8/7/2019 International Conference PPT
http://slidepdf.com/reader/full/international-conference-ppt 1/22
An Econometric Assessment on
Relational Interaction between HigherEducation and Economic Growth in
India
Nishant Joshi, Assistant Professor, Prestige Institute of Management and
Research, Indore
Dr.R.K.Sharma, Professor and Director , Prestige Institute of Management
and Research, Indore.Neha Joshi, PhD Research Scholar, School of Future Studies, DAVV, Indore.
8/7/2019 International Conference PPT
http://slidepdf.com/reader/full/international-conference-ppt 2/22
BEFORE WE MOVE AHEAD
8/7/2019 International Conference PPT
http://slidepdf.com/reader/full/international-conference-ppt 3/22
T
he relational interaction between educationand economic growth has been the subject of
public debates, enjoying a wide interest since
the era of Plato. According to Dikens et all.
(2006), Zoega (2003) and Barro (1991), theeducation has a high intrinsic economic value
since the investments in education led to the
formation of human capital, which is one of the cause of economic growth.
8/7/2019 International Conference PPT
http://slidepdf.com/reader/full/international-conference-ppt 4/22
Amongst the BRIC countries (i.e. Brazil, India,
China and Russia) successful studies has been
conducted in Brazil and china and the result
amongst the two variables showed that higher
education had a significant impact on theeconomic growth for Brazil and China. But a
similar study in Chile showed no causal
relationship amongst the variables.
8/7/2019 International Conference PPT
http://slidepdf.com/reader/full/international-conference-ppt 5/22
8/7/2019 International Conference PPT
http://slidepdf.com/reader/full/international-conference-ppt 6/22
Podrecca and Carmeci (2002) analyzed the causality between
education and economic growth using Granger causality, for a
set of 86 countries over the period 1960-1990.T
heir resultsshow that both education investment and the educational
stock had an impact to growth rates, both individually and
jointly with physical capital investment. There is also a reverse
causality that runs from growth to investment in education.
Jaoul (2004) analyzed causality between higher education andeconomic growth in France and Germany in the period before
the Second World War. The obtained results demonstrate that
higher education has an influence on gross domestic product
just for the case of France. For Germany, education does not
appear as a cause of growth. Kui (2006) analyzed the causality
and co-integration between education and GDP, showing that
economy development is the cause of higher education
8/7/2019 International Conference PPT
http://slidepdf.com/reader/full/international-conference-ppt 7/22
Hunang, Jin, and Sun (2009) analyzed the
causality between scale evolution of higher
education and economic growth in China,
between 1972 and 2007. The empirical results
show that there is a long-term steadyrelationship between variables of enrollment
in higher education and GDP per capita. For
the analyzed period, with the growth of theeconomy, scale of higher education exhibits an
ascending trend.
8/7/2019 International Conference PPT
http://slidepdf.com/reader/full/international-conference-ppt 8/22
Objectives
To analyze the causality between higher
education and economic growth for India.
8/7/2019 International Conference PPT
http://slidepdf.com/reader/full/international-conference-ppt 9/22
Hypothesis In the first step of the analysis the stationarity of the variables was to be examined.
If all the variables are stationary I(0), if there is no problem to estimate the
coefficients using the variables with initial specification. However, most of the
main macroeconomic variables are non-stationary, integrated of order higher than
zero. If the series are non-stationary but co-integrated, then the estimation as an
autocorrected model is admissible. If the variables are non-stationary and are not
co-integrated then the specification of variables as differences is necessary.
We used the Augmented Dickey-Fuller test (ADF) to test the existence of unit roots
and to determine the order of integration of the variables. The test was conducted
with and without a time trend. In order to test the co-integration of the analyzed
variables, we used the maximum likelihood estimation method of Johansen and
Juelius (1990, 1995), since the cointegration test procedures of Johansen and
Juselius (1990) can distinguish between the existence of one or more cointegrating
vectors and also generate test statistics with exact distributions (Van den Berg and
Jayanetti 1993);
8/7/2019 International Conference PPT
http://slidepdf.com/reader/full/international-conference-ppt 10/22
it is hereby appropriate to utilize. Thus, assuming a vector
autoregressive (VAR) model:
xt=ixt-i + xt-1 + + t
Where xt is a vector of non-stationary variables p x 1 and (i =
1,.,k). In essence, the JJ (Johansen and Juselius) method
tests whether the coefficient matrix reflects the
fundamentals of long run equilibrium among the non-stationary variables. As a result, if 0 < rank = r < p, then
there are matrices and of dimension p x r where =
and r cointegrating relations among elements of xt; where
and are cointegration vectors and error correction
parameters, respectively. Both Eigen value and trace tests,without a trend and with a trend were estimated. According
to Sims et all (1990), if the times series are non-stationary and
not co-integrated, then they obtained F statistics used to
detected Granger causality are not valid. Finally we shall applythe Granger causality to test the hypothesis.
8/7/2019 International Conference PPT
http://slidepdf.com/reader/full/international-conference-ppt 11/22
Yt = i xt-i + jYt-j + 1t . (1)
Xt = i xt-i + jYt-j + 2t . (2)
with the assumption that the disturbances t1 and t2 are uncorrelated. We distinguish four
cases: H0 1: Their exists stationarityin data
H0 2: The variables are co-integrated
H0 3: Unidirectional causality from X t to Y t is indicated if the estimated coefficients on the
lagged X t in (1) are statistically different from zero.
H0 4: Unidirectional causality from Y t to X t is indicated if the estimated coefficients on the
lagged Y t in (2) are statistically different from zero as a group. H0 5: Bilateral causality is indicated when the set of X t and Y t coefficients are statistically
different from zero in both regression equations (1) and (2).
H0 6: Independence occurs when the set of X t and Y t coefficients are not statistically
significant in both regression equations (1) and (2).
In all four cases it is assumed that the two variables Xt andY
t are stationary. In a stochasticprocess stationary means that the statistical characteristics of the process do not change in
time. As Granger and Newbold (1974) and Cheng (1996) point out, Granger causality on non-
stationarity time data may lead to spurious causal relation. The stationarity of a non-stable
time series can be obtained with the help of certain mathematic procedure, such as
differentiation of variables (Gujarati, 2004).
8/7/2019 International Conference PPT
http://slidepdf.com/reader/full/international-conference-ppt 12/22
Research Methodology
In the first step of the analysis the stationarity
of the variables was to be examined. If all the
variables are stationary I(0), if there is no
problem to estimate the coefficients using the
variables with initial specification. However,
most of the main macroeconomic variables
are non-stationary, integrated of order higherthan zero.
8/7/2019 International Conference PPT
http://slidepdf.com/reader/full/international-conference-ppt 13/22
If the series are non-stationary but co-
integrated, then the estimation as anautocorrected model is admissible. If the
variables are non-stationary and are not co-
integrated then the specification of variables
as differences is necessary.
We used the Augmented Dickey-Fuller test
(ADF) to test the existence of unit roots and to
determine the order of integration of the
variables. The test was conducted with and
without a time trend.
8/7/2019 International Conference PPT
http://slidepdf.com/reader/full/international-conference-ppt 14/22
In order to test the co-integration of the
analyzed variables, we used the maximumlikelihood estimation method of Johansen and
Juelius (1990, 1995), since the cointegration
test procedures of Johansen and Juselius
(1990) can distinguish between the existence
of one or more cointegrating vectors and also
generate test statistics with exact distributions
(Van den Berg and Jayanetti 1993); it is herebyappropriate to utilize.
8/7/2019 International Conference PPT
http://slidepdf.com/reader/full/international-conference-ppt 15/22
Thus, assuming a vector autoregressive (VAR) model:
xt=ixt-i + xt-1 + + t
Where xt is a vector of non-stationary variables p x 1 and (i =1,.,k). In essence, the JJ (Johansen and Juselius) method
tests whether the coefficient matrix reflects the
fundamentals of long run equilibrium among the non-
stationary variables. As a result, if 0 < rank = r < p, then
there are matrices and of dimension p x r where =
and r cointegrating relations among elements of xt; where
and are cointegration vectors and error correction
parameters, respectively. Both Eigen value and trace tests,
without a trend and with a trend were estimated. Accordingto Sims et all (1990), if the times series are non-stationary and
not co-integrated, then they obtained F statistics used to
detected Granger causality are not valid. Finally we shall apply
the Granger causality to test the hypothesis.
8/7/2019 International Conference PPT
http://slidepdf.com/reader/full/international-conference-ppt 16/22
Results
The results of Augmented Dickey-Fuller test
(ADF) to test the existence of unit roots and to
determine the order of integration of the
variables both tests showed that the variables
higher Education and GDP are non-stationary,
at the 5% significance level. However, the non-
stationary problem vanished after seconddifference
8/7/2019 International Conference PPT
http://slidepdf.com/reader/full/international-conference-ppt 17/22
8/7/2019 International Conference PPT
http://slidepdf.com/reader/full/international-conference-ppt 18/22
Upon the application of Johansen co-integration tests both Eigen value and
trace tests. Trace test indicates 1 co integrating equation(s) at both 5% and
1% levels; Max-eigen value test indicates 1 co integrating equation(s) at both
5% and1%levels
8/7/2019 International Conference PPT
http://slidepdf.com/reader/full/international-conference-ppt 19/22
The fact that the analyzed variables are co-integrated is very important for the
validity of Granger causality test results. According to Sims et all (1990), if the
times series are non-stationarity and not co-integrated, then they obtained F
statistics used to detected Granger causality are not valid.
In order to determine if there is a Granger causality between education and gross
domestic product per capita, Granger causality between GDP per capita and
Higher Education,. The obtained results are presented in Table 4.
8/7/2019 International Conference PPT
http://slidepdf.com/reader/full/international-conference-ppt 20/22
8/7/2019 International Conference PPT
http://slidepdf.com/reader/full/international-conference-ppt 21/22
8/7/2019 International Conference PPT
http://slidepdf.com/reader/full/international-conference-ppt 22/22
Conclusion
The aim of this article was to analyze the causality between
education and economic growth for India, in between 1994 -
95 and 2009-10. Using data gathered from IMF website and
from the websites of Index Mundi Using VAR methodology,
we found out that there is empirical evidence of a long-run
relationship between higher-education and gross domestic
product per capita in India, during the analyzed period. , Like
in case of many other BRIC countries like Brazil and China and
very close to the results obtained in Romania by Daniela andLucian (2005) Granger test showed a unidirectional causality
running from gross domestic product per capita to higher
education
Top Related