aids-induced orphanhood as a systemic shock: magnitude, impact, and program interventions in africa

AIDS-Induced Orphanhood as a Systemic

Shock: Magnitude, Impact, and

Program Interventions in Africa

KLAUS DEININGER, MARITO GARCIA and K. SUBBARAO *The World Bank, Washington, DC, USA

Summary. — According to many descriptive accounts, the orphan crisis in Africa has assumedalarming proportions, largely due to AIDS-related deaths. Using household panel data fromUganda to confirm this and assess the impact on affected households and children, we find that (a)receiving a foster child leads to a significant reduction of investment; (b) initial disadvantages infoster children�s access to education were largely eliminated by the introduction of a program ofUniversal Primary Education; and (c) new inequalities have emerged in foster children�s access tohealth services. Even though this suggests that specific programs could help to alleviate some of thenegative impacts of orphanhood, the policy response in many African countries has remainedpiecemeal. We use data from existing programs to estimate the cost of a concerted policy responseand highlight implications for further research.� 2003 Elsevier Science Ltd. All rights reserved.

Key words — AIDS, Uganda, orphans, health, education, investment

1. INTRODUCTION

In the past, vulnerable children as a groupwere comprised largely of street children, chil-dren exposed to strenuous labor, children en-gaged in trafficking, and children affected byarmed conflict. The AIDS crisis has swelled thenumber of children at risk and changed the na-ture of risks that children face. It has expandedthe human development agenda of countries andis beginning to threaten the care-giving capacityof communities. Will it also threaten the capa-city of governments to handle the crisis? Is thecrisis of orphans so serious as to become a sys-temic shock? The answer to this question de-pends largely on the extent to which the familiesand individuals taking care of the orphans willbe able to cope and provide orphans with accessto social services, and on the degree, quality andduration of the support the rest of society andbroader government policies provide.The incidence and growth in the share of

children who are orphaned, i.e., under the ageof 14 and having lost at least one parent, inAfrica is indeed frightening. While in the early1980s barely 2% of African children were or-phans, recent estimates put this proportion atas high as 15–17% in some countries (Table 1).

Children whose parents have been affected byAIDS make up the vast majority of this in-crease; recent estimates put their total numberin the range of 12–13.2 million (UNAIDS,2000; USAID, UNAIDS & UNICEF, 2002). 1

Household surveys are consistent with thispicture; a recent DHS survey reports that, inUganda, every fourth family is hosting anorphan and that the number of orphans hasreached 1.4–1.7 million out of a total popula-tion of 21 million.The dramatic increase in the number of or-

phans is an important policy issue for two

World Development Vol. 31, No. 7, pp. 1201–1220, 2003� 2003 Elsevier Science Ltd. All rights reserved

Printed in Great Britain0305-750X/03/$ - see front matter

doi:10.1016/S0305-750X(03)00061-5www.elsevier.com/locate/worlddev

*The authors thank the UNHS team from the UgandaBureau of Statistics, especially Jackson Kanyerezi and

James Muwonge. The paper also benefited from helpful

comments by H. Alderman, M. Grosh, T. Marchione, E.

Skoufias, seminar participants at the IDB and the World

Bank, and anonymous reviewers. Support from the

World Bank�s Social Protection Network Anchor, theBelgian Consultant Trust Fund and the African Statis-

tical Capacity Building Initiative is gratefully acknowl-

edged. The views expressed in the paper are those of the

authors and do not necessarily reflect those of the World

Bank, its Board of Executive Directors, or the countries

they represent.

1201

reasons. Many countries, for example, Cooted�Ivoire and Nigeria, have not yet reached thepeak of the AIDS epidemic. As such, the num-ber of orphans will increase significantly for along time even after rates of new infections withAIDS have been brought under control in somecountries. Taking these two factors together,USAID estimates that, by 2010, there will be35 million AIDS orphans in Africa. Failure toprovide adequate education and health care forthis large segment of the population is likely tohave long-term impacts on social and economicinfrastructure as well as on productive capacityfor the foreseeable future.In this paper we use a case study from

Uganda to assess the extent to which there isa need for action, specifically whether undercurrent regimes orphans face a particular dis-advantage. We then draw on a wider range ofcase study material to discuss the various policyoptions available to respond to this challenge.The paper is structured as follows. Section 2provides a brief review of the literature and aconceptual framework that motivates the em-pirical analysis. Section 3 explores the extent towhich, in the case of Uganda, accommodationof an orphan has affected household welfareand whether there is any evidence for fosterchildren being disadvantaged in terms of theiraccess to public services in education andhealth. Section 4 discusses the evidence withrespect to various strategies in order to providespecific assistance to orphans and Section 5concludes by pointing out areas that requiremore detailed study.

2. CONCEPTUAL FRAMEWORK

The literature has identified a number ofpossible impacts of a child becoming orphaned,

both in the short term and in the long term,and for the child, its immediate caregivers, andthe broader community (Foster & Williamson,2000; Urassa, Boerma, & Ties, 1997). Roughquantifications of the different channels areoften used to provide an estimate of the mac-roeconomic impact of AIDS and the ways todeal with this crisis. While the literature offersa moving account of the short-term effects oflosing a family member due to AIDS, there isvery little information on the impacts that arisefor the remaining orphans and family members(who may, in fact, disintegrate) in the mediumto longer term. Furthermore, to identify theextent to which this phenomenon would war-rant interventions that are distinct from ageneral targeting of poor households, it will becritical to have a more rigorous quantificationof the impact that would allow a comparison oforphans to children who live with their par-ents. 2 To motivate our subsequent analysis webriefly discuss the different channels that havebeen identified in the literature and then high-light how our empirical study adds to these.For the orphan herself or himself, the impact

of losing one or two parents is likely to increasethe risk of having to leave school and/or engagein child labor in order to generate resources, apossible decline in nutritional status, and therisk of being discriminated against and ex-ploited. One particularly serious risk that hasbeen variously described is the risk of losingassets, especially land, either because these willbe appropriated by others in the extendedfamily or because they had to be sold off inorder to provide for medical care.The decreased school enrollment for orphans

is well-documented (see Figure 1). UNICEFdata for 1994–99 for a wide range of Africancountries show that there exist systematic dif-ferences in enrollment between children who

Table 1. AIDS orphans and the dependency ratio in selected countries

AIDS orphans, 2000

Country Estimated number Percentage of total Age dependency ratio

Botswana 66,000 10 0.82

Burundi 230,000 7 0.94

Coote d�Ivoire 420,000 6 0.87

Kenya 730,000 6 0.90

Namibia 67,000 9 0.84

Uganda 1,700,000 15 4.0

Zambia 650,000 15 3.0

Zimbabwe 900,000 17 0.80

Sources: UNAIDS (2000).

WORLD DEVELOPMENT1202

have lost both parents and children who havelost neither or only one parent (UNICEF &DHS, 1994–99). Data on over 10,000 childrenin Burundi confirm that a significantly lowerpercentage of children with both parents de-ceased are in school compared with childrenwho have lost neither parent. Thus, at the na-tional or macrolevel, the orphan crisis poses achallenge to the preservation of Africa�s humancapital.Various studies demonstrate that being an

orphan affects children�s nutritional status. Forexample, stunting is significantly higher amongorphans than other children, even if other fac-tors are controlled for (Ainsworth & Semali,2000). While nonpoor families are reported tobe able to cope with this risk, loss of a par-ent raises the incidence of stunting to levelsfound among the poor; the impact is particu-larly severe for poor households. Recent datafrom Burundi also demonstrate that a higherpercentage of double and maternal orphansare malnourished compared to children withboth parents alive (Subbarao, Mattimore, &Plangemann, 2001).While all of this suggests that orphanhood on

the scale found in Africa today is likely to haveeconomy-wide impacts, the fact that most ofthe case studies are from relatively small areasmakes it difficult to document this and/or to

draw broader conclusions. Doing so would beimportant to identify how government pro-grams interact with the orphan crisis. It wouldbe indispensable if one wants to make a case forusing scarce public resources to help individualsand families cope with the orphan crisis, and inparticular to point toward specific interventionsthat would help in accomplishing this goal.

3. THE CRISIS OF ORPHANS INUGANDA: A CASE STUDY

In this section, we analyze the relevant data,discuss ways of testing hypotheses, and providequantitative evidence to show that, in Uganda,foster parenting not only imposes a seriousburden on communities but also has a consid-erable and quantifiable negative impact on theeconomy as a whole. Uganda is a good casestudy for analyzing the broader impacts ofAIDS as it is one of the few countries in Africawhere, as a consequence of quick and decisiveaction by the government, the incidence ofnew infections is now generally believed to bedeclining (Ntozi & Ahimbisibwe, 1999). Thisprovides an opportunity to explore indirect ef-fects that occur after the death of a specificperson and at the same time focus on thelonger-term impact of the disease. Doing so

Figure 1. Orphaned and unorphaned children in school in selected countries. Note: Countries are shown in decreasingorder of disparity between children whose parents are living and orphaned children. Source: UNICEF, DHS (1994–99).

http://www.unicef.org/pon00/outof.htm.

AIDS-INDUCED ORPHANHOOD 1203

explicitly acknowledges that the longer-termimpact of AIDS on surviving adults and chil-dren may be as important as the loss associatedwith a specific death. The changes in children�sschool attendance and nutrition (Ainsworthet al., 2000; Ainsworth & Semali, 2000) takentogether may well have a social impact that isat least equal to that of the death of anadult household member. On the other hand,the ability to draw on extended networks tosmooth consumption may help to somewhatcompensate for the immediate loss (Lundberg& Over, 2000). The ability to assess the impactof AIDS deaths in a longer-term perspective isof great relevance in terms of the persistence ofthe shock and therefore the possible justifica-tion for government intervention.

(a) Data and descriptive statistics

Evidence on foster children from two sur-veys collected in 1992 and 2000, respectively,provides an opportunity to address these issues,in three ways drawing partly on Deininger,Crommelynck, and Kempaka (2001). First, weassess the extent to which the number of fosterchildren has increased between the two periods,even though the overall incidence of AIDS hasdeclined. Second, we aim to identify the extentto which the addition of a foster child to ahousehold is equivalent to a shock that resultsin a reduction in long-term productive invest-ment, in addition to the reduction of per capitaconsumption which one would intuitively ex-pect to be associated with such a phenomenon.Finally, to assess the longer-term impact ofadult death on the accumulation of humancapital and thus the future productivity of thelabor force, and at the same time to demon-strate the interaction between public policiesand the fate of individual orphans, we focus onchanges in foster children�s access to health andeducation services.To distinguish short-term adjustments from

longer-term impacts, it is critical to have lon-gitudinal information, preferably collected forthe same household at different periods in time.In Uganda, adding additional rounds to initialsurveys has provided greatly improved infor-mation on households� coping mechanisms aswell as the incidence of AIDS-related orphan-ing (Ntozi & Ahimbisibwe, 1999). We use arelatively long panel data set consisting ofthe 1992 Integrated Household Survey (IHS)and the 1999/2000 Uganda National House-hold Survey (UNHS). The IHS is a compre-

hensive multipurpose household survey basedon a nationally representative sample of 9,886households. In addition to the standard socio-economic and expenditure information, thesurvey contains detailed information on eco-nomic activities. The UNHS is a nationallyrepresentative survey of 10,696 households,fielded between August 1999 and September2000. The questionnaire was developed tomatch closely the one used in 1992 and a panelof approximately 1,300 households who wereinterviewed in both periods was included. 3 Useof the panel enables us to not only to makeinferences about changes in the incidence ofAIDS and of foster children over time but alsoto assess the impact of the addition of a fosterchild to the household during the relevant timeperiod.The data on general socioeconomic charac-

teristics for the sample provided in Table 2point towards a considerable, though region-ally differentiated, improvement in living stan-dards during the period, albeit from a very lowstarting point. A clear indicator is the im-provement in housing conditions. The share ofhouseholds who live under a thatched roofdecreased from almost 70% to less than 50%during the period although there are markedregional differences and the share remains high,at about 94% in the North. In addition, theshare of households with cement walls morethan doubled, from 5% to 13%. This is cor-roborated by the decrease in the share ofstunted children, a figure that dropped from53% in 1992 to 43% in 2000.Despite a marked increase in overall living

standards, the incidence of foster care inUganda has increased rapidly, implying that,even for the age group below six years, themagnitude of the orphan crisis has indeed as-sumed dramatic proportions. This suggeststhat, although the number of new infections ison the decline, major long-term impacts of theAIDS crisis remain. The increase in foster carehas been particularly dramatic for childrenbelow school age where, during 1992–2000, theshare of foster children increased from 10% toalmost 20%, with an even stronger increase inthe North (from 7.2% to 19.3%). With almostevery fifth child not living with its biologicalparents, the share of households who have atleast one foster child has also increased. Whilein 1992 only 5% of households hosted a fosterchild, this figure tripled to 15% in 2000. Othersources support this trend; according to the2001 National Service Delivery Survey, 37% of


Table 2. Socioeconomic characteristics, Uganda 1992 and 1999–2000

Whole country Central East North West

1992 2000 1992 2000 1992 2000 1992 2000 1992 2000

Population characteristics

Total no. of individuals

(million)

18.50 22.04 5.25 6.40 4.81 5.85 3.89 4.19 4.54 5.59

Total no. of households

(million)

3.73 4.11 1.14 1.29 0.95 1.08 0.74 0.77 0.90 0.97

Share of population rural 87.4% 86.5% 74.0% 70.3% 90.3% 91.3% 93.9% 94.9% 94.0% 93.9%

Rate of population

growth

2.2% 2.5% 2.5% 0.9% 2.6%

Female headed

households

25.6% 26.3% 28.8% 27.8% 21.3% 23.7% 31.1% 34.9% 22.1% 20.8%

Housing & infrastructure

Roof thatched 68.7% 47.9% 46.5% 21.3% 70.8% 54.2% 93.9% 93.5% 67.6% 29.7%

Cement walls 5.4% 12.8% 12.0% 24.7% 5.7% 16.1% 2.0% 1.7% 1.2% 6.9%

Health status and shocks

Stunted children 53.1% 42.5% 53.6% 38.2% 54.5% 42.6% 49.5% 39.4% 54.3% 48.2%

Households w foster

child < 14 years

16.6% 28.1% 20.5% 31.3% 16.8% 26.7% 15.5% 27.9% 12.3% 25.4%

Share of foster

children< 14 years

13.7% 21.8% 18.5% 28.3% 13.9% 19.2% 11.8% 20.8% 9.9% 18.0%

Households w foster

child < 6 years

5.5% 15.0% 6.3% 16.9% 6.6% 13.2% 4.4% 15.9% 4.4% 13.7%

Share of foster

children< 6 years

10.5% 19.3% 14.5% 24.4% 12.0% 15.8% 7.2% 19.3% 8.0% 17.7%

Source: Own computation from the 1990–2000 UNHS and 1992 HIS.

AIDS-INDUCEDORPHANHOOD

1205

households hosted at least one orphan (i.e., achild who had lost either of both parents), thusconfirming the magnitude of the phenomenon.With more than one-third of Uganda�s

households hosting a foster child, the phe-nomenon of foster care appears to be signifi-cant enough to have broader macroeconomicconsequences. We explore the extent to whichthis has been the case by looking at the in-vestment behavior of households who did ordid not receive a foster child in the period be-tween our two surveys and the access of fosterchildren to health services and education.Information on household income, expendi-

ture, and investment provides a first impressionof the possible economic impact of the orphan

crisis. 4 Table 3 presents evidence on thechanges in real per capita expenditure and in-come as well as the mean annual rate of in-vestment of households that, during 1992–2000,received a foster child below the age of 14.These data are based on the approximately1,300 panel households for which informationon initial conditions (including presence of afoster child) is available. Both in the aggregateand if only foster children in the 7–14 year agegroup are considered, one notes large and sta-tistically significant differences between the twogroups. While it is not surprising that the ad-dition of a foster child would reduce per capitaconsumption or income, 5 mechanisms to copewith such shocks seem to be of limited effec-tiveness in offering relief for such crises, as canbe seen from the fact that the impact of addinga foster child to the household is still visibleafter a relatively long time. 6 Data on the meanannual investment in business assets, as de-scribed below, are consistent with this evidenceand suggest that households who had to ac-commodate an orphan invested significantlyless than those who did not. This suggests that

shocks of this type draw resources away fromeconomically productive pursuits and thus re-duce a household�s income generating capacityin the longer term. 7

Turning from the households who host fosterchildren to the children themselves, we find thatthat foster children were highly disadvantagedwith respect to service access, as defined byoutcome variables in the area of health andeducation (Table 4). For health, outcomevariables included use of vitamin A capsules (avariable which is unfortunately available onlyfor 2000), and vaccination against diphtheriaand measles, for all children below the age ofsix. In terms of education, we focus on netprimary and secondary school enrollment forchildren in the 6–12 and 12–18 age groups. Theevidence suggesting a general decline in theshare of children with access to vaccination andan increase in those having access to education,reinforces the importance of having a controlgroup that allows one to distinguish shock-induced changes in service access from thoseoccurring as a consequence of broader changescompletely unrelated to the specific shock underscrutiny. Specifically, comparison between fos-ter and nonfoster children indicates that, eventhough the formers were effected by the generaldecline of health-service access, they had sig-nificantly less access to all three types of healthservices in 1999–2000 than those who lived withtheir biological parents. 8

The picture is quite different in the case ofeducation. In contrast to the significant worsen-ing of the situation over time for foster childrenin terms of health, we find a marked increasein overall enrollment and very little differencebetween foster and nonfoster children either inthe initial or in the second period. While thiswould prima facie suggest that foster childrendo not suffer from disadvantages in educational

Table 3. Growth of per capita consumption and income for households receiving foster childrena

Change in per capita

expenditure

Change in per

capita income

Rate of investment

No foster child received 4.91%�� 8.34%�� 2.40%��

Foster child receivedb 2.95% 5.81% 1.88%

Total sample 4.42% 7.70% 2.27%

aAll rates are mean annual growth rates.b The indicator used is whether the household received a foster child less than 14 years old during 1992–2000.** Significant at 5%.*** Significant at 1%.


access, none of these descriptive statistics holdsconstant other factors that may have an impacton the observed outcomes. 9 Controlling forother factors, such as initial asset ownership orcontemporaneous changes in household com-position that may also have an impact onhouseholds� investment in a regression frame-work, is important in order to avoid drawingspurious conclusions. Doing so will also allowone to gain an idea of the order of magnitude ofthe effect caused by the shock as compared toother factors that are normally thought toaffect household investment. Exploring theseissues in more detail is the purpose of theeconometric analysis.

(b) Impact of fostering on household investment

To assess the impact of foster care onhousehold welfare, we are interested in twovariables, namely the investment effect of hav-ing to accommodate an additional member ona household�s long-term productive capacityand, second, the public good access effect, i.e.,the extent to which not being directly related tothe head of the household reduces foster chil-dren�s access to social services. Since policyregimes for health and education which were inforce in 1992 are significantly different fromthose in 2000, this also provides an opportunityto assess the extent to which policy can have animpact on such outcomes.The intuition underlying our econometric

test for an investment effect of receiving a foster

child is simple; households subjected to an un-anticipated shock in the form of having to ac-commodate an additional foster child are likelyto reduce not only consumption but also in-vestment. 10 Based on a standard investmentmodel it is straightforward to derive the opti-mal amount of investment in business assets ineach period t, I�t . Assume that the household isfaced with a sudden shock in the form of re-ceiving an additional foster child. Distinguish-ing between households which, in the periodunder concern, experienced such a shock andthose who did not, allows us to estimate a re-duced form investment equation which can beestimated empirically as follows:

Ikt ¼ a þ bXkt þ cZkt þ ekt ð1Þ

where Xk is an indicator for whether thehousehold received a shock during the previousperiod, Zk is a vector of household specificcharacteristics, and ek is an error term. In thiscontext, the estimate of b will provide a directmeasure for the impact of the shock. Of course,selectivity could be an issue, in particular ifthere are systematic initial differences in thecharacteristics of households who take in or-phans. If, for example, the propensity of takingin orphans is higher for households whothemselves suffer from AIDS, it may well bethis characteristic that would lead to an ob-served negative investment outcome. Unfortu-nately, we do not have a good instrument todeal with this issue. 11 The fact that we wereunable to find significant differences in initial

Table 4. Access to services by children with and without their parentsa

All children Own children Foster children

1992 1999 1992 1999 1992 1999

Health indicators for children 0–6 years old

Share vaccinated against

measles

76.9% 67.8% 76.1% 69.1% 80.9%�� 61.9%��

Share vaccinated against

diphtheria

88.3% 82.5% 88.1% 84.0% 89.3% 75.3%��

Share using vitamin A

capsules

57.4% 59.0% 50.2%��

School enrollment indicators

Share of 6–12 year olds in

primary

62.1% 83.7% 61.8% 83.7% 63.0% 83.9%

Share of 12–18 year olds

in secondary

10.3% 14.8% 10.4% 15.0% 10.0% 14.3%

a Stars indicate significant differences between foster and own children in 1992 and 1999, respectively.*** Significant at 1%.


conditions between households who receivedorphans and those who did not suggests thatthis may not be too important an issue inpractice.To estimate Eqn. (1), we use the increase

in enterprise assets over 1992–2000 as thedependent variable. This variable includesagricultural assets, structures, and transportequipment, but excludes household durablesand other consumer goods. Due to problems inthe comparability of asset stocks between thetwo surveys, the dependent variable, invest-ment (in percentage terms) is obtained fromretrospective information contained in the1999–2000 survey. 12

Results, reported in Table 5, with changes inthe number of foster children below 14 and inthe 7–14 age group entered separately as right-hand side variables, support the conclusionfrom the descriptive statistics. Adding onefoster child is estimated to reduce individualhousehold investment by between 0.59 and0.51% points. In addition, we find that both theage of the head of household (in squared form)and initial asset value are negative, indicatingreduction of investment over the life-cycle anda decrease in the propensity to invest withhigher initial levels of assets which could indi-cate conditional convergence in asset endow-ments. Note that the sign of the addition offoster children is diametrically opposite to‘‘natural’’ growth in the household size throughaddition of own children. In fact, higher initialhousehold size, as well as the addition ofmembers to the household during the period, isestimated to be associated with an increase ininvestment of between 0.09 and 0.11% pointsper person. Thus, everything else constant,households with two children who received afoster child will be much worse off than thosewho had a third child of their own. The mag-nitude of the estimates is unlikely to be affectedby the fact that we implicitly assume that theincrease in foster children, e.g., through deathof a household member is an exogenousshock. 13

To illustrate the magnitude of the shock, notethat overall annual investment in the samplewas only about 2.2% and that, with an averageof almost four children below 14 per house-hold, AIDS-related deaths normally createmore than only one case of foster care. Com-paring the estimated magnitude of the impactto that of other variables such as education andregional dummies suggests that even addingone orphan is equivalent to a reduction in the

head of household�s level of education by aboutthree years. Noting that the median head ofhousehold in the sample has only four yearsof schooling, this suggests quite a significantimpact. Similarly, due to combination of lackof access to infrastructure, low agro-ecologicalpotential, and civil unrest, the North is char-acterized by significantly lower investment (by0.75% points) than the rest of the country.Although with 0.58 or 0.50% points still slightlylower, the impact of receiving one foster child isalmost equal to the difference between theNorth and other regions of the country.All of this suggests that the increase in the

phenomenon of foster care observed in thedescriptive statistics will have a significant im-pact on reducing investment. To the extent thatsuch reduced investment is likely to affect the

Table 5. Impact of increase in foster children on house-hold investment

Age category of foster

children

Below 14

years

6–14 years

only

Change in no. of foster

children, 1992–2000a

)0.586�� )0.509�

(2.08)a (1.73)

Change in no. of own

children 1992–2000a

0.108�� 0.095��

(2.20) (2.02)

Head�s education (yearscompleted)

0.194�� 0.195��

(2.25) (2.25)

Education squared )0.007 )0.007(1.02) (1.00)

Head�s age (years) 0.054 0.046

(1.18) (1.03)

Head�s age squared )0.001�� )0.001��

(2.12) (1.99)

Assets (US$1,000) in

1992

)0.087�� )0.086��

(2.20) (2.20)

Urban dummy 0.112 0.107

(0.09) (0.09)

Household size in 1992 0.092�� 0.088��

(2.03) (1.98)

Eastern region )0.278 )0.286(0.83) (0.85)

Northern region )0.748� )0.757�

(1.84) (1.87)

Western region )0.415 )0.395(1.34) (1.28)

Observations 1056 1056

R-squared 0.08 0.07

aRobust t statistics in parentheses.* Significant at 10%.** Significant at 5%.


capital stock of the economy, it could be ofbroader relevance for economic growth. Al-though further and more specific researchwould be required to identify mechanisms andpolicy options that could help prevent such anundesirable outcome, the rapidly growing inci-dence of foster care will warrant attention bypolicy makers.

(c) Impact of fostering on human capitalinvestment and service access

Increased incidence of foster care may notonly reduce a household�s level of productiveinvestment but may also have a direct impacton the children themselves, in terms of humancapital investment and access to health services.From a policy perspective, this is of particularinterest because, during the period under con-cern (i.e., 1992–2000), policies in the educationand health sector differed considerably fromone another. The program of Universal Pri-mary Education (UPE) was introduced in 1997.By eliminating the costs of schooling for up tofour children per household (of which at leasttwo had to be girls), this program aimed toboost enrollment and thus human capital ac-quisition especially by the poor. In addition tothe elimination of fees, publicity campaigns andmobilization drives at the local level were con-ducted and appear to have been quite successful(Watkins, 2001). In the health sector, the periodcoincides with an increase in user fees as ameasure to improve availability and quality ofsupplies. While the latter appears to have beenachieved, household surveys suggest that thepolicy also led to a considerable increase inhealth spending by the average household

(Deininger, 2001). 14 Use of the household dataallows us to identify empirically whether thesepolicies had differential impacts on foster chil-dren who are likely to be among the mostvulnerable groups in society.In addition to deriving the households� opti-

mum level of investment, the standard house-hold model yields demand functions for healthand education, respectively. If, with an overallresource constraint, a household has two chil-dren, one own and one fostered, it is likely thatspending for child m will be lower than for childl. This implies that the observed health or edu-cational status of each child becomes a functionof household characteristics, supply of services,and whether the child is own or fostered. Thiscan be utilized to estimate cross-sectional

reduced form equations for health and educa-tion outcomes for child j in household k suchthat:

Hktj ¼ a þ bXkt þ cZkjt þ bTXktT

þ cTZkjtT þ vT þ ekjv ð2Þ

where Hkj is the health outcome (e.g., beingvaccinated or not) or the educational outcome(i.e., school attendance) and Xk is a vector ofhousehold specific characteristics that includehousehold income, the head�s age and educa-tion, etc., Zj is a vector of child specific char-acteristics including an indicator for whether ornot the child is a foster child, and ekj is an errorterm that is composed of a household-specificeffect gk and a random white noise term mkj. Asfor most of the outcome variables, there aretwo observations at different points in timeavailable either for two cross-sections or for atrue panel. We add a time dummy T equalingzero if the observation is from 1992 and one ifthe observation is from 2000. The coefficient vthen denotes the magnitude of an independenttime trend and b þ bT or c þ cT are the coeffi-cients for household or individual characteris-tics, respectively, in the second period. Thisspecification allows us to evaluate overallchanges in health or educational outcomes thatcan be attributed to general policy shifts sepa-rately from changes in other household char-acteristics, including whether the householdhad to care for foster children. It further allowsthat statistical tests can be conducted to deter-mine the significance of specific variables.While Eqn. (2) can be estimated using stan-

dard ordinary least squares (OLS), presenceof unobserved household characteristics, ifcorrelated with other right-hand side variables,may result in biased estimates. This can beeliminated by focusing on variation within thesame household, thus estimating a fixed-effectsequation

Hkjt � Hk ¼ cðZjtkt � ZikÞ þ cT T ðZjkt � ZikÞþ vT þ mkjt ð3Þ

As the sample includes about 1,300 householdswho were observed in both periods, the coeffi-cient v can be identified, something that wouldnot be possible if we had only two pooled cross-sections.Results for educational enrollment are re-

ported separately for different age groups inTable 6. Both cross-sectional and householdfixed-effect regressions suggest that foster chil-dren faced a distinct disadvantage in 1992 but


that, in parallel with a considerable expansionof enrollment opportunities for the wholepopulation, this disadvantage has disappearedover time. As the table illustrates, the coefficientof the foster child dummy is highly significantand negative both for primary education of 6–12 year olds as well as for secondary educationof 6–18 year olds, both in the cross-section and,for a smaller sample, with household fixed ef-fects. This suggests that in 1992 foster childrenwere at a distinct disadvantage in terms ofaccess to opportunities, for primary and second-ary education. For primary, the time dummy issignificant and highly positive in both equa-tions, suggesting that the implementation of

UPE has helped to increase primary enrollmentacross the population irrespective of socio-economic status. A slightly smaller but againhighly significant coefficient in the case of sec-ondary education for the cross-section (thoughnot for the fixed-effect regression) implies thatwe cannot exclude the possibility that, in ad-dition to directly improving primary levels ofenrollment, UPE has generated spillover effectsthat helped to improve enrollment at the sec-ondary level as well.These results lead us to conclude that in ad-

dition to improving access to education for thepopulation as a whole, UPE and the associatedpolicy reforms have been particularly beneficial

Table 6. Logit regression for children�s school attendance

Primary, 6–12 year olds Secondary school

Cross-section Fixed effects Cross-section Fixed effects

Foster child )0.213�� )1.249�� )0.130�� )1.279��

(6.40) (7.33) (2.82) (4.66)

Foster child � year 0.237�� 1.177�� 0.154� 0.700�

(5.00) (4.90) (2.53) (2.08)

Household income 0.404�� 0.547� 0.566�� 0.658

(18.63) (2.05) (19.55) (1.23)

Income � year )0.230�� )1.048�� )0.119�� )0.523(7.03) (3.52) (3.08) (0.98)

Female dummy )0.161�� )0.583�� )0.281�� 0.181

(5.71) (5.86) (6.52) (0.99)

Female dummy � year 0.135�� 0.457�� 0.355�� 0.080

(3.36) (3.20) (6.43) (0.35)

Father�s education 0.083�� 0.140�� 0.072�� 0.104

(8.82) (2.80) (6.00) (1.38)

Father�s education � year )0.042�� )0.068 )0.034�� )0.056(4.20) (1.27) (2.66) (0.71)

Mother�s education 0.108�� 0.181�� 0.049�� 0.009

(10.85) (3.51) (4.12) (0.11)

Mother�s education � year )0.083�� )0.135� 0.006 0.103

(7.59) (2.43) (0.46) (1.30)

Year dummy 3.576�� 14.545�� 1.688�� 6.202

(8.94) (3.96) (3.44) (0.91)

Western region 0.089�� 0.016

(3.11) (0.44)

Eastern region 0.229�� 0.098��

(7.73) (2.73)

Northern region )0.177�� )0.099�

(5.71) (2.18)

No. of observations 24,216 7,424 15,535 2,560

No. of households 2,281 859

Pseudo R2 0.2276 0.2167

Log likelihood )10109.91 )1466.71 )5655.28 )868.09

Absolute value of z statistics in parentheses. Age dummies (in years) included but not reported.* Significant at the 5% level.** Significant at the 1% level.


to foster children. Foster children have not onlybenefited from increased nationwide availabil-ity of educational opportunities, the positiveand significant interaction between the fosterchildren and year dummies (for the wholegroup) indicates that they have been able tocompensate for their initial disadvantage inregard to both primary and secondary educa-tion as well. Indeed, in 2000, i.e., after theadoption of UPE, being a foster child no longerconveyed a disadvantage with respect to accessto education.

The equations for vaccination against diph-theria and measles suggest that, in the case ofhealth, changes have been quite different fromthose observed in education. Specifically, wefail to find any evidence for discriminationagainst this group in the initial period, asevidenced by the lack of significance of thedummies for both the cross-section and thefixed-effects regression 15 (Table 7). At the sametime, the negative coefficient of the interactionbetween foster and time dummies suggests that,in contrast to improved access to education,

Table 7. Logit regressions for children�s access to vaccinationsa

Diphtheria Measles

OLS Fixed effects OLS Fixed effects

Foster child 0.036 )0.065 0.132 0.343

(0.33) (0.14) (1.35) (0.86)

Foster child � year )0.392�� )1.233�� )0.428�� )1.656��

(2.98) (2.36) (3.60) (3.64)

Female head 0.417�� 0.327��

(6.81) (6.15)

Household income )0.038 0.028

(0.47) (0.39)

Male dummy 0.028 )0.212 0.001 )0.116(0.37) (0.89) (0.01) (0.59)

Male dummy � year )0.114 0.142 0.014 0.293

(1.17) (0.48) (0.16) (1.17)

Father�s education 0.111�� 0.085��

(3.68) (3.33)

Father�s education � year )0.053� )0.049�

(1.71) (1.84)

Mother�s education 0.153�� 0.113��

(4.48) (3.87)

Mother�s education � year )0.071�� )0.054�

(2.02) (1.79)

Year dummy )0.250 )0.765�� )1.087 )0.665��

(0.25) (2.58) (1.25) (2.56)

Western region 0.770�� 0.758��

(11.40) (12.35)

Eastern region 0.401�� )0.051(6.34) (0.92)

Northern region 0.594�� 0.518��

(7.94) (7.59)

Constant )2.940�� )1.944��

(3.43) (2.63)

No. of observations 16,578 1,635 16,577 3,875

No. of households 652 1,541

Adj. R2 0.0972 0.253

Log likelihood )382.305 )493.079

aAbsolute value of z statistics in parentheses. Age dummies (in months) included but not reported.* Significant at the 10% level.** Significant at the 5% level.*** Significant at the 1% level.


access of young foster children to health ser-vices has actually worsened disproportionatelyover time, compared to the overall population.While foster children were not disadvantaged interms of vaccination rates in 1992, they wereless likely to be vaccinated in 2000. Tests revealthat, with the exception of the fixed-effects re-gression for diptheria (column 3), the slopecoefficient for foster children is indeed signifi-cantly different from zero at the 5% level. Thisis a first indication suggesting that governmentpolicies can make a difference in terms of accessto services and household welfare. While partof this phenomenon may be due to supply-sidefactors (there are reports that clinic staff are,for various reasons that range from socialstigma to fear of contracting diseases them-selves, averse to dealing with foster children),the result clearly points toward the fact that, inthe health sector, there was no policy in placeto help foster children overcome their naturaldisadvantage. In the absence of such a policy, adistinct worsening of access to health servicesensued.Restricting the sample to the same house-

holds for a fixed-effects estimation as discussedabove confirms this result and at the same timeyields a negative and significant coefficient onthe time dummy, suggesting a more universaldecline in households� ability to access vacci-nation services between the two periods (col-umns 2 and 4 of Table 7). Discrimination inaccess to the use of Vitamin A capsules is alsovisible (Table 8). To what extent this is corre-lated with the simultaneous adoption of in-creased user fees for the provision of healthservices remains an issue for further research.Nonetheless, it is clear that, in the absence ofgovernment policies to broaden access to suchservices, the ability of foster children to gainaccess to health services has seen a markeddeterioration.

4. INTERVENTIONS IN SUPPORT OFORPHANS

The above discussion has demonstrated thedimensions of the orphan crisis in Uganda,shown that it affects outcomes at the individualand household level, and that policies do havean impact on such outcomes. The latter finding,i.e., that the impact of the crisis will be accen-tuated or moderated by sectoral policies sug-gests that appropriate interventions may havean important role to play in helping those af-

fected cope with their situation. We use this asa motivation to describe and discuss in moredetail the different types of interventions thatare currently under consideration or imple-mentation in different African countries to dealwith the orphan problem. As available evidenceon cost, effectiveness, and impact of such pro-gram is rather thin, the discussion will belargely at the conceptual level.

(a) Types of programs

Following Subbarao et al. (2001), we cate-gorize interventions to support orphans intoseven broad groups, ranging from household-based ones to government interventions (seeTable 9). We discuss the advantages and dis-advantages of the different public policy andinterventions to mitigate the impact of thesystemic shock of AIDS in the most affectedcountries, provide some tentative estimates ofthe likely scale and cost of mounting a credibleeffort to help orphans and families cope withthe problem, and use this as a basis to identifypromising areas for future research.

Table 8. Logit regression for children�s vitamin A capsuleusea

Specification

Cross-

section

HH fixed

effects

Foster child )0.156� )1.680�

(4.28) (3.07)

Income (log) 0.070�

(2.66)

Male dummy 0.005 )0.116(0.20) (0.54)

Father�s education 0.019�

(5.19)

Mother�s education 0.006

(1.47)

Western region 0.384�

(9.99)

Eastern region 0.632�

(16.26)

Northern region 0.345�

(7.34)

Observations 9,044 1,407

No. of households 574

Pseudo R2 0.1134

Log likelihood )5469.67 )163.02

aAbsolute value of z statistics in parentheses. Agedummies (in months) included but not reported.* Significant at the 1% level.


Fostering of orphans by relatives in an ex-tended family continues to be the most preva-lent practice in Africa, as illustrated by the highincidence of fostering in the case study reportedabove. Data from other East African coun-tries suggest that in most cases, grandparents

assume this function (23% in Uganda, 43% inTanzania, and 38% in Zambia) although theextended family is involved as well (see Table10). This practice is more attuned to Africansocio-cultural conditions than most otheroptions, allowing foster children to be better

Table 9. Interventions for foster families and children in vulnerable circumstances

Intervention Advantages Disadvantages

Fostering ––Family members are most likely to act in

child�s best interest.––Discrimination in food allocation, work-

load, education, etc., may exist.

––Family integration promotes psychologi-

cal and intellectual development of children.

––Fostered children are integrated into

society more readily than children in or-

phanages.

Subsidies distributed

through the

family

––Encourages even poor families to foster

orphans with the additional costs of caring

for orphans borne by the government.

––Difficult to monitor.

––Subsidies sometimes benefit head-of

household only.

––Subsidies may be shared among too many

family members, thus diluting the amount

of support going to the orphan.

––Subsidies exclusively for the orphan may

stigmatize the orphan

Subsidies distributed

through the com-

munity

––Communities will better know needs of

family.

––May not work in urban areas where sense

of community is weak.

––If distributed by churches, stigma may be

reduced.

––May not work in communities where

ethnic tension or discrimination exists.

School vouchers/

subsidies; health

vouchers re-

deemed by clinics

––School subsidies are easy to monitor. ––May entail horizontal inequity, to the

extent children with parents alive but living

in abject poverty do not receive any subsidy.

––Most likely to prevent future loss of

human capital.

Income-generation

schemes for fos-

tering families

––Increase short-term incentives of house-

holds to adopt children.

––Rarely succeed without training, follow-

up, and leadership.

––If successful, improve the welfare of

orphans.

––Provide no long-term incentive for caring

for orphans.

Family tracing ––Being reunited with family members

brings psychological benefits.

––May not be viable in post-conflict situa-

tions, in areas where a large percentage of

the population has died or is missing, or in

war-torn economies where family members

are unable to care for orphans.

Orphanages ––Better than child-headed households or

being a street child.

––Lack incentive to act on behalf of orphan.

––Orphanages run by religious groups may

reduce stigma and attract donor and char-

itable funds.

––May harm psychological development of

orphans.

––Not cost-effective.

––Can easily become commercial institu-

tions rather than welfare institutions.

––May not meet the emotional needs of

children.

Source: Subbarao et al. (2001).


integrated into society than they would be hadthey been placed in an institutional setting suchas an orphanage. As it is also beneficial to thepsychological and intellectual development ofthese children, this is considered the most de-sirable type of intervention. As we have seenabove, however, having to host an orphanplaces additional strains on the household�slivelihood and, unless sectoral policy is veryfavorable, can lead to disadvantageous humancapital and health outcomes for the orphan himor herself, e.g., by orphans being more likely todrop out of school, or less likely to have accessto healthcare. Part of this may be due to thefact that orphans often live with caregivers whoare either too old or too young, and often tooimpoverished, to adequately provide for them.The burden of caring for orphans thus putsenormous pressure on the extended family,raising the question whether support by eitherthe community or the state can help householdsdeal with this issue and thus improve the well-being of orphans.

Community-level support comes from vari-ous sources including national and interna-tional nongovermental organizations (NGOs),religious groups, community-based organiza-tions and communities themselves. Support isgiven primarily to the foster families, althoughit is sometimes provided through institutionalcare in the form of children�s homes, or or-phanages. Communities actively participate inactivities intended to benefit the orphans, suchas contribution of labor and materials, man-agement of orphans programs, and sometimesoversight of households headed by orphansthemselves.Although the level of such interventions is

increasing, the scale of interventions still re-mains a poor match for in light of the scale ofthe problem. As Table 11 illustrates, only about4.8% of the roughly 1.7 million orphans inUganda receive any support (Uganda AIDSCommission). In other countries, various kindsof intervention to assist orphans are being at-tempted, some with success, but most are in the

Table 10. Who is caring for orphans?

Sample covered Caregivers

Uganda Luweero district, Survey of 732 orphans 32% grandparents

50% surviving parent

16% extended family

5% community

Zambia National Survey (1996) 38% grandparents

55% extended family

11% older orphan

6% non-relative

Rural Tanzania 297 rural orphans in Mawezi Regional Hospital 43% grandparents

27% surviving parent

15% extended family

10% older orphan

5% community

Sources: Monk (2001), Government of Zambia (1999), and O�Gara, Huffman, and Lusk (2000).

Table 11. Uganda: orphan support programs, 2000a

Type of organization Number of agency units Budget (US$ mn) (1998–2000) Orphans assisted

International NGOs 29 6.05 27,200

National NGOs 32 7.10 33,855

CBOs 62 0.52 6,300

Religious groups 31 1.22 10,100

Local government agencies 22 0.94 7,210

Private sector 7 0.65 3,400

Total 183 17.40 83,100

Source: Based on data from Uganda AIDS Commission (2002).a Total number of orphans in the country: 1.7 mn; percent assisted: 4.8%.


form of pilots or smaller scale interventions.For example, in Malawi, NGOs are slowlyspreading out. Some communities have wit-nessed the establishment of village orphancommittees to monitor the local situation. InZambia, the government, NGOs and UNICEFcollaborated to establish the Children in NeedNetwork (CHIN) which provides support toregistered NGOs working with vulnerablechildren. In Zimbabwe and Kenya, church-sponsored orphanages are operated with thehelp of some private sector support, thoughtheir outreach is limited compared to theorphans in need. Many mosque-sponsored in-stitutions also work to protect orphans, par-ticularly in North Africa and in Muslim areas.In these countries, many of the larger NGOsare devoting a significant part of their pro-grams toward orphan care. In Botswana anumber of NGOs and CBOs have taken thelead in support of extended and foster families.These groups provide services throughout thecountry ranging from family counseling andday care for orphans to providing for basicneeds including food, clothing and education.As many of these programs are supported byNGOs who rely heavily on donor funding,sustainability is, however, a big problem; in thepast programs have often disappeared whendonor funds dried up.Public intervention is appropriate if there is

market failure or a clear case for redistribution.On both counts, programs for orphans arelikely to be justified––individuals can normallynot borrow for acquisition of human capitaland households who host orphans are generallyamong the poorest (Case et al., 2002). In fact,recent years have seen increased activity on thisissue in many countries, although the actualresources allocated to orphan support mayvary.For example, Botswana has set up a National

Orphans Program in 1999, as a partnershipamong government departments, NGOs, com-munity-based organizations, and the privatesector. Policies were developed to strengtheninstitutional capacity, provide support for wel-fare services, support community-based ini-tiatives and monitor and evaluate programs.This program is responsible for coordinating,for example, the registration of orphan datathrough a national database, identifying andaddressing the needs of foster children and theirfoster parents, training of community volun-teers in basic childcare, and reviewing govern-ment child protection policies.

In Eritrea, the government aggressivelysought to reintegrate orphans with their nearestrelatives. A one-time economic support to thefoster families, for example, through provisionof economic assets such as oxen, is providedto help the families sustain support of theorphaned child. While this program is fundedfrom national sources, Zambia, Uganda, Tan-zania, Kenya, Malawi, Ethiopia, and Burundiare borrowing for AIDS prevention and care,providing an opportunity to include orphancare as a component of a broader program.The above evidence allows three conclusions.

First, the orphan crisis has elicited a multitudeof responses, from NGOs, religious groups,and increasingly also from governments. Sec-ond, given the differences in the severity of theproblem as well as culture and institutionsacross countries measures to deal with theproblem will need to be country specific. Third,while it is safe to say that the responses thus farhave often been inadequate to address theenormity of the problem, it is not clear whatlevel of effort would constitute a ‘‘satisfactory’’response that is in line with the potentialdamage wrought by the crisis. We use this asa motivation to explore in more detail theamount of resources needed to deal with themost urgent aspects of the orphan crisis andthen formulate some questions for further re-search.

(b) Cost estimates

The cheapest alternative for recurrent sup-port to orphans is to provide them withschooling and nutritional supplementation.Current estimates put the cost of this at about$105 per orphan per year in Uganda and $148in Burundi. Given its likely impact on the futurewelfare of orphans, assistance to schooling hasemerged as the preferred option in Zimbabwe,Swaziland and Zambia. 16 The attraction ofthis form of assistance is that, with assistancefrom the community in identifying childrenwho need help, it can be implemented throughexisting institutions and that therefore scalingup is relatively easy. Even the cost of such a‘‘minimalist’’ approach are significant; in thecase of Uganda, and assuming that 20% of theestimated 1.5 million orphans are vulnerable,the annual cost would be $31.5 million. For thewhole of Africa, taking the lower estimate andrestricting the target group to the bottom 20%of double and maternal orphans the total an-nual cost would amount to about $250 million,


in both cases excluding the cost of administer-ing the subsidy. 17

In countries with very high levels of orphanswhere community and household-based fostercare has reached its limits, there may be a casefor expanding the menu of options to includenonfamily-based options. Orphanages havebeen established on a larger scale in Eritrea,Uganda and Zambia. But, expansion of thisoption in Africa is prevented by the high unitcosts, which range from $649 and $689 perchild and year in Tanzania and Eritrea, re-spectively to $1,350 per child and year in Eri-trea. These figures put scaling beyond the reachof most African economies. In this context, thenotion of a ‘‘children�s village,’’ which convertsthe concept of orphanage and thus allowsbetter integration of children with the nearestcommunity in a way that is particularly suitableto African culture, has evolved. Some of theadvantages of this option include that childrenare located within communities who provideoversight, economies of scale can be realized,and NGOs are easily attracted. Though data oncosts of such intervention are not available, thepresumption is that it costs a lot less than atypical ‘‘orphanage,’’ thus possibly providingan attractive and cheaper alternative.Compared to the above, the costs of tracing

and re-integrating orphans with their extendedfamily, an option that is of relevance in post-conflict situations, is estimated to be between$228 and $305 per child for the case of Burundiand Eritrea, respectively. Of course, being aone-time cost, this is not really comparablewith the options discussed earlier and may notbe feasible in all circumstances.

(c) Areas for further research

The evidence from Uganda presented earlierclearly demonstrates that foster children havemore limited access to some public services butdoes not allow to make inferences on the im-pact or efficacy of different types of interven-tions, or the type and nature of interventionsthat would be needed to deal with the problem.The fact that governments are increasingly be-coming aware of the need to deal with the or-phan crisis and initiate programs to deal with itprovides an important opportunity to advanceknowledge in this area, both from a policy anda research point of view. A number of areaswhere research might fruitfully be undertakenare the following.

(i) Identification of high risk groups andsituationsIdentifying the specific risky situations in

which an orphan could be disadvantaged, suchas being fostered by a distant relative or anonrelative, is an area that provides ampleopportunity for further study. This has impor-tant implications for targeting of assistance.While there seems to be consensus that themost needy groups, i.e., double and maternalorphans in low-income households, should bethe first priority for assistance, learning moreabout the coping strategies adopted by thesecould provide valuable knowledge to improvetargeting and timing of government programsand NGO interventions.

(ii) Asset dynamics of affected householdsOur paper illustrates that the impact of or-

phanhood goes beyond the individual or evenfamily affected. Households providing fostercare to orphans reduce their savings and in-vestment, and their vulnerability is likely toincrease. Research to explore the dynamics andmechanisms involved in more detail could helpidentify critical points when support could helpprevent a permanent downward-spiral and thusbe most effective in preventing asset loss andthus safeguarding at least a minimum level ofconsumption for orphans and their hosts. Onestrategy of addressing this issue, which alsowould allow to deal with many of the short-comings in our analysis that are due to using alarge national household survey rather thana more specialized instrument tailored to theissue at hand, could be to re-interview a sampleof households who did receive orphans duringthe period under concern.

(iii) Impact of existing interventionsGiven that a wide variety of strategies and

delivery mechanisms are currently used toprovide assistance to orphans, the differences inprogram design and implementation could bevery useful in order to learn more about theimpact of such programs and the relative ad-vantages and disadvantages of various deliverymechanisms. Even though it may take sometime before the ultimate impact can be evalu-ated, laying the basis for this in terms of con-sistent baseline surveys needs to be done nowand is an appropriate task for joint support byinternational donors who are sponsoring theseprograms.


(iv) Potential and limits of community targetingThe literature clearly shows that there are

many potential obstacles, e.g., ethnic cleavages,that might make exclusive reliance on commu-nity targeting undesirable (Abraham & Plat-teau, 2001). At the same time, without somedegree of reliance on communities and oncommunity-based monitoring, most of the in-terventions discussed earlier are likely tobecome infeasible. Further research on the prosand cons of community-based approaches totargeting, including delineation of minimumconditions for community-targeting to reachvulnerable orphans, would not only be desir-able but could also provide quick evaluationresults in the short to medium term.

5. CONCLUSIONS AND POLICYIMPLICATIONS

Throughout sub-Saharan Africa, the crisis ofHIV/AIDS, and with it the number of AIDS-induced orphanhood has reached alarmingproportions. Longitudinal data for Uganda for1992 and 2000 allow us to explore the impactof receiving orphans at the household level inorder to make inferences on the broader con-sequences the crisis may have, as well as thescope and level of response that would beappropriate.A first finding is the sheer magnitude of fos-

ter care in Uganda and the extent to which thephenomenon has increased over time. Thissuggests that, even if the immediate impact ofAIDS on mortality is in decline, the legacy oflonger-term negative welfare impacts will con-stitute a formidable challenge for the fore-seeable future. The dramatic increase in thenumber of foster children and the householdshosting them observed during the period sug-gests that the crisis may adversely affectbroader social systems and human develop-ment in African countries.

Second, we note the negative impact ofreceiving foster children on households� capitalaccumulation. The magnitude of the effect,which is estimated to reduce investment bybetween one-fifth and one-fourth relative to theaverage, is important in an environment whereinvestment is unequivocally considered crucialfor future growth, and where returns to botheducation and productive assets have consid-erably increased during the last years (Deinin-ger & Okidi, 2001).Regarding the accumulation of human capital

by foster children themselves, the case studysuggests that foster children face difficultiesin accessing services but that sectoral policies cango a long way toward helping them to overcomethese barriers. Thus, foster children in Ugandadid not experience a serious disadvantage ineducation since the country�s education sectorpolicies have been fairly inclusive, equitable andeffective. On the other hand, they were at a dis-advantage in terms of health outcomes includingimmunization. This implies that even thoughNGO programs and donor-assisted programsare important, they may not yield desirableoutcomes in a policy environment that is lessenabling for foster children.What can we learn from interventions and

specific programs thus far to guide design ofsuch programs? Most of these efforts to datehave been piecemeal and inadequate relative tothe crisis at hand. Estimates for Uganda, as justone example, suggest that the totality of effortsthus far (NGOs, governments and donors)reaches no more than 5% of orphans. In addi-tion, interventions need to be carefully chosento address the specific risks faced by orphans ina given country environment, and for strength-ening the existing community coping strategies.Finally, the cost of assisting orphans and theirextended families is not negligible; assumingthat assistance is reserved to the 20% of or-phans who are most vulnerable, the annualrequirement would be as much as $300 millionfor sub-Saharan Africa.

NOTES

1. The estimates are based on a number of known and

assumed parameters: the age pattern of HIV/AIDS

infection, the age pattern of fertility, perinatal transmis-

sion rates, the average survival time after infection, the

mortality rate for those under 15 (for orphans and

nonorphans), the morality rates for adults from other

causes, and the population distribution by age and sex.

The literature distinguishes maternal orphans, i.e., chil-

dren under 15 years of age whose mother has died;

paternal orphans, i.e., children under 15 whose father has

died; and double orphans, i.e., children under 15 who

have lost both parents.


2. One current in the debate argues that poverty

targeting will be enough to take care of the orphan

problem (Ainsworth, Beegle, & Godlike, 2000) while

others have found a systematic difference between

orphans and nonorphans in access to education even

within the same family (Case, Paxton, & Ableidinger,

2002). Unfortunately, available evidence on this issue is

thin and further research would be highly desirable.

3. The ability to use a panel of households that spans a

long-time period comes at the cost of not being able to

identify true orphans. We therefore focus on foster

children, i.e., children who are not physically related to

the household head, throughout. Three arguments sug-

gest that this may be less of a limitation than one would

think. First, while the lack of an overlap would be a

serious limitation if we were concerned only about the

impact of AIDS, whether the physical parents are still

alive is less of an issue if the interest is to adopt a broader

perspective that includes other shocks (e.g., foster care

that is induced by the displacement of children or families

caused by civil strife in various parts of the country). In

fact, even for AIDS victims, children may be sent to

relatives some time before the person actually dies,

especially in the case of single-parent households. Second,

under the assumption that themagnitude ofmeasurement

error does not change over time (i.e., the share of foster

children that is not related to AIDS in the respective age

group does not change), the variable of interest, i.e., the

change in AIDS-related foster children will still be

measured accurately. Finally, economically motivated

foster care (e.g., mothers leaving their children in the

village to be able to pursue urban careers) is normally

associated with a transfer of wealth to the household

taking care of the child, which is generally expected to do

better in school than its peers (Ainsworth, 1996). Any

estimates of economic hardship caused by the receipt of a

foster child derived from such a sample would thus

underestimate the effect of crisis- or shock-related foster

care, implying that the figures given would constitute a

lower bound for the true shock-induced impact of foster

care. Also note that, especially for children below the age

of six, it is unlikely that economic motivations are the

most decisive factor. Note also that our sample includes

only children who have found a foster family. To the

extent that there are many orphans who are not actually

taken up by a household, our results would significantly

underestimate the impact of being an orphan.

4. While income and consumption are defined in a

standard way (the latter including home production), the

value of assets for both 1992 and 2000 was computed

based on retrospective information provided by the 2000

survey.

5. In fact, the decline in the growth rate is somewhat

larger than is accounted for just by the increase in the

denominator, suggesting that addition of a foster child

has a slight welfare-reducing impact.

6. Unfortunately, the data do not contain information

on changes in family composition which would allow us

to identify the length of time for which a foster child has

been with a particular family, a variable that would have

allowed to study in more detail the process of adjust-

ment by different types of households.

7. The lack of information of higher frequency as well

as an indication of precisely when the foster child

entered the household prevents us from identifying how

households were able to cope and in particular whether

the addition of the foster child resulted in a permanent

reduction of welfare or whether it led to a big immediate

decline from which the household was able to gradu-

ally recover. This would be of considerably interest and

constitutes an important area for follow-up research.

8. This is particularly surprising since foster children

were, in 1992, significantly more likely to be vaccinated

against measles.

9. For example, if, as is frequently reported in the

literature (and in fact supported by our data), foster

children and their parents live in areas where access to

education (but not to health services) is systematically

better, they may still be less likely to be sent to school

than own children.

10. As will be discussed in more detail below, the fact

that the shock may not be completely unexpected will

not affect the results from the econometric estimation.

11. The ideal instrument would be whether there is a

fatal illness or whether a recent death occurred in the

household�s extended family. Although an attempt wasmade to collect information on recent deaths in the

household, many enumerators skipped the question in

order to avoid the suffering and emotional hardship it

caused many respondents. While a large-scale household

survey is clearly not the right vehicle for dealing with

such sensitive issues, it would in principle be possible to

conduct follow-up interviews with a smaller and specif-

ically selected sample to explore such issues in more

depth. In the absence of such a survey, the estimates of

the investment effect presented above may be subject to

some bias. The fact that our results for education and

health are robust to use of household fixed effects

suggests that the bias might be limited.


12. As it was impossible to obtain precise retrospective

information on the value of each of the assets, respon-

dents were asked to provide a qualitative rank for the

value of the asset in question in 1992. Based on

experience in the field, the five rankings (about equal,

somewhat more, somewhat less, none, and much more)

were transformed into percentage increases (0, +25%,

)25%, )100%, +50%) in order to be able to obtain thevalue of assets in 1992.

13. In the case of an AIDS-related death, households

are likely to have some period of forewarning and thus

canmake adjustments implying that this assumptionmay

not hold. But, to the degree that households are able

to anticipate this shock and take precautionary mea-

sures, the measured impact would be biased downwards,

implying that our estimate provide a lower bound.

14. In fact, user fees for health services were aban-

doned in run-up to the 2001 elections. It remains to be

seen to what extent this will affect availability of supplies

and staff motivation as well as households� ability toreceive adequate services.

15. As rightly observed by a reviewer, parents of

fostered children may not actually know whether the

child has been vaccinated, especially if the vaccina-

tion occurred before the child entered the household.

As the 1999–2000 survey contained information on

whether the child had a card, we re-estimated the

cross-sectional regression using only the cases where

a card was actually seen. Results are very similar,

suggesting that this is not a significant source of

bias.

16. Over the last two years, Zimbabwe, one of the most

affected countries in Africa, has, through a Basic

Education Assistance Module (BEAM), launched a

large effort to prevent orphans and other vulnerable

children from dropping out of school.

17. If there is a need to provide similar levels of

assistance to other members of the same household in

order to avoid stigma, the cost would, of course,

increase correspondingly. Very little is currently known

about the stigmatization of orphans in different African

cultures.

REFERENCES

Abraham, A., & Platteau, J. P. (2001). Participa-tory development in the presence of endogenouscommunity imperfections, Mimeo. Department ofEconomics and CRED, University of Namur, Bel-gium.

Ainsworth, M. (1996). Economic aspects of childfostering in Cote d�Ivoire. In T. Schultz (Ed.),Research in Population Economics: Vol. 8 (pp. 25–62). Greenwich, CT: Greenwood.

Ainsworth, M., Beegle, K., & Godlike, K. (2000). Theimpact of adult mortality on primary school enroll-ment in Northwestern Tanzania. World Bank PolicyResearch Working Paper, The World Bank, Wash-ington, DC.

Ainsworth, M., & Semali, I. (2000). The impact of adultdeaths on children�s health in Northwestern Tanza-nia. The World Bank Development Research Group,Poverty and Human Resources, The World Bank,Washington, DC.

Case, A., Paxton, C., & Ableidinger, J. (2002). Orphansin Africa. Research Program in Development StudiesPrinceton University, Center for Health and Well-Being, Mimeo, Princeton, NJ.

Deininger, K. (2001). Determinants of health service useand health outcomes: Evidence from Uganda,Mimeo. World Bank, Wasington, DC.

Deininger, K., Crommelynck, A., & Kempaka, G.(2001). Long term welfare and investment impactsof AIDS-related changes in family composition,Mimeo. World Bank and Makerere University,Washington, DC/Makerere.

Deininger, K., & Okidi, J. (2001). Growth and povertyreduction in Uganda 1992–2000: Household levelevidence, Mimeo. World Bank, Wasington, DC.

Foster, G., & Williamson, J. (2000). A review of currentliterature on the impact of HIV/AIDS on children insub-Saharan Africa. AIDS 2000, 14(Suppl. 3), S275–S284.

Government of Zambia (1999). Situation analysis oforphans in Zambia, 1999: A joint USAID/UNICEF/SIDA study. Lusaka.

Lundberg, M., & Over, M. (2000). Transfers andhousehold welfare in Kagera, Mimeo. World Bank,Washington, DC.

O�Gara, C., Huffman, S., & Lusk, D. (2000). Assessmentand improvement of care for AIDS-affected childrenunder 5. Washington, DC: Academy for EducationalDevelopment.

Monk, N. (2001). Understanding the magnitude of amature crisis: dynamics of orphaning and foster carein rural Uganda. In International perspectives onchildren left behind by HIV–AIDS (pp. 7–13). Asso-ciation Francois-Xavier Bagnoud. Available: http://www.albinasactionforofphans.org/learn/ORPHAN-ALERT1.pdf.

Ntozi, J. P. M., & Ahimbisibwe, F. E. (1999). Somefactors in the decline of AIDS in Uganda. In: J. C.Caldwell, I. O. Orubuloye, J. P. M. Ntoza (Eds.), Thecontinuing African HIV/AIDS epidemic (pp. 93–107).Canberra: Health Transition Centre, AustralianNational University. Available: http://nceph.ani.edit.au/htc/html/continuing.html.


http://www.albinasactionforofphans.org/learn/ORPHANALERT1.pdf



http://nceph.ani.edit.au/htc/html/continuing.html

http://nceph.ani.edit.au/htc/html/continuing.html

Subbarao, K., Mattimore, A., & Plangemann, K. (2001).Social protection of Africa�s orphans and othervulnerable children: issues and good practice pro-gram Options. AFR HD working paper.

Uganda AIDS Commission (2002). The HIV/AIDSepidemic: prevalence and impact. Available: http://www.aidsuganda.org.

UNAIDS (2000). Report on the global HIV/AIDSEpidemic. Geneva, UNAIDS, June.

UNICEF & DHS (1994–99). Available: http://www.unicef.org/pon00/outof.htm.

Urassa, M., Boerma, T., & Ties, T. J. (1997). Orphan-hood, child fostering, and the AIDS epidemic inrural Tanzania. Health Transition Review, 7(Suppl.2), 141–153.

USAID, UNAIDS & UNICEF (2002). Children on thebrink 2002: A joint report on orphan exiimaies andprogram strategies. USAIDS. Washington, DC.Available:http://www.usaids.org/barcelona/prresskit/childrenonthebrink.html.

Watkins, K. (2001). The Oxfam Education Report.Oxford: Oxfam.


http://www.aidsuganda.org

http://www.aidsuganda.org

http://www.unicef.org/pon00/outof.htm

http://www.unicef.org/pon00/outof.htm

http://www.usaids.org/barcelona/prresskit/childrenonthebrink.html

http://www.usaids.org/barcelona/prresskit/childrenonthebrink.html

aids-induced orphanhood as a systemic shock: magnitude, impact, and program interventions in africa

Documents