Essays on Benefit Incidence and Efficiency of Public Spending on Education in Kenya
Abstract
Between 2002103 and 2009110 fiscal years, the Kenya government spent an average 0/6.4 percent 0/ GDP
and 26 percent 0/ total government outlays on education and training. Households cover such costs as
boarding expenses, user charges and private schooling costs. However education outcomes are not only
low but also vary across counties. About 8.6 percent and 68 percent of primary and secondary school age
children, respectively, are not in school. Survival to tertiary education level is relatively low with only 2
percent 0/ the pupils enrolled at primary grade one surviving to first year university education. This calls
for a detailed analysis 0/ levels 0/ benefit incidence; technical efficiency and external efficiency 0/ public
education spending in the country. The thesis consists 0/ three interlinked essays that empirically
investigate the nexus between education inputs; and participation, learning processes, outputs and
outcomes. The first essay estimates the average and marginal benefit incidence 0/ public education
spending and the associated/actors. The second essay focuses on technical efficiency of public education
spending and its correlates using the Data Envelopment Analysis (DEA) double bootstrap procedure. The
DEA two-step procedure is used to address the main criticism 0/ the standard DEA approach. As an
example, the technical efficiency indices obtained from the standard DEA may suffer from serial
correlation. The standard DEA approach takes no account 0/ the underlying data-generating procedures,
while violating the basic model assumption 0/ independence within sample. DEA double bootstrap process
combines the DEA model with bootstrapping techniques in estimating bias corseted efficiency scores. The
analysis was undertaken at county level with a view to mapping inequalities in education benefit incidence
and efficiency. The third essay focuses on external efficiency 0/ education and provides a link between
education and training to the country's labour market outcomes.
Using 200516 Kenya integrated budget household data: the thesis illustrates that education enrollment rates
are biased with children from medium and high income groups more likely to enroll at all levels 0f
education and hence benefit more from any expansion in public education spending in Kenya. The bias is
more pronounced at post primary level where enrollment rates are higher among the medium and high
income groups compared to the low income group. The level 0/ benefit incidence varies across counties.
The results point to the importance of increasing benefit incidence by improving socio-economic welfare 0/
households while addressing the socio-economic factors that constrain the poor and some counties from
benefiting from public education spending. Gender disparities and over-age enrolment in schools need to
be addressed.
Technical efficiency scores estimated using DEA bootstrap in f.ssay 2 c;re I. 24, 1.12 and 3.04, for primary,
secondary and tertiary education levels, respectively. Total/actor productivity change was estimated at
0.95 depicting a decline between 200516 and 2009110. The decline could be attributed to a decrease in
technical progress during the review period. The findings indicate that there is scope for improving
efficiency in education resource utilization. Outputs can be increased by 24 percent, 12 percent and 204
percent for primary, secondary and tertiary education, respectively, without increasing inputs. The
magnitude of resource saving is estimated at 17 percent, 10 percent and 52 percent for primary, secondary
and tertiary education, respectively. The level of savings depicts potential level of cut back in input use to
produce same output. Personnel emoluments as proportion of education expenditure have a negative and
statistically significant effect on efficiency of education. Female literacy has a positive and statistically
significant effect on education efficiency. The study demonstrates that rates of return to schooling increase
with level of education and vary across employment sectors. However, external and within household
externalities are higher than private or individual rates of return to schooling.
The study has various implications for policy. First, it is necessary to deal with pre-labour market
inequalities including disparities in access and benefit incidence of schooling across counties, gender and
income groups. Second, there is need to improve efficiency in allocation and utilization of education
expenditures. National and county governments will need to allocate resources more equitably while taking
into account other socio-economic factors that constrain demand for schooling. Some of these factors
include poverty and over-age enrollment. Further, an equitable resource allocation framework should be
developed under which counties with lower education outcomes receive higher portion of shared revenue
than the better performing regions. The government may consider linking education inputs and financing to
outcomes such as learning achievements and appropriate skills development. A qualifications framework
that provides a mechanism for measuring and linking educational attainment to the labour market should
be developed. It is important to increase access to schooling in secondary and tertiary levels of education
whilst addressing inequalities; in order for all Kenyans to benefit from the positive and substantial internal
and external effects of schooling.
This thesis has five contributions that relate to application of most current techniques in linking education
inputs to learning processes, outputs, outcomes and education externalities. First, the study is one of the
few studies to estimate technical efficiency of education , in Kenya. In doing so the study uses DEA double
bootstrap procedure which is a superior technique 10 the standard DEA to estimate technical efficiency
measures across all education levels and to determine correlates of technical efficiency of education.
Confidence intervals for technical efficiency measures obtained from the DEA double bootstrap procedure
enable statistical inference to be conducted. Second, the study applies the marginal benefit incidence
technique using cross sectional data-,to determine potential beneficiaries of expansion of public education spending in the Country. The study provides analysis on disparities in education sector performance,
technical efficiency and average benefit incidence at county le~l; 'and factors-affecting benefit incidence of
I education. Third, the study estimates comprehensive return! to education including external benefits of
schooling; within household social externalities; returns for technical education; and returns of education
in the agriculture sector. Fourth, the study addresses important policy issues including inequalities,
efficiency and benefit incidence of education. Addressing these policy issues is important for the country in
terms of meeting the aspirations of Kenya Vision 2030 and the need to enhance decentralization in
education management as stipulated in the 20 I0 Constitution. Fifth, the research contribute to existing
academic work on efficiency in public expenditure management, externalities of education and benefit
incidence analysis by providing additional empirical evidence at county sub-national level. Finally,
findings from the three essays would inform policy on improving efficiency in public expenditure
management and effective resource utilization within decentralized education system; and targeting of
education spending on education inputs
Citation
Doctor of Philosophy in EconomicsPublisher
University of Nairobi Department Of Economics