EDUCATION AND ECONOMIC GROWTH OF

ETHIOPIA

(1975-2014)

Dire Dawa University

College of Business and Economics Department of
economics

A senior Essay Submitted to Department of Economics for Requirement in

Partial Fulfillment of Bachelor Degree of Arts in Economics

BY: TAHIR TUNE

ADVISOR: MR. FAHM ABAS

Dire Dawa, Ethiopia

JUNE 2017
Table of Contents
Acknowledgement

Acronyms and Abbreviations

Lists of tables

Abstract.................................................................................................................................................;…….IV

CHAPTER ONE………………………………………………………………………………………………………………………………………….1
INTRODUCTION...........................................................................................................................................2
1.1 Background of the study.................................................................................................................2
1.2Statement of the problem..................................................................................................................3
1.3Objective of the Studies..................................................................................................................4
1.3.1 General Objective.......................................................................................................................4
1.3.2Specific Objective........................................................................................................................4
1.4Significance of the Study....................................................................................................................4
1.5Scope of the Study..............................................................................................................................4
1.6Limitation of the Study.......................................................................................................................5
1.7Organization of the Study...................................................................................................................5
CHAPTER TWO………………………………………..……………………………………………………………………………………………….6
LITERATUREE REVIEW..................................................................................................................................6
2.1Theoretical Literature Review.............................................................................................................6
2.1.1Theories on Determinants of Growth..........................................................................................6
2.1.2. Solow Model..............................................................................................................................7
2.1.3. Endogenous Growth Models.....................................................................................................8
2.1.4. Neo-classical Theories................................................................................................................9
2.2. Empirical Literature Review........................................................................................................10
CHAPTER THREE...................................................................................................................................14
Methodologies of the Study......................................................................................................................14
3.1 Data Source and Type......................................................................................................................14
3.2 Data Analysis Method....................................................................................................................14

2
 3.3 MODEL SPESIFICATION AND METHODOLOGY...............................................................................14
3.4 MODEL SPESIFICATION..................................................................................................................15
3.5 Definition of Variables and expected sign of coefficients...............................................................16
3.5.1. Explanatory Variable................................................................................................................16
3.5.2. Expected Sign of Coefficients...................................................................................................16
3.6. Tests Used......................................................................................................................................16
3.6.1.Testing for unit roots................................................................................................................16
3.6.2. Co- integration analysis...............................................................................................................17
3.6.3. Error Correction Model................................................................................................................17
3.7. Post estimation tests.....................................................................................................................17
3.7.1.Hetroskedasticity..........................................................................................................................17
3.7.2. Testing for normality...............................................................................................................17
3.7.3.Autocorrelation........................................................................................................................18
3.7.4. Multicollinearity.......................................................................................................................18
CHAPTER FOUR.....................................................................................................................................20
DISCUSION AND RESULTS........................................................................................................................20
4.1. Descriptive statistics and distributional issues................................................................................20
4.2.1. Test for stationarity: unit root test..............................................................................................22
Table 4.2 DF unit root test.................................................................................................................23
4.2.2. Co-integration Test..................................................................................................................23
4.2.3. The Long Run Model................................................................................................................24
Table 4.4 Long Run Model Results.....................................................................................................24
4.2.4. THE SHORT RUN MODEL..............................................................................................................27
Table 4.5 short run model result...........................................................................................................28
Conclusion and Recommendation.............................................................................................................30
5.1. Conclusion..................................................................................................................................30
5.2 Recommendation............................................................................................................................30
References.................................................................................................................................................32

3
Acknowledgement

First I would like to thank ALLAH for giving me the strength and capacity to reach this stage. I
am also thanks my essay advisor FAHMI ABAS(msc) for his professional advises and
constructive comments and valuable guidance which have a positive contribution to the final
stage of this paper. Moreover I owe my deeper thanks to my father and my mother for their
endless support throughout my stay in the university

Finally, I also like to thank all my friends who I would not forget in my life Without the help of
them , I would face many difficult while doing this paper.

4
 List of Acronyms and Abbreviations

CSA Central Statistics Agency

DF Dickey Fuller

ECM Error Correction Model

EEHE /GDP Education Expenditure plus Health Expenditure as a ratio

to GDP

GDP Gross Domestic Product

K Gross Capital Formation

L Labor Force

MoE Ministry of Education

MoFEC Ministry of Finance and Economic Cooperation

OLS Ordinary List Square

5
 ABSTRACT
The study is meant to fill the gap on the literature of education and economic growth in
Ethiopia.This study investigates the impact of education on economic growth in Ethiopia
over the period 1975-2014 using an OLS methodology. The study is conducted by entering
education expenditure and health expenditure as proxy of education apart from gross
capital formation and labor force in the growth model. The empirical result of the study
reveals that education expenditure and health expenditure are statistically significant in
the long run but they are not statistically significant in the short run.

6
INTRODUCTION

7
1. Background of the study
Education defined as broadly as "all deliberate learning activities are usually used as an
approximation to human capital. It is believed to create a productive by in calculating important
and useful knowledge in to the minds of people thereby speeding up economic development. In
other words education transforms the raw human being in to production human capital by
installing the skill required by both the traditional sector and modern sector of the economy, and
makes the individuals more productive not only in the market place but also in the household."
(Tilak, 1999)

Education satisfies a basic human need for knowledge, adds greater sense of participation, and
provides a means meeting other basic needs, helps cultural and civil dimensions to an individual
existence. That is why most of the studies in the area of human capital use formal education
(usually year of schooling ) and informal education (age experience) to analyze the contribution
of human capital to growth unlike this study which use macro study to analyze the contribution
of education to economic growth.

Ethiopia is one of the least developed countries in the world. Its economy rebates mainly on
agriculture, which contribute about 59% of GDP, over 75% of total exports and 85% of
employment. In Ethiopia the education sector has not been performing well in the past decades,
the declining quality of education has become a serious problem in Ethiopia, illiteracy rate is
higher,and nearly 57.2% of men and 42.8% of women are illiterate MoE (2014). . This implies
that the majority of the population has not access to education due to culture, economic and
social reasons.

Regarding economic growth, for many years’ people had a concept that educated and skilled
labor force is necessary condition of sustained economic growth. The contribution of education
to economic growth is by creating more productive labor force and endowing it with increased
knowledge and skill, providing wide spread employment income earning opportunities for
teachers, school construction workers and others ; creating a class of educated leaders to feel the
class of others, promoting literacy and creating modernity. (Todaro, 2006)

1.2 .Statement of the problem

Education is the most important element for a country either developing or developed countries,
it is true that one cannot get a country developed without better education. Because it’s through
8
education that a research, which is a basic for any type of development, can be conducted this
seems the fact that almost all countries in the world give greater attention for it. An important
evidence for that is education being made one primary objectives that developing countries are
expected to achieve in the millennium development goals.

The development of skilled labor force makes an important contribution to economic growth. An
educated, trained and more productive work force contributes to a greater economic growth.
Until recently the experience of East Asian countries confirmed the importance of education and
human capital formation. This was most apparent in fast growing Asian economies. It may
therefore be argued that investment in education pays off, investment in human capital may play
a more persistent role in the growth process (Wubshet, 2006).

Tekola(2007)studied the contribution of primary education to economic growth in Ethiopia .

The method he usedto analyze the data is econometrics approach using time series data for the
period 1971-2007. He found that the long-run static regression reveal that the primary level of
education center in explaining output. The contribution of primary education is larger compared
to short-run dynamic regression show that growth of primary enters positively and significant in
explaining economic growth. Moubet (2006) also investigated the impact of education on
economic growth in Ethiopia through the application of error correction model and on her
finding she state that the average level of education has significant impact on the evolution of
total level output.

Education has been major force taken in accelerating the process of economic growth in the
country; it has yet generated sufficient inertia to spread the nation’s development contribution on
the area of education on national economic growth. Even though the varieties of studies have
been conducted in the studyarea,theyanalyzed the education and economic growth without
considering human capital. This paper focuses on human capital as determinant of economic
growth by including education expenditure and health expenditure as proxy of human capital

The whole literature on education and economic growth has been ultimately concerned with the
specific problem of measuring the effect of enhanced educational inputs upon economic outputs.
But this must be seen with a broader evolutionary and conceptual perspective, which attempts to
examine the problematic relationship between education and economic growth in the widest
sense.

9
1.3 Objective of the Studies

1.3.1 General Objective

The main objective of the study is to show the impact of education on Ethiopia’s economic
growth.

1.3.2Specific Objective
The study has the fallowing specific objectives.

(i) To examine contribution of education to economic growth.

(ii) To discuss educational policies of Ethiopia.

1.4Significance of the Study

The study tried to provide comprehensive evidence on contribution of education to economic
growth in Ethiopia during period of 1975-2014. Even if the study area is in debate this study
serve as a bench mark for policy makers, individuals, institutions and government. It also serves
the researcher as a source among other sources and it may also initiate others for further and
detailed studies in the area. Besides, the theoretical argument is usually based on empirical
evidences which study the direct and indirect impacts of education on economic growth.

1.5Scope of the Study

The study covers from 1975-2014 years of data recorded by Ministry of Finance and Economic
Cooperation (MoFEC), Ministry of Education (MoE) and Central Statistical Agency (CSA).

1.6 Limitation of the Study

The main limitation of the study is availability of sufficient data, accurate records and
time.,despitethe fact that, maximum efforts would be given to get relevant information

1.7 Organization of the Study

The remainder of the thesis is organized as follows. Chapter two reviews both the theoretical and
empirical studies related to the topic. Chapter three discusses the model specification, data types
and sources, and also estimation techniques. The fourth chapter includes econometric result of
the data collected and the last chapter contains conclusion and recommendation.

10
 CHAPTER TWO:

LITERATUREE REVIEW
2.1Theoretical Literature Review
2.1.1Theories on Determinants of Growth
]The relationship between economic growth and education has been one of central threads of
economic analysis. Both Adam Smith in the 18th century and Alfred Marshal in the 19th century,

11
two important figures for the economics profession, addressed the question of how individual
investment in “education” influences the wealth of nations. Modern professional economists
have been attempting to develop empirical estimates of the relationship between education and
economic growth. Some of the most famous names in the late 20 th century economic made their
populations studding the question of individual returns to investment in education Jacob Mincer
(1974), Gary Becker (1964) and a long list of researchers inspired by the work have produced
hundreds of books and papers. Much of this literature is highly technical in a sense that it uses
formal econometric models to test hypothesis using empirical data.

Sub-Saharan countries are struggling to establish the basic socio-economic foundation. From
European Union in the case of Eastern Europe the highest returns may be at the secondary and
tertiary levels of education, while in less developed nation there may be a high pay off-not just in
income-from primary schooling. Indeed there are considerable evidences from the development
literature. Regarding the importance of increasing girl’s participation rates in primary education
that underscores how differences in stages of economic and social development can alter the
impact of schooling (UNESCO, Education for all, 2005).

According to kifle(2006) over the last decades the issue of economic growth has attracted
increasing attention in both theoretical and empirical research. Despite the lack of a unifying
theory, there are several partial theories that discuss the role of various factors in determining
economic performance and growth. Two mainstream strands can be distinguished: the
neoclassical, formalized by Solow (1970), which emphasizes the importance of capital
accumulation and, the more recent, theory of endogenous growth, pioneered by Habib (1986,
1990) and Lucas (1988), which has drawn attention to human capital and innovation capacity.

2.1.2. Solow Model

Robert Solow and Trevor Swan developed a growth model in 1956 that can help to think about
the approximate causes and the mechanics of the process of economic growth and cross-country
income differences. The model is called Solow-Swan model or simply Solow model. This model

12
has shaped the way we approach not only economic growth but the entire field of
macroeconomics

The Solow model focuses on four variables: output(Y), capital (K), labor (L), and “knowledge”
or “the effectiveness of labor” (A). At any time the economy has some amount of capital, labor
and knowledge, and these are combined to produce output.The production function takes the
following form:

Y (t) = f (k (t),A(t),L(t)),

Where, t denotes time.

The features of the production should be noted. First, time does not enter in the production
function directly, but only through K, L, A. that is output changes over time only if the input into
production change. In particular, the amount of output obtained from given quantities of capital
and labor rises overtime there is technological progress only if the amount of knowledge
increases. (Romer,, 1996).

Second, A and L enter multiplicatively. AL is referring to as effective labor, and technological

progress that enters in this fashion is known as labor augmenting or Harrod-neutral. This way of
specifying how A enters together with other assumptions of the model, will imply that the ratio
of capital to output, K/Y, eventually settled down. In practice capital output ratio do not show
any clear upward or down ward trend over extended periods. In addition to building the model so
that the ratio is eventually constant make the analysis much simpler. Assuming that A multiplies
L therefore it is very convenient. The central assumption of Solow concerns the properties of the
production function and the evolution of the three inputs into production (capital, labor and
knowledge) overtime.

More fundamentally, the model does not identify what the “effectiveness of labor” is; it is just a
catch all for factors other than labor and capital that effect output .to precede, we must take a
stand concerning what we mean by the effectiveness of labor and what cause it to vary. One
neutral possibility is that the effectiveness of labor corresponds to abstract knowledge. To
understand worldwide growth it would then be necessary to analyze the determinant of the stock
of knowledge overtime. To understand cross country difference in real income, one would have

13
to explain why firms in some countries have access to more knowledge than firms in other
countries, and why the greater knowledge is not rapidly transferred to poorer countries.

The principal result of this model is a negative oneif earnings reflects its contribution to output
and if its share in total income is the modes, then capital accumulation cannot account for a large
part of either long run growth or cross-country income differences .and the only determinant of
income in the model other than capital is a mystery variable, the “effectiveness of labor” (A),
whose exact meaning is not specified and whose behavior is taken as exogenous.

2.1.3.Endogenous Growth Models

The endogenous growth models developed by Lucas–Romer challenged the old neoclassical
model by emphasizing the role of endogenous factors (i.e., human capital stock and R&D
activities) as the main engines of economic growth. Whereas early neoclassical models assumed
total factor productivity growth (or technical progress) as exogenously given, the newer
endogenous growth models attributed this component of growth to the “learning by doing” effect
that occurs between physical and human capital, which results in increasing returns to scale in
production technology (Lucas, 1988).

The most distinctive difference between neoclassical exogenous and endogenous growth theories
is that the former assumes constant returns to scale, whereas the latter generally assumes
increasing returns to scale. Making the assumption of increasing returns to scale provides a
possible path to long-run sustained growth in endogenous growth theories.

This theory investigates the fundamental questions of growth theory more deeply. It considers
two broad views. The first view is that the driving force of growth is the accumulation of
knowledge. This view agrees with the Solow model that capital accumulation is not central to
growth. But it differs from thus models in explicitly interpreting the effectiveness of labor as
knowledge and in formally modeling its evaluation overtime. The second view is that, contrary
to the Solow model, capital is central to growth. Specifically, we will consider models that take a
broader view of capital that we have considered so far most importantly extending it to include
human capital.

In the mid 1980‘s, a new paradigm was developed in the literature, mostly due to Hagos (1986).
Which is known commonly endogenous growth model discard neo-classical assumption of

14
diminishing return to capital in this model capital is broadly defined as human and physical
capital embodied in material objects. The main argument of the new growth model for the
existence of increasing return is due to deliberate innovation which is fostered by allocation of
resources in research and development (Ray, 1998).

In the endogenous growth approach human capital affects total factor productivity in two
channels. First, higher level of human capital directly influences productivity via its impact on
domestic innovation. Secondly, higher level of human capital causes improvements in total
factor productivity by facilitating the adoption and the implementation of foreign technology.
The factor leading to endogenous growth (in particular technology change) is explicitly related to
the stock of human capital. Technology and human capital are both “endogenous” to the system
(Tallman and Wanga , 1990) .

2.1.4. Neo-classical Theories

The application of human capital concept to economic growth and to labor economics was
initially pioneered independently by Jacob Mincer (1958), Theoder W. Schultz (1961) and Garry
S,Becker(1961). At microeconomics level, Mincer(1958) has shown that difference in individual
human capital stocks and their growth can explain much of the observed variation in the wage
structure and in the personal distribution of income. Jacob Mincer developed a model for
examining the nature and cause of inequality in personal incomes. According to Mincer (1981),
the microeconomic analysis of investment in human capital is the underpinning of our
understanding of the contribution of human capital to the aggregate level of income and it’s the
rate of growth.

At macro level education yields return to the society that must be reflected at the level of
economy. Schultz (1961) and Denison (1962) using computationally different although
conceptually similar approaches introduced the quality of labor or human capital in to traditional
production function. They showed that education contributes directly to the growth of national
income by improving the skills and productive capacities of the labor force. In this theory,
Schultz (1961) used macroeconomic approach; the emphasis however placed on the former.
While Schultz predominantly asserted the prime relationship of education to human capital
formation, the breadth of intellectual understanding he brought to the field encompassed many
other type of human capital investment.

15
2.2. Empirical Literature Review
The prove role of education in economic growth and entail that certain proportion of
social wealthy must be allotted for education country fail to do is deemed to remain under
developed. To build up a base of easy activities and enter more demanding activities and enter
more demanding activities call for increasing level and technical specialization on education. To
low level of human resource relevant to industry in Africa including Ethiopia suggests why the
region presents a general picture of poor technological masters and a dynamic in industry. The
reason was whose Africa countries with relatively high literacy rates and secondary schooling
enrollment ratio are those with the best industries. Records such as Kenya, Maturities and
Zimbabwe .

When emphasizing the importance of enrollment and education attainment an initial

condition for growth for East Asians said “once the initial level of enrollment miraculous about
the high performance Asian economies” growth with experience (Roderick 1994) Education is
lived to create production citizen by including and use full knowledge into minds of people there
by spending up economic development. Education demanded as a producer good because it is
expected to enhance the future productivity of an educated investment (Schultz, 1985).

According to world bank estimates the proportion of growth national product devoted to
education in developing countries rose on the average of 2.7 percentage in 1980’s to 5.2 percent
in 1990.

Denison (1977) estimates using data for 1970-1989 and 1990-1999, he found that 48
percent and 50 percent rate respectively in the per capital income due to education in America ,
and the same writer 1946- 1973duringthis21 percent of growth due to the increasement of
education of the labor force in 1979.

Based on a panel data of 19 sub Saharan countries for a year 1982-2000, this study explores the
determination of economic growth in the region. The study result indicates that physical capital
formation, a vibrant export sector and human capital formation significantly contribute to the
economic growth among sub-Saharan countries. However, government expenditure, nominal
discount rate, and foreign aid significantly lead to negative economic growth. (Ndambiri, 2010)

16
 PASCHAROPOULOS (1984) estimates for 21 developing counters shows that 12 to 23
percent in economic growth can be attributed to education and that education explained the
economic growth from 0.8 percent in Mexico to 25 percent in Canada.Tilak(1988) using 29
country data and analyzing 66 less developed country, He found that a significant negative
correlation between reduction in poverty and an attainment of higher enrollment of education.

In most economists conclude that investment in education contributes positively and

significantly to the development of the economy both in developed and developing
countries ,particularly tertiary education had a significant positive growth effect. An increase in
education of 0.09 years raises annual growth by 0.5 percent a year. An interaction between initial
GDP and human resources, so that countries of the that large behind tend to grow faster if they
have high level of human resources [H.patrinos, 2002]

Lustig (2006) examined a direct relationship between health and growth in Mexico uses 1970-
1995 data and uses life expectancy and mortality rates for different age groups as health
indicators. He observed that health is responsible for approximately one-third of long term
economic growth. He considered health to be an asset with an intrinsic value as well as
instrumental value. According to him, good health is a source of wellbeing and highly valued
throughout the world.

Musibau and Rasak (2005) carried out a study on the long-run relationship between education
and economic growth in Nigeria using the Johansen’s Co-integration techniques. It was
discovered that there is a long-run relationship between education and economic growth in
Nigeria. They also found out that a long-run effect of a 1 percent increase of average years of
schooling on output per worker while keeping the other variable constant is approximately 0.86
percent while the long-run elasticity of capital is 0.139 percent.

Bakare (2006) investigated the growth implications of human capital investment in Nigeria using
the vector autoregressive error corrections mechanism. The study revealed that there is a
significant functional and institutional relationship between the investments in human capital and
economic growth in Nigeria. It was revealed that 1% fall in human capital investment led to a
48.1% fall in the rate of growth in gross domestic output between 1970 and 2000. Dauda (2010)

17
examined investment in education and economic growth in Nigeria using annual time series data
from 1977 to 2007. The paper employs Johansen co integration technique and error correction
methodology. Empirical results indicate that there is a long-run relationship between investment
in education and economic growth. The main variable of interest, the growth rate of educational
expenditure had positive and significant effect on economic growth in Nigeria. The result
indicates that educational investment plays a crucial role in developing an economy and it
enhances growth in the nation’s income. The coefficient of growth of gross fixed capital
formation has positive and statistically significant effect on the Nigerian economy.

Adelakun (2011) conducted a study in Nigeria on human capital development and economic
growth using OLS technique. It evaluates human capital using the GDP as proxy for economic
growth; total government expenditure on education and health, and the enrolment pattern of
tertiary, secondary and primary schools as proxy for human capital. He concluded that there is a
positive relationship between government expenditure on education and health as well as pattern
of enrolment in primary, secondary, and tertiary institutions in enhancing economic growth in
the long run.

In Ethiopia, little researcher have been done on measuring the relationship between education
and economic growth. Some of these are; Netsanet (2012) (cited in Azmeraw, 2011) has
established the facts of Ethiopia. He used error correction model to look in to short run and co-
integration for long run dynamics to explain the impact of education on economic growth, he
used enrolment ratio at different level. And he found out that both in the short run and long run
education enters positively and significantly in explaining output. The contribution of primary
education is found to be relatively large when compared with higher levels of education in both
cases.

Mandfrew (2000) studied the contribution of primary education to economic growth in Ethiopia.
The methods employed to analyze the data are economic approach, using cross sectional data for
the period 1971-2005. He found that the long run static regression reveal that the primary level
of education enters positively in explaining output. The contribution of primary education is
larger compared to short run dynamic regression and show that positively and significantly in
explaining economic growth.

18
Abela(2004) investigated the impact of education on economic growth in Ethiopia through the
application of error correction model and on his finding he stated ‘’the average level of human
capitals appears to have no significant impact on the evolution of total level of output. Therefore
one reason low performance of the economy in terms of output growth may be attributed to
Ethiopia’s poorly developed human capital base lagging far behind Sub-Saharan African
average.”

CHAPTER THREE

Methodologies of the Study

3.1 Data Source and Type
All data used in this paper span from secondary sources. Those data were taken from journals,
books, annual reports, and bulletin of different organization mainly from Ministry of Education
(MoE), Ministry of Finance and Economic Cooperation (MoFEC), and Central Static Authority
(CSA).

19
3.2 Data Analysis Method
This paper uses quantitative econometric analysis. So it uses time series regression to find the
impact of education on economic growth. Since the approach is to investigate how education
affects growth of an economy, education expenditure and health expenditure are taken as proxy
of human capital

3.3 MODEL SPESIFICATION AND METHODOLOGY

This research is based on secondary data obtained from National Bank of Ethiopia (NBE) for
gross capital formation variable, Central Statistical Agency (CSA) for labor force variable and
Ministry of Finance and Economic Cooperation (MoFEC) for education expenditure and health
expenditure variable. Econometric analysis uses an ordinary least square method (OLS), to
investigate the relationship between education and economic growth. The study is based on time
series annual data for 40 years (1975-2014).

3.4 MODELSPESIFICATION
The model is specified with the objective of measuring the impact of education on economic
growth by using the augmented Solow model. The Augmented Solow Model extends the basic
framework to allow human capital as an extra input to inter to production function. As it stands,
the neoclassical growth model, relaying as it does on differences in capital-labor ratio across
countries to explain the wide disparities in level of percapita output, can’t satisfactorily explain
world income differentials

. In response to this deficiency Mankiw at al. (1992) “augment” the Solow model by including
the accumulation of human capital as well as physical capital. It is represented in a Cobb-
Douglas production function with constant return to scale:
α β β
Y =K t HK t ( A t Lt ) … … … … … … … … … … … … … … … … … … … … … … … … … .(1)

To estimate the parameters the function is transformed to logarithm:

lnY =α ln K t + β ln HK t + β ln At … … … … … … … … … … … … … … … … … … … .(2)

For the purpose of this study Education expenditure plus Health expenditure as a ratio of GDP is
used as a proxy of human capital. Health expenditure is used because in poor countries,

20
investment in health is major contributor to human capital. And it is called cost based approach
of studying human capital. This method also takes all costs of forming human capital into
account wheret almost every aspect of human capital has to be calculated separately (education,
finance, food, health, etc.)

The variables of this study are specified as follows:

RGDP= f(K,L,EEHE/GDP)

Where RGDP- Real Gross Domestic Product

K- Gross capi

L- Labor force

EEHE/GDP- education expenditure plus health expenditure as ratio of GDP

To estimate parameters the model can be specified in log=linear formed by taking the natural
logarithm of GDP and the other variables taken with their respected values. The other rationale
for this specification is the log=linear form of specification enables the researcher to interpret the
coefficient of the dependent variable directly as percent with respect to the independent
variables. In addition it is also useful for accommodating the hetroskedasticity problem. It is also
compress extreme values .To this end the model is specified as follows:

EEHE
lnRGDP=β 0+ β1 K + β 2 L+ β 3 +U t --------------------------------------------------------------- (3)
GDP

Where, Ut- error term

3.5 Definition of Variables and expected sign of coefficients

3.5.1. Explanatory Variable
Independent variables are factor of production such as labor, capital; education expenditure plus
health expenditure as a ratio of real gross domestic product are included in factor input.

3.5.2. Expected Sign of Coefficients

Theindependent variables included in the model are expected to have impact on economic
growth: education expenditure plus health expenditure as a ratio of RGDP are expected to have
crucial impact on economic growth, and impact of education on economic growth can be

21
positive. As a result, we expect positive sign for variables gross capital formation, labor force
and education expenditure plus health expenditure as a ratio of real gross domestic product.

3.6. Tests Used

3.6.1.Testing for unit roots
Most time series data is non-stationary at level. So the first and important step in time series
analysis is to checkfor unit root. There are different methods for unit test; this study uses Dickey
Fuller (DF) test because it account for autocorrelation between error terms.

If a time series differenced once and the difference series is stationary, we say that the original
(random walk) series is integrated of order I denoted by I (1). In general, if a time series has to be
differenced (d) times, it is integrated of order d or d (I). Thus, any time we have an integrated
time series of order1 or greater, we have a non stationary time series. By convention if d=0, the
resulting I(0) process represents a stationary time series .

3.6.2. Co- integration analysis

Regressing one non- stationary time series leads to spurious result. There may be a possibility of
solving level non- stationary problem by regressing the first or higher order difference, if their
difference is found to be stationary. However, in taking the first (or higher order) difference we
may lose valuable long term relationship between time series variable that is given by the levels
(as against the first difference) of the variables.

3.6.3. Error Correction Model

Although there is a long term equilibrium relationship between dependent and independent
variables i.e. (co integration), in the short run there may be disequilibrium. The error correction
model is employed to correct for disequilibrium and determine the short run relationship between
variables. The analysis of short run dynamics is often done by first eliminating trend in variable
usually by differencing. The dynamic short run equilibrium is obtained by regressing the first
difference of dependent variable with the first difference of explanatory variable and one period
lagged error term i.e. ECM model is incorporates the equilibrating error (ECM t-1)to capture the
adjustment of toward the long run equilibrium.

22
3.7. Post estimation tests

3.7.1.Hetroskedasticity
An important assumption of classical linear regression model is that the disturbance term, ui
appearing in the population regression function is homoskedastici.e.they all have the same
variance. But if there exist an outlying observation in relation to the observation in the sample
the assumption of constant variance is violated and this violation is referred to as
hetroskedasticity.

The problem would lead to the least square estimator to be inefficient and the estimate of the
variance also biased which invalidate the test of significance. Breusch-Pagan or Weisberg test for
hetroskedasticiy is used with the null hypothesis of constant variance and it is possible to reject
this hypothesis when the p-value is greater than the 5% significance level.

3.7.2. Testing for normality

Normality test is used to determine if the data is well modeled by a normal distribution and to
compute how likely it is random variable is underlying the data is set to be normally distributed.
Normality of this model data is tested using graphical assessment. In descriptive statistics terms,
one measure of goodness of fit of normal model of data. If the fit is poor then the data are not
well modeled in that respect by a normal distribution, without making a judgment on any
underlining variables (madala, 1992).

3.7.3.Autocorrelation
The correlation between the residual is referred as autocorrelation which is included by the
transformation of original data and manipulation of the data through interpolation and
extrapolation. The simple and most widely used model is one where the error term U tandECMt-1
have a correlation p for this model one can think of testing hypothesis about p on the basis of
estimated correlation coefficient between the residual. A commonly used statistics for this
purpose is the Durbin-Watson (DW) statistics which is denoted by d. when the DW statistics is
zero (d=0) the estimated correlation coefficient is 1 and d=4 when the correlation coefficient
which is estimated is -1. If d is closed to 0 or 4, then the residual are highly correlated the
k2standard d statistics that serves as rule of thumb is that d=2 which indicate that the estimated
correlation confident is 0 hence the residual are not correlated.

23
As explained earlier DW test is the most often used test for existence of autocorrelation, and if
there exist autocorrelation, it is customary to transform the data on the basis of the estimated first
order autocorrelation and use ordinary least square with the transformed data, if I is pure
autocorrelation one can use appropriate transformation of the original model so that we do not
have the problem of autocorrelation in the transformed model. In this regard we will have to use
the generalized least square (GLS) method (madala1992).

3.7.4. Multicollinearity
Multicollinearity refers to the condition that independent variable is intercorrelated and it is the
future of sample not for the population. The classical linear regression model assumes that there
is no multicollinearity among the explanatory variables. If perfect multicollinearity exists the
regression coefficients of the explanatory variables are indeterminate and their standard errors
are infinite. On the other hand if multicollinearity is less than perfect, in the regression
coefficients, although determinates. Have a larger standard error (in relation to coefficient
themselves), which means the coefficient cannot be estimated with great precision or accuracy.
Variance inflation factor (VIF) is used to determine the problems of multicolliniearty in making
inferences. When VIF>10, there is a problem of multicolliniearty and hence we cannot reject the
hypothesis of no correlation among explanatory variables. There are some remedial measures
suggested to the problem of multicollinearity such as prior information from previous empirical
works, combining both cross section and time series data. One of the simplest things to do is
dropping the variables which are highly correlated and specification bias. The variable should
also be transformed when they tend to move the same direction (Gujarat,

24
 CHAPTER FOUR

DISCUSION AND RESULTS

4.1. Descriptive statistics and distributional issues
Before going into detail discussion of the model, it is natural to study the characteristics and
distribution of the dataset for the variable that are included in the model of study. The summary
statistics (means, standard devotions, coefficient of variation, minimum and maximum values)

25
 and the summary distribution (skewness and kurtosis) of all variables at their respective value are
presented in the table below.

Table 4.1 Summary Statistic

Variable Obs Mean Std.dev Min Max Skewness Kurtosis Coeff. Of

variation

LnRGDP 40 10.89836 4553425 10.36682 11.81653 0.8598898 2.527677 4.8%

K 40 10.066826 97725 8.531298 11.58976 2.6237 8.952447 206.21%

L 40 712510 7555145 22973 2367 0.397195 1.88339 37.8%

EEHE/ 40 1395.594 2358.8 46.321 8486.2 1.783872 5.651271 108.5%

RGDP

As we can observe the result above the mean which is the average of the data or the sum of all
observations divided by the number of observation and the standard deviation which is the most
common measure of dispersion or how spread out the data are about the mean shows that
10.066826 and 9772598 for gross capital formation and 1395.594 and 2358.87 for education
expenditure plus health expenditure as a ratio of real gross domestic product respectively, Which
signifies a larger sample size results in smaller standard error of the mean and a more precise
estimate of the population mean and the higher standard deviation value indicates greater spread
in the data. Coefficient of variation in this table which is the measure of spread that describes the
variation in the data relative to the mean and the larger the coefficient of the variation the greater
the spread in the data, as a result from the result the variables relatively more dispersed with the
highest coefficient of variation 206.2% for gross capital formation, 108.5% for education
expenditure plus health expenditure as a ratio of real gross domestic product. But other variables
relatively less dispersed and not more lower coefficient of variation. The data sets are assumed to
be positive distributed; a picture of the distribution of the data can be analyzed using the concept
of skewness and kurtosis

26
. According to (Gupta, 1974), skewness is the degree of a symmetry and it shows the departure
from normal distribution. Looking in to the skewness of the data, they are all positively skewed
because the calculated value of skewness is greater than zero. In a positive skewened
distribution, smaller observation are more frequent than large observation, that means the
majority of observation have below an average and it has a long tail in the positive direction. So
that we conclude from this Gdp has a positive relation with given observation.
Kurtosis, on the other hand, is the degree of peakinessof a distribution relative to the normal
distribution. The kurtosis shows most of variables, although not higher, are picked or flat toped
indicating that they are normally distributed. A large number of observation have the same value
so that capital is the same value within the number of observation .There also large number of
observation which have low frequent are spread in middle interval.This means labor and gdp are
low frequency in a given observation.

So as we observed in the above4.1 table it show that real GDP growth rate is
positively skewed or skewed to the right and according to Gupta (1979) rule
of thumb; the data for the variable is relatively symmetrically skewed
because 0.859 is found greater than zero. The result of kurtosis tells us that,
2.52 are lower than 3 which means, the data has lower and slower central
peak and shorter and thinner tail which is known by the name platykurtic.

Labor force has positively skewed which means the right tail is longer than
the left one. Its result shows that, the data for the variable is relatively
symmetric with skewedness of 0.397 and has lower and slower central peak
and shorter and thinner tail, which is also known as platykurtic with kurtosis
of 1.88 (<3). Kapital formation is positively skewed which means the right
tail is longer than the left tail. The data for the variable is highly skewed with
skewedness of 2.62 and has higher and narrower central peak and longer
and thicker tail with kurtosis of 8.95 (>3). Education expenditure plus health
expenditure is another variable which has positively and highly skewed data
with skewedness of 1.78 and with kurtosis of 5.65 it has higher and
narrower central peak and longer and thicker.
27
4.2.1. Test for stationarity: unit root test
It is necessary to conduct unit root test to check stationarityof the data before estimating a time
series data. This is mainly because the model may include non-stationary variable that would
make the problem of spurious regression result or non sense regression. Since data used in this
study is time series, a unit root is conducted to find out the order of integration of the variables
used in the regression. The result from the Dickey Fuller (DF) test is presented in the table
below.

Table 4.2 DF unit root test

28
Variable DF test at level DF at the first Order of
difference Integration

LnRGDP 3.551** I(0)

K 8.381*** I(0)

_
L 2.447 4.510*** I(1)

EEHE/RGDP -1.840 -6.437*** I(1)

***= stationary at 1% critical values: 1%= -3.655

**= stationary at 5% 5%=-2.961

*= stationary at 10% 10%=-2.613

The essence of testing for unit root is because if the series is not stationary then all the results
from the classical linear regression analysis are not valid.The null hypothesis of this test is that
there is a unit root. The rejection of the null hypothesis is based on Mackinnon (1996) critical
values. The decision rule of DF test is when the absolute value of the computed DF --statistics is
greater than the absolute value of the critical statistics. In this case we can conclude that the
given variable is stationary otherwise we say non stationary. Based on DF tabulated above gross
domestic product (LNRGDP) and gross capital formation (K) are stationary at level 5% and all
critical values respectively. On the other hand, labor force (L), education expenditure plus health
expenditure as a ratio of real gross domestic product are stationary at first difference and at all
critical value.

4.2.2. Co-integration Test

An important issue in econometrics is the need to integrate short run dynamics with long run
equilibrium. The analysis of short run dynamics is often done by first eliminating trends in the
variable usually by differencing. The procedure, however, throw out potential long run
relationship about which economic theories have a lot to say (Maddalla1992).

So far, there are two major procedures to test for the existence of co-integration, namely, the
Engle-Granger two step procedures and the Johansone maximum likelihood estimation

29
procedure. However, the study will uses the Engle-Granger two step procedure in particular.
Here, the variables entering the co-integrating vector are tested for integration of same order. The
first step is to estimate the long run static model of the same level of integrated variables and
obtain the residual. If this residual, which is the linear combination of the variables are said to be
co-integrated, that they do have long run relationship.

This study has followed the EG approach to perform the co-integration test using DF test.
Accordingly, the residual from the long run model is tested for its stationarity; having a null
hypothesis there is unit root and it’s found that the residual is stationary at 1%, 5% and 10%
critical values which are shown in the table below. The fact that the residual is stationary implies
that the variables are co-integrated or there exists a long run equilibrium relationship among the
variables of the model.

Table 4.3 Co-Integration Test

Variable DF at Critical values

level

Residual -4.148 1%=-3.655 5%=-2.961 10%=-2.613

30
4.2.3. The Long Run Model

Table 4.4 Long Run Model Results

Explanatory Dependent variable is

variable LNRGDP

K 0.886 (0.883) (0.000)

L 0.0480599 (0.0027848)
(0.000)

EEHE/GDP 3.916768 (1.128658)

(0.001)

Constant 10.95999 (0.0476432)

(0.000)

F-Static 601.78

R2 0.9702

Adj_R2 0.968

DW(transformed) 1.66

Mean VIF 3.32

Prob>Chi2 0.9665

 The figures in parenthesis are coefficients, standard errors and probabilities in the table
and diagnostic tests respectively.

The regression is done by taking the observation of 40 years annual data from the stated above
organizations. R-square is an overall measure of the strength of association and does not reflect
the extent to which any particular independent variable is associated with the dependent variable.

31
As the OLS regression signifies the R-Square 0.978 of the long-run variation in real gross
domestic product is explained by explanatory variables included in the model. Adj R-squared is
an adjustment of the R-squared that penalizes the addition of extraneous predictors to the model.
Adjusted R-squared is computed using the formula 1 - ((1 - Rsq)((N - 1) /( N - k - 1)) where k is
the number of predictors.

As we can see from the above table results, gross capital formation, labor force, education
expenditure plus health expenditure as a ratio of real gross domestic product at all levels are
consistent with their prior expectatedsigns; mean that they have positive impact. In 1957, Nobel
Laureate Robert Solow described the growth of national income as having three sources: stock of
physical capital, increase in size of labor force and using the same approach as solow, but
explicitly accounting for the role of education, Becker stated that increasing the level of
education is the source of the total potential output. The significance of the coefficients for
capital, education expenditure plus health expenditure as a ratio of real gross domestic product
indicate that, in the long run, the two variables have high contribution to the economic growth as
compared to other variables included in the model. As we observed in the above 4.4 table result
1 unit increase in gross capital formation increases national output (RGDP) by 88.6%. Similarly
according to the long run result, a 1 unit increase in education expenditure plus health
expenditure as a ratio of as a ratio of output (RGDP) increases national output (RGDP) by
3.91%. From this it can be understood that gross capital formation, education expenditure plus
health expenditure as a ratio of real gross domestic product, have a significant contribution for
the Ethiopian economic growth in the long run. The findings of the empirical analysis show the
importance ofincreasinggovernment expenditure on education and health as a means to achieve
the long term economic growth in Ethiopia. This may directly affect the economic growth of the
country. This study suggests that it is important to increase government expenditure on education
to achieve sustainable economic growth. The coefficient of labor force is statically significant
but has relativelyless contribution to economic growth.The Durbin-Watson statistics indicates
little severity of auto-correlation. In some situations we can continue to use the OLS method
(Gujarat, 2007). However by applying praise wingsten regression it can be corrected and then
Durbin Watson statistics (transformed) is used and the same for short run model in this study.

32
4.2.4. THE SHORT RUN MODEL
To explain the short run relationship between independent and dependent variable that are co-
integrated, error correction model is used. So to complete the analysis of this section we consider
the error correction mechanism (ECM) which has short-run information. It has been shown
previously from the test of co-integration on the error term that the variables have long-run
equilibrium relationship. Thus after testing for co-integration the dynamic (short run) equilibrium
is obtain by regressing the first difference dependent variable on first difference of the
explanatory variables and the one period lagged error term..

dlnRGDP=β 0 + β 1 dK + β2 dL+ β 3 d ( EEHE

GDP )
+ ECM t −1 +U t

33
Table 4.5 short run model result
Explanatory Dependent variable is dlnRGDP
variables

DK 4.77 (1.88) (0.016)

DL 0.072382 (0.0173769) (0.179)

DEEHE/RGDP 0.4205266 (1.072456) (0.697)

ECMt-1 -0.3194835 (0.156892) (0.050)

Constant 0.0237801 (0.0152948) (0.129)

F-Static 2.64

R2 0.23

Adj_R2 0.14

DW(transformed) 1.83

Mean VIF 1.2

Prob>Chi2 0.6144

Note:- D= difference

 The figures in parenthesis are coefficients, standard errors and probabilities in the table
and diagnostic tests respectively.

Based on the above table 4.5 the R2 and adj_R2 for short run dynamics are 0.23 and after adjusted
0.14 of the variation in national output (RGDP) is explained by the variation in the explanatory
variables in the short-run. The error correction model (ECM) shows that a onetime deviation in
the equilibrium can be attained next period.

Regarding the individual significance of variables, the coefficients of gross capital formation,
labor force are found to be significant in the short run in Ethiopia based on this study. The result

34
here suggests that holding other variables constant under the period of analysis, a unit increase in
gross capital formation and labor force raise national output (RGDP) by 4.77 and 2.3
respectively in the short run.

On the other hand, education expenditure plus health expenditure as a ratio of national output
(RGDP) is statistically not significant. This is as a result of schooling has no short-run effect.
And the above short run estimation result shows the coefficient of the lagged error correcting
term (ECMt-1) is significant with the expected negative sign of -0.3194835. Its magnitude
indicates that deviation from the long-run equilibrium is adjusted where only 32% of
disequilibrium is removed in each year, in other word, the value 0.319 imply about 32% of the
discrepancy between the actual and long-run value of growth rate is corrected in one year and it
takes almost three years to avoid the whole disequilibrium.

35
CHAPTER FIVE

5 Conclusion and Recommendation

5.1. Conclusion
The main objective of the study is to analyze the impact of education on economic growth in
Ethiopia (using RGDP) as a proxy for economic growth and education expenditure plus health
expenditure as a ratio of RGDP as a proxy foreducation.The study use OLS approach to test for
co-integration and error correction model (ECM).

Before making an econometric estimation, the variables are tested for their order of integration
and it is found that they are integrated of order zero I (0) and integrated order one I (0). The
stationary of the residual also indicated that the variables are co-integrated. That is, there exist a
long run relationship among the variables OLS was used to estimate the long run and the short
run coefficients of the model.

According to this result, the variables gross capital formation, labor force and educational
expenditure plus health expenditure as a ratio of national output (RGDP) entered as explanatory
variables in the growth equation. The result from the long run static regression shows that the
three variables enter positively and explain national output (RGDP) and they have positive
impact to economic growth both through increased individual productivity brought about by
acquisition of skills and attitudes and through accumulation of knowledge.

The result in the short run, the coefficient of error term is –0.3194 suggesting about 31 percent
annual adjustment towards long run equilibrium. This is another proof for the existence of a
stable, long run relationship among variables. But unlike long run significant impact, education
and health have no significant short run impact on the economy. This could be due to the reason
that schooling has no short run effect and that of health expenditure may have a big impact on
the people who have no positive impact on the economy.

36
5.2 Recommendation
As it can be observed from the result education expenditure and health expenditure are the major
cause of growth of Ethiopian national output. The result of this study have important policy
implications in order to improve economic growth, public expenditures needs to be better
prioritized towards health service and education. In other words, the more people become
educated and healthy; they will increase their productivity in the long run.

The policy makers and government should center on securing more resources and structures that
are essential and appropriate for better education and improved basic health service provision.
There must be attraction of investment into the system of education, securing tax and customs
benefits for legal entities and individuals whose activities aimed at educational institutions
development.

37
References
Adelakun, O. J. (2011). Human Capital Development and Economic Growth in Nigeria.

Azmeraw, A (2011), the role of education to economic growth in Ethiopia.

Tekola (2007) the contribution of primary education to economic growth in Ethiopia

Bakare, A.S. (2006). The Growth Implications of Human Capital Investment in Nigeria: An
Empirical Study.Journal of Economics and Social Studies.AdoEkiti, Pp110-125

Tilak(1999) the traditional sector and modern sector of the economy in Ethiopia

Branson W.H. (2005), “Macro economic theory and policy”, 3rdedition, nice printing press,Delhi.

Wubshet (2006) investment in education and human capital in Ethiopia

Debraj Ray (1998): Development Economics, New Jersey.

Dornbush, R and Fisher, S. (1994). “Macro Economics” 6th Edition.

Education Statistics Annual Abstract (2014/15).of Ethiopia

Dauda(2010) investment in education and economic growth in Nigeria using annual time series
data

Ministery of Education( 2014 ) of Ethiopia

Greene W.H. (2006), “Econometrics Analysis” 5thEdition,Kinderselypvt.Ltd, India.

Gujarati, Damodar N (2004), “Basic econometrics,” forth edition, rata McGraw-Hill, New
Delhi.Abela(2004) investigated the impact of education on economic growth in Ethiopia

Gupta (1973) An introduction to statistical methods,” ninth edition.

38
Lucas, Robert E.; 1988, On the Mechanics of Economic Development, Journal of
MonetaryEconomics, Vol. 22, No. 1.

Lustig, Nora (2006). Investing in Health for Economic Development: The case of
Mexico”Midfrew Research Paper No. 2006/30.
Maddala, G.S (1992). Econometrics, University of Horida.

Mankiw .G 1992, “Macro Economics” 6th Edition.

Mincer, Jacob; 1974, Schooling, Experience, and Earnings, New York: Columbia
UniversityPress
Musbau, A.B. &Rasak A.A. (2005). Long Run Relationship between Education and
EconomicGrowth in Nigeria: Evidenceeconomics, and policy in Africa, University of Pretoria,
from the Johansen’sCo-integration Approach”. Paperpresented at the Regional Conference on in
West Africa, Cornell University, Senegal.Ndambiri H.K. Center for environmental

Netsanet, W (1997), education and economic growth in Ethiopia, proceeding of the

seventhannual conference on the Ethiopian economy,

Schultz, Theodore W.; 1961, Investment in Human Capital, American Economic Review, Vol.
51.
Becker, Gary S.; 1964, Human Capital: A Theoretical and Empirical Analysis, With
SpecialReference to Education. New York, National Bureau of EconomicResearch.
Taiwan." Working Paper 90-9.FederalReserve Bank of Atlanta.

Tallman, E.W. and P. Wang. 2000. "Human Capital and Endogenous Growth: Evidence from

Todaro, M.P (2006), Economic development in third world, fifth edition, Harlaw, New York.,
Education UNESCO for all, (2012)

HagosandHabib(1986)Theory of Endogenous growth model

Kifle (2006): Human Capital and Economic growth in Ethiopia,.

39
40