Working-Age Adult Mortality, Orphan Status, and Child Schooling in Rural Zambia
November 5, 2011 - David Mather
Print
EmailIDWP 118. David Mather. 2011. Working-Age Adult Mortality, Orphan Status, and Child Schooling in Rural Zambia
EXECUTIVE SUMMARY:
During the last decade, the Zambian government has dramatically increased expenditures on
primary and secondary schooling, and enrollment rates have risen dramatically. At the same
time, Zambia has faced the challenge of rising HIV prevalence and the possibility that recent
gains in long-term human capital development could be eroded if households which suffer
the death of a working-age (WA) adult pull their children out of school due to family labor
shortages or financial constraints. This paper uses panel survey data from rural Zambia to
measure the impact of WA adult mortality and morbidity on primary school attendance and
school advancement, and separately tests the extent to which orphan status affects these
schooling outcomes. There are five principal findings from our analysis.
First, we find that a homogenous conceptualization of WA adult mortality and morbidity
shocks are not by themselves a reliable indicator of poor child schooling outcomes. For
example, while we find that the effect of WA adult mortality and morbidity does not have a
significant negative effect on primary school attendance using the full sample of children, we
do find significant negative effects in some cases when we consider the gender of the child,
the pre-death wealth level of the household, and/or the gender and household position of the
deceased or ill adult.
Second, we find that effects of chronic adult illness on child school attendance depend upon
the household position and gender of the ill adult. For example, when we disaggregate
morbidity shocks by household position of the ill adult, we find that the presence of a
chronically ill head or spouse reduces attendance by 4.1%. We also find that the presence of a
chronically ill male adult reduces attendance of girls by 8.5%.
Third, we find that households in rural Zambia are more likely to respond to adult mortality
or morbidity shocks by reducing the attendance of girls relative to boys. For example, when
we stratify the sample by gender of the child, we find a recent WA adult death (0-4 years
ago) reduces girls’ probability of attending school by 7.9%, while the presence of a
chronically ill male adult in the household reduces girls’ school attendance by 8.5%. The fact
that negative impacts of WA mortality on girls’ schooling are larger in magnitude than those
for boys, and are significant for girls from both poor and less-poor households (while
insignificant for boys), suggests that there is a clear gender bias in rural Zambia in how
households respond to the death or chronic illness of a working-age adult.
Fourth, we find that the effects of adult mortality and morbidity on girls’ attendance are of
larger magnitude and more likely to be significant for girls from poorer households. For
example, while a recent WA adult death reduces girls’ probability of attending school by
7.9%, this effect is stronger among girls from poorer households where a recent WA death
reduces attendance by -10 points or roughly 12.8%. Likewise, while we find that the presence
of a chronically ill male adult reduces girls’ attendance by 8.5%, this effect is stronger among
poorer households, where it reduces girls’ schooling by 12.5%. The fact that we find
significant or larger impacts among children from poor households suggests that the
opportunity costs of children in such households become high during the illness or following
the death of a WA adult. It is likely that the financial constraints and increased labor demands
faced by poorer households who suffer a WA adult death leads them to reallocate the time of
children from school to family labor following the death of a WA adult.
Fifth, although we find evidence that mortality and morbidity shocks reduce attendance for
some children – namely, for girls, and especially girls from poor households – this negative
effect on attendance does not appear to result in delayed grade progression, as we do not find
evidence that these shocks result in significant losses for either school advancement or
highest grade completed. Nevertheless, the magnitude of the reductions in attendance for girls
due to mortality and morbidity shocks are large enough to warrant the concern of
policymakers, though measuring the potential effects of these shocks on a child’s actual
learning would require a much different and more in-depth methodological approach.
Sixth, we find that there are no significant negative effects of orphan status (parental,
maternal or double-parent orphans) on either child school attendance or school advancement,
regardless of whether we use the full sample or samples stratified by household wealth or
gender of the child. Because orphans in our sample are just as likely to be found in relatively
ex ante poor or wealthy households, this appears to rule out the possibility that insignificant
effects of orphan status on schooling outcomes is due to the orphans’ migration from their
original household. Rather, this suggests instead that orphan status is simply not a good
indicator of potential schooling disadvantage in the context of rural Zambia.
There are several policy implications from these results. First, because the extent to which
childrens’ schooling outcomes are affected by adult mortality or morbidity is specific to the
gender of the child, the household’s wealth level, characteristics of the deceased or ill adult,
and the timing of the adult death, it is inappropriate to categorize all children in Zambia who
are directly or indirectly affected by HIV/AIDS-related morbidity and mortality as being
especially vulnerable and in need of targeted school subsidies. Second, it follows that social
protection and education policymakers concerned with primary school under-enrollment in
Zambia need to tailor mitigation measures to the specific needs and situation of children in
rural Zambia. The evidence in this paper suggests that girls from households with a currently
ill head/spouse or male adult, as well as girls from households with a recent WA adult death
(i.e., within in the past 0-4 years) – especially those from poorer households – are most likely
to face losses in school attendance and advancement. Mitigation measures appropriate for
rural Zambia may therefore include conditional cash transfers targeted to girls from poorer
households which have incurred these mortality/morbidity shocks. Such assistance might not
only ensure that these girls attend school but could also enable poorer households to hire
additional labor rather than pulling other children from school to meet family labor demands.
Third, although Zambia abolished primary school fees over a decade ago, there may still be
barriers to enrollment such as continued household demand for child labor, additional
educational expenses for transport, school uniforms and books, and declining school quality
if enrollment outpaces new school construction and teacher hiring. These additional barriers
to enrollment may explain why we have found evidence of negative effects of adult mortality
and morbidity on girls’ schooling, even in a time period after the government had abolished
primary school fees. In addition, targeted schooling subsidies alone may not reduce schooling
deficits of some orphans, in the event that their poor schooling progress is due to the
emotional and psychological trauma of losing one or both parents or a lack of interest by their
adult guardians in their schooling.
Fourth, Zambia should continue to provide universal free primary schooling, as this policy
has been found in a number of countries to improve the enrollment and schooling progress of
those children most likely to suffer from poor schooling – namely children from poorer
households, both orphan and non-orphan alike. For example, evidence from Malawi and
Uganda suggest that improvements in enrollments among the poor through universal
abolition of primary school fees can substantially raise the enrollment of orphans, even to the
point of eradicating orphan schooling deficits (Ainsworth and Filmer 2006). Finally, it should
be noted that because of the well-established positive correlation between educational
attainment and safer sexual behavior (World Bank 1999), Education for All is itself an
important policy that can help reduce the spread of HIV/AIDS and thus the potential for
negative shocks to child schooling (Ainsworth and Filmer 2006).
