2,620,175/= shillings per year. This includes the contribution of staff time of district officials, the contribution of fuel made available for JNSP activities, the value of the labor of day care attendants and the value of food consumed. Under the same assumptions about the level of participation In the child care litervention, nationally-financed external inputs came to 16,79(/= shillings per village and 317/= per child per year. No breakdown is provided for the distribution of national external inputs for child care by source, but for the entire JNSP program it has been estimated that two-thirds of the national contributions was from the villages themselves. The JNSP estimates for national sources are thus far below the estimated annual operating costs per child detailed in the Table below. The value of the food and labor Inputs alone come to between 2,7001= and 6,6001= per child per year. These dramatic differences cannot be accounted for by devaluation or inflation between 1987 and 1989. Tota aea= co The total average costs include annual operating costs and the external Inputs financed by donors. External costs financed nationally are excluded, since the JNSP document does not distinguish between the part contributed by villages and districts, and most of these costs are village-financed Inputs already included in the estimation of annual operating costs. Conceivably donors may have financed some of the items included in the annual operating costs. However, the most commonly financed article - cooking pots - amounts to only 1-3 percent of estimated operating costs. With all of the caveats about costs not included in the calculations, the very rough estimate of total average costs per child ranges from US $21 - $41 per child per year. Total averae costs per child per year. feeding post/child care center Annual operating costs per 3,338 7,394 child ($17) ($37) External assistance from $4 $4 donors TOTAL $21 $41