HEALTH CARE FINANCING AND HEALTH OUTCOMES IN PACIFIC
ISLAND COUNTRIES*
Azmat Gani
Macmillan Brown Centre for Pacific Studies
University of Canterbury
Christchurch, New Zealand
Email: [email protected]
ABSTRACT
This paper provides empirical evidence on the relationship between per capita public
health expenditure and three measures of health outcomes (infant and under five
mortality rate and crude death rate) using cross-country data from seven Pacific Island
countries for selected years between 1990 and 2002. The results of the fixed effects
estimation procedure correcting for AR (1) errors provide strong evidence that per
capita health expenditure is an important factor in determining health outcomes. The
elasticity of infant mortality rate with respect to per capita health expenditure is -0.66.
Based on this elasticity, a 10 percent increase in per capita health expenditure means
that a country such as Papua New Guinea (PNG), with high infant mortality rate,
would face a reduction of 3.6 infant deaths per 1,000 live births and a reduction of 2.0
infant deaths per 1,000 live births on average for the Pacific Island countries. The
empirical results also provide strong evidence of per capita incomes and
immunisation as additional core factors that determine health outcomes. Some policy
implications are drawn.
Key words:
Pacific Island countries, health expenditure; incomes; immunisation;
mortality.
This project is supported by the University of Canterbury’s Macmillan Brown Centre for Pacific
Studies Research Scholarship awarded to the author. The author gratefully acknowledges this financial
support. The findings and discussions herein are those of the authors. The author is responsible for all
errors and omissions.
*
1
1.0 INTRODUCTION
Of the 11 million children dying annually, 90 percent of them happen to be under the
age of 5 (Bokhari et. al., 2007) and in poor countries 30 percent of deaths are amongst
children compared with less than 1 percent in rich countries (Cutler, et. al., 2006). At
least as many as 10 million children under the age of five die each year, mainly from
preventable (or curable) conditions that seldom kill children in rich countries (Jones
et. al., 2003). These statistics provide clear evidence that poor health being is
concentrated amongst poor people in poor countries (see also Bhalotra, 2007). Like
developing countries elsewhere, the island countries in the South Pacific region are
poor. Measured in terms of the per capita incomes, the Pacific Island countries fall in
the low and lower-middle-income categories and on the basis of human development
index; they fall in the low and medium human development categories. A number of
countries in the region rank poorly in terms of their populations health status. For
example, many countries reveal high incidence of infant and under five mortality and
prevalence of preventable diseases. Children continue to die annually in the Pacific
Island countries, majority of who are under the age of five (Figures 1 and 2). At the
country level, under-five mortality rates in 2005 ranged from 18.0 per 1,000 in Fiji to
74.4 per 1,000 in PNG. In the same year, infant mortality rates ranged from 15.7 per
1,000 live births in Fiji to 55.2 per 1,000 live births in PNG. Neonatal causes have
been the major cause of deaths among children under the age of five in several Pacific
Island countries. In additions, diarrhoeal diseases, pneumonia and measles are other
causes of deaths among children under the age of five (Table 1).
2
Figure 1.
Infant Mortality Rates - 2005
60
Per 1,000 Live Births
50
40
30
20
10
V a
nu
a t
u
T u
va
lu
T o
ng
a
n d
s
la
oa
Is
on
S a
m
ew
G
ui
ne
a
u
P a
la
N i
u e
N a
ur
u
ne
si
a
nd
s
ic r
o
S o
lo
P a
pu
a
N
m
M
Is
la
at
i
K i
rib
F i
ji
M
C o
ok
ar
sh
al
l
Is
la
n d
s
0
Countries
Source of data for Figure 1: World Health Organisation (2006).
Figure 2.
Under Five Mortality Rate - 2005
80
70
Per 1,000
60
50
40
30
20
10
0
Fiji
Kiribati
Marshall
Islands
Micronesia Papua New
Guinea
Countries
Samoa
Solomon
Islands
Tonga
Vanuatu
Source of data for Figure 2: World Bank (2006).
3
Table 1. Causes of death among children under five years of age in 2000
(percentages).
Countries
Neonatal
HIV/AIDS Diarrhoeal Measles
Malaria
Pneumonia
diseases
Cook Islands 96.1
0.0
0.7
0.5
0.0
1.1
Fiji
41.2
0.2
10.6
0.0
0.0
9.2
Kiribati
22.1
0.0
21.9
2.6
0.7
11.5
Marshall Is.
37.1
0.3
14.1
0.5
0.0
13.5
Micronesia
49.2
0.3
8.0
1.5
0.0
11.3
Nauru
7.0
0.0
37.8
5.5
0.0
30.3
Palau
47.0
0.3
9.7
0.7
0.0
12.4
PNG
35.4
0.3
15.3
2.1
0.8
18.5
Samoa
49.2
0.3
9.7
0.1
0.1
10.2
Solomon Is.
49.5
0.3
8.8
0.5
0.1
9.5
Tonga
57.2
0.0
10.0
1.8
1.3
7.3
Tuvalu
40.0
0.3
13.2
1.2
0.0
13.5
Vanuatu
42.3
0.3
11.5
0.3
0.6
13.0
Source: World Health Organisation (2006).
While relevant interventions (immunisation and oral rehydration therapy) are
found to cost low (see Deaton, 2006) and aids in minimising deaths among children,
the incidence of under five mortality in the developing world as well as the Pacific
Island countries raises the question whether national budgetary allocation to these
forms of health care services among others are adequate and effective in terms of its
impact on infant or under five mortality. Available evidence from the high-income
4
countries provides strong support that main contributors to mortality decline in
children were improved nutrition and medical technological progress (see Cutler and
Miller, 2005 and Cutler et. al., 2006). Other than advances in medical science,
improved immunisation, improved water, improved sanitation and education; public
health expenditure is also considered to be an important factor that can influence
health outcomes particularly in terms of reducing the incidence of infant and child
mortality rates in developing countries. In their study on government health
expenditure and health outcomes, Bokhari et. al., (2007) found that government
spending is an important contributor to health outcomes. Similar effects of health
spending were also noted by Bhalotra (2007) in her study on state health expenditure
and infant mortality in India. Using lagged effects and controlling for trended
unobservable and restricting sample to rural households on India, Bhalotra (2007)
found significant effects of health expenditure on infant mortality rates.
The examination of health care expenditure and health outcomes have been a
subject of an ongoing inquiry (see for example, Newhouse, 1992; Hitris, 1997; Di
Matteo and Di Matteo, 1998; Gerdtham and Jonsson, 2000; Hitris and Nixon, 2001;
Gianonni and Hitris, 2002; Bokhari, et. al., 2007; and Costa-Font and Pons-Novell,
2007). It is clear from a number of these studies that fiscal policy and the composition
of public spending are important ingredients to improving health outcomes. The
emphasis on increasing public spending on primary health care is generally justified
on the basis that such spending ameliorates the impact of disease on the productive
life years of the population (Gupta, Verhoeven and Tiongson, 1999). It has been also
shown that expenditure allocations in favour of health can also boost economic
growth while reducing poverty (see for example, Barro, 1991 and Temple, 1999).
5
While large developing country case studies provide evidence of health care
spending improving health outcomes such as under five mortality rates (see for
example, Bokhari, et. al., 2007 and Bhalotra, 2007), not much is written in terms of
Pacific Islands health care financing and health outcomes. As such, an investigation of
the relationship between per capita health expenditure and health outcome indicators
becomes increasingly important in of appropriate policy intervention so as to ensure
continued improvements in infant and under five mortality rates in the Pacific Island
countries.
Thus, the purpose of this study is to examine whether public expenditure
allocations to health sector improves health outcomes in Pacific Island countries. In
doing so, this study utilises cross-country data from seven Pacific Island countries
(Fiji, Kiribati, PNG, Samoa, Solomon Islands, Tonga and Vanuatu) for selected years
between 1990 and 2002 for public spending on health together with selected control
variables that determine health outcomes. The empirical methodology utilises a fixed
effects estimation procedure correcting for AR(1) errors. The results obtained from
this method of estimation indicate that per capita health expenditure together with
income and immunisation as core factors that determine infant and under five
mortality rates and crude death rates.
The rest of the paper is organised as follows. The next section discusses the
patterns and trends in health expenditure in Pacific Island countries. Section three
outlines the empirical model. Section four discusses the empirical findings. The
limitations of this study are highlighted in section five. A summary and a discussion
of policy issues follow.
6
2.0 HEALTH EXPENDITURE IN PACIFIC ISLAND COUNTRIES:
PATTERNS AND TRENDS
In this section, the basic statistics on central government expenditures for key
functional areas are summarised together with an overview of long-term trend in
health care financing across the Pacific Island countries. The intention here is to
provide an assessment of the trends and not to review the causes of progress or
stagnation. The government expenditure as a share of total expenditure by functional
category is summarised in Table 2. Not all of the countries listed in Table 1 appear in
Table 2 due to incomplete or absence of published data. Health expenditure statistic
for each country is shown relative to government expenditure in other key functional
areas with the year of statistics indicated in parentheses. It is important to keep in
mind that there are other functional areas such as defence, housing and social security
which are not included in Table 2. It is clear from Table 2, that for all countries,
health expenditure does not represent the largest share of total government
expenditure. In Kiribati and Tonga, health is the second largest government
expenditure while in the rest of the countries it is the third largest expenditure. It is
also interesting to note that in all countries except Tonga, health expenditure is even
less than education expenditures. While educational spending gets priority over health
and economic services in Fiji and Kiribati, spending on economic services has priority
over health and education in the remaining countries.
7
Table 2. Government expenditure by function (percentage of total).
Country
General public
Education Health Economic
services
services
Cook Islands (2005)
20.8
13.9
11.3
38.9
Fiji (2002)
26.1
29.4
14.3
18.3
Kiribati (2005)
5.0
14.0
9.3
8.7
PNG (2002)
11.3
10.0
5.7
12.1
Samoa (2005)
26.5
22.1
16.7
24.6
Tonga (2002)
40.8
12.9
13.9
18.8
Vanuatu (2005)
15.3
22.7
11.1
43.4
Source: Asian Development Bank (2007).
Figure 3 depicts the trends in health expenditure as a share of GDP for 19902001 for seven Pacific Island countries. Other than Kiribati, health expenditure as a
share of GDP for the remaining six countries remained below 5 percent between 1990
and 1996. While health expenditure as a share of GDP rose slightly after 1997 for Fiji,
PNG, Samoa, Solomon Islands and Tonga, it declined in case of Vanuatu. It is quite
difficult to offer much more than a basic description of trends across each country.
Indeed clear potential explanations for the observed trend is to do with governments
budgetary allocation where other functional areas such as public services, economic
services and education are given more priority and thus budgetary allocations done
accordingly. Among the national policies that might explain the trends are economic
growth, public sector reforms and the demand for health care services.
8
Figure 3.
Percentage of GDP
Health Expenditure
20.0
15.0
10.0
5.0
0.0
1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001
Years
Fiji
Samoa
Vanuatu
Kiribati
Solomon Islands
Papua New Guinea
Tonga
Source of data for Figure 3: Asian Development Bank (2007).
Figure 4 depicts per capita health expenditure for seven Pacific Island
countries. Using 2003 as the cut off year, per capita health expenditure ranged from a
low of US$8 in PNG to a high of US$104 in Fiji. Of all the seven countries depicted
in Figure 4, Fiji spends the most on a per capita basis while PNG spends the least.
Between 1990 and 1994, per capita health expenditure remained below US$50 for all
except Fiji. Kiribati, Samoa and Tonga’s per capita spending rose in the post-1995
period. For the Solomon Islands, per capita spending remained stagnant between
1990-2002. Surprisingly, PNG’s per capita health expenditure has been overt, falling
gradually since 1997. Significant improvements in per capita spending are noted for
Kiribati between 2001 and 2005. The statistics presented above provide confirmation
that the Pacific Island governments have accorded low priority to improving health
conditions of their populations. Every country has a ministerial portfolio, the Ministry
of Health, but it has little bureaucratic effect and receives a smaller share of budget
9
allocation. Many countries have regarded expenditure on economic services as an
increasingly important for investment as opposed to investments in health care
services that can have direct effect on national human capital formation.
Figure 4.
Per Capita Health Expenditure
200.0
180.0
US Dollars (Current)
160.0
140.0
120.0
100.0
80.0
60.0
40.0
20.0
0.0
1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006
Years
Fiji
Solomon Islands
Kiribati
Tonga
Papua New Guinea
Vanuatu
Samoa
Source of Data: Authors calculation based on data extracted from Asian Development
Bank (2007).
3.0 HEALTH EXPENDITURE AND HEALTH OUTCOMES: LITERATURE
AND THE EMPIRICAL MODEL
Public spending for health care is important for health outcomes of the poor as the
poor are more likely to obtain health care from publicly provided facilities (Gwatkin,
2000). Past studies show that the poor are significantly less healthy than the rich (see
Wagstaff, 2000 and Gwatkin, 2000) and that the rich are more likely to obtain medical
10
care when sick (Makinen et. al., 2000). Hence, it is the level of health care financing
that can bridge the gap between the health status of the poor and the rich.
A number of studies on health care financing and health status reveal that
public expenditure on health care minimizes the poor rich differences in health
outcomes. In a study involving 70 developing and transition economies, Gupta,
Verhoeven and Tiongson (2001) found that the poor are more strongly affected by
health care in comparison with the non poor and that the difference in the impact of
spending between the poor and non-poor could be substantial. In the study by
Gakidou and King (2000), health expenditures per capita, among other variables were
found to be negatively correlated with health inequality.
The effect of public spending on health is usually measured by health outcome
variables such as infant or child mortality rates and life expectancy. The effect of
government health expenditure on infant and under five mortality has been
investigated by several researchers. While some studies do not find any support for
health expenditure reducing mortality rates, other reveal that health care spending has
beneficial outcomes in terms of reducing infant and under five mortality. For
example, Anand and Ravallion (1993) using cross-sectional data for 22 developing
countries in 1985 find that health expenditure raises life expectancy. In a study on
Philippines, the World Bank (1995) reported that public expenditure on health in the
Philippines contributed to the reduction in infant mortality rates in the poorer regions,
but not in the richer regions. In their study involving 50 developing countries, Gupta,
Verhoeven and Tiongson (1999) found empirical evidence to support the claim that
greater public spending decreased infant and child mortality rates. In a further study
by these authors (see Gupta et al., 2001) relating to public spending on health care for
a larger sample (70 developing countries), the authors find some evidence that health
11
expenditure reduces childhood mortality. Hojman’s (1996) study involving Central
American and Caribbean countries revealed that public health spending has a
statistically significant effect on health status. The study by Bidani and Ravallion
(1997) also find that public spending has beneficial impact on health condition for the
poor.
There are studies that do not find any support or statistically insignificant
support for health expenditure reducing mortality rates. Kim and Moody (1992) and
Musgrave (1996) found that the effect of public spending on health status as measured
by health outcome variables, infant and child mortality rates, to be statistically
insignificant. Le Grand (1987) found weak and a negative correlation between health
inequality and share of public spending in health care. Filmer, Hammer and Pritchett
(1998) attempted to address the issue of allocations within the health sector by
including a measure of government spending on primary health care in their crosssection analysis of the causal factors of infant mortality and fail to find a statistically
significant impact of primary health care spending on infant mortality rates. Filmer
and Pritchetts’s (1999) further study on government health expenditure on infant and
under five mortality in 98 developing countries reveal statistically insignificant effect.
Deolalikar (2005) using a state panel for 1980-99 for India found no effect of current
health expenditure on mortality rates.
While the above review provides evidence of some studies support for health
expenditure reducing mortality rates (and some studies without any support), no such
study exists for Pacific Island countries. The effect of health care financing on health
outcomes in the Pacific Island countries certainly deserves an investigation in light of
the prevalence of preventable diseases, high incidence of infant and under five
mortality rates and low levels of budgetary allocations to the health sector. The case
12
of Pacific Island countries health expenditure and health outcome links can be
examined through an empirical framework where the key issue relating to health
outcomes and per capita health expenditure is unfolded. Thus, the structural equation
to examine the impact of public spending on health care in the Pacific Island countries
takes the following general form:
Yit = f ( H it , X it )
(1)
where, Y is a health outcome indicator reflecting health status of country i, H is per
capita public spending on health care; X is a vector of socio-economic control
variables, i is the country and t is the time.
The infant mortality, under five mortality and crude death rates are considered
to be better indicators of the health status of the population (see also Sen, 1998).
Changes in national health policies and in particular budgetary allocations to health
sector that directly affect investment in health capital are likely to show quicker
effects in terms of health status of the population. It has been noted that the long-term
improvements in the health status of populations are best reflected in infant mortality
and life expectancy rates (see for example, Gupta and Mitre, 2004).
Health outcomes are presumed to be primarily a function of per capita health
expenditure as well as several other variables. The variables tested are predominantly
the main conventional variables as used in many cross-country studies. The control
variables include per capita income, immunization rates, urbanisation rates and calorie
intake. In the regression analysis, equation (1) is expressed in three sets of reduced
forms as follows:
13
ln IMRit = α 0 + α 1 ln Yit + α 2 ln PCH it + α 3 ln IMU it + α 4 ln URBit + ln α 5 CI it + υ it (2)
ln U 5M it = β 0 + β1 ln Yit + β 2 ln PCH it + β 3 ln IMU it + β 4 ln URBit + ln β 5 CI it + υ it (3)
ln CDRit = χ 0 + χ 1 ln Yit + χ 2 ln PCH it + χ 3 ln IMU it + χ 4 ln URBit + ln χ 5 CI it + υ it (4)
where,
IMR
=
U5M =
infant mortality rate;
under five mortality rate;
CDR
=
crude death rate;
Y
=
per capita income;
PCH
=
per capita health expenditure;
IMU
=
immunisation (against measles);
URB
=
urbanisation;
CI
=
calorie intake;
ln
=
logs;
i
=
country;
t
=
time;
The error term in the above equation is υ it with the assumption
that υ it ≈ iid ( 0, σ 2 ) . The expected effects are: Y (-); PCH (-); IMU (-); URB (-); and
CI (-).
All variable names and their definitions are presented in Table 3. The sample
years for health outcome indicator infant mortality rate were 1990, 1995, 1998, 2000
and 2002. For the health indicator outcome under five mortality rates, the sample
years were 1990, 1995, 2000 and 2002. For the health indicator crude death rate, the
14
sample years were 1990, 1992, 1995, 1997, 2000 and 2002. All data were were
extracted from various publications from a number of sources as indicated in Table 3.
Table 3. Variable definition and data sources.
Variable
Infant mortality
rate
Definition
The number of infants dying before
reaching one year of age per 1,000
live births in a given year
Source of data
World Health
Organisation
Under five
mortality rate
The probability that a new born baby
will die before reaching age five
expressed as a rate per 1,000
Crude death rate
The number of deaths occurring
during the year per 1,000 population
estimated at mid-year.
World Bank
Per capita
income
The gross national income per capita
(adjusted for purchasing power parity)
in US dollars
World Bank
Per capita health
expenditure
The per capita health expenditure in
US dollars
Asian Development
Bank
Immunisation
The percentage of children ages 12-23
months who received one dose of
vaccine against measles before 12
months
World Bank
Urbanisation
The estimated urban population as a
percentage of total population
World Bank and Asian
Development Bank
Calorie intake
The average daily per capita total
calorie supply in grams
World Health
Organisation and Asian
Development Bank
World Bank
4.0 ESTIMATION PROCEDURE, RESULTS AND DISCUSSION
Equation (2) includes 7 countries and 5 time periods; equation (3) includes 7 countries
and 4 time periods; and equation (4) includes 7 countries and 6 time periods. While
equations (2) to (4) can be estimated with ordinary least squares, the result is likely to
15
be biased if the error terms are correlated within each time series unit and are
heteroscadastic across each cross sectional unit given that the data utilised here is
cross-country. Combining these assumptions means estimating a cross-sectionally
heteroskedastic and time-wise autoregressive model. Hence, the initial estimation
procedure begins with the estimation of a cross-sectionally heteroskedastic and timewise autoregressive model. This procedure of estimation is also equivalent to the
generalised least squares (GLS) estimation. The results of this estimation procedure
are recorded in Table 4.
It should be noted that the GLS equivalent estimation does not take into
account of the country-specific factors. While the sample of countries falls in same
geographical latitudes and somewhat similar socio-economic structures, the extent of
health outcomes differs from one country to another. To take into account of countryspecific differences, a fixed effects estimation procedure including country-specific
dummy variables is adopted. In total there are 7 dummies, D1 to D7, (see Table 5) for
each of the equations (2) to (4). The no-constant option is adopted in the estimation
procedure so as to avoid the commonly known dummy variable trap. Given the nature
of data, the possibility of AR (1) errors are likely and so the fixed effects estimation
procedure corrected for AR (1) errors is adopted. The results of this procedure of
estimation is reported in Table 5 and considered to be most robust as the statistical
significance and R-square improved significantly compared to the GLS estimation in
Table 4. The discussion of the results of the right hand side variables of equations (2)
to (4) follows.
16
Table 4. Empirical results based on generalised least squares estimation.
Ln IMR
-0.705
(0.199)
ln U5M
0.518
(0.169)
Ln CDR
0.331
(0.109)
ln Y
-0.483
(7.922)*
-0.154
(3.005)*
-0.105
(1.632)***
ln PCH
-0.611
(6.185)*
-0.005
(0.092)
0.036
(0.615)
ln IMU
-0.490
(2.077)**
-0.478
(3.183)*
-0.506
(3.244)*
ln URB
0.068
(0.434)
-0.110
(0.841)
-0.125
(1.058)
ln CI
1.562
(3.090)*
0.656
(1.538)
0.644
(1.528)
Number of
observations
35
28
42
R-square
0.88
0.49
0.46
Durbin Watson 1.80
statistic
1.75
1.81
Constant
t - statistics are in parentheses.
*, **, and *** indicates significant at the 1, 5 and 10 percent levels respectively.
17
Table 5. Empirical results of fixed effects estimation corrected for AR (1) errors.
Ln IMR
-0.409
(8.416)*
Ln U5M
-0.416
(2.425)**
Ln CDR
-0.047
(0.753)
ln PCH
-0.655
(9.897)*
-0.069
(1.089)
0.123
(1.969)**
ln IMU
-0.884
(5.888)*
-0.446
(2.216)**
-0.643
(3.863)*
ln URB
0.139
(0.979)
-0.098
(0.716)
-0.163
(1.319)
ln CI
1.983
(4.390)*
0.793
(1.619)***
0.855
(1.784)***
D1
-2.850
(0.905)
-0.474
(0.133)
-1.049
(0.308)
D2
-2.863
(0.905)
-0.558
(0.156)
-1.137
(0.333)
D3
-2.989
(0.945)
-0.488
(0.136)
-1.125
(0.328)
D4
-2.826
(0.902)
-0.540
(0.150)
-1.184
(0.346)
D5
-2.898
(0.917)
-0.614
(0.171)
-1.244
(0.363)
D6
-2.902
(0.917)
-0.585
(0.163)
-1.217
(0.355)
D7
-3.280
(1.033)
-0.692
(0.193)
-1.337
(0.389)
Number of
observations
35
28
42
0.54
0.51
ln Y
R-square
0.94
t – statistics are in parentheses.
*, **, and *** indicates significant at the 1, 5 and 10 percent levels respectively.
18
Per capita health expenditure
The coefficient for the per capita health expenditure has the correct sign as expected,
negative for infant and under five mortality rates (Tables 4 and 5). The coefficient for
infant mortality rate is statistically significant at the 1 percent level. The best point
estimates for the coefficients of infant and under five mortality are -0.66 and -0.07,
respectively. There is a large difference between the estimated coefficients of the
infant and under five mortality rates. This suggests that government health care
funding does impact infant mortality rate more strongly than the under five mortality
rate. This outcome is consistent with the pattern of the governmental budgetary
allocation to health care in the Pacific Island countries which are largely targeted for
basic primary health care services such as immunisation, control of diarrhoea,
nutritional training programs for mothers and some antenatal care. Based on the
elasticity of infant mortality rate, a 10 percent increase in per capita health
expenditure would mean a reduction in infant mortality rate by approximately 6.6
percent. For a country such as PNG with high infant mortality rate, it means a
reduction in approximately 3.6 infant deaths per 1,000 live births and an average
reduction of 2.0 infant deaths per 1,000 live births for the Pacific Island countries.
The coefficient for the crude death rate is positive and statistically insignificant and
indicates that health care expenditure does not matter much.
Per Capita Income
The sign on the coefficient of income is as expected, negative for infant and under
five mortality rates and death rates. Elasticity with respect to per capita income is 0.41 for infant and under five mortality rates and -0.05 for crude death rate (Table 5).
The income coefficient is statistically significant at the 1 percent level, both for infant
19
and under five mortality rates. There are large differences between the estimated
coefficients for per capita income for infant and under five mortality rates and death
rates. This outcome also suggests that governmental expenditure is more so targeted
towards primary health care. The elasticity of the under five mortality rate of -0.41 is
very close to that obtained by Bokhari et. al. (2007) in a study involving several large
developing countries where the elasticity of income was -0.40. The results obtained
for the income variable provides strong support that the level of income matters
strongly for infant and under five mortality and death rates: rising income means
falling mortality rates. On the basis of elasticity’s obtained for infant and under five
mortality rates, it can be argued that a 10 percent increase in per capita incomes
implies a reduction in approximately 3.0 under five deaths per 1,000 live births in
PNG and an average reduction of approximately 1.7 deaths per 1,000 live births for
the Pacific Island countries.
Immunisation
The coefficient immunisation has the expected negative sign for infant and under five
mortality rates and crude death rates. In all these three health indicator outcomes the
coefficients of immunisation are statistically significant at the 1 percent (infant
mortality and death rates) and 5 percent (under five mortality rate) levels. The
elasticity of infant and under five mortality rates is -0.88 and -0.45 respectively. The
elasticity’s obtained indicate that a 10 percent increase in immunisation rates would
reduce infant and under five mortality rates by 8.8 and 4.5 percent respectively. On
the basis of the elasticity of infant mortality rate, a 10 percent increase in
immunisation rates implies a reduction of approximately 4.8 infant deaths per 1,000
live births in PNG and an average reduction of approximately 2.6 infant deaths per
20
1,000 live births for Pacific Island countries. These are in fact large effects for
countries such as Kiribati, Marshall Islands, Micronesia and PNG as they have high
infant and under five mortality rates. The results obtained for the immunisation
variable provides strong evidence that immunisation programs are vital and it
certainly contributes towards reducing mortality rates. Comparing the size of the
elasticity’s of per capita health expenditure, per capita income and immunisation; it is
clear that immunisation has a much stronger impact in reducing mortality rates.
Urbanisation and calorie intake
The model here also controls for urbanisation and calorie intake. In all the three health
outcome indicators, the coefficient of urbanisation is statistically insignificant. For
under five mortality rates and crude death rates, the sign on the coefficient is as
expected, negative. However, the sign on the coefficient of urbanisation for infant
mortality rate is positive and contrary to the prior expectations. Similarly, calories
intake as the positive sign on its coefficient and contrary to a priori expectations.
5.0 LIMITATIONS AND SUGGESTIONS FOR FUTURE RESEARCH
The study has its limitations and it is worthy to point some of the specific drawbacks.
The major limitation surrounds the choice of variables and data. Published data on
some of the key variables of interest largely to do with socio-economic characteristics
that may affect health outcomes were not available. For example, a potential
limitation lies in the fact that this study does not control for private health
expenditure. Past studies have indicated that private health expenditure and private
health insurance does play a role in supplementing public health care (see for
example, Costa-Font and Garcia, 2003). Some Pacific Island countries do host private
21
health care facilities through private hospitals or private health practitioners. Data on
these forms of private health care services is not available. Hence, the models tested
here do not include private health care expenditure.
Another variable that is not controlled for is the level of education or literacy
rates. Past studies have shown that literacy, particularly of females,’ impacts on health
status of infants and children (see for example, Shultz, 1961 and Deolalikar, 2005).
Previous studies analysing the effect of health expenditure on health outcomes have
controlled for literacy or education levels (for example, Gupta, et. al., 1999 and
Bokhari, et. al., 2007). The models here do not include any measures of adult female
literacy or education. Data on level of education is available for some countries and in
some countries this data is missing. Given the nature of cross-country estimations,
consistent set of data was needed for the specified time periods. Due to missing data
for some countries, it was impossible to control for this variable. It is worthy to note
that while infant mortality is expected to be lowered by improvements in education,
the recent study by Bhalotra (2007) testing a model with a very large sample from
India and controlling for state education expenditure while revealing the expected
negative effect was statistically insignificant.
Further, the empirical analysis here does not compare health care outcomes
between rich and the poor. It can be strongly argued that the rich may be able to
access better health care services than the poor when governmental budgetary
allocations to health care are squeezed. On the contrary, the poor may access health
care services in times of budgetary cuts through foreign aid or forms of in-kindtransfers from family and friends and charity groups. The data utilised here are
national aggregates that do not differentiate between rich and the poor. Hence, such
data limitations constrain further analysis in this line.
22
Once data becomes available, future research should examine the effects of
some of the core control variables mentioned here such as the effect of literacy rates
and private per capita health expenditure. In addition, it would be useful to examine
the effects of health care funding on the rich and poor segments of the population in
future once data becomes available. Household survey data would be more useful for
this kind of inquiry and the collection of such data is highly encouraged. Further,
more country-specific studies are encouraged so as to account for the in-country
variations in the socio-economic structures that determine health status with greater
strength and meaning.
6.0 SUMMARY AND POLICY IMPLICATIONS
The central focus of this study was to examine whether governmental expenditure
allocations to health sector improves health outcomes in Pacific Island countries (Fiji,
Kiribati, PNG, Samoa, Solomon Islands, Tonga and Vanuatu). In doing so, this study
utilised cross-country data for these countries for selected years between 1990 and
2002 for public spending on health on a per capita basis together with selected control
variables that determine health outcomes. Three indicators of health outcomes: infant
mortality rate, under five mortality rate and crude death rate were chosen for
empirical analysis.
The regression results of a fixed effects model correcting for AR (1) errors
provide strong confirmation that per capita health expenditure matters for health
outcomes in Pacific Island countries. Based on the elasticity of infant mortality rate, it
is argued that a 10 percent increase in per capita health expenditure would mean a
reduction in infant mortality rate by approximately 6.6 percent. For a country such as
PNG with high infant mortality rate, it means a reduction in approximately 3.6 infant
23
deaths per 1,000 live births and an average reduction of 2.0 infant deaths per 1,000
live births for the Pacific Island countries. The regression results also provide strong
evidence that other than per capita health expenditure, per capita incomes and
immunisation rates are other two core factors that determine health outcomes in the
Pacific Island countries. Based on the elasticity’s obtained, it is also argued that a 10
percent increase in per capita incomes would mean a reduction in approximately 3.0
under five deaths per 1,000 live births in PNG and an average reduction of
approximately 1.7 deaths per 1,000 live births for the Pacific Island countries. In
terms of the elasticity’s obtained for immunisation, it is argued that a 10 percent
increase in immunisation rates would mean a reduction of approximately 4.8 infant
deaths per 1,000 live births in PNG and an average reduction of approximately 2.6
infant deaths per 1,000 live births for the Pacific Island countries.
While the limitations of this study is acknowledged and results may be
interpreted with caution, the empirical findings of this study nevertheless strongly
indicates that governments per capita health expenditures do matter much for
improvements in health outcomes across the Pacific Island countries. This study
points to the fact that governments of the Pacific Island countries certainly need to
boost their budgetary allocations to health care.
The findings here also reveal that health outcomes are also determined by
factors other than just per capita health expenditure. Per capita incomes and
immunisation rates are found to be other two core variables that influence health
outcomes. Given that the Pacific Island governments have faced hard long-term
budget constraints, there is an urgent need to devote more resources in terms of
uplifting per capita incomes and immunisation levels.
24
Like elsewhere, Pacific Island governments are always pressed to bring budget
into balance and raise productivity. In this process government budgetary allocation
may be cut without much thought to health sector specific outcomes. Human health is
a major determinant for national human capital formation among other factors. Like
several of the governmental functional areas of spending, the budgetary allocation to
the health sector need to be guarded against expected national revenue shortfalls and
unexpected downward revisions of budgets. Pacific Island governments should
account for the match between health expenditures and the associated health
outcomes over the long-term. This should be of concern for policy makers and
Finance Ministers. While Pacific Island government’s annual budgets are usually
intended to ensure the macroeconomic sustainability of total government
expenditures, measuring outcomes is a serious problem and budgets don’t take these
into consideration. There is an urgent need to take into account of health expenditures
and health outcomes in budgetary formulation process and formulate budgetary
allocations to the health sector accordingly.
REFERENCES
Asian Development Bank. (2007). Key Indicators. Manila.
Anand, S. and Ravallion, M. (1993). Human development in poor countries: on the
role of private incomes and public services, Journal of Economic Perspectives,
Vol. 7, No. 1, pp. 133-150.
Barro, R.J. (1991). Economic growth in a cross-section of countries, Quarterly
Journal of Economics, Vol. 106, No. 2, pp. 407-44.
Bhalotra, S. (2007). Spending to save? State health expenditure and infant mortality in
India, Health Economics, Vol. 16, pp. 911-928.
25
Bidani, B. and Ravallion, M. (1997). Decomposing social indicators using
distributional data, Journal of Econometrics, Vol. 77, pp. 125-139.
Bokhari, F.A.S., Gai, Y., and Gottret, P. (2007). Government health expenditures and
health outcomes, Health Economics, Vol. 16, pp. 257-273.
Costa-Font, J. and Garcia, J. (2003). Demand for private health insurance: how
important is the quality gap? Health Economics, Vol. 12, No. 7, pp. 587-599.
Costa-Font, J. and Pons-Novell, J. (2007). Public health expenditure and spatial
interactions in a decentralised national health system, Health Economics, Vol.
16, pp. 291-306.
Cutler, D. Deaton, A. and Lleras-Muney, A. (2006). The determinants of mortality.
Journal of Economic Perspectives, Vol. 20, No. 3, pp. 97-120.
Cutler, D. and Miller, D. (2005). The role of public health improvements in health
advances: the twentieth century United States, Demography, Vol. 42, pp. 1-22.
Deaton, A. (2006). Global patterns of income and health: facts, interpretations and
policies. WIDER Annual Lecture, September. WIDER, Helsinki.
Deolalikar, A. (2005). Attaining the Millennium Development Goals in India: How
Likely and What Will it Take to Reduce Infant Mortality, Child Malnutrition,
Gender Disparities and Hunger-Poverty and to Increase School Enrolment
and Completion, Oxford University Press: New Delhi.
Di Matteo, L. and Di Matteo, R. (1998). Evidence on the determinants of Canadian
provincial government health expenditure 1965-1991. Journal of Health
Economics, Vol. 17, pp. 211-228.
Filmer, D. and Pritchett, L. (1999). The impact of public spending on health: does
money matter? Social Science and Medicine, Vol. 49, No. 10, pp. 1309-1323.
Filmer, D., Hammer, J. and Pritchett, L. (1998). Health policy and poor countries:
26
weak links in the chain, Policy Research Working Paper No. 1874, World
Bank: Washington, D.C.
Gakidou, K. and King, G. (2000). An individual level approach to health inequality:
child survival in 50 countries, Global programme on evidence for health
policy discussion paper no. 18 (Geneva: World Health Organisation).
Available on line at:
http://www-nt.who.int/whosis/statistics/discussionpapers/pdf/paper18.pdf.
Gerdtham, U.G. and Jonsson, B. (2000). International comparisons of health
expenditure: theory, data and econometric analysis. In Handbook of Health
Economics, Vol. 1, Culyer, A.J. Newhouse, J.P. (eds), Elsevier, pp. 11-53.
Gianonni, M. and Hitris, T. (2002). The regional impact of health care expenditure:
the case if Italy. Applied Economics, Vol. 34, pp. 1829-1836.
Gupta, I. and Mitra, A. (2004). Economic growth, health and poverty: an exploratory
study for India, Development Policy Review, Vol. 22, No. 2, pp. 193-206.
Gupta, S., Verhoeven, M., and Tiongson, E. (2001). Public spending on health care
and the poor, IMF Working Paper No. WP/01/127, International Monetary
Fund: Washington, D.C.
Gupta, S., Verhoeven, M., and Tiongson, E. (1999). Does government spending buy
better results in education and health care? IMF Working Paper No.
WP/99/21. International Monetary Fund, Washington, D.C.
Gwatkin, D.R. (2000). Health inequalities and the health for the poor: what do we
know? What can we do? Bulletin of the World Health Organisation, Vol. 78,
No.1, pp. 3-18.
Hitris, T. (1997). Health care expenditure and integration in the countries of the
European Union, Applied Economics, Vol. 29, pp. 1-6.
27
Hitris, T. and Nixon, J. (2001). Convergence of health care expenditures in the EU
countries, Applied Economics Letters, Vol. 8, pp. 223-228.
Hojman, D.E. (1996). Economic and other determinants of infant and child mortality
in small developing countries: the case of Central America and the Caribbean,
Applied Economics, Vol. 28, pp. 281-90.
Jones, G., Steketee, R., Bhutta, Z., Morris, S., the Bellagio Child Survival Study
Group. (2003). How many child deaths can we prevent this year? The Lancet,
Vol. 362, pp. 65-71.
Kim, K. and Moody, P.M. (1992). More resources better health? A cross-national
perspective, Social Science and Medicine, Vol. 34, No. 8, pp. 837-42.
Le Grand, J. (1987). Inequalities in health: some international comparisons, European
Economic Review, Vol. 31, pp. 182-91.
Makinen, M., and Others. (2000). Inequalities in health care use and expenditures:
empirical data from eight developing countries and countries in transition,
Bulletin of the World Health Organisation, Vol. 78, No. 1, pp. 55-65.
Musgrave, P. (1996). Public and private roles in health: theory and financing patterns,
World Bank Discussion Paper No. 330, World Bank: Washington, D.C.
Newhouse, J.P. (1992). Medical care costs: how much welfare loss? Journal of
Economic Perspectives, Vol. 6, pp. 3-21.
Schultz, T.W. (1961). Investment in human capital, American Economic Review, Vol.
51, No. 1, pp. 1-17.
Sen, A. (1998). Mortality as an indicator of economic success and failure, Economic
Journal, Vol. 108, pp. 1-25.
Temple, J.R.W. (1999). The new growth evidence, Journal of Economic Literature,
Vol. 37, No. 1, pp. 112-156.
28
Wagstaff, A. (2000). Socioeconomic inequalities in child mortality: comparisons
across nine developing countries, Bulletin of the World Health Organisation,
Vol. 78, No. 1, pp. 19-29.
World Bank. (1993). World Development Report: Investing in Health. New York:
Oxford University Press for the World Bank.
World Bank. (1995). Philippines: public expenditure management for sustained and
equitable growth, World Bank Report No. 14680-PH (Washington: World
Bank).
World Bank. (2006). World Development Indicators CD ROM 2006, World Bank:
Washington, D.C.
World Health Organisation. (2006). Regional Databases. Regional Office of the
Western Pacific. Available online at:
http:www.wpro.who.int/information_sources/databases
29