Arshad
Ali, University Utara Malaysia. Email: arshadswata@yahoo.com
Abstract. The main
focusing of this study is to examine empirically the connection between female
secondary school enrollment and economic growth of Pakistan taking the period
of 1975-2014. The variables of the series passed the test of stationary by the
first difference as evaluated by the ADF and PP test. Therefore, by employing
the Johansen test of cointegration, the result shows that female secondary
school enrollment and labor employment
have insignificantly long run positive influence on economic growth, however, capital formation has significantly
positive impact on economic growth of Pakistan. The Granger causality test
based on VECM shows that female secondary school enrollment and GDP have long
run two-way causality, however, the short
run bidirectional causality does not exist but unidirectional causality, which
is running from GDP to female secondary school enrollment.
Key words: Female
secondary education, Employment, capital formation, economic growth, Pakistan
Introduction
The
opportunity of secondary education when provides especially to girls tends to
get out a country from extreme poverty and enhance economic growth through high
chances of achieving quality work place,
tends to have low fertility level, securing good health condition of the women
and increase productivity level by the human capital development (The World Bank, 2008). Girls Secondary education completion have most of
the advantages such as to boost up drastically the life time earnings of the girls, similarly dramatically reduce the
rate of fertility and mortality. Each year 0.58 percent long-run economic growth can be achieved by the addition of
secondary schooling per year. Further,
the 100 countries study as demonstrated by the World Bank that one percent
increase in girls secondary education tends to enhance 0.3 percent income per
capita. (UNGEI, 2014). A recent study conducted n a panel of Asian
economies to explore empirically the influence of female education at primary,
secondary and territory level on economic growth. The result of the analyses
indicated a significant contribution of the female primary, secondary and
territory education to the economic
growth of the selected Asian countries (Cooray & Hassan,
2013).
Pakistan has
a large number of out of school children
around 6.7 million, of which girls are 5.5 percent, this becomes a big obstacle
from a long time in the way of economic
and social development of the country. The secondary education proportional
marginally decreases with the increasing of educational level. Total enrollment in the secondary school is 2.8
percent, of which 42 percent are female and 58 percent are male. (Malik et al., 2015). According to Pakistan, economic survey (2014)
Education determines significantly the development of the economy of a country.
High literacy rate guarantees the sustainable economic growth, economic
prosperity and high productivity of the labor
force. Equal opportunity of education to boys and girls eliminates gender
discrimination and strengthens to compete for
the emerging and modern challenges to adopting
new technology and upgrading the intellectual power. The main aim of the gender
equality of education is to remove gender disparity at all level of education
such as primary, secondary and territory education. Pakistan gender disparity
ratio indicates 47 percent literacy rate of female and 70 percent of male,
which demonstrates that more financial and human resources are needed to
achieve gender parity. According to World Bank (2015)
the female secondary school enrolment in Pakistan is sufficiently lower than
the male enrolment of secondary education. The gross enrolment ratio of
secondary education in 2014 is 36.6 percent for female and 46.3 percent for a male. Similarly, UNESCO (2016) has also
demonstrated the lower enrolment ratio of 39.20 percent for girls secondary
education as compared to 49.45 percent for a male
in 2015.
The study
has following two specific objectives.
1. To examines the influence of
female secondary education on the economic growth of Pakistan; and
2. To examines the bilateral
relationship between female secondary education and economic growth of
Pakistan.
Mankiw, Weil
and Romer (1992) conducted a research based on the growth model of endogenous
to find out the interrelationship between education and economic growth. Human
capital, capital formation, government policies, political stability, market
distortion and technology have considered as factors of the growth model of
endogenous etc, which have a significant contribution to overall economic
growth. The main objective of the study is to explore the several countries
enrolment rate modification. The result of the research found a significant
effect on the economic growth with the enhancing of school enrolment, The
conclusion of the study indicated a significant influence of the schooling on
economic growth, which reflects that secondary school enrolment for girls will
tend to low fertility and mortality, high employment opportunities, promoting
human capital development. The efficient and skilful
labor then access to utilize the resources efficiently, which
consequently increase the productivity level and economic growth.
Lucas (1988)
was of the view that education is one of the important factors, which
determines to stimulate the development process by the accumulation of human
capital, where the role of human capital is the main factor in the process of
production. As argued by this study that Human development by the education produces skilful
and efficient labors that have a positive
impact on the production and better economic performance of a country.
Dollar and
Gatti (1999) investigated the interrelationship between gender disparity of
education and economic growth. They were of the view that the status of women
in developed countries is stronger than
developing the world. Mostly in
developing countries, the Government spending on female education and health is
quite less than the investment on male; therefore, girls have very less access
to education, political power and to their legal rights in the society. The
main findings of the research are that
education gander disparity has
significantly a negative influence on economic growth, the result of the study
also indicates a significantly positive influence of female secondary school
attainment and insignificantly a negative impact of male attainment of
secondary school on the economic growth.
Klasen
(2002) contended that girls education immensely has a significant contribution
to the economic growth of developing the world.
The study also stressed on the equal access to education is required for both
girls and boys. If more opportunities of education provide to girls then their
outcome is higher than the return of the boys. It means that in developing
countries the girls education has a marginal
increasing return.
Hassan and
Cooray (2013) employed Bounds testing approach to finding out empirically the impact of male and female education at
primary, secondary and territory level on the economic growth of a penal of
Asian economies. The findings of the empirical analysis of the study indicate
that the enrolment ratios of the male and female at primary secondary and
territory level have a significantly positive influence on the economic growth
of the Asian countries.
Self and
Grabowski (2004) found the influence of male and female primary, secondary and
territory education separately on the economic growth of India covering period
1966-1996. The conclusion of the study demonstrated that male primary education
has a strong positive influence on the economic growth of India, where
secondary education has a week positive effect on growth. However, the female
education at all level has a significantly positive influence on the economic
growth of India.
Knowles,
Lorgelly and Owen (2002) examined the impact of gender gap of education on
economic development, especially to find out whether the boosting of female
enrolment in school enhances the productivity level of labor in the long run. The empirical findings of the study
demonstrated a positive and significant influence of female education on the
productivity level where the unclear
result has shown by the effect of male education on the economic productivity.
The specific
econometric model for this study is based on the growth model of endogenous
presented by Mankiw, Romer, and Weil
(1992), which indicates the linkage between educational attainment and economic
growth. Mankiw included in his model the endogenous factors such as human
capital, political instability, market distortion, capital accumulation, public
policies, modern machinery and technology have a significant impact on the
overall growth and development process. The growth model of Mankiw et al.
(1992) is based on the Solow (1956) standard neoclassical model of growth.
Solow (1956) included three factors of production in his standard growth model.
Y = f (A, K,
L)
..
..(1)
Y shows total output or production by bringing
together the three production factors, A stands for the productivity of total
factors or the promotion or advancement of the training or education
technically, K indicates capital accumulation physically and L stands for the
employment ratio of the labor force. When
taking derivative of the above equation 1, we get the following equation 2.
It has been hypothesized that the production function of
Solow standard growth model is based on production function of Cobb-Douglas
indicates constant return, meaning that by adding altogether the share of each
factor such as physical capital, technology or technical training and labor equal to one as demonstrated in the
following equation.
γ + α + β = 1
..(3)
α shows the share
of capital, β indicates the share of labor
and γ is the shares of technology or knowledge.
By adding equation 2
and 3 we get the following basic equation of Solow.
We get the following
equation when takes natural logs of the equation 4.
LnYt = LnAt + αLnKt
+ βLnLt +΅t
(5)
Mankiw,
Romer and Weil (1992) modified the Solow standard mode of growth by adding
human capital then the model general form
can be written as;
Yt = At Ktα
Etβ Lt1-α-β eit t = 1, 2, 3
.
(6)
Yt indicates the
aggregate outcome by making all together
the share of each economic factor, At shows the development of
education or training technically, Kt indicates capital
accumulation, Et shows the number of female enrolment in secondary
school, Lt indicates the employment ratio of the labor force and eit
is the error term.
We get the following equation 7 when taking natural logs both sides of the
above equation 6.
lnYt
= lnAt + αlnKt + βlnEt + γlnLt
+ ejt t = 1, 2, 3
(7)
lnAt shows constant or stable parameter, α is the capital
production elasticity, β is the
elasticity of production by the female secondary education, γ shows labor production elasticity and ejt is the error term or
the influence of external factors, which are outside the model.
The study
has chosen the country of Pakistan using Time series data of the period
1975-2014. Gross fixed capital formation
(Current LCU) proxies for the variable of capital accumulation (K). GDP
(Constant LCU) proxies for the aggregate output or total economic growth (Y). Et
denotes the number of female enrolment in secondary school, which
reflects the development of human capital. Lt shows the employment
of labor force. The data for gross fixed
capital formation and GDP have collected from World Bank (2015). However, the
data of labor employment and female
secondary school enrolment have collected from Pakistan economic survey (2015).
This study
has employed Dickey and Fuller (1979) and Perron (1988) unit roots test to
check stationary of the variables, the Johansen (1988, 1991) approach for long
run cointegration, and the Granger test for bilateral causality. It has been
assumed by the analysis of classical regression that gross domestic product,
female secondary school enrolment, labor and gross fixed capital formation should
be stationary, meaning that the variables should have a constant mean and
variance. However, if the variables have
no constant mean and variance over time, meaning that they are non-stationary then the result of the classical
regression analysis will be considering as invalid or spurious (Thomas, 1997).
Despite if, the variables have a significant association but still, their result will be invalid. Therefore,
the study first employed the PP and ADF test to find out unit roots in the
selected variables of the model. However, the Granger (1987) has pointed out
that if the residuals of ordinary least square estimation of the non-stationary variables are stationary then
the non-stationary variables shows long
run cointegration association. Thus the study has used Johansen (1988, 1991)
approach to examine the long-run cointegration among the variables. Finally, the
test of Granger (1969) causality has been employed by this study to find out
the bilateral association between
enrolment number of female secondary education and GDP. The standard causality
test of Granger will be used only if all the variables have no unit roots,
meaning that the variables are stationary at level or I(0). The following
equations having a lag length of k will
be used for the standard Granger test for the bilateral causality among the
variables.
GDPt
= a1 + b1GDPt-1 +..+ bkGDPt-k +
c1Et-1+..+ ckEt-k + e1
(8)
Et
= a2 + b1Et-1+..+ bkEt-k+
c1GDPt-1+..+ ckGDPt-k+ e2
(9)
a1
and a2 indicates constants, b1
..bk and c1
..ck
are the slope coefficients.
The
causality test of Granger will be employing for the joint hypothesis by using
Wald test.
c1 =
c2 = c3 =
..ck = 0
. (10)
The null
hypothesis of equation 8 indicates that female secondary school enrolment in
Pakistan does not have a unidirectional causality on GDP. Conversely, the equation 9 demonstrates that
GDP does not have a one-way causation on
the number of female enrolment in secondary school. However, the alternative
hypothesis indicates the existence of bilateral causality between female
secondary school enrolment and GDP. The optimum lag length will be selecting by
minimising the criteria of Akaike
information.
If the
variables of the model have long run cointegration relationship and they are
integrated of the same order of I(1), then for the bilateral causality this
study will use Granger causality test based on VECM, which is based on the
following equation.
ΔGDPt =
a1+b1ΔGDPt-1+..+bkΔGDPt-k+c1ΔEt-1
+..+ckΔEt-k +d1ECt-1+΅1
..(11)
ΔEt
= a2 +b1ΔEt-1+..+bkΔEt-k+c1ΔGDPt-1+..+ckΔGDPt-k+d2ECt-1+΅2
(12)
∆
stands for the change or difference and ECt-1 abbreviates the error
correction term shows the speed of adjustment to the long run equilibrium from
short run shock.
The VECM
type of causality test is more beneficial than the standard causality test of
Granger. The causality test based on VECM determines to find out not only long
run causality but short run too. The short run causality among the variables
will be finding out by using Wald test. However,
the negative coefficient along with significant of ECt-1 indicate
the long run causality.
The four
possible causality between female secondary enrolment and GDP have mentioned
below.
If the two
variables do not show any interdependency between them demonstrate independence,
meaning that there is no causality exist between female secondary enrolment and
GDP.
The
association between the two variables shows unidirectional or one-way causation, meaning that there is
unidirectional causality running from female secondary school enrolment to GDP.
This type of
linkage between the two variables also shows unidirectional or one-sided
causality, meaning that there is unidirectional granger cause from GDP to
female secondary school enrolment.
This type of
relationship between the two variables indicates bilateral causality, Such as
female secondary school enrolment influences GDP but conversely, GDP also affects
female secondary school enrolment.
This section
exhibits the findings of the analysis by employing the econometrics tests to
find out the linkage between female secondary school enrolment and GDP of
Pakistan covering the period 1982-2014.
First, the
ADF and PP unit roots test have been employed by this study to find out the
stationary order of all the data of the variables.
The
following table 1 shows the result of ADF and PP tests together, identifying
that all the variables have passed the stationary test by the first difference,
meaning that by the employing of ADF and PP test together we found that gross
domestic product, female secondary school enrollment,
labor employment and gross fixed capital formation got unit roots at level but
they become stationary when converted to first difference or I (1).
Table 1 Result of ADF and
PP test
|
|
ADF
Test statistics |
PP
test |
||
|
Variables |
Level |
First
Difference |
Level |
First
Difference |
|
InGDPt |
-1.6743 |
-4.752* |
-1.267 |
-4.722* |
|
lnKt |
-1.336 |
-6.143 * |
-2.323 |
-9.605* |
|
lnLt |
0.563 |
-7.315 |
0.813 |
-7.120 |
|
lnEt |
-0.637 |
-4.718* |
-0.9713 |
-4.726* |
Note: *denotes1% significance of the test statistics, value inside the
parentheses is probability
As we have
confirmed from the above ADF and PP tests together that all the variables have
integrated of the same order I(1) therefore, the Johansen approach will be
using to find out the long run
association among the variables. But before employing Johansen test of
cointegration we should select the optimum lag length by the criteria of VAR
lag such as LR test statistics in the sequential
modified form, final prediction error or simple FPE, Hanan Quinn criterion
(HQ), Akaike information criterion (AIC), and the Schwarz criterion (SC). The
criterions lag selection have identified in the following Table 2 that Schwarz
criterion (SC) and Hanan Quinn criterion (HQ) have chosen 1 lag, however, the Akaike information criterion (AIC)
and the criterion of final prediction
error (FPE) together have selected 4 optimum lags. We select 4 lags by giving more preference to the choice of
AIC and FPE.
Table 2 The Result of Optimum lags Selection by the
Unrestricted VAR Criteria
|
Lag |
LogL |
LR |
FPE |
AIC |
SC |
HQ |
|
0 |
70.28 |
NA |
0.000 |
-3.68 |
-3.51 |
-3.62 |
|
1 |
280.36 |
361.80 |
0.000 |
-14.46 |
-13.59* |
-14.16* |
|
2 |
291.59 |
16.85 |
0.000 |
-14.2 |
-12.62 |
-13.65 |
|
3 |
317.13 |
32.63* |
0.000 |
-14.73 |
-12.44 |
-13.93 |
|
4 |
338.31 |
22.35 |
0.000* |
-15.02* |
-12.03 |
-13.97 |
*shows the
criterion selection of the optimum lag
We put the optimum lag into the Johansen
approach to finding out long run
cointegration among the variables by the Trace and Eigenvalue tests. The following Table 3 demonstrates that the null
hypothesis of no cointegration has been rejected by the Trace test, indicating
that the value of Trace statistics is significant and greater than the critical
value. Similarly, the null hypothesis of one cointegration vector and two
cointegration vectors respectively have been rejected by the Trace test,
implying that their values of Trace statistics are significant and greater than
their critical values. However, The Trace test accepted the null hypothesis of
at least three cointegration vectors, implying that the value of Trace
statistics is insignificant and lower than it is respective critical value.
Therefore, the series indicate a long run cointegration relationship.
Similarly, the Eigenvalue test rejects
the null hypothesis of no cointegration relationship among the variables,
indicating that the Eigenvalue statistics
is significant and greater than it is critical value. However, the null of the hypothesis of at least one cointegration
association has been accepted by the Eigenvalue
test, indicating that the Eigenvalue
statistics is insignificant and lower than it is respective critical value.
Therefore, it indicates that the variables have long run cointegration
relationship.
Table 3 Results of Unrestricted Trace
Test of Johansen
|
Hypothesized No. of CE(s) |
Eigen value |
Trace Statistic |
0.05 Critical Value |
Prob.** |
|
None* |
0.651 |
72.64 |
47.86 |
0.000 |
|
At most 1* |
0.425 |
35.85 |
29.80 |
0.001 |
|
At most 2* |
0.307 |
16.50 |
15.50 |
0.035 |
|
At most 3 |
0.099 |
3.66 |
3.84 |
0.056 |
Trace test shows 3 cointegrating vectors
at the level of 0.05
* indicates the hypothesis rejection at the
level of 0.05
Table 4 Results of Unrestricted Eigen test of
Johansen
|
Hypothesized No. of CE(s) |
Eigen value |
Max-Eigen Statistic |
0.05 Critical Value |
Prob.** |
|
None * |
0.651 |
36.796 |
27.584 |
0.003 |
|
At most 1 |
0.425 |
19.350 |
21.132 |
0.087 |
|
At most 2 |
0.307 |
12.835 |
14.265 |
0.083 |
|
At most 3 |
0.100 |
3.663 |
3.842 |
0.056 |
Max-Eigen value test shows 1 cointegrating
vectors at the level of 0.05
*denotes
rejection of the hypothesis at the 0.05 level
The
following Table 5 shows the estimated result of the long run coefficients
interpreted by the VECM
Table 5 The result of the estimated long-run coefficients by the VECM
|
Cointegrating Eq: |
CointEq1 |
|
LGDP(-1) |
1.000 |
|
lnK(-1) |
-0.319 |
|
(0.051) |
|
|
[-6.334] |
|
|
lnL(-1) |
-0.080 |
|
(0.198) |
|
|
[-0.403] |
|
|
lnE(-1) |
-0.1224 |
|
(0.096) |
|
|
[-1.275] |
|
|
C |
-19.726 |
Note: The values inside ( ) and
[ ] are the standard error and t-statistics
respectively
The long run
Johansen coefficients in the normalized form have identified in the equation
below.
lnYt
= 19.72615 + 0.319662 lnKt + 0.122398 lnEt + 0.079602 lnLt
+ ejt
...(13)
[-6.334] 1.275] [-0.403]
We conclude
from the above equation that the coefficient sign of capital accumulation is
positive and statistically significant, meaning that capital formation has a
significantly positive impact on the economic growth of Pakistan. Similarly, female secondary school enrollment and labor
employment have also positive coefficients sign but statistically
insignificant, indicating that female secondary education and labor employment have an insignificantly
positive influence on the economic growth of Pakistan.
The findings
of the pairwise bilateral causality
between female secondary education and economic growth based on VECM are
identified below in Table 6 and 7.
Table 6 Results of Granger Causality Based on VECM
|
|
Coefficient |
Std. Error |
t-Statistic |
Prob. |
|
ECMt-1 |
-0.169 |
0.057 |
-2.974 |
0.006 |
|
D(lnGDP)(-1) |
0.196 |
0.156 |
1.228 |
0.230 |
|
D(lnGDP)(-2) |
0.207 |
0.170 |
1.220 |
0.233 |
|
D(lnK)(-1) |
-0.09 |
0.041 |
-2.178 |
0.038 |
|
D(lnK(-2) |
-0.102 |
0.040 |
-2.527 |
0.018 |
|
D(lnL)(-1) |
-0.083 |
0.083 |
-0.999 |
0.327 |
|
D(lnL)(-2) |
-0.167 |
0.086 |
-1.940 |
0.063 |
|
D(lnE)(-1) |
-0.005 |
0.054 |
-0.087 |
0.931 |
|
D(inE)(-2) |
0.107 |
0.052 |
2.042 |
0.051 |
R-squared
0.533 Prob
(F-statistic) 0.006
Hannan-Quinn criteria 4.997
F- statistics 3.422
ECMt-1
is the error correction term shows the speed
of adjustment from short run shock to long-run
equilibrium. The coefficient of ECMt-1 is -0.168625, which is
negative and significant, indicating that 16 percent short run current
deviation can get an adjustment in the
next period, meaning that there is long-run
causality running from explanatory variables to GDP. However, we use the Wald
test to find out short run causality, the findings show that the chi-square statistics are insignificant for the
female secondary school enrollment and labor employment; therefore, we could not found
short run unidirectional Granger cause from female secondary education and labor employment to GDP respectively. However,
there exists unidirectional short run causality from capital formation to GDP,
because the chi-square statistics for capital formation is significant.
Now we take
female secondary school enrolment as dependent variable and GDP, labor employment and capital formation are
included in the series as independent, therefore we get the following result of
causality based on VECM in Table 7.
Table 7 Results of Granger Causality Based on VECM
|
|
Coefficient |
Std. Error |
t-Statistic |
Prob. |
||
|
ECTt-1 |
-0.028 |
0.007 |
-3.783 |
0.001 |
||
|
D(lnE)(-2) |
-0.248 |
0.150 |
-1.650 |
0.111 |
||
|
D(lnE)(-3) |
-0.170 |
0.153 |
-1.106 |
0.279 |
||
|
D(lnGDP)(-2) |
0.978 |
0.476 |
2.054 |
0.050 |
||
|
D(lnGDP(-3) |
-1.222 |
0.504 |
-2.423 |
0.023 |
||
|
D(lnK)(-2) |
0.101 |
0.133 |
0.758 |
0.455 |
||
|
D(lnK)(-3) |
0.130 |
0.118 |
1.100 |
0.281 |
||
|
D(lnL)(-2) |
0.080 |
0.263 |
0.304 |
0.764 |
||
|
D(lnL)(-3) |
-0.439 |
0.274 |
-1.603 |
0.121 |
||
|
R-squared |
0.558 Prob (F-statistic) 0.005 |
||||
F-statistic 3.652 Hannan-Quinn criter. -2.801
The error
correction term in the above table is, again, significantly negative, which
shows long run causality from gross domestic product, labor employment and capital formation to female secondary school
enrolment. Where the short run causality
have been analyzed by the Wald test,
indicating that the chi-square statistics for the variables of capital
formation and labor employment are
insignificant, meaning that there is no short-run
causality from these variables to female secondary school enrolment. However,
the chi-square statistics for the variable of GDP is significant, implying that
there exists short-run causality from GDP
to female secondary school enrolment.
The summary conclusion of the Causality test of
Granger based on VECM demonstrates that female secondary school enrolment and
GDP have long run bilateral causality, meaning that female secondary school
enrolment affects GDP in the long run but conversely GDP also has long run
influence on female secondary school enrolment. However, the following Table 8
indicates that there is no short run bilateral causality between female
secondary enrolment and GDP but unidirectional short run causality, which is
running from GP to female secondary school enrolment.
Table 8 Results of Granger Causality Tests Based on VECM
|
Dependent variables |
lnGDP |
lnL |
LnE |
lnK |
|
lnGDP |
-- |
4.407 |
4.195 |
10.538 |
|
-0.110 |
-0.123 |
-0.005 |
||
|
lnE |
8.065 |
2.984 |
-- |
1.626 |
|
|
-0.018 |
-0.225 |
-0.444 |
Inside the braces is the probability value
This study
shows the empirical findings of the association between female secondary school
enrolment and economic growth of
Pakistan, taking the period 1975-2014. The ADF and PP tests together found at
level non-stationary of all the data but they become stationary by converting
into the first difference. The
cointegration technique of Johansen approach has been used in this study due to the same integration level
of the variables. The findings show that the variables have long run
cointegration relationship. The normalized
long run Johansen coefficients have been extracted from the VECM indicate that
female secondary education and labor
employment have an insignificantly
positive influence on the economic growth
of Pakistan, however, the capital
accumulation has a significantly positive
effect on economic growth of Pakistan.
The causality test of Granger based on VECM found long pairwise causality between female secondary
school enrolment and GDP; however, the Trace test does not show short run
bilateral causality but unidirectional causality, which is running from GDP to
female secondary school enrollment.
Cooray, A., &
Hassan, G. M. (2013). Effects of male and female education on economic growth. Working
Paper in Economics.
Dickey, D. A., & Fuller, W. A. (1979). Distribution of
the estimators for autoregressive time series with a unit root. Journal of
The American Statistical Association, 74(366), 427431.
Dollar, D., & Gatti, R. (1999). Gender inequality income
and growth: Are good times good for women? The World Bank, (1), 142.
Engle, R. F., & Granger, C. W. J. (1987). Co-integration
and error correction: Representation, estimation, and testing. Journal of
the Econometric Society, 55(2), 251276.
Gemmell, N. (1996). Evaluating The impacts of human capital
stocks and accumulation on economic growth: some new evidence. Oxford
Bulletin of Economics and Statistics, 58(1), 928.
Granger, C. W. J. (1969). Investigating Causal Relations by
Econometric Models and Cross-spectral Methods. Journal of Economitric
Society, 37(3), 424438.
Hassan, G. M., & Cooray, A. (2013). Effects of Male and
Female Education on Economic Growth : Some Evidence from Asia Using the
Extreme Bounds Analysis. Working Paper in Economics, 10(13),
128.
Johansen, S. (1988). Statistical analysis of cointegration
vectors. Journal of Economic Dynamics and Control, 12(2-3),
231254.
Johansen, S. (1991). Estimation and hypothesis testing of
cointegration vectors in gaussian vector autoregressive models. The
Econometric Society, 59(6), 15511580.
Knowles, S., Lorgelly, P., & Owen, P. D. (2002). Are
educational gender gaps a brake on economic development? Some cross-country
empirical evidence. Oxford Economic Papers, 54(1), 118149.
Lucas, R. E. (1988). On The Mechanics Of Economic
Development. Journal of Monetary Economics, 22, 342.
Malik, B. A., Amin, N., Ahmad, K., Mukhtar, E. M., Saleem,
M., & Kakli, M. B. (2015). Pakistan Education for All Review Report 2015.
Mankiw, N. G., Romer, D., & Weil, D. N. (1992). A
Contribution To the Empirics of Economic Growth. Nber Working Paper Series,
407437.
PES. (2015). Education. Retrieved from
http://www.finance.gov.pk/survey/ chapters_15/10_Education.pdf
Phillips, P. C. B., & Perron, P. (1988). Testing for a
unit root in time series regression. Oxford University Press on Behalf of
Biometrika Trust Stable, 75(2), 335346.
Razmi, M., Falahi, M., Abbasian, E., & Salehifard, M.
(2015). The relationship between womens education and human development. European
Scxientific Journal, 1(4), 177183.
Self, S., & Grabowski, R. (2004). Does education at all
levels cause growth? India, a case study. Economics of Education Review,
23(1), 4755.
The World Bank. (2008). Girls Education in the 21st
Century.
Thomas R. L. (1996). Modern Econometrics: An Introduction
Harlow: Addison-Wesley.
UNESCO. (2016). Education: Gross Enrolment Ration by Level
of Education (Vol. 37).
UNGEI. (2014). Accelerating Secondary Education for
Girls : Focusing on Access and Retention. Global Partnership for
Education.