Post on 13-Jan-2017
SCHOOL OF ECONOMICS
UNIVERSITY OF NAIROBI
COURSE NAME: ADVANCED MICROECONIMICS THEORY 11
COURSE CODE: XET 502
LECTURER: PROF MWABU. G
PRESENTED BY: MAJUNE KRAIDO SOCRATES
REGISTRATION NUMBER: X50/68921/2013
CONCEPT PAPER TOPIC: THE RETURNS TO SCHOOLING IN KENYA
DATE OF SUBMISSION: 9th JUNE, 2014
ii
TABLE OF CONTENTS
TABLE OF CONTENTS ................................................................................................................ ii
1. INTRODUCTION ............................................................................................................... 1
1.1 Background ...................................................................................................................... 1
1.1.1. External Expenditure .................................................................................................... 1
1.1.2 Domestic Expenditure (Kenya) ................................................................................. 1
1.1.3 Household Expenditure ............................................................................................. 2
1.2 Problem Statement and Objectives .................................................................................. 2
2 LITERATURE REVIEW ........................................................................................................ 3
2.1 Theoretical Literature Review .......................................................................................... 3
2.2 Empirical Literature Review ............................................................................................ 3
3 MODEL ................................................................................................................................... 4
3.1 Conceptual Framework .................................................................................................... 4
3.2 Data .................................................................................................................................. 5
3.3 Estimation shortcomings .................................................................................................. 5
4 DATA ANALYSIS ................................................................................................................. 7
4.1 Regression results and heteroscedasticity test .................................................................. 7
4.2 Discussion of results......................................................................................................... 7
5 SUMMARY AND CONCLUSION ........................................................................................ 9
5.1 Areas of further research .................................................................................................. 9
REFRENCES ................................................................................................................................ 10
APPENDIX: Summary Statistics .................................................................................................. 12
1
1. INTRODUCTION
1.1 Background
Education is one of the major pillars of development often poised as a means of achieving both
social and economic benefits at individual and country level. Education is also a human right and
a foundation for development as declared by the United Nations, the reason for the push for the
adoption of Education For All (EFA) yardsticks (UNESO, 2014; Youseff, 2005). Therefore the
benefits of education can be generalized as follows; improvement of equity, acceleration of
technological change, promotion of human rights, enhancing social cohesion and reducing
poverty especially in developing countries (Fasih et al., 2012; Orodho et al., 2013). These
benefits have compelled several stakeholders at household, country and international levels to
prioritize education as shown by their expenditure levels. This study is interested in determining
the returns to these expenditures.
The next sections of this background indicate different levels of expenditures at the aforesaid
levels with an inclination to Kenya.
1.1.1. External Expenditure
Donor funding highly promotes resource allocation in education mostly in developing countries
(UNESCO, 2014)1. UNESCO (1998) shows that the World Bank alone financed about a third of
all external lending for education between 1989 and 1996. The World Bank has also been critical
in setting of international education policy agendas. However, donor funding has declined in
recent times and this trend is likely to persist over time (UNESCO, 2014). For instance, overall
education aid has decreased from $10.7 billion in 2004 to $8.3 billion in 2005 coupled with a 1%
decline from 2010 to 2011 and a further 4% decline in the 2011/2012 period (UNESCO, 2014).
1.1.2 Domestic Expenditure (Kenya)
Kenya like many developing countries has increased her commitment towards enhancing
education thereby leading to increased enrolment and government expenditure on education. The
government introduced Free Primary Education in 2003 and Free Day Secondary Education in
2008, which have led to increased enrolment mostly at Secondary level, i.e. 65% between 2001
and 2013 (ROK, 2013). Moreover, net primary enrolment rate increased to 95.3% in 2012, up
from 92.5% in 2008 (ROK, 2013). As a remedy to declining external funding, total education
expenditure as percentage of GDP has stabilized above 6% in recent years as per the analysis on
the following page but it is encouraging to note that it has been above the average of most Sub-
Saharan Africa countries.
1 Donor funding accounts for a quarter of education budgets in some countries (UNESCO, 2014).
2
1.1.3 Household Expenditure
Besides government, households also account for a significant level of education expenditure
whose burden is larger in developing countries. For instance, households are responsible for 32%
of secondary education expenditure in Malawi (UNESCO, 2014) while it can be assumed that
households meet most of the remaining University expenditure in Kenya given the government
only covers 60% (ROK, 2012). According to (Mariara et al., 2007), the monthly cost of sending
a child to primary school is roughly Ksh 90 per child (with a high standard deviation) while
secondary school ranks at Ksh 277 per month with a high standard deviation of Ksh 1200.
1.2 Problem Statement and Objectives
With this in mind, the question raised in this study is whether there are returns to these huge
investments in education. Returns to education can be addressed from both a macroeconomic and
a microeconomic level. At the macro level, returns to education can be computed from the
amount of human capital that is released to the economy. Conversely, at the micro level, returns
to schooling are depicted through wages of individuals whose externalities spread to the whole
population. Education can also be viewed as a consumption good where by returns are in form of
utility to the consumer who in this case is the person bearing the burden of meeting the costs of
education.
This study seeks to estimate returns to schooling in Kenya as the main objective by addressing
the following specific objectives: seeking to estimate the effect of sex on the level of earnings in
6.1 6 6.2 6.3 6.25.3 4.8 5.7 4.3 3.8
28 2623.2
25 26.7
0
5
10
15
20
25
30
Percentage
Years
Education as % of GDP
External financing as % of Education Expenditure
Education as % of GOK Total Expenditure
Enrolment
Primary 5,941,600(2001)-10,182,600(2013)
Secondary 763,245(2001)-2,104,262(2013)
University 203,566(2006)-324,560(2013)
Number of Universities has escalated by 27%
between 2006 and 2013 (24-33)
Total public expenditure on education as a percentage of
GNP was 6% in 2007 above the average for Sub-Saharan
countries in 2011 i.e. 5.0% with a 6.7% share of
Education on GDP in 2011
Figure 1: Domestic and External expenditure on education Table 1: Enrolment
Source: UNESCO, 2014; ROK, 2012; KIPPRA, 2010; and authors’ calculations
3
Kenya which will offer great insights in meeting Millennium Development Goals. In addition
this study will also seek to analyze the effect of experience on private remuneration in Kenya.
2 LITERATURE REVIEW
This section reviews studies in and outside Kenya alongside theories on returns to schooling.
2.1 Theoretical Literature Review
The human capital theory is based on the principle that expected future gains on earnings
determine an individual’s current investment in education. Thus, Investment in education
continues until marginal cost equals marginal return to education (Walker et al., 2001). Equally,
schools are taken as a good platform for acquiring relevant labour market skills at the expense of
foregone earnings from the time spent in schools (Wambugu, 2003). Nevertheless, a detailed
analysis of this theory will be discussed in the next chapter when the Mincer (1974) model is
reviewed.
Besides the human capital theory, signaling theory is also a vital explanation of returns to
schooling. This theory by Michael Spence views schooling level as an indication of someone’s
innate ability which employers consider in hiring labour and not necessarily skills (Frazis, 2002).
Therefore, individuals are compelled to seek education for perception that it creates.
2.2 Empirical Literature Review
Several studies have established that education positively impacts an individual’s future
earnings. For instance Krueger and Lindahl (2000) note that an additional year of schooling
increases earnings by 10% in the US with 13% in Sub-Saharan Africa (UNESCO, 2014). This is
certainly in line with the initial projections of 5% in developed countries and 29 % in developing
countries (Psacharopoulos, 1994)2. However, empirical evidence shows that not all countries
have met this standard. For instance, Oyelere (2010) found that that the average returns to
education in Nigeria was only 2.8 %. Similar Chirwa and Matita (2009) found that an additional
year of schooling increases life time earnings by 10% in Malawi, with more benefits accruing to
women than me, 11.4% and 9.7% respectively.
Studies on returns to schooling in Kenya resonate with most findings from other places. Kimenyi
et al. (2006) while analyzing private returns and human capital externalities in Kenya found that
returns were higher among females than males. This was supported by a study by Mariara and
Mwabu (2007) which concurred that girls had a high education affection ratio than boys using
household survey data. Wambugu (2003) further notes that all levels of education reduce chances
of agricultural employment while higher education levels are skewed to formal employment and
vice versa for lower education levels.
2 Primary education has highest returns in developing countries while tertiary education has high returns in
developed countries (Fasih et al., 2012)
4
This study adds to this pool of knowledge by reviewing returns to schooling with a major
emphasis of sex and experience on remuneration.
3 MODEL
3.1 Conceptual Framework This study formulates the basic earning function based on Mincer (1974) which estimates a semi-
log ordinary square regression with the natural logarithm of earnings as the dependent variable.
Independent variables include years of schooling, potential years of labour market experience
and its square. This can formally be shown in the following equation:
ln 𝑊 = 𝑎 + 𝑏1𝑆 + 𝑏2𝐸 + 𝑏3𝐸2 + 𝜀 (1)
Where ln 𝑊 is the natural logarithm of earnings, S is individual’s years of schooling or different
levels of schooling, E is work experience which is age minus years of schooling, 𝐸2 is square of
work experience, 𝑎, 𝑏1, 𝑏2 𝑎𝑛𝑑 𝑏3 are a constant and coefficients to be estimated respectively
with 𝜀 as the error term. Work experience is squared to capture the declining effects of
experience as an individual ages (Kifle, 2007).
This model can further be modified to include control variables like area of residence, gender
and dummies as follows:
𝑙𝑛(𝑊𝑗) = 𝑎 + ∑ 𝑏𝑘 𝑆𝑗𝑘 + 𝛾𝐴𝑗+∝ 𝑍𝑗 + 𝜀𝑗 (2)
Where W is the hourly earnings for worker j (used hourly wages other than monthly earnings to
avoid labour supply effects on returns), 𝑆𝑘 is a dummy variable for being educated at least up to
level k, A is the experience variable, Z is a vector of j control variables such as gender and area
of residence, α,b, 𝛾 and ∝ are parameters to be estimated, and 𝜀 is the error term.
Using the analysis of Heckman et al. (2003), equation 2 can be derived from Mincer’s
assumption that potential earnings today (at time t i.e.𝐸𝑡) depends on investment in human
capital made yesterday. Hence, an individual invests in human capital a share 𝑘𝑡 of his/her
potential earnings with a return of 𝑟𝑡 in each period t.
𝐸𝑡+1 = 𝐸𝑡 (1 + 𝑟𝑡𝑘𝑡) which with repeated substitution yields 𝐸𝑡 = ∏ (1 + 𝑟𝑗𝑘𝑗)𝐸0𝑡−1𝑗=0 or
basically ln 𝐸𝑡 = ln 𝐸0 + ∑ ln (1 +𝑡−1𝑗=0 𝑟𝑗𝑘𝑗).
This specification postulates that schooling is based on the number of years spent in full-time
investment (𝑘𝑡=1), the return to post-schooling investment in terms of potential earnings is
constant over time, the return to formal schooling in terms of potential earnings is also constant
over time and that formal schooling takes place at the beginning of life. This study objects this
5
last assumption since there is no schooling between the ages of 0 to 6 in Kenya according to the
current regulations unless it is Early Childhood Education which is ages 4-5 (ROK, 2012).
To establish a relationship between potential earnings and years of labor market experience,
Mincer (1974) approximated the Ben Porath (1967) model besides assuming a linearly declining
rate of post-school investment which in turn gives ln 𝑊(𝑠, 𝑥) ≈ 𝑙𝑛𝐸𝑥+𝑠 − 𝐾 (1 −𝑥
𝑇) where 𝑥 =
𝑡 − 𝑠 ≥ 0 is the amount of work experience as at age t. T is the length of working life and it is
assumed to be independent of years of schooling. It is from this that the model in Equation 1 is
found.
To fulfill the objective of finding returns to education, the following equation can be used as
suggested by (Amin et al., 2005);
Rate of Return to Education= 𝑒𝑥𝑝(𝜏𝑖−𝜏ℎ)
𝐷𝑖−𝐷ℎ -1 (3)
With 𝜏𝑖 being the estimated coefficient of a higher level of education; 𝜏ℎ is the estimated
coefficient of a lower level of schooling; 𝐷𝑖 is the total number of years taken to attain a
particular level of higher education while 𝐷ℎ is the total number of years spent schooling at a
lower level of the education system.
Nevertheless, it is significant to note that both years of education and schooling are expected to
contribute to high earnings given that they foster skill acquisition and there by enhance
productivity. Education squared is expected to affect incomes negatively because it is expected
that as one ages their earnings reduce.
3.2 Data
This study uses cross-sectional from the Kenya Integrated Household Budget Survey of
2005/2006 which was collected among 13,430 households (8,610 rural and 4,820 urban) for a
period of 12 months in 70 Districts in Kenya.
Heteroscedasticity (unequal variance) is tested to avoid any wrong inference on the estimated
coefficients (Gujarati, 2004)3.
3.3 Estimation shortcomings
It is important to note that due to data limitation, this study fails to account for omitted variables
and endogeneity of schooling:
i. Omitted variables; These basically include unobservable characteristics such as ability and
family background whose omission can give biased OLS estimates (Kimenyi et al., 2006).
Several studies have diversified and included additional variables in their analysis of returns
to education which has given better results. For example, Wambugu (2003) included parent’s
3 Heteroscedasticity is prevalent in cross-sectional data (Gujarati, 2004).
6
education and income to depict family background and found out that the two variables had a
positive impact on wages and that returns to education decline when family background
variables are included in the earnings regressions.
ii. Endogeneity of schooling; By treating schooling as exogenous implies that investment in
schooling is independent of earnings which should be the contrary. Accounting for
endogeneity in schooling means that an individual takes into account expected earnings in
making the decision to invest in education (Kimenyi et al., 2006).Two remedies can be
suggested i.e. constructing a ‘selectivity-correction’ term from a schooling attainment
equation and then including the correction term in the earnings equation or using exogenous
variation in educational attainment (such as differences in educational attainment across
siblings) to provide instrumental variable (IV) estimates of returns to education (Kimenyi et
al, 2006).
7
4 DATA ANALYSIS
This section contains regression results as per equation (2) and a test for heterogeneity.
4.1 Regression results and heteroscedasticity test4
Running a regression on equation (2) gives the following results: Where natural logarithm of
earnings is the dependent variable with years of education, working experience (age minus 6
minus years of schooling), working experience squared and, sex (where 1=female, 0=male) are
the respective independent variables.
Log wage coefficient std. error t-ratio P>|t| [95% Conf. Interval]
Years of
education
0.0427027 0.0368713 1.16 0.247 -0.0296867 0.1150921
Experience 0.0057483 0.0054169 1.06 0.289 -0.0048868 0.0163834
Experience
squared
-0.0000434 0.0001001 -0.43 0.665 -0.00024 0.0001533
Sex -0.2940651 0.1034785 -2.84 0.005 -0.4972245 -0.0909056
Constant 4.75398 0.2253118 21.10 0.000 4.311625 5.196335
Number of observations=712 F(4,712)=2.89 R-squared=0.0160 Adj R-squared=0.0105
Breusch-Pagan test for heteroscedasticity
Null hypothesis: heteroscedasticity not present
Test statistic: LM = 13.1061,
with p-value = P(Chi-square(4) > 13.1061) = 0.0107688
Critical Chi-square value (at 1%)= 13.2767
t-critical (n=712)=1.96
F-critical (4,707,0.05)= 2.38453
4.2 Discussion of results
From the Breusch-Pagan test results in Table 2, the calculated value is lower than the critical
value which indicates that we accept our null hypothesis and conclude that homoscedasticity
exists. This means that we can credibly make inferences from the regression results.
The regression results linked to equation (2) yields the following results:
log 𝑤𝑎𝑔𝑒 = 4.75398 + 0.0427027𝑌𝑒𝑎𝑟𝑠 𝑜𝑓 𝑒𝑑 + 0.0057483 Experience −
0.0000434 Experience Squared − 0.2940651𝑆𝑒𝑥 + 𝜀 (4)
With t-statistic values ranging from 21.10, 1.16, 1.06, -0.43 and -2.84 respectively. Therefore, it
can be noted that:
Both experience and years of schooling boost earnings by 0.6% and 4.27% respectively for an
additional year of schooling and an additional year of experience. Hence, workers pursue higher
4 The Breusch- Pagan test was used to detect heteroscedasticity. It follows a Chi-Square distribution where the
calculated and critical values are compared. If the calculated value is greater than the critical value, we reject the
null hypothesis and conclude that there is heteroscedasticity. Homoscedasticity only exists where the calculated
value is less than the ctitical value (Gujarati, 2004).
Table 2: Regression results & Breusch-Pagan test results
8
incomes when they have been educated and at the same time accumulate skills which results in
improved output leading to higher remuneration. However, they are both insignificant at 5%
level because there t-calculated values are below the t-critical value of 1.96. Sex is a significant
determinant of earnings with female workers earning less by about 30%. This could be attributed
to the notion that women are still perceived in a low status because studies such as (Kimenyi et
al., 2006) used a male dummy and found out a positive relationship.
The F test shows that our model is significant at 5% level because the calculated value is greater
than the critical value. There is need to include more explanatory variables in the model because
the current explanatory variables only explain 1.6% of variations in wages according to the R-
squared value.
9
5 SUMMARY AND CONCLUSION
From the analysis of returns to schooling depicted through earnings, this study has shown the
importance of considering sex in determining levels of earning. Even though the female
coefficient was negative, indicating declined earnings, there is still a need to promote education
of females. This will not only promote gender equality but also female empowerment. The
positive impact of years of schooling and experience indicates the importance of both in
fostering personal development.
However, it can generally be noted that perhaps better results5 would have been obtained had the
study solved the two estimation problems, omission of variables and endogeneity of schooling.
5.1 Areas for further research There is need to expound on this topic of returns to education by considering a rural set-up of a
developing country where in most cases land or cattle6 is sold to finance education. Two critical
points arise from this: (1) whether this inter-temporal investment is worthwhile because the
recipients of this education work towards buying inter alia land or cattle that was used to fund
their studies in future. (2) The first puzzle begs the question whether there is need to seek
alternative ways of financing education in a rural set-up other than merely relying on the stated
assets.
5 As seen, the R-Squared value is too low, only 1.6%. 6 Other household assets are also sold.
10
REFRENCES
Amin, A.A., & Awung, J.W. (2005). Economic Analysis of Private Returns to Investment in
Education in Cameroon. African Institute for Economic Development and Planning, 1-16.
Chirwa, W.E., & Matita. M.M (2009). The Rate of Return on Education in Malawi. Department
of Economics University of Malawi, Working Paper No. 2009/01.
Frazis, H. (2002). Human Capital, Signaling, and the Pattern of Returns to Education. Oxford
Economic Papers, Vol. 54, No. 2, 298-309.
Gujarati, D.N. (2004).Basic Econometrics (4th Edition). New York, USA: The McGraw-Hill
Companies.
Kifle, T. (2007). The Private Rate of Return to Schooling: Evidence from Eritrea. Essays in
Education, Vol. 21, 83.
Kimenyi, S.M, Mwabu, G., & Manda, D.K. (2006). Human Capital Externalities and Private
Returns to Education in Kenya. Eastern Economic Journal, Vol. 32, No. 3, 493-513.
Kenya Institute of Public Policy Research and Analysis. (2010). Kenya Economic Report 2010:
Enhancing Sectoral Contribution towards reducing poverty, unemployment and inequality in
Kenya. Nairobi, Kenya: KIPPRA.
Krueger, A.B., & Lindahl, M. (2001). Education for Growth: Why and For Whom? Journal of
Economic Literature, Vol. 39, No.4, 1101–1136.
Mariara, J.K., & Mwabu, D.K. (2007). Determinants of School Enrolment and Education
attainment: Empirical evidence from Kenya. South African Journal of Economics, Vol. 75, No.3,
572-593.
Mincer, J. (1974). Schooling, Experience and Earnings. New York, National Bureau of
Economic Research.
Republic of Kenya. (2012). Towards a globally competitive quality education for sustainable
development. Government Printer.
Republic of Kenya. (2013). Vision 2030 Second Medium Term Plan, 2013 – 2017. Government
Printer.
Orodho, J.O., Waweru, P.N, Ndichu, M., & Nthinguri, R. (2013). Basic Education in Kenya:
Focus on Strategies Applied to Cope with School-based Challenges Inhibiting Effective
Implementation of Curriculum. International Journal of Education and Research, Vol. 1, No.11,
1-20.
11
Oyelere, R. U. (2010). Africa's education enigma? The Nigerian story. Journal of Development
Economics, Vol. 91, No.1, 128-139.
Psacharopoulos, G. (1994). Returns to Investments in Education: A Global Update. World
Development, Vol. 22, No. 9, 1325–1343.
UNESCO. (2014). Teaching and Learning: Achieving quality for all (First edition). Paris,
France, United Nations Educational, Scientific and Cultural Organization.
Wambugu, A. (2003). Essays on earnings and human capital in Kenya (PhD Thesis). Goteborg
University, Gothenburg, Sweden.
Youssef, C. (2005). World Bank priorities in education lending: An ever-changing
endeavor. Paterson Review, Vol. 6, 1-29.
12
APPENDIX: SUMMARY STATISTICS
Variable Mean Median Minimum Maximum
log wage 4.90451 4.60517 0.000000 10.2036
Experience 10.5270 6.00000 -21.0000 90.0000
Experience
Squared
461.464 81.0000 0.000000 8100.00
age 22.1716 17.0000 0.000000 99.0000
sex 0.493338 0.000000 0.000000 1.00000
years_ed 5.66071 5.00000 2.00000 15.0000
Variable Std. Dev. C.V. Skewness Ex. Kurtosis
Log wage 1.36929 0.279190 1.12295 2.42045
Experience 18.7257 1.77882 1.20886 1.28860
Experience
Squared
945.528 2.04898 3.65492 16.8101
age 18.5964 0.838748 1.21117 1.27432
sex 0.499959 1.01342 0.0266490 -1.99929
years_ed 1.38857 0.245299 2.48990 10.9230