World Applied Sciences Journal 24 (3): 408-413, 2013

ISSN 1818-4952
IDOSI Publications, 2013
DOI: 10.5829/idosi.wasj.2013.24.03.1087

Impact of Human Capital Development on

Economic Growth of Pakistan: A Public Expenditure Approach

Maria Javed, Sadia Abbas, Alvina Fatima,

Muhammad Masood Azeem and Sania Zafar
Institute of Agricultural and Resource Eonomics,
University of Agriculture Faisalabad, Pakistan

Submitted: Jul 4, 2013; Accepted: Aug 10, 2013; Published: Aug 25, 2013
Abstract: This study was undertaken to ascertain the relationship between human capital development and
economic growth of Pakistan. Particularly, we were interested to find out the impacts of expenditures on
education and health on the growth performance of the country. We attempted to estimate the direction and
magnitude of the coefficients for both short run and long run. The co-integration and error correction
techniques were applied on the time series data ranging from 1978 to 2008. The results of the study indicate that
expenditures on health have positive and statistically significant effects on the economic growth rate in the
short run. On the other hand, expenditures on education have significantly positive long run impacts.
Moreover, primary school enrolment has positive while secondary school enrolment has negative impacts for
both short and long runs. The magnitude of the long run coefficients is higher than the short-run coefficients.
The results of the study suggest that there is a vast yet unfulfilled potential for Pakistan to move to higher
trajectory of growth by investing in people in terms of education and health.

Key words: Human capital Economic growth Education expenditures Health expenditures
Co-integration Error correction

INTRODUCTION important variable that determine growth [2]. The

aggregate production function now came to be
Human capital of a nation is understood from the represented as Y=F (K, L, H, A), where K is the capital, L
perspective of health, education and life expectancy of is the labour force, H is the human capital and A is the
the population. Education and health are closely related technology. This production function differs from the
components of the human capital that work together to neoclassical one in the sense that it incorporates both
make individuals more vigorous and productive. physical and human capital as two distinct factors of
Expenditures on any one of these reinforce the other. production [3]. This new growth theory was further
Improving the quality of education and health is not only extended to explain the determinants of technology itself.
an end itself; it also positively affects the future growth This came to be known as endogenous growth theory
prospects of a country. wherein H was represented by education and on job
Economists around the world keep on understanding trainings [4].
various determinants of economic growth. The Harrod- A paradigm shift in the focus of applied research
Domar growth model was based on the accumulation of resulted with the evolution in the concept of economic
physical capital stock which in turn depends on saving growth theory. Researchers started investigating the
rate [1]. With the emergence of new growth theory in relationship between human capital development and
1990s, the earlier emphasis on the accumulation of economic growth, for example, [5-10]. However, findings
physical capital shifted to the human capital as an of various studies on this issue are not always similar. For

Corresponding Author: Muhammad Masood Azeem, Institute of Agricultural and Resource Economics,
University of Agriculture, Faisalabad, Pakistan.
Cell # +61-469-164499.
408
 World Appl. Sci. J., 24 (3): 408-413, 2013

example, [8 and 11] found positive impacts of health MATERIAL AND METHODS
expenditures on the economic growth. In terms of
education, [12] found positive impacts on economic Secondary data, ranging from 1978 to 2008, were
growth in the case of Tanzania and Zambia. He, further, gathered from various issues of Economic Survey, Federal
concluded that these benefits can be maximized if heavy Bureau of Statistics and State Bank of Pakistan. Following
investment in physical capital is made. However, [13] [18] below is the final form of the model used in this
found negative impacts of public expenditure on study.
education and health on the economic growth. Similarly,
using cross national data [14] found that education LnGDP = 0+ 1 LnHEexp + 2 LnEDexp + 3 LnPE
contributed much less to economic growth than was + 4 LnSE+ 5 LnTI+ 6 LnLB+U t
expected. He gave three possible reasons for the cross
In this equation gross domestic product (GDP) is
country differences in the impacts of education on the
used as the proxy for economic growth. Public
economic growth. These are the differences in the
expenditures on health and education are given by HEexp
institutional environment, differences in the rate of growth
and EDexp, respectively. Primary and secondary school
of demand for labour and failure of schools in imparting
enrolments are represented by PE and SE, respectively.
skills to the students. Similarly, [15] also explains that
Total investment is denoted by TI and LB indicates labor
although education inculcates knowledge and skills which force in the above equation. All these variables are taken
raise productivity, yet it can also perpetuate societal in logarithmic form.
inequalities.
Above discussion highlight that the studies Empirical Consideration: Empirical considerations
regarding the impacts of human capital on the consist of following steps.
economic growth present mixed findings. These variations
in the results are due to the differences in socio-economic Unit Root Test: There are number of approaches to test
status of various countries. It is thus imperative to the unit root hypothesis but the Dicky-Fuller test is
investigate this issue in each individual country specific most commonly used [19-20]. The same was applied in
context. the present study to know the stationarity or non
What has been the direction and magnitudes of these stationarity of time series data.
impacts in the case of Pakistan? We attempt to answer
this question in the present study. By employing public Co-Integration: Johansens full information maximum
expenditures approach, we attempt to investigate both likelihood (FIML) approach is used for testing the co-
short and long run impacts of education on the economic integration [21-22]. This method allows the estimation of
growth of Pakistan. all possible co-integrating vectors and how they work
The state of public sector expenditures on education in the system. If the series is integrated of same order,
and health is not satisfactory in the case of Pakistan. The Johansens procedure can be used to test the presence of
trend of investment in education in terms of GDP has been a co-integrating vector among GDP, public expenditures
2.24, 2.50, 2.47, 2.10 and 2.05 percent from 2005-06 to 2009- on health and education, primary and secondary
10, respectively [16]. In cross country comparison, enrolments, total investment and labor force.
There are different steps to test co-integration. In the
Pakistans investment in education is lower than
first step, order of stationarity is determined. Variable
Bangladesh (2.6%), India (3.3%) and Vietnam (5.3%) in the
must be stationary at same level. Second step involves
year 2009-10. The status of health in Pakistan is also not
choosing the optimal lag length. To determine the lag
different from education. Public sector expenditures on
length VAR model is used. According to AIC criteria, we
health remained in the range of 0.5-0.8 % of GDP from
determined the lag length of one for the model used in this
1970-2010, [16]. study. Next step deals with determining the number of co-
Even though the public sector expenditures on integrating vectors. For this purpose, both trace statistic
education and health are low, yet marginal returns on and eigenvalue statistic were used in this study.
investment in both of these sectors may be higher [17].
We test this hypothesis in this research by quantitatively Error Correction Model: Error correction model indicates
estimating the impacts of education and health the speed of adjustment towards the long run equilibrium
expenditures on GDP growth of Pakistan. after a short run shock. According to [23] co-integrated

409
 World Appl. Sci. J., 24 (3): 408-413, 2013

variables must have an ECM representation. The major After selection of VAR with order 1, the second step
advantage of the ECM representation is that it avoids the in the Johnsens procedure was to test the presence and
problems of spurious correlation between dependent and number of co-integrating vector among the series of each
explanatory variables. model. For this purpose Maximal Eigen Value and Trace
tests were used and results are presented in table 3.
RESULTS AND DISCUSSION The results show, that the null is not rejected when
r=1. Similarly, the trace statistic did not reject the null
Unit Root Test Results: The results of second differenced hypothesis for the first time when r=1. There is one
ADF unit root test are given in table 1. The second co-integrating vector at 95 percent critical value because
differenced absolute values of test-statistics are well first statistical value of the trace test is greater than its
above the 95 percent value of their critical values. 95 percent critical value. The trace test indicates that
Therefore null hypothesis is rejected. This means that all there is one co-integrating vector in the concerned series.
variables are stationary after taking second difference. Therefore, this is the appropriate model for further
analysis.
Johansens Co-Integration Results: The first step in the If variables are co-integrated, then dynamic ECM
Johnsens procedure is the selection of order of Vector framework is an ideal basis for estimation of growth
Auto Regressive (VAR). Adjusted LR-test on the VAR response because it provides information about the speed
with a maximum of three lags was carried out. The results of adjustment to long-run equilibrium and avoids the
are presented in Table 2. The adjusted LR test does not spurious regression problem between the variables [23].
reject the order one because p-value is greater than 0.05. The coefficients in growth model (Table 4) represent
Therefore, the chosen order of VAR is one. Schwarz estimates of long run elasticities of GDP with respect to
Bayesian Criterion also has maximum value at order one. health expenditures, education expenditures, primary and
The AIC has maximum at third order but finally Adjusted secondary school enrolment. All these variables carry
LR-Test supports the result at order 1. expected signs except LSE which has negative sign.

Table 1: Second-differenced ADF unit root test results

Variables Trended Model Non-Trended Model Conclusion
LnGDP 5.02 4.47 Stationary
LnHE 5.55 2.99 Stationary
LnEE -4.61 -4.38 Stationary
LnPE -4.95 -4.94 Stationary
LnSE -3.91 -3.87 Stationary
Critical Values at 5 percent -3.58 -2.97
Note: critical values (95 percent confidence level) are taken from Dickey and Fuller (1979)

Table 2: LR-test on VAR with maximum of three lags

List of variables included in the unrestricted VAR
LGDP LHE LEE LPE LSE
List of deterministic and/or exogenous variables
LI LBF
Order AIC SBC Adjusted LR-test
3 157.25 100.63 ----------
2 128.09 88.13 42.54(0.016)*
1 139.27 115.09 53.40(0.345)
0 35.68 29.09 154.44(0.000)
Critical values -3.58 -2.97
AIC = Akaike information criterion
SBC = Schwarz Bayesian criterion
*values in parenthesis are p-values

410
 World Appl. Sci. J., 24 (3): 408-413, 2013

Table 3: Co-integration LR Test Based on Maximal Eigenvalue and Trace Test

List of variables included in the unrestricted VAR
LGDP LHE LEE LPE LSE
List of deterministic and/or exogenous variables
LI LBF
Null Alternative Statistic 95%CriticalValue 90%CriticalValue
Maximal Eigen value Test
r=0 r=1 52.943 34.4000 31.7300
r<= 1 r=2 22.661 28.2700 25.8000
r<= 2 r=3 11.616 22.0400 19.8600
r<= 3 r=4 6.419 15.8700 13.8100
r<= 4 r=5 4.722 9.1600 7.5300
Trace Test
r=0 r>= 1 98.355 75.9800 71.8100
r<= 1 r>= 2 45.420 53.4800 49.9500
r<= 2 r>= 3 22.758 34.8700 31.9300
r<= 3 r>= 4 11.142 20.1800 17.8800
r<= 4 r=5 4.722 9.1600 7.5300

Table 4: Johansens Normalized Estimates for Growth Model

List of variables included in the unrestricted VAR
LGDP LHE LEE LPE LSE
List of deterministic and/or exogenous variable
LI LBF
GDP = 7.543+ 0.198 LHE + 0.128 LEE+ 0.865 LPE+ -0.369LSE+7.543
(6.881)* (0.424) (2.389) (5.436) (2.184)
* t-ratios are in parenthesis.

Table 5: The Error Correction Model Estimates for GDP growth Table 5 reports the results of estimated coefficients
Variables Short-Run Long-Run for both short run and long run. The coefficient of real
Constant -0.0016(0.258)* 7.543(6.881) expenditure on health is consistent with a priori
DLHE 0.0414(4.028) 0.198(0.424) expectation implying a positive relationship between
DLEE 0.0459(1.642) 0.128(2.389) government expenditure on health and economic growth
DLPE 0.1810(1.825) 0.865(5.436) in long run. These results indicate that one percent
DLSE -0.1605(1.619) -0.369(2.184)
increase in health expenditures increases GDP growth by
LI -0.0424(3.505)
0.198 percent in the long run. In the short-run, value of the
coefficient is 0.041. However this is not statistically
LBF 0.1031(3.374)
significant. The values of the coefficient suggest that the
ECM1 -0.555(6.755)
magnitude of the impact on GDP is not too large in the
Diagnostic Tests
case of Pakistan. This could be because of the fact the
R2 0.73
health sector remains neglected in this country over
DW-statistics 1.97
several years. The results of present study are in line with
LM-test-x2 (1) 0.0911[.924]** [7 and 24].
RESET test-x2 (1) 13.545[.000] Real expenditures on education are found to have
Jarque-Bera positive relation with economic growth. The estimated
Normality-x2 (2) 2.651[0.266] long-run elasticity of education expenditures is 0.12. In
*Values in parenthesis are t-ratios. the short run, the elasticity is further low as 0.045. These
**Values in square brackets are p-values. results are consistent with the findings of [18 and 25].

411
 World Appl. Sci. J., 24 (3): 408-413, 2013

The short run and long run elasticities of primary school 3. Garbowski, R., S. Self and M.P. Shields, 2007.
enrolment found to be 0.180 and 0.865, respectively. Economic Development: A Regional, Institutional
These results are similar to [26]. The elasticity of and Historical Approach. Prentice Hall of India, New
secondary school enrolment was found to be -0.160 in Delhi.
short run and -0.369 in long run, which implies negative 4. Romer, P.M., 1990. Endogenous Technological
relation between secondary school enrolment and Change. The Journal of Political Economy.
economic growth. The study of [27] also concludes that 98(5): 71-102.
primary education contributes more than the secondary 5. Psacharopoulos, G., 1994. Returns to Investment in
education. Education: A globe Update. World Development.
Our results indicate that public sector 22(9): 1325-1345.
expenditures on both education and health contribute 6. Imran, M., S. Bano, M. Azeem, Y. Mehmood and
positively to economic growth of Pakistan. Although, A. Ali, 2012. Relationship between Human Capital
the magnitude of the coefficient is not very high in and Economic Growth: Use of Cointegration
both education and health expenditures, yet positive Approach. J. Agric. Soc. Sci., 8(4): 135-138.
and statistically significant results mean that the 7. Aurangzeb, 2003. Relationship between Health
country can reap benefits by investing in human Expenditure and GDP in an Augmented Solow
capital. It is also worth mentioning that the coefficients Growth Model for Pakistan: An Application of Co-
of long run elasticities are higher as compared to the integration and Error-Correction Modeling. The
short-run. This is understandable because returns from Lahore Journal of Economics, 8(2): 1-18.
investment in human capital are not immediate. The 8. Bloom, D.E., D. Canning and J. Sevilla, 2001. The
Effect of Health on Economic Growth: Theory and
results of this study also support the findings of [12] who
Evidence. World Development, 32(1): 1-13.
maintain that investment in education can give maximum
9. Samimi, A.J., M. Madadi and N. Heydarizadeh, 2012.
benefits only when combined with heavy physical
An Estimation of Human Capital Share in Economic
investment.
Growth of Iran Using Growth Accounting Approach.
Middle-East Journal of Scientific Research,
CONCLUSION
11(1): 90-93
10. Khilji, B.A., Z.K. Kakar and S. Subhan. 2012. Impact
The main objective of this paper was to investigate
of Vocational Training and Skill Development on
the relationship between government expenditure on
Economic Growth in Pakistan. World Applied
education and health and economic growth of Pakistan.
Sciences Journal, 17(10): 1298-1302.
To attain that objective, co-integration and error
11. Bose, N. and M.E. Haque, et al. 2007. Public
correction techniques were used. The results of this study
Expenditure and Economic Growth: A Disaggregated
confirm that investment in health and education
Analysis for Developing Countries. The Manchester
contributes significantly and positively in the GDP growth School., 75(5): 533-556.
of Pakistan. As the results are statistically significant 12. Jung, H.S. and E. Thorbecke, 2003. The Impact of
mainly in the long run, we propose that long term Public Education Expenditure on Human Capital,
investments in these sectors should be made to acquire Growth and Poverty in Tanzania and Zambia: A
maximum benefits in the future. It is also suggested that General Equilibrium Approach. Journal of Policy
both quality and quantity of health services and Modeling, 25(8): 701-725.
education should be improved so that their effect on the 13. Devarajan, S. and V. Swaroop, et al. 1996. The
overall economic performance of the nation could be Composition of Public Expenditure and Economic
magnified. Growth. Journal of Monetary Economics,
37(2): 313-344.
REFERENCES 14. Pritchett, L, 2001. Where has all the education
gone? The World Bank Economic Review.,
1. Hussain, I., 2007. Pakistan: The Economy of an Elitist 15(3): 367-391.
State. Oxford University Press. 15. Todaro, M.P. and S.C. Smith, 2000. Human Capital:
2. Saeed, K.A, 2007. The Economy of Pakistan. Oxford Education and Health in Economic Development.
University Press. Economic Development. Pearson Education, Inc.

412
 World Appl. Sci. J., 24 (3): 408-413, 2013

16. Government of Pakistan (GOP). 2010. Economic 23. Engle, R.F. and C.W.J. Granger, 1987. Cointegration
Survey of Pakistan. Ministry of Finance, Islamabad. and Error-correction: Representation, Estimation and
17. Akram, M. and F.J. Khan, 2007. Health Care Services Testing. Econometrica, 55(2): 251-276.
and Government Spending in Pakistan. Pakistan 24. Siddiqui, R. and M.A. Mahmood, 1994. The
Institute of Development Economics. Working Paper Determinants of Health Status: A Cross-Country
pp: 32. Analysis. The Pakistan Development Review.
18. Oluwatoyin, M, 2011. Human Capital Investment and 33(4): 745-758.
Economic Growth in Nigeria: The Role of Education 25. Musila, W.J. and W., Belassi. 2004. The Impact of
and Health. Manager., 14(2011): 266-277. Education Expenditures on Economic Growth in
19. Dickey, D.A. and W.A. Fuller, 1979. Distribution of Uganda: Evidence from Time Series Data. Journal of
the Estimators for Autoregressive Time Series with a Developing Areas., 32(1): 124-133.
Unit Root. Journal of American Statistical 26. Gupta, S., M. Verhoeven and E. Tiongson, 2002. The
Association, 74(366a): 427-431. Effectiveness of Government Spending on Education
20. Dickey, D.A. and W.A. Fuller, 1981. The Likelihood and Health Care in Developing Economies. European
Ratio Statistics for Autoregressive Time Series with Journal of Political Economy, 18(4): 717-737.
a Unit Root. Econometrica, 49(4): 1057-1072. 27. Papageoriou, C., 2003. Distinguishing Between the
21. Johansen, S., 1988. Statistical Analysis of Co- Effects of Primary and Post-Primary Education on
Integrating Vectors. Journal of Economic Dynamic Economic Growth. Review of Development
and Control., 12(2): 231-254. Economics, 7(4): 622-635.
22. Johansen, S. and K. Juselius, 1990. Maximum
Likelihood Estimation and Inferences on Co-
Integration with Application to the Demand for
Money. Oxford Bulletin of Econometrics and
Statistics, 52(2): 169-210.

413