Documente Academic
Documente Profesional
Documente Cultură
INTRODUCATION
The rationalization of higher government expenditure on basic education is often
based on its impact on individual life time earning i.e. social rate of return. Different
studies indicated that social return for primary education is higher than secondary and
tertiary education but expenditure on tertiary education is inappropriately high in most
of the countries (Gupta et al.1999).
Higher budget allocation for primary health care is justified on the basis that such
expenditures ameliorates the impact of diseases on productive years of people. Many
studies suggested that burden of disease could be minimize in developing countries if
government ensure the availability of basic and cost effective health services for all
population(World Bank1993). Preventive measures from diseases are more cost
effective but in developing countries mostly resources are allocated for curative
services (Sahn et al.1993; Pradhan 1996).
Impact of public spending on education attainment and basic health care is
inconclusive. It is possible that public spending on education and health crowed out
the private spending, or government resources are used inefficiently and inequitably.
Infant mortality rate, child mortality and life expectancy are used by many researchers
as a proxy for health care, likewise for education attainment, primary, secondary and
tertiary school enrollment are used as an indicators. Beneficial impact of sufficient
resource allocation on health and educational outcomes are mixed according to the
social, political and economic conditions of the country.
Health is vital elements of human capital. A healthier worker can contribute more in
the production process than his unhealthy counterpart. There are several channels
that define the contribution of health in production and output. For a given level
of all other factors, the economy can produce higher output if it has higher
levels of health. Health is an important factor for determining the level of
returns from education. Improvement in health increases output due to increased
strength and also due to more learning from a given level of education.
The relationship between health care expenditure and health status has received some
attention in developing regions. At the country level, Akinkugbe and Mohano (2004)
performed time series analysis using the error correction model (ECM) and found that
in addition to public health care expenditure, the availability of physicians, female
literacy and child immunization significantly influenced health outcomes in Lesotho.
At the regional level, (Anyanwu and Erhijakpor, 2007) in a panel data analysis and
using a fixed effect model found that total health expenditures are a significant
contributor to health outcomes with a 10 percent increase in total health care
expenditure per capita resulting in21 percent and 22 percent decrease in under-five
and infant mortality rates respectively. Similarly (Rajkumar and Swaroop 2008;
Craigwell et al. 2012) confirm the positive impact of government spending on infant
mortality rate, child mortality rate and life expectancy.
;Howitt , 2005). In most of the studies education or health related indicators are
employed as a proxy for human capital. Studies undertaken on both developed and
developing countries have indicated that efficient and sufficient government resource
sectors of the Pakistans economy. This study therefore attempts to analyze the
discussion on the role of government expenditure in education and health care in
Pakistan.
1.1 OBJECTIVES OF THE STUDY
To investigate the effects of the public education expenditures on primary and
secondary school enrolment in Pakistan for period of 1980 to 2012.
To determine the effect of government expenditure on health status measured
by infant mortality rate and child mortality rate
1980-2012.
1.2 SIGNFICANE OF THE STUDY
To the best of my knowledge, a very few studies have been done to investigate the
impact of government expenditure on education and health care in Pakistan with these
variables. The significance of the study is to bridge this knowledge gape and fill this
empty field of research in Pakistan for policy implication
Chapter: 2
LITERATUR REVIEW
5
2.1 INTRODUCATION
This chapter focuses on the previous views of the researchers about the impact of
government expenditure on health status and education attainment. Previous studies
used different proxies to measure health care, like life expectancy, infant mortality,
under five mortality rate and maternal mortality rate. Similarly primary school
enrollment, secondary and tertiary school enrolment is used as a proxy for education
attainment by many researchers.
2.2 REVIEW OF PREVIOUS LITERATURE
Without use of empirics (Schultz, 1961) argues that human capital has been the basis
of the faster growth in Western countries. So investment in direct expenditure on
health is necessary to achieve economic growth through increase in level of
productivity. According to (Schultz ,1961) access to education plays a very important
role in equipping persons with opportunities that shape their character and develop
their personal, economic, socia and cultural status. This is demonstrated by
educations progressive impudence on health; income, family structure and political
participation
Using the sample of 40 countries for the years 1985-1990 (Carrin and Politi, 1995)
analyze the impact of poverty reduction and government health expenditure on health
care in developing countries. Dependent variable, Health status is measured by life
expectancy, infant mortality and under-ve mortality. Public health expenditure to
gross national product ratio, incidence of total absolute poverty and per capita income
is used as explanatory variables. Study concluded that per capita income and
reduction of poverty have significantly positive impact on health status while Public
health expenditure is found to be statistically insignicant in regression analysis.
To establish the link between per capita income and several indicators of educational
development (Mingat and Tan, 1998) use the large sample of 125 developing and
developed countries for year 1993.results indicate that per capita income have greater
influence on literacy rate then public spending on education.
Using cross-sectional data of 98 developing countries, (Filmer and Pritchett, 1999)
examine the impact of government health expenditure on infant and under-5 mortality
rate. Authors find
spending over the period of 1992/3. They suggested that 95% of the variation in infant
and child mortality is explained by income inequality, income per capita, female
literacy and ethnic fractionalization.
To support the evidence that government expenditure positively influence health and
education indicators (Tiongson et al .1999) employ 2SLS for cross section data of 50
developing and transition economies. Study confirms that education investment
increase school enrollment and health expenditure reduce the infant child mortality
rate.
(Mello et al .2003) investigate the social outcomes of health and education
expenditure for 94 developing countries in the period of 199698.Findings of the
study show that public spending is major determinant of social outcomes in education
sector particularly but not in health sector.
To show the effectiveness of government spending on education (Baldacci et al .2003)
uses a panel data of 94 developing countries. By employing covariance structure
model for the period 1996 to 1998 empirical findings reveal that government spending
on education alone does not advance social outcomes. Gender inequality deteriorates
social outcomes so government needs to remove these unfavorable social conditions
along with increase in public spending to accelerate human development.
(Roberts ,2003) did comprehensive global survey of the literature on the determinants
of education in developing countries, findings of the study suggested that despite the
fact that developing countries need to assign more resources to primary education,
they also
simultaneously. Although since 1970 developing countries have been spending more
(relative to GDP) on education, Roberts examine that education expenditure has no
strong relationship with primary school enrolment.
For the fifteen states of India (Kaur and Misra 2003) have done empirical analysis to
analyze the impact of public expenditure on primary Intermediate, and secondary
school enrollment rates. Regression analysis for the period of 1985-86 and 2000-01
point out that government expenditure on education is effective especially in poorer
states. Study also reveals that government expenditure has a greater outcome in
primary education than secondary. The authors Hypothesize that private funding plays
a greater role in secondary education therefore role of public spending decreases at
higher stages of education.
increase the school enrollment however public spending are inefficient in countries
with poor governance.
(Rajkumar and Swaroop ,2008) used annual data of 1990, 1997 and 2003 for 91
developed and developing countries to find out the impact of public spending on
health status. By employing Ordinary least square regression on cross-section data,
results show that public expenditure on health is inversely related with child mortality
in countries with high quality of bureaucracy, good governance and low corruption
levels. Similarly, government expenditure on education is more effective to increase
primary school enrollment in countries with good governance.
(Mello and Pisu ,2009) explore the impact of government expenditure on health and
education outcomes by combining data of census, household survey and budget of 4
000 Brazilian municipalities for year 2000.By employing two stage least square
(2SLS) findings of the study suggest that education expenditure increase the
education outcome, but on the other hand health expenditures are ineffective.
By utilizing the primary data of 115 districts across three states in India( Iyer and
Tarozzi, 2009) investigate the effectiveness of public spending in education. They
employ fix effect model and concluded that government expenditure on education
have negligible impact on primary enrollment.
10
ensure long run growth, special attention should be given to health and education
sector.
For
conducted a very
on child mortality
over the period 1986 to 2007. As compare to other studies, impact of improved water
and sanitation is smaller but still positive on reduction of mortality. The authors also
find that the positive result of clean water is slighter and affect only children between
1 and 12 months.
11
1980 to 2009 study concluded that health expenditure has significant positive
outcomes while education spending have slight impact on school enrollment.
By using the data set of 177 countries (Obrizan and Wehby, 2012) examined the
influence of health expenditure on life expectancy. Results of regression analysis
show that longevity and public health expenditure have direct relationship.
(Ijaz, 2012) analyzes the impact of female literacy rate in 35 districts of Punjab
Pakistan. By simple regression analyses it is concluded that female literacy rate has no
significant impact o reducing the child mortality in Punjab while male literacy rate is
effective in year 2007-2008.it is also suggested that quality of service delivered and
presence of better institutions are the major factors to decrease the infant mortality
rate.
Improved water and sanitation access are key strategies to reduce child and maternal
mortality. (Cheng et al, 2012) abstracted the data of 193 countries from global data
base and linear regression analysis was used for the outcomes. Results suggested that
both clean water and sanitation negatively influence the infant and maternal deaths.
(Gitau , 2012) investigate the impact of health aid expenditure on child mortality over
the period of 1980 and 2010 for Kenya. They employ semi log regression analysis on
the Model and later an Error- Correction methodology on time series data of thirty
year. Results of the study reveal that immunization coverage and health aid
expenditure negatively impact the under five mortality in Kenya.
12
for analysis. They suggested insignificant relationship between health expenditure and
childhood mortality rate while per capita income, female literacy rate and poverty
have significant impact on reduction of mortality rate in India and EAG states.
One of the prime benefits of educating women is healthier children. (Shetty and
Shetty, 2014) found the inverse relationship between female literacy rate and infant
mortality rate in India. Data was collected for 28 Indian states for year 1981 to 2001.
States which have high female literacy rate front with lower infant mortality so
government should encourage female education in India.
Manoux et al used the data of 26 states of India over the period of 1998-1999 to
explore the relationship of adult education, cast, wealth and urbanization with child
mortality. By utilizing a two-level multilevel logistic regression model they suggested
that adult education decrease the child mortality but household wealth and
urbanization have no significant relation with mortality rate in India.
2.3 CONCLUSION OF PREVIOUS LITERATURE
Previous findings of the studies show that impact of government expenditure on
education and health care is mixed. Some researchers concluded the positive
outcomes of health and educational expenditures done by government and some
studied suggested the insignificant and negligible impact of government spending.
Therefore impact of government spending can be different according to the economic,
political and environmental conditions of the country
Table: 2.1 SUMMERY OF LITERATURE REVIEW
Author
Schultz
Year
1961
Key findings
Investment in direct
expenditure of health
13
1995
1998
1999
is
necessary
to
achieve
economic
growth
through
increase in level of
productivity.
Access to education
plays
a
very
important role in
equipping
persons
with
opportunities
that
shape
their
character and develop
their
personal,
economic, socia and
cultural status.
Per capita income and
reduction of poverty
have
significantly
positive impact on
health status.
Public
health
expenditure is found
to be statistically
insignicant
in
regression analysis.
Per capita income has
greater influence on
school
enrollment
then public spending
on education.
Very
small
and
insignificant impact
of
government
spending on infant
and child mortality
rate.
95% of the variations
are
explained
income
by
inequality,
Year
14
Mello et al
2003
Public spending is
major determinant of
social outcomes in
education
sector
particularly but not in
health sector.
Baldacci et al
2003
Government needs to
remove unfavorable
social
conditions
along with increase in
public spending to
accelerate
human
development.
Roberts
2003
Education
expenditure has not
strong relation with
primary
Gupta et al
2003
2004
15
school
enrolment.
Government spending
on
education
is
effective especially in
poorer states.
Government spending
has a greater outcome
in primary education
than secondary.
To accelerate the
economic
growth,
government need to
assign recourses for
education efficiently.
Author
Year
2007
Key findings
2007
2007
Baldacci et al
2008
2008
16
Life expectancy is
positively influenced
by
increment
in
health expenditure.
Literacy
rate,
urbanization rate and
per capita calorie
availability also add
to health status, while
per capita carbon
dioxide
emissions
negatively impact the
health status.
Health
expenditure
reduce the infant
mortality and under
five mortality.
Female literacy and
higher number of
physicians
are
inversely related with
health outcomes.
Significantly positive
relationship between
public expenditure on
education and school
enrollment.
Government spending
increase the school
enrollment however
public spending is
inefficient
in
countries with poor
governance.
Public expenditure on
health is inversely
related with child
mortality in countries
with high quality of
Author
Year
2009
bureaucracy,
good
governance and low
corruption levels.
Government spending
on education is more
effective to increase
primary
school
enrollment.
Key findings
Education
expenditure increases
the
education
outcomes.
Health expenditures
are ineffective to get
desired results.
2009
Government spending
on education has
negligible impact on
primary enrollment.
Pueyo et al
2009
Life expectancy is
positively influenced
by public health
spending.
Abhijeet
2010
Economic
growth
contributes to the
government spending
on
education
irrespective of any
lag
effect
but
investment
in
education accelerate
the economic growth
after some time leg.
Craigwell et al (2012)
2012
17
education spending
has slight impact on
school enrollment.
Obrizan and Wehby
2012
Author
Year
Ijaz
2012
Key findings
and
presence
of
institutions
major
better
are
the
factors
to
2012
Gitau
2012
Immunization
coverage and health
aid
expenditure
negatively impact the
infant
and
child
mortality rate.
Kaushal et al
2013
Insignificant
relationship between
health
expenditure
and
childhood
18
2014
mortality rate.
per capita income,
female literacy rate
and poverty have
significant impact on
reduction of infant
and child mortality
rate
High female literacy
rate front with lower
infant mortality rate.
Chapter: 3
19
Data is collected from the World Bank, The United Nations Educational Scientic and
Cultural Organization (UNESCO) database, State bank of Pakistan, Federal Bureau of
Statistics Government of Pakistan, world Development indicator
and WHO.
Ordinary least square for health model and ARDL approach for education model are
used to statistically analyze the data. An E view 6 is used for estimation in present
study.
Female
literacy rate
Government
expenditure on
health
Income per
capita
Carbon dioxide
emission
Access to sanitation
and clean water
Immunization
DPT3 and
measles
20
DEPENDENT VARIABLE
Infant mortality rate and under five mortality rate.
INDEPENDENT VARIABLES
Government expenditure on health, per capita income, female literacy rate,
immunization DPT3 and measles, carbon dioxide emission, access to improved
sanitation and clean water, urban population.
t = 19802012.
Where:
Htj = is health care, proxied by infant mortality and under five mortality rate.
Xt = is a vector of investment variables comprising of public expenditure spent on
health, income per capita and female literacy rate.
Zt = is a vector of accessibility indicators composed of urban population as a percent
of total population, Carbon dioxide emissions and percent of population with access
to sanitation facilities and clean water sources.
21
product (GDP).
Female literacy rate is the percentages of females ages 15 and above who can,
with understanding, read and write a short, simple statement on their
INVESTMENT VARIABLES
In terms of the a priori signs of the explanatory variables, many studies have indicated
that government spending on health care is pertinent for health enhancement and
human development (especially for those who have lower incomes) and consequently
22
economic growth (Schultz, 1961; Anand and Ravallion, 1993; Swaroop, 1996; Gupta
et al., 2004). Therefore, it is expected to reduce infant and child mortality rate.
Income per capita measured by gross domestic product per capita (purchasing power
parity) suggests that as household income increases, a countrys health position should
improve. If people have more disposable income then they will have the capacity to
personally invest more in health care and caloric intake per capita may increase which
improves health status (Greenidge and Stanford, 2007). Thus, a priori the coefficient
on income per capita is negative.
For female literacy rate, many studies show that a negative relationship exists
between female literacy and infant and child mortality rate. ).Female literacy reduces
the infant mortality by allowing them to read and understand the necessary
information for healthy living. As suggested by (Schultz, 1993; Ijaz, 2012) that one
prime benefit of educating women is healthier children.
b)
ACCESSIBILITY VARIABLES
With respect to the accessibility variables, increased access to sanitation facilities and
water creates a more salubrious environment thus improving health status (Gupta et
al., 2004). Deprived access to sanitation and water promote the spread of health
problems like hepatitis and diarrheal diseases like cholera and a weakened immune
system (World Health Organization (WHO, 2011)). Evidence has suggested that
water-poor and sanitation facility deprived communities are typically simultaneously
economically poor. This variable is expected to be negatively related to infant and
under five mortality rates.
23
With respect to urbanization, dened as the percent of the entire population existing in
urban areas, it is believed that in such areas access to health facilities is much easier
than rural areas (Greenidge and Stanford, 2007) and related to improved health status
(Schultz, 1993). Though, Thornton (2002) states that urban areas are characteristically
polluted with carbon dioxide emissions (metric tons per capita) and thus have positive
impact on health indicators measured as infant and child mortality rate. Consequently,
the relationship between urbanization and health expectancy depend on the overall
effect of pollution. On other aspect urbanization expected to reduce the infant and
under five mortality rate and carbon dioxide emissions increase.
c)
IMMUNIZATION VECTOR
24
Government
expenditure on
education
School aged
populatio
Income per
capita
Adult literacy
rate
Pupil teacher
ratio
25
DEPENDENT VARIABLES
Primary and secondary school enrollment
INDEPENDENT VARIABLES
Government expenditure on education, Income per capita, adult literacy rate, School
aged population, Pupil teacher ratio and urban population.
3.11 MAIN HYPOTHESIS FOR EDUCATION MODEL
H0: Government educational expenditure has significantly positive impact on primary
and secondary school enrollment
3.12 ECONOMETRIC SPECIFICATION OF EDUCATION MODELS
The education equation is modeled as follows:
Etj= tj+ 1Xtj + 2Zt+ 3Ytj + 4Atj+tj
t = 19802012.
Where
Etj = is education attainment for j enrollment where j is primary and secondary
percentage gross school enrollment, respectively.
Xt j= represents a vector of investment variables consisting of public expenditure
spent on education as a percentage of GDP, income per capita ,per pupil public
spending and adult literacy.
Zt = is an accessibility indicator measured by urban population as a percent of total
population.
26
27
28
increase in public education expenditure. Additionally, (Mingat and Tan ,1998) found
that a reduction in this variable has a small impact on student learning and has a long
run effect of lowering levels of education attainment levels. It is expensive and
difficult to increase enrollment rates when the population is relatively young (Mingat
and Tan, 1992). (Gupta et al .2002) claim that a high incidence of young people
(population aged 5-14) should have a negative a coefficient.
3.15 BOUND TESTING APPROACH:
The use of the bounds technique is based on three validations. First, Pesaran et al.
(2001) advocated the use of the ARDL model for the estimation of level relationships
because the model suggests that once the order of the ARDL has been recognised, the
relationship can be estimated by OLS.
Second, the bounds test allows a mixture of I (1) and I (0) variables as independent,
the order of integration may not necessarily be the same. Third, this technique is
suitable for small or finite sample size (Pesaran et al., 2001).
Following Pesaran et al. (2001), we assemble the vector auto regression (VAR) of
order p, denoted VAR (p), for the following growth function:
p
Z t i z t i t
i 1
...................................... (1)
xt
defined as school enrolment primary and secondary ,
represents a set of explanatory variables i.e. per capita income ,Adult literacy rate ,
29
school aged population, public spending on education, pupil teacher ratio primary and
secondary and urban population, and t is a time or trend variable. According to
yt
Pesaran et al. (2001),
xt
must be I(1) variable, but the regressor
p 1
i 1
i 1
z t t z t 1 t y t i t xt i t
where
................................. (2)
as:
YY YX
XY XX
The diagonal elements of the matrix are unrestricted, so the selected series can be
YY 0
YY 0
, then Y is I(0).
The VECM procedures described above are imperative in the testing of at most one co
yt
integrating vector between dependent variable
xt
and a set of regressors
. To
derive model, we followed the postulations made by Pesaran et al. (2001) in Case III,
that is, unrestricted intercepts and no trends. After imposing the restrictions
YY 0, 0
and
30
10 ( PCI )ti
i=0
i=0
i=1
11 (PSE)t i
i=0
13 ( SAP)t i
8 ( SE)jt i
9 ( ALR)ti
i=0
12 ( PTR) jti
i=0
14 (UP )ti
i=0
(1)
Where
is a white-noise disturbance
term.
SE
ALR
PCI
PSE
31
PTR
SAP
UP
Urban population
Equation (1) also can be viewed as an ARDL of order (p, q, r). Equation (1) indicates
that education tends to be influenced and explained by its past values. The structural
lags are established by using minimum Akaikes information criteria (AIC). From the
estimation of UECMs, the long-run elasticises are the coefficient of one lagged
explanatory variable (multiplied by a negative sign) divided by the coefficient of one
lagged dependent variable (Bardsen, 1989). For example, in equation (3), the long-run
2 / 1
3 / 1
) and (
) respectively.
The short-run effects are captured by the coefficients of the first-differenced variables
in equation (3).
After regression of Equation (1), the Wald test (F-statistic) was computed to
differentiate the long-run relationship between the concerned variables. The Wald test
can be carry out by imposing restrictions on the estimated long-run coefficients of
school enrolment, Adult literacy rate, Per capita income, public spending on
education, Pupil teacher ratio in primary and secondary schools, school age
population and urban population.
H0: 1 =2 =3 =4 = 5 =6 = 7 = 0
32
The computed F-statistic value will be evaluated with the critical values tabulated in
Table CI (iii) of Pesaran et al. (2001). According to these authors, the lower bound
xt
critical values assumed that the explanatory variables
xt
or I(0), while the upper bound critical values assumed that
one, or I(1). Therefore, if the computed F-statistic is smaller than the lower bound
value, then the null hypothesis is not rejected and we conclude that there is no longrun relationship between school attainment and its determinants. Conversely, if the
computed F-statistic is greater than the upper bound value, then school attainment and
its determinants share a long-run level relationship. On the other hand, if the
computed F-statistic falls between the lower and upper bound values, then the results
are inconclusive.
Chapter: 4
33
4.1 INTORDUCATION
We used ordinary least square for health models and ARDL approach for education
models based on the previous studies as discussed in methodology section. Results of
the estimation by e views are discussed in this chapter.
4.2 UNIT ROOT TEST FOR HEALTH MODELS
Application of conventional econometric methods for estimation of coefficients by
using time series data is based on assumption that the model variables are stationary.
A time series variable is stationary only if its mean value, variance and correlation
coefficients remain constant through the time. If time series variables used in
estimation of coefficients are non-stationary, then its R square coefficient may be of a
high value and can cause an incorrect understanding about level of relation between
variables although there may be no significant relation between variables Econometric
software e-views6 was used for estimation of this study. To check the order of
integration, standard Augmented Dickey-Fuller (ADF) unit root test was exercised for
all the variables included in the study.
Constant
Pci
3.906
Level
1st difference
Constant linear Constant
Constant linear Decision
trend
1.256
(-3.455)**
trend
-5.019
Stationery
at
difference
CO2
1.000
-0.881
0.999
-2.572
0.016
(-6.917)***
34
0.001
-6.884
Stationery
at
difference
0.781
-1.695
Dpt3
0.294
-3.120
0.000
(-4.671)***
0.000
-5.029
Stationery
at
difference
0.423
-0.466
Flr
0.118
-1.723
0.000
(-6.753)***
0.001
-6.859
Stationery
at
difference
0.885
-2.190
Im
0.717
-2.462
0.000
(-5.075)***
0.000
-5.348
Stationery
at
difference
Imr
0.213
-2.029
0.343
(-2.757)***
0.000
-6.847
0.000
-8.149
Stationery at level
Mru5
0.273
(-3.331)**
0.007
-2.477
0.000
-1.499
0.000
-1.901
Stationery at level
Psh
0.022
-0.408
0.335
-2.269
0.518
(-5.054)***
0.628
-5.224
Stationery
at
difference
0.896
1.481
Up
0.437
0.283
0.000
(-5.657)***
0.001
-6.759
Stationery
at
difference
0.998
0.997
0.000
0.000
Note: *, ** & *** indicate the rejection of the null hypothesis of non-stationary at
10%, 5% and 1% significant level, respectively.
The results are reported in Table 4.1. Based on the ADF test statistic, it was initiate
that out of nine variables, seven have unit root i.e. PCI,CO2, DPT3,FLR,IM,PSH, UP
and stationary at first difference, while our dependent variables IMR,MRU5 is I(0).
These results imply that OLS provides consistent estimate for health models.
CO2
DPT3
FLR
IM
IMR
ASF
ACW
MRU5
PCI
PSH
123.04
551.595
0.71
0.69
50.96
34.00
52.60
95.22
34.51
S
81.77
35
Median
0.70
54
32.8
52
95.1
34.3
87.1
122.8
453.494
0.73
Maximum
8
0.96
86
48
83
121.3
49
92
160.4
1290.36
1.19
Minimum
0.40
19.6
69.3
19.3
35
85.9
296.179
0.23
Std. Dev.
0
0.17
24.09
7.651
22.39
16.36
9.083
13.86
23.487
269.890
0.20
33
33
Observatio 33
33
33
1
33
33
33
33
ns
Heteroskedasticity
Test, Breusch- Godfrey serial correlation LM test and Ramsey RESET specification
test . All the tests disclosed that the model has the aspiration econometric properties, it
has a correct functional form and the models residuals are normally distributed,
homoskedastic and serially uncorrelated. Therefore results reported by OLS are valid
for reliable interpretation.
Test-stats
p-values
Heteroskedasticity Test
0.724
0.681
Normality test
0.923
0.630
2.592
0.125
1.926
0.176
36
33
Test
Test-stats
p-values
Heteroskedasticity Test
1.434
0.233
Normality test
0.832
0.659
0.415
0.527
1.780
0.198
Coefficient
Standard
T statistic
Probability
error
(ACWS)
(-0.043)**
-2.341
-2.341
0.029
(ASF)
(-1.998)**
0.945
-2.115
0.046
(PCI)
(-0.488)
0.297
-1.642
0.115
(FLR)
(-2.498)**
1.175
-2.125
0.048
(-0.212)**
0.100
-2.116
0.046
(IM)
(PSH)
(-0.194)***
0.042
(DPT3)
(-0.096)
0.124
37
0.000
-4.613
-0.772
0.450
(CO2)
(1.156)**
0.499
0.030
2.315
MA(2)
0.922
0.096
0.000
9.541
Durbin-Watson
R-squared
Adjusted
F-statistic
15.525
Prob(F-
0.000
stat
1.969
0.916
R-squared
0.877
statistic)
Explanation
Access to clean water sources
Access to sanitation facility
Per capita income
Female literacy rate
Immunization measles
Public spending on health
Vaccination against diphtheria, pertussis and tetanus
Carbon dioxide emissions
38
Our Results are consistent with the previous literature and signs of the coefficients are
similar as expected. Coefficient value of improved water source indicates that 1%
increase in population access with clean water decrease the infant mortality by
0.043%, likewise improved sanitation facility reduce the infant mortality by 1.99 %
in Pakistan. These results are similar as (Kim and moody, 1992; hojman, 1996; Cheng
et al .2012; Fink et al .2011). Better access to sanitation facilities and clean water
creates a more hygienic environment thus improving health status (Gupta et al.,
2004). According to WHO ,2011 Deprived access to sanitation and clean water
endorse the spread of health problems like hepatitis ,cholera and a weakened immune
system .Almost one tenth of the global disease burden could be prevented by
improving water supply, sanitation, hygiene and management of water resources.
Worldwide, 1.4 million children die each year from preventable diarrheal diseases and
some 88% of diarrhea cases are related to unsafe water, inadequate sanitation or
insufficient hygiene.
Female literacy rate lessen the rate of infant mortality by 2.49%. As suggested by
(Ijaz, 2012; Schultz, 1993) one prime benefit of educating women is healthier
children. Improvement in literacy status of women results in a downward trend in
infant mortality rate, (Shetty and Shetty, 2014).female literacy reduce the infant
mortality by allowing them to read and understand the necessary information for
healthy living. They always try to bring up their children in hygienic conditions and
39
Increase in immunization measles lessens the rate of infant mortality by 0.21% while
co2 emissions per capita positively impact the infant mortality. Research from other
studies has demonstrated substantial reductions in mortality associated with measles
immunization programs (Aaby, 1995; Koenig, Fauveau and Wojtyniak, 1991). We
therefore considered it important to assess the independent contribution of measles
immunization to survival in this population. Likewise polluted environment can affect
the respiratory system of the human and cause many diseases so clean environment
is essential for infant survival in Pakistan.
40
In case of per capita income we can say that PCI shows the average income of the
country and in presence of huge income disparities it cannot be significant
determinate. Most of the infant deaths occurred in ruler areas of Pakistan where
income level of the people is less than average income shown by the PCI so it is not
considerable to show variation in infant mortality rate.
Variables
Coefficient
Standard
T statistic
Probability
(IWS)
(-0.074)**
error
0.032
-2.283
0.033
(ISF)
(-0.067)*
0.034
-1.932
0.067
(PCI)
(-0.615)**
0.290
-2.119
0.046
(FLR)
(-2.786)**
1.325
-2.102
0.050
(-0.240)**
0.112
-2.137
0.044
(-0.172)***
0.065
-2.612
0.016
-1.038
0.314
2.281
0.033
257.618
0.000
(IM)
(PSH)
(DPT3)
(-0.145)
0.139
(CO2)
(1.285)**
0.563
AR(1)
0.999
0.0038
41
1.965
Adjusted
R-squared
Prob(F-statistic)
0.000
0.905
42
null and alternative hypothesis while the rejection of the null hypothesis is based on
MacKinnon (1996) critical values.
0.998
0.997
0.000
43
0.000
Note: *, ** & *** indicate the rejection of the null hypothesis of non-stationary at
10%, 5% and 1% significant level, respectively.
Null hypothesis
44
Coefficient
Standard error
T statistic
Probability
(0.309)**
0.131
2.357
0.027
LOG(SEP(-1))
(-0.508)***
0.126524
-4.017
0.000
LOG(ALR(-1))
(0.393)***
0.123
3.185
0.004
LOG(PCI(-1))
LOG(PSE(-1))
(0.093)**
(0.095)***
0.043
0.025
2.177
3.665
0.040
0.001
(-0.162)
0.148
-1.089
0.287
LOG(SAP(-1))
(-0.479)***
0.134
3.564
0.001
LOG(UP(-1))
DLOG(ALR(-1))
(1.850)***
0.626
2.953
0.007
(0.427)**
0.177
2.403
0.024
(0.079)
0.049
1.612
0.120
(0.099)***
0.030
3.309
0.003
LOG(PTRP(-1))
DLOG(PCI(-1))
DLOG(PSE(-1))
DLOG(PTRP(-
(-0.085)
0.128
-0.669
0.509
1))
DLOG(SAP(-1))
(-0.558)
0.857
(-36.955)**
15.907
F-statistic
15.123
-0.650
0.521
-2.323
Durbin-
0.029
1.917
DLOG(UP(-1))
R-squared
0.878
Watson
stat
45
Adjusted
R-
squared
Prob(F-statistic)
0.000
0.820
The estimation of Equation (3) using the ARDL model is reported in Table 4.9 Using
Hendrys general-to-specific method, the goodness of fit of the specification that is,
R-squared and adjusted R-squared, is 0.87 and 0.82 respectively. Several diagnostic
tests were exercised to ensure the robustness of the model such as Breusch- Godfrey
serial correlation LM test, Breusch-Pagan-Godfrey Heteroskedasticity Test JacqueBera normality test and Ramsey RESET specification test. All the tests disclosed that
the model has the aspiration econometric properties, it has a correct functional form
and the residuals of the model are serially uncorrelated, homoskedastic and normally
distributed and Therefore, the outcomes reported are serially uncorrelated, normally
distributed and homoskedastic. Hence, the results reported are valid for reliable
interpretation.
Test-stats
Heteroskedasticity Test
1.572
0.193
Normality test
2.187
0.334
0.132
p-values
0.719
1.198
0.325
46
Table 4.9 illustrate the short run as well as long run impact of explanatory variables
on primary school enrolment of Pakistan. Adult literacy rate (ALR) has a significant
impact on primary school enrolment at 5% and 1% in short run and long run
respectively in Pakistan. Our results are similar with (Gupta et al .2004 and Craigwel,
2012) which indicating the direct relationship between the variables. If persons in the
household are educated they will definitely acknowledge the importance of education.
They will try their level best to educate their children according to their recourses and
hence school enrolment will increase. Uneducated people are less likely to enroll their
children in school.
Per capita plays a significant role to improve primary school enrolment in long run
but in short runs it is not significant detriment of enrolment in Pakistan. These results
are confirmed by many other studies .i.e. (Mingat and Tan, 1998; Gupta et al. 2004;
Craigwel, 2012). As PCI go up the relative cost of enrolling children into school is
decreased indicating that growing income s expand school enrollment in Pakistan.
Parents incur direct and indirect costs when they send their children to school which
include uniforms, supplies, transportation and the forgone income of the childs work
in the labor market (McEwan, 1999). In addition, if education is a normal good, at
higher income level the demand for education increases (Gupta et al. 2002).
47
number of schools and teachers increases, access to school will be easy and
inexpensive so school enrolment will increase.
Coefficient of Pupil teacher ratio is negative as suggested by theory but not significant
both in short run as well as in long run. As Pakistan is developing country and most of
the population is illiterate so pupil teacher ratio is not considered both by government
due to lack of resources and by parents due to lack of understanding and education.
However, the decrease in this ratio necessitates an increase in public education
expenditure. (Craigwel, 2012) found the same results for Caribbean countries.
School age population (SAP) do not have a significant relationship with school
enrolment in short run because increase in school age population is only possible in
long run. In short span of time SAP cannot lessen the school enrolment in Pakistan
Craigwel (2012).in long run our results are consistent with the previous findings. As
in long time period of time school age population increases so school enrollment
diminish in Pakistan. It is expensive and difficult to increase enrollment rates when
the population is relatively young (Mingat and Tan, 1992). Gupta et al. (2002) claim
that a high share of young people (population aged 0-14) should have a negatively
impact the enrollment.
Last variable included in the model is urban population (UP) which is significant at
5% in short run while in long run it is highly significant at 1% of level. According to
plank (1987) urbanization increase the school enrolment because access to education
is typically better in cities. Quality of education is also comparatively better in urban
48
areas than ruler, among all other reasons transportations cost is low for urban
household so they are most likely to send their children to school. Gupta et al (1999)
In Table given below the results of the bounds co-integration test demonstrate that the
null hypothesis of against its alternative is easily rejected at the 1% significance level.
The computed F-statistic of 15.05396 is greater than upper l bound value of 5.06, thus
indicating the existence of a steady-state long-run relationship among SEP, ALR,
PCI.PSE, PTRP, UP and SAP.
Test Statistic
Value
Probability
F-statistic
15.053
0.000
Critical value
1%
5%
10%
Note: Computed F-statistic:
Values are cited from Pesaran et al. (2001), Table CI (iii), Case 111: Unrestricted
intercept and no trend.
The estimated coefficients of the long-run relationship between SEP, ALR, PCI, PSE,
PTRP, SAP and UP are expected to be significant, that is:
49
D log (SEP)t
=0.309**
0.187***log(PSE)t--0.3193
log
(PTRP)t
-0.9436***log(SAP)t
+3.641***
log(UP)t(4)
Equation (4) indicates that adult literacy rate, public spending on education and urban
population are highly significant to determine the primary school enrolment in long
run.1% increase in adult literacy rate increase the primary school attainment by 0.77%
,likewise public spending on education and increase in urbanization enhance the
school enrolment by 0.18% and 3.64% respectively. Per capita income is also
significant determinant of primary school enrolment i.e. 1 % increase in per capita
income improve the enrolment by 0.18%. School age population negatively impact
the enrolment by 0.94% but pupil teacher ratio is not significant to decline the
primary enrolment in long run.
Coefficients
0.773
0.184
50
PSE
PTRP
SAP
UP
0.187
-0.319
-0.943
3.641
Variable
F-statistic
Probability
DLOG(ALR(-1))
13.472
(0.001)***
DLOG(PCI(-1))
4.111
(0.056)**
DLOG(PSE(-1))
3.208
(0.087)*
DLOG(PTRP(-1))
4.227
(0.051)**
DLOG(SAP(-1))
0.034
(0.855)
DLOG(UP(-1))
1.591
(0.223)
Adult
51
To sum up the findings we can say that public spending on education, and adult
literacy rate, pupil teacher ratio and per capita income granger cause in short run.
statistics
Mean
PTRP
38.181
PTRS
29.112
SAP
41.042
SEP
67.766
SES
26.032
UP
32.166
PSE
2.407
Median
38.339
28.655
43.062
66.891
27.654
32.096
2.398
Maximum
41.62
42.266
43.634
94.809
36.600
36.549
3.022
Minimum
32.999
16.898
34.320
47.886
16.504
28.066
1.837
Std. Dev.
2.442
10.050
3.066
14.659
5.960
2.523
0.335
33
33
33
33
33
33
Observation 33
s
Variables
Coefficient
Standard
T statistic
Probability
-2.072
-3.493
0.049
0.002
error
C
LOG(SES(-1))
0.352**
-0.596***
0.169
0.170
0.053
LOG(ALR(-1))
LOG(PCI(-1))
0.588**
0.174***
LOG(PSE(-1))
52
0.289
0.055
2.035
3.120
0.004
0.049
2.742
0.011
LOG(PTRS(-1))
LOG(SAP(-1))
0.135***
-0.338***
0.092
-1.188***
0.334
-3.680
(0.001
0.002
-3.548
LOG(UP(-1))
0.868
DLOG(ALR(-1))
3.864***
0.504*
0.275
DLOG(PCI(-1))
0.322***
0.103
4.452
1.833
0.079
0.004
0.068
3.122
2.829
0.009
0.192***
-0.558***
0.171
-3.248
0.003
1.054
1.133
0.930
0.361
0.016
DLOG(PSE(-1))
DLOG(PTRS(-1))
DLOG(SAP(-1))
0.000
DLOG(UP(-1))
1.170
3.041***
F-statistic
9.744
2.598
Durbin-
2.051
Watson
R-squared
Adjusted R-
0.822
squared
0.738
stat
Prob(F-statistic)
0.000
Test
Test-stats
p-values
Heteroskedasticity Test
1.511
0.212
Normality test
3.225
0.199
53
0.568
0.460
0.418
0.663
Our findings in case of secondary school enrolment are also consistent with the
previous literature as discussed in model 1 where our dependent variable is primary
school enrolment. All the references of previous literature discussed in model 1 to
support our findings are equally applicable when we take secondary school enrolment
as a dependent variable because all the researchers used both primary and secondary
school enrolment in their analysis.
Adult literacy rate positively impact the secondary school enrolment in short run at
10% of significance level while in long run it is significant at 5%.As literate parents
know the importance of educating their children so with the increase in adult literacy
arte secondary school enrolment increases.
Per capita income is significant determent of secondary school enrolment both in
short and long run in Pakistan. As income level of the household increases, they have
to allocate relatively small percentage of total income for children education.
Government expenditure on education is highly significant to improve the secondary
school enrolment in short run as well as in long run.
54
In Table given below the results of the bounds co-integration test exhibit that the null
hypothesis of against its alternative is easily rejected at the 5% significance level. The
computed F-statistic of is 4.418 greater than upper bound value of 4.01, thus
indicating the existence of a steady-state long-run relationship among SES, ALR,
PCI.PSE, PTRS, UP and SAP.
Test Statistic
Value
Probability
F-statistic
4.418
0.003
The estimated coefficients of the long-run relationship between SES, ALR, PCI, PSE,
PTRS, SAP and UP are expected to be significant, that is:
D log (SES)t
=0.352**
Equation (4) exhibits the long run estimate coefficients of the secondary school
enrolment.
Coefficient of adult literacy rate show that 1% increases in adult literacy rate increase
the enrolment by 0.84%
56
Coefficients
0.8467
0.2929
0.2269
-0.5684
-1.9934
6.4828
F-statistic
Probability
(7.675)***
0.011
(25.222)***
0.000
(4.741)**
0.040
(4.682)**
0.041
(5.104)**
0.034
(5.414)**
0.029
DLOG(ALR(-1))
DLOG(PCI(-1))
DLOG(PSE(-1))
DLOG(PTRP(-1))
DLOG(SAP(-1))
DLOG(UP(-1))
57
Short run causality test for secondary enrolment show that per capita income and
adult literacy rate is statistically significant at 1% to Granger-caused the primary
school enrolment public spending on education and pupil teacher ratio, school age
population and urban population is significant at 5%.cocluding our findings we can
say that all variables included in the model are significantly granger cause in short
run.
58
Chapter: 5
CONCLUSION AND RECOMMENADTIONS
5.1 SUMMARY
Many researchers advocated the greater public expenditures on basic education and
primary health care but little empirical support exists on the beneficial impact of such
expenditure on social outcomes. Using time series of 33 years for Pakistan we
investigated the effectiveness of government expenditure in health and education
sector. Infant and child mortality rate is used as health indicators, while educational
outcomes are expressed by gross primary and secondary school enrollment in
Pakistan. Study employ simple regression analysis for Health models but impact of
government spending on education sector is estimated by ARDL methodology as per
the requirement of stationery level of the variables, included in the study. The
evidence is stronger both for health and education sector in Pakistan.
59
5.2 CONCLUSION
Our investment variables for health models are government expenditure on health, per
capita income and female literacy rate. Impact of government expenditure and female
literacy rate on infant and child mortality rate is same as prescribed by the literature
but per capita income shows the insignificant relationship with infant mortality rate in
Pakistan. Provisions of efficient and sufficient resources for health sector reduce
infant and child mortality both, which are universally accepted as a measure of health
status. Literate women can have a better awareness of child growth, diseases, clean
and healthy food and sanitation therefore with the increase in female literacy child
and infant mortality reduces. Insignificance of per capita income may be the result of
highly unequal distribution of income in Pakistan.
Improved water source and sanitation facility plays a vital role to diminish the infant
and child deaths in Pakistan. DPT3 do not illustrate the results as prescribed by the
previous literature yet this result can be justified in case of Pakistan. Coefficient of
DPT3 is negative but not significant both for infant and child mortality rate in our
analysis. Among all other reasons one source of this insignificant relation is due to
lack of awareness and access about immunization in ruler areas. People do not
immunize their children against these diseases because of their conservative and
ignorant Attitude towards these vaccinations.
Immunization measles is an important component of health status in Pakistan .Co2
emission is positively related with infant and child mortality as indicated by the
literature.
Increase in government spending on education directly affects the primary and
secondary school enrollment in Pakistan. Most importantly, increase in government
expenditure alone do not ensure desired level of school enrollment, adult literacy rate,
60
higher per capita income and urbanization level are significant indicators to achieve
greater school enrollment. Decline in pupil teacher ratio and school age population
can also play a significant role to boost enrolments, which are treated as quality
variables in our study
Efficient and sufficient public expenditure on education and health care in the
Pakistan is imperative. As resources are limited and economic growth is necessary to
sustain economic development, and thus improve standards of living and human
development.
5.3 RECOMMENDATIONS
According to the findings of this study we suggest the following recommendations.
Pakistan have lowest health budget and higher child deaths in the region so
government should increase its health budget to ensure basic health care for
all.
Special attention should be given to educate the women by government and
public both.
Government should provide adequate drainage and sewerage systems to
people; moreover government should ensure easily accessible clean water
61
teachers.
To improve the educational system, government should allocate the money in
Different proxies can be used to measure the health status like life expectancy
at birth.
Tertiary and higher education attainment, primary and secondary pass out ratio
etc can be employed as proxies for education attainment as suggested by
previous literature.
63
Chapter: 6
REFERENCES
20:12 UTC
Baldacci, E., Clements, B., Gupta, S., and Cui, Q. (2008) Social spending,
human capital, and growth in developing countries, World Development,
36(8), 1317Baldacci, E., Guin-Siu Maria, T. and De Mello, L. (2003), More on the
effectiveness of public Spending on health care and education: a covariance
64
pp. 717-37.
Greenidge, K. and Stanford, S. (2007), What are the determinants of the
health status in LatinAmerica and the Caribbean, Central Bank of Barbados
105-143.
Kaushal et al (2013) Public Spending on Health and Childhood Mortality in
India, MPRA Paper No. 48680, posted 29. July 2013 11:33 UT
Mingat, A. and Tan, J. (1998), The mechanics of progress in education:
evidence from Cross-country data, World Bank Policy Research, pp. 1-45.
65
Manoux et al, Adult education and child mortality in India: the influence of
reconsideration
Roberts, J. (2003). Poverty reduction outcomes in education and health:
Public expenditure and aid (Working Paper No. 210). London: Overseas
Development Institute.
Rajkumar, A.S. and Swaroop, V. (2008), Public spending and outcomes:
does govern matter? Journal of Development Economics, Vol. 86, pp. 96-
111.
Schultz, T.W. (1961), Investment in human capital, The American
Mortality Rate in Indian States, Curr Pediatr Res 2014; 18 (1): 49-56
Tiongson .E ,verhoeven.M and Gupta.S (1999) Does higher governmrnt
spending buy better results in education and health care ? , international
66