05 Tangcharoensathien (jl/d)
20/10/99 8:29 am
Page 342
HEALTH POLICY AND PLANNING; 14(4): 342–353
© Oxford University Press 1999
National Health Accounts development: lessons from Thailand
VIROJ TANGCHAROENSATHIEN,1 ADIT LAIXUTHAI,2 JITPRANEE VASAVIT,3 NUAN-ANAN TANTIGATE,1
WIPHANT PRAJUABMOH-RUFFOLO,2 DUANGKAMOL VIMOLKIT4 AND JONGKOL LERTIENDUMRONG1
1Health Systems Research Institute, 2Chulalongkorn University, 3National Statistical Office, and 4National
Economics and Social Development Board, Thailand
National Health Accounts (NHA) are an important tool to demonstrate how a country’s health resources are
spent, on what services, and who pays for them. NHA are used by policy-makers for monitoring health expenditure patterns; policy instruments to re-orientate the pattern can then be further introduced. The National
Economic and Social Development Board (NESDB) of Thailand produces aggregate health expenditure data
but its estimation methods have several limitations. This has led to the research and development of an NHA
prototype in 1994, through an agreed definition of health expenditure and methodology, in consultation with
peer and other stakeholders. This is an initiative by local researchers without external support, with an
emphasis on putting the system into place. It involves two steps: firstly, the flow of funds from ultimate
sources of finance to financing agencies; and secondly, the use of funds by financing agencies. Five ultimate
sources and 12 financing agencies (seven public and five private) were identified. Use of consumption expenditures was listed under four main categories and 32 sub-categories.
Using 1994 figures, we estimated a total health expenditure of 128 305.11 million Baht; 84.07% consumption
and 15.93% capital formation. Of total consumption expenditure, 36.14% was spent on purchasing care from
public providers, with 32.35% on private providers, 5.93% on administration and 9.65% on all other public
health programmes. Public sources of finance were responsible for 48.79% and private 51.21% of the total
1994 health expenditure. Total health expenditure accounted for 3.56% of GDP (consumption expenditure at
3.00% of GDP and capital formation at 0.57% of GDP). The NESDB consumption expenditure estimate in 1994
was 180 516 million Baht or 5.01% of GDP, of which private sources were dominant (82.17%) and public
sources played a minor role (17.83%). The discrepancy of consumption expenditure between the two estimates is 2.01% of GDP. There is also a large difference in the public and private proportion of consumption
expenses, at 46:54 in NHA and 18:82 in NESDB.
Future NHA sustainable development is proposed. Firstly, we need more accurate aggregate and disaggregated data, especially from households, who take the lion’s share of total expenditure, based on amended
questionnaires in the National Statistical Office Household Socio-Economic Survey. Secondly, partnership
building with NESDB and other financing agencies is needed in the further development of the financial information system to suit the biennial NHA report. Thirdly, expenditures need breaking down into ambulatory
and inpatient care for monitoring and the proper introduction of policy instruments. We also suggest that in
a pluralistic health care system, the breakdown of spending on public and private providers is important.
Finally, a sustainable NHA development and utilization of NHA for planning and policy development is the
prime objective. International comparisons through collaborative efforts in standardizing definition and
methodology will be a useful by-product when developing countries are able to sustain their NHA reports.
Introduction
The World Health Organization developed a manual for the
estimation and data collection of health expenditure in
developing countries in 1983 (Mach and Abel-Smith 1983). It
provides several useful dummy tables for adaptation to suit
each country’s health system and policy needs. However, few
developing countries have attempted to develop a sustainable
mechanism to estimate and report health expenditure on a
regular basis. Health expenditure information in these countries is based mostly on ad hoc surveys (Newbrander et al.
1994). The World Bank recently reiterated the importance of
having a system to monitor efficiency and equity in health
care spending (World Bank 1993).
Myers et al. were pioneers in exploring health care expenditure in Thailand (Myers et al. 1985). They found that twothirds of health expenditure was financed directly by
households. Third-party payment represented a minuscule
portion. Most public health care finance was paid by the Ministry of Public Health (MOPH) which financed one-fifth of the
total. Based on the National Economic and Social Development Board (NESDB) data on health expenditure from 1979
to 1983, they predicted that health expenditure would increase
05 Tangcharoensathien (jl/d)
20/10/99 8:29 am
Page 343
National Health Accounts in Thailand
from 4.6% of the Gross National Product (GNP) in 1983 to
7.9% and 6.4% of GNP in 1991, assuming low and high GNP
growth rates respectively. However, information to allow a
breakdown of both public and private expenditures was
lacking. Definitions of health expenditures, inclusion and
exclusion criteria were not spelled out in this report. A study
by Wibulpolprasert (1987) showed that health expenditure
was 5.1% of GNP in 1985, or 1060 Baht per capita (25 Baht =
1US$), of which 69% was paid by private, 29% by public and
2% by external sources (National Account Division 1988).
Health expenditure was growing at a higher rate than GNP.
This work was based totally on the official report produced by
the NESDB National Account Division.
Information on health expenditure is crucial for health care
financing policy development and monitoring. More recently,
it has become a crucial tool for monitoring health care reform
initiatives in both developed and developing countries. Rapid
increases in health expenditure under the Civil Servant
Medical Benefit Scheme (CSMBS), based on a fee-for-service
retrospective reimbursement model, prompt policy-makers
to introduce capitation payment for a private-sector employees’ Social Health Insurance scheme.
The purpose of this paper is to share our experience on the
development of NHAs in a pluralistic health system like Thailand, the problems encountered and discrepancies between
the NHA and the current estimate made by NESDB, as well
as how to sustain NHAs in a developing country. Our lessons
might be useful for similar NHA development in other
developing countries.
The paper starts with a methodological review of the health
expenditure estimation done by NESDB, and its major weaknesses. Based on NHA methodology available to date and
realizing NESDB’s limited ability to provide a meaningful,
detailed expenditure breakdown, the authors developed the
NHA, based on primary data collection and criteria for
expenditure breakdown. Finally, we propose further research
and development of NHA to facilitate international comparison, especially with members of the Organization of Economic Cooperation and Development (OECD).
NESDB health-expenditure estimation
methodology
The NESDB has been producing the national income and
expenditure account since 1967, based on guidelines proposed by the United Nations System of National Account
(UNSNA). It annually reports private (households and
private enterprises) and government (central, local and
parastatal) consumption expenditures under several important categories (e.g. food, clothing, housing, personal care,
health expenses for private consumption; and general
administration, defence, justice, education, health services,
transport for public consumption). Capital formation was
lumped under public and private categories for all sectors –
defence, transport, health, education, etc. This report is not
useful for health financing monitoring or policy re-orientation (Tangcharoensathien 1991), as it cannot breakdown
health expenditure by functions or types of providers. It does
343
not demonstrate expenditure spent by detailed sources of
finance, but only two broad categories of public and private.
Private consumption expenditure on health was indirectly estimated based on drug and non-drug (other medical services)
consumption. Consumption of domestic drugs was estimated
based on total production less exports. The Food and Drug
Administration (FDA) is responsible for data compilation on
the value of imported and domestic production at wholesale
prices. Drug consumption by end users at retail prices was
computed by assuming 184–320% mark-up of the wholesale
price at private pharmacies (Jaidee 1987). A small-scale survey
found that 5.5% of drugs were damaged during transportation;
this was subtracted to attain the consumption of domestic
drugs. Consumption of imported drugs was estimated using
data from the Customs Department. Costs, including freight
plus import duty plus the standard profit margin set by the
Ministry of Commerce, were calculated for consumption of
imported drugs. Finally, free drugs subsidized by the government to low-income households were deducted to produce
real household, total private drug-consumption expenditure.
Household consumption of medical services, e.g. laboratory,
radiology, and other related medical services, was further
estimated. Household medical-service expenditure in the
private sector was estimated based on a small sample survey
of private hospitals and clinics in 1979 and 1980 on revenue
generated from non-drug services. After 1980, there were no
more surveys; household non-drug consumption was
adjusted, based on 1980 benchmark figures, by the medicalcare consumer price index (CPI) produced by the Department of Business Economics, assuming that the consumption
pattern had not changed since 1980.
Government expenditure on medical services was estimated
indirectly. Total drug consumption at retail prices was
deducted, i.e. household consumption at private pharmacies
and private clinics and hospitals, estimated at 21%. A proportion of 70/30 was used to estimate drug and medical service
expenditure at government facilities. Government consumption expenditure other than drug and medical services, such as
public health programmes, was estimated from Ministry of
Finance Comptroller General Department budgetary reports.
NESDB estimation
The NESDB approach in the late 1970s was a good start when
the National Statistical Office Household Socio-Economic
Survey (NSO-SES) was not in place (the first SES was
launched in 1981). However, the NESDB estimation had two
weaknesses: the methodologies and utility.
Estimation methodologies
The use of drug and medical service approaches to estimate
household expenditure posed several limitations.
• In-depth interviews with key informants in the FDA found
under-reporting of domestic and import drugs at a magnitude of 2–3 (Tangcharoensathien 1991). Wibulpolprasert
further estimated an under-reporting of 48% for domestic
05 Tangcharoensathien (jl/d)
344
•
•
•
•
20/10/99 8:29 am
Page 344
Viroj Tangcharoensathien et al.
products (Wibulpolprasert 1994). The magnitude of underreporting of imports to the FDA was unknown, although
the FDA should have verified its data with import data at
the Customs Department.
The single small-scale survey finding of 320% mark-up at
private pharmacies was not verified by subsequent studies,
but was used by NESDB ever-since. There are wide variations in drug mark-up for private clinics and public and
private hospitals.
The survey of private clinic and hospital revenue from
drugs and medical services suffers from under-reporting.
The resultant proportion of revenue from drugs and services has been used ever-since without subsequent surveys
and other verification, assuming no changes in consumption patterns of drugs and medical services by households
despite significant growth of the private sector and extensive acquisition of high medical technologies during the
late 1980s and early 1990s (Tangcharoensathien and Nittayaramphong 1994; Bennett et al. 1994).
Employer-arranged medical benefit (in addition to social
security contributions) as part of an employment contract
is not insignificant, especially among white-collar workers.
This source is not captured by the NESDB estimation.
Using the 1980 benchmark and CPI adjustment can greatly
distort the true magnitude and pattern of health financing
within the public and private sectors in the health system.
Indirect estimation of public consumption expenditure may
cause errors. The private pharmacy, clinic and hospital market
share of 21% of total drug consumption does not hold true at
present (Wibulpolprasert 1994). Government consumption
expenditure on salaries and wages did not include other fringe
benefits. Medical fringe benefits are a significant proportion, at
4.5–7.19% of total salaries and wages during 1988 to 1997
(MOF 1997). There is a significant under-estimation when
health services provided by the Ministry of Defence and university hospitals under the Ministry of University Affairs are
counted as defence and education expenditure. Health expenditure by local government is increasingly important due to
decentralization policy, but was missing from the estimate.
Comparing reports by NESDB on household consumption
expenditures on health with the National Statistical Office
Household Socio-Economic Survey (NSO-SES), which is a
biennial national survey (NSO 1986), we found significant discrepancies: the NESDB estimates were double SES data.
However, the SES may suffer from non-sampling bias such as
under-reporting by household respondents, as well as sampling bias, especially as households in urban areas do not
cooperate well.
Utility for health sector policy application
As a result of NESDB estimation methods, application for
the health sector is limited, as only aggregate figures are produced to show how much was spent by public and private
sources on consumption expenditures. The changing pattern
and magnitude of public and private investment in health
care, especially during the economic boom in the early 1990s,
could not be detected.
Thailand has a pluralistic health care system, whereby the
private sector plays an increasing role in service provision.
This has major implications for health expenditures. Figure 1
Figure 1. Private hospital beds and full-time doctors, 1977–96
05 Tangcharoensathien (jl/d)
20/10/99 8:29 am
Page 345
National Health Accounts in Thailand
345
shows two phases of private hospital growth in terms of beds.
There was a 140% increase in the 10-year period from 1977 to
1987, and a 192% increase in the nine years from 1987 to 1996.
The share of private hospital beds increased from one-tenth
to one-quarter of total beds in two decades. A similar trend
emerged for private full-time physicians (Lertiendumrong
1998). NESDB estimates, failing to capture health expenditure spent in the public and private sectors, have limitations
for finance monitoring, policy orientation and strategy
development.
framework was well demonstrated in several low- and middleincome countries.
Further, NESDB defines medical fringe benefits for government and state enterprise employee and dependants as part
of income transfer, not health expenditure. We regard this as
a payroll tax similar to social health insurance; this should be
counted as a public source of finance.
(1) How much was spent in terms of per capita and percentage of GDP?
(2) What were the proportions of consumption expenditure
and capital formation?
(3) What types of services were purchased, e.g. curative, promotion and prevention, and other public health programmes?
(4) At what levels of care were funds spent, e.g. primary,
secondary, tertiary care; public and private providers?
(5) Who paid for it – public and private sources?
Concern about the validity of NESDB indirect estimation
using the drug and non-drug approach, its inability to produce
useful breakdowns of expenditure, and the discrepancies
compared to SES, prompted us to develop National Health
Accounts in Thailand.
International methodological review
The historical evolution of health expenditure surveys has
been described in greater detail by Griffiths and Mills (1983).
Very useful manuals for health expenditure surveys in
developing countries were also proposed by Griffiths and
Mills (1982), and Mach and Abel-Smith (1983). They defined
health expenditure as spending on activities whose primary
purposes (regardless of effect) are health improvement.
Mach and Abel-Smith (1983) proposed a dummy matrix of
source of finance and area of expenditures. Sources of finance
consisted of public (MOPH, other ministries, local government, etc.) and private sources (industry, direct private
payment, insurance, self help, foreign aids, etc.). Areas of
expenditure consisted of central and district hospitals,
primary care centres, mental hospitals, dispensaries, and
public health programmes (e.g. disease control, transport,
training of doctors). Careful attention to transfer payments is
proposed to solve the problems of financial transfer from one
source to others, such as the MOPH subsidies to charitable
NGOs, then used by NGOs in providing care directly to the
communities. The transfer of funds among sources of finances
can cause confusion if there are many transactions.
Recently, Berman recommended solving the ‘financial transfer’ problem among financing agencies by using a two step
approach – the ‘source and uses matrix’ method (Berman
1996, 1997). The first step is the transfer of finance from ultimate sources to financing agencies. Ultimate sources, such as
ministry of finance, households and foreign aid, generally do
not purchase health care services directly. They play a purely
financing role. But financing agencies, after receiving funds
from ultimate sources of finance, purchase services directly
from public and private providers. In some cases, financing
agencies, such as MOPH, directly operate public health programmes. The second step is the use of funds from financing
agencies according to type of provider and function (administration, curative care, public health programmes). Berman’s
NHA development objective
When primary and secondary data on health expenditures are
available, it is more accurate to estimate from these sources
than indirect estimation based on an approach such as that
used by NESDB. In the immediate term, we aimed to develop
the national health account to answer the following questions.
The breakdown of personal health care (into ambulatory and
inpatient services at both public and private providers), selfprescribed drugs and others (such as public health programmes) is the long-term NHA development goal. This is to
facilitate international comparison.
Methodology
We reviewed health expenditure definitions and sought
consultations among users in the country to agree the operational definitions. Based on Berman’s methods (Berman
1997), we defined a list of ultimate sources of finance and
financing agencies. The first step is to develop a Dummy
Table to demonstrate the flow of funds from ultimate sources
to financing agencies. The second step was to develop a
second Dummy Table to demonstrate the flow of funds from
financing agencies to type of providers (public and private
institutions), functions (administration, curative care, public
health programmes) and type of expenditure (consumption
and capital formation).
Secondary data on expenditure were collected from financial
records at each financing agency for the fiscal year for public
sources (October to September) and calendar year for private
sources (January to December). The 1994 figures were used
as our first NHA model. Verification of data sources was done
rigorously. Where there was a severe lack of secondary data,
primary data collection through census and sampling survey
was carried out. Household health expenditure data mainly
relied on the 1994 SES. The table in the Appendix summarizes the detailed methodology used. Extrapolation of survey
data to national figures was made with caution.
The 1994 NHA findings were then compared with the 1994
NESDB estimates. Discussions were held with NESDB technical staff to reconcile figures when we found significantly
different results between the two methods. We also ran two
peer review workshops to scrutinize and discuss the validity
and reliability of NHA calculations, and one dissemination
4.55
49.35
0.23
0
0
0
0
33.58
326.03
100
2 834.99
2 500.01
4 588.07
56 941.77
2 716.02
141 818.06
2.00
1.76
3.24
40.15
1.92
100
actual expenditure 2 266.22
budget = expenditure
actual expenditure 1 909.57
budget = expenditure
actual expenditure 2 086.75
actual expenditure 36 644.67
actual expenditure 4 784.61
budget = expenditure
actual expenditure 5 573.62
budget = expenditure
actual expenditure 3 494.08
actual expenditure 392.89
1.18
2 545.19
0
4 588.07
56 941.77
2 575.61
69 993.52
26.97
3.92
7.02
4.04
1.18
7.02
0.80
44.69
289.80
2 500.01
0
0
0
6 449.38
38 243.95
5 561.79
9 954.00
5 728.07
1 666.91
9 949.93
1 132.55
0
0
0
0
0
1 666.91
292.45
0
0
0
0
0
0
0
0
0
0
106.83
63 382.23
807.37
0
0
0
0
2 535.51
0
0
0
0
0
1 666.91
0
0
37 144.13
5 561.79
9 954.00
5 728.07
0
4 881.20
6.20
1. MOPH
2. Other ministries
3. MOF: CSMBS
4. Local gov’t/municipality
5. State enterprises
6. Social Security Scheme
7. Workmen Compensation
13. Scheme
8. Private Insurance Plan
9. Employer-provided benefit
10. Traffic accident insurance
11. Households
12. Non-profit institution
13. TOTAL ULTIMATE
13. SOURCES OF FINANCE
14. Percentage
346
0
0
0
0
0
2 533.22
1 126.35
Ultimate sources of finance
––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
1. MOF
2. State enterprise
3. Employer
4. Households
5. Donors
6. TOTAL
7. %
8. Technical notes
20/10/99 8:29 am
Financing
Agencies
Table 1. Flow of funds from five ultimate sources of finance to 12 financing agencies, Thailand, 1994; budget figures (million Baht)
05 Tangcharoensathien (jl/d)
Page 346
Viroj Tangcharoensathien et al.
05 Tangcharoensathien (jl/d)
20/10/99 8:29 am
Page 347
National Health Accounts in Thailand
workshop for staff in NESDB and all other financing agencies
as well as users of NHA in the MOPH and universities.
Results
Health expenditure is defined as spending on activities whose
primary objectives are health improvement (Mach and AbelSmith 1983). Expenditure on health manpower production
was excluded as it was counted as education expenditure, but
on-the-job training of existing health workers was included.
Rural water sanitation programmes under the MOPH
Department of Health were included, but not the Water
Works Authority or Municipality sewage systems.
Not unexpectedly, we found a severe lack of expenditure
breakdown by type of provider and detailed functions to
suit dummy table two. Criteria were then developed to allocate aggregate figures into detailed breakdowns using two
rounds of peer review. For example, the MOPH budgets to
provincial and district hospitals are apportioned not only to
curative care but also to other public health functions such
as health promotion, disease control, immunization and
personnel training. State enterprise expenditure was given
70% to private and 30% to public hospitals, based on indepth interview with some enterprises. Health expenditure
by the Workmen Compensation Fund for work-related conditions are allocated 90% to private and 10% to public hospitals.
A manual with detailed descriptions of data validity in all
financing agencies, techniques of data verification, census and
survey methodology, and methods of extrapolating survey
data to national figures, was produced for subsequent versions of the NHA. This manual serves as a stepping stone for
long-term NHA development.
A list of five ultimate sources of finance was finalized.
(1) Ministry of Finance (MOF) transfers funds to MOPH,
other ministries and local governments, and also acts as
the financing agency responsible for the Civil Servant
Medical Benefit Scheme (CSMBS).
(2) State enterprises which act as both ultimate source and
financing agencies, providing fringe benefits to their
employees and dependants.
(3) Employers contribute to the Social Security Scheme
(SSS) and Workmen Compensation Scheme (WCS) and
also provide medical benefits to their employees.
(4) Households act as both ultimate source and financing
agencies. They contribute to SSS, pay premiums for
private insurance plans, third-party car insurance coverage and also directly purchase health care services.
(5) Donor organizations play a minor financing role through
the MOPH and other ministries.
An exhaustive list of 12 financing agencies was then finalized.
Seven agencies are classified as public sources: MOPH, other
ministries, CSMBS, local governments and municipalities,
state enterprises, Social Security Scheme, Workmen Compensation Scheme. The other five are classified as private
sources of finance: private health insurance plan, employer
347
medical benefit, third-party car insurance plan, households
and non-private institutions.
In our Dummy Table 2, there were four broad categories of
consumption expenditures: I. Administration; II. Care provided by public institutions; III. Care provided by private
institutions; IV. Other public health programmes. Under
these four categories, we further broke down expenditures
into 32 sub-categories. This was successfully done using criteria mentioned above.
The flow of funds from five ultimate sources of finance to
twelve financing agencies is shown in Table 1. In 1994, a total
of 141 818 million Baht1 was available for health, largely
financed by households (49.35%) and the MOF (44.69%).
Table 1 shows that households play a major role, at 40.15% of
the totals – MOPH 26.97%, CSMBS 7.02%, SSS 7.02% and
local government 4.04%. Overall, public sources of finance
(agencies 1–7) had a 50.95% share of the total budget and
private sources (agencies 8–12) had 49.07%. The technical
notes in Table 1 show a balance of unspent budget by the end
of the fiscal year, especially among public financing agencies.
We use budget figures in Table 1 to contrast actual expenditure in Table 2. Public financing agencies, such as the MOPH,
other ministries and local government, could not spend all
their allocated budgets by the end of the fiscal year.
Use of funds by 12 financing agencies is shown in Table 2. For
simplicity, we show in Table 2 only four major consumption
expenditure categories which were derived from 32 subcategories. We found 32 sub-categories to be feasible but this
breakdown required great effort. In 1994, the total health
expenditure was 128 305.11 million Baht (with an unspent
budget of 13 512.96 million Baht), of which 84.07% was consumption expenditure and 15.93% was capital formation. Of
the total consumption expenditure, 36.14% was spent on purchasing care from public providers, 32.35% on care from
private providers, 5.93% on administration and 9.65% on all
other public health programmes.
Of the total health expenditure of 128 305.11 million Baht in
1994, households paid 44.38%, MOPH 28.56%, CSMBS
7.76%, local government 4.34%, other ministries 3.8%, and
Social Security 2.7%. Others were insignificant.
In Table 2, the budget in row 0 could apply precisely with
public sources of finance when there is variation in budget
absorption capacity. Unspent but obligated budget will be
transferred for the next year’s spending, and there is then a
discrepancy between budget and expenditure. CSMBS and
state enterprise benefit employed a ‘fee-for-service retrospective reimbursement model’, thus expenditure is then
equal to the budget. The budget for Social Security refers to
tripartite contributions to the Social Security Fund, whereby
WCS is employer contribution to the Workmen Compensation Fund. Positive balances to the fund arise when less is
spent than the contribution collected. However, this budget
may not apply well to private sources of finance, such as
employer-provided benefit and household expenditure, when
expenditure equals budget. It could apply well to private
insurance and traffic insurance where the budget refers to the
0
8 248.52
1 705.48
0
9 954.00
100.00
0
0
9 954.00
7.76
0.00
1.00
818.92
138.44
0
4 331.78
5 289.13
94.90
284.49
5.10
5 573.62
4.34
154.45
0.97
0
483.26
1 127.62
0
1 610.88
96.64
56.03
3.36
1 666.91
1.30
0.00
1.00
525.44
1 156.08
1 734.12
0
3 415.64
97.76
78.44
2.24
3 494.08
2.70
6 455.85
0.35
4.82
38.55
346.96
1.18
391.51
99.65
1.38
0.35
392.89
0.31
739.66
0.35
1 132.55
567.00
254.89
1 444.33
0
2 266.22
100.00
0
0
2 266.22
1.77
568.77
0.80
2 834.99
119.05
531.98
1 848.98
0
2 500.01
100.00
0
0
2 500.01
1.95
0.00
1.00
2 500.01
917.61
615.02
376.95
0
1 909.57
100.00
0
0
1 909.57
1.49
2 678.50
0.42
4 588.07
0
16 768.40
32 908.09
0
49 676.49
87.24
7 265.28
12.76
56 941.77
44.38
0.00
1.00
56 941.77
298.20
1 114.74
13.56
171.32
1 597.82
76.57
488.93
23.43
2 086.75
1.63
629.27
0.77
2 716.02
7 611.08
5.93
46 372.42 36.14
41 506.09 32.35
12 378.01
9.65
107 867.59 84.07
84.07
20 437.51 15.93
15.93
128 305.11 100.00
100.00
13 512.96
0.90
141 818.06
11
2 945.46
0
218.38
3 174.84
65.13
1 699.77
34.87
4 874.61
3.80
687.18
0.88
9 949.93
4 349.04
14 077.08
0
7 655.35
26 081.47
71.17
10 563.20
28.83
36 644.67
28.56
1 599.28
0.96
1 666.91
348
5 728.07
5 561.79
38 243.95
0. Budget
Consumption Expenditure
I Administration
II Public institutions
III Private institutions
IV Public health programmes
Total consumption
% consumption expenses
Total capital formation
% capital expenses
Total health expenses
%
Balance
Exp/Budget ratio
14.
%
20/10/99 8:29 am
9 954.00
FINANCING AGENCIES
––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
Public Sources
Private Sources
–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
–––––––––––––––––––––––––––––––––––––––––––––––––––
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
MOPH
Other
MOF/
Local
State
Social
WCS
Private
Employer Traffic
Household Non-profit TOTAL
ministries CSMBS
Gov’t
enterprise Security
insurance benefit
insurance
institutes
Flow of funds from 12 financing agencies to types of expenditure, Thailand, 1994; actual expenditure figures (million Baht)
Budget &
Types of
Expenditure
Table 2.
05 Tangcharoensathien (jl/d)
Page 348
Viroj Tangcharoensathien et al.
05 Tangcharoensathien (jl/d)
20/10/99 8:29 am
Page 349
National Health Accounts in Thailand
349
total net premium collected, then expenditure less than the
budget reflects the profit margin.
with public and private proportions at 46:54 in NHA and
18:82 in NESDB.
It is worth noting that the MOPH, other ministries and
local governments did not purchase care from private institutions. In contrast, other financing agencies purchased care
both from public and private providers to a varying degree.
The state enterprise medical-benefit scheme, workmen compensation scheme, private insurance plans, employerprovided benefit and households purchased care more
from private than from public providers. In particular,
households purchased care twice as much from private than
from public providers. CSMBS spent more money purchasing care from public than private providers. Interestingly,
public health programmes were mainly paid by public
sources of finance, especially the MOPH, local government
and other ministries, with the exception of non-profit institutions. Other major private sources spent almost entirely on
curative care.
We spoke with NESDB technical officers to explain and reconcile the findings. Discrepancies in public and private proportions are due to definitions of public and private sources.
NESDB defined CSMBS, and state enterprise medical
benefit, as income transfers (fringe benefit) to household
members, and regarded this as a private source.
Table 3 provides a macroscopic view: public sources of
finance were responsible for 48.79%, and private for 51.21%,
of the total 1994 health expenditure. Total health expenditure accounted for 3.56% of GDP; of which consumption
expenditure was 3% and capital formation 0.57%. In contrast, the NESDB consumption expenditure estimate in 1994
was 180 516 million Baht or 5.01% of GDP, of which private
sources dominated (82.17%) and public sources played a
minor role (17.83%). The discrepancy in consumption
expenditure between the two estimates is 2.01% of GDP,
Table 4 presents results from reconciliation between NHA and
NESDB estimates. Technical staff in NESDB re-calculated
their data by adjusting down private drug consumption – the
drug industry was critical of a double estimation on drug consumption. Private consumption expenditure thus went down
from 148 334 to 77 000 million Baht.2 NESDB adjusted public
capital formation, assuming 20% of total public expenditure
(8045 million) was capital formation and assuming that private
capital formation was equal to the NHA estimates of 7265
million (see household capital formation in column 11 in Table
2). At the same time, we reclassified CSMBS and state enterprise medical benefit as private sources to be comparable with
NESDB estimates.3 After reconciliation, the two estimates
came up with a similar result, a total health expenditure of
3.56% of GDP for NHA and 3.46% of GDP for reconciled
NESDB. The proportions for public and private sources are
40:60 for NHA and 32:68 for NESDB. This finding shakes the
previously understood private financing domination (more
than 80%) in the Thai health system, and the official records
of more than 5% of GDP being spent on health.
Table 3. Comparison between NHA results and NESDB (million Baht)
Consumption exp.
%
Capital formation
%
Total
%
NHA
–––––––––––––––––––––––––––––––––––––––––––––––
Public
Private
Total
% GDP
NESDB estimates
–––––––––––––––––––––––––––––––––––––––––––––––
Public
Private
Total
% GDP
49 917.47
46.28
12 683.31
62.06
62 600.78
48.79
0.57
32 182
17.83
na
148 334
82.17
na
180 516
100
na
3.56
na
na
na
57 950.11
53.72
7 754.21
37.94
65 704.32
51.21
107 867.58
100
20 437.52
100
128 305.10
100
3.00
5.01
GDP in 1994 was 3 600 907 million Baht.
Table 4. Comparison of NHA results and NESDB data after reconciliation (million Baht)
Type of expenditure
NHA
–––––––––––––––––––––––––––––––––––––––––––––––
Public
Private
Total
% GDP
Reconciled NESDB
–––––––––––––––––––––––––––––––––––––––––––––––
Public
Private
Total
% GDP
Consumption exp.
%
Capital formation
%
Total health exp.
%
38 352.59
35.56
12 627.28
61.78
50 979.87
39.73
32 182
29.48
8 045
52.55
40 227
32.31
69 514.99
64.44
7 810.24
38.22
77 325.23
60.27
107 867.58
100
20 437.52
100
128 305.10
100
3.00
0.57
3.56
77 000
70.52
7 265
47.45
84 265
67.69
109 182
100
15 310
100
124 492
100
3.03
0.43
3.46
05 Tangcharoensathien (jl/d)
20/10/99 8:29 am
350
Page 350
Viroj Tangcharoensathien et al.
Discussion
Our NHA development using a two-step approach makes it
possible to analyze the flow of funds from source of finance to
financial intermediaries and the use of funds by types of
providers. It facilitates an identification and assessment of
resource allocation mechanisms between sources and intermediaries, and between intermediaries and providers, as well
as direct payment by users to providers.
Similar exercises in China have also demonstrated discrepancies between the NHA and National Statistics Bureau estimations. This has lead to a recent national household health
expenditure survey in China (World Bank 1995). In the Thai
case, the major cause of discrepancy is the inaccurate NESDB
estimation of private consumption expenditure, based on
drugs which suffered greatly from under-reporting of both
domestic and import drugs and the over-estimation of percentage mark-up. Further, there was less than comprehensive
coverage of financing agencies, such as local government,
defence and university ministries, employer-provided
medical benefit and non-profit institutions. The re-calculation
of private consumption from 148 334 to 77 000 million Baht
(48% reduction) reflects methodological problems. However,
NHA private consumption relied solely on the NSO-SES,
which may also suffer from sampling and non-sampling bias.
There is a tendency towards under-estimation due to recall
biases.
Our NHA figures are low as a share of GDP in relation to
other countries, and also much lower than previous NESDB
estimates. Continuous dialogue and partnership between the
Health Systems Research Institute and NESDB at the policy
and technical levels call for further development with financing agencies and the fine tuning of estimation methodology,
instead of argument on whose work is correct. It is too early
to make the assertion that the NHA figure is more accurate.
It needs further research and development. However, NHA
has a more rational methodology than the NESDB estimation in several ways (see Table 5).
• NHA public consumption expenditure is reliable, with
extensive coverage of all sources; the NESDB public
expenditure tends to be under-estimated.
• NHA private consumption expenditure is more comprehensive than NESDB as other sources such as private
health insurance plans, third-party traffic insurance,
employer-provided benefit and non-profit institutions are
captured.
• NHA produces accurate capital expenditure in public
(from direct data collection) and private (from estimation
of private medical institutions, new establishments and
expansion using MOPH Medical Registration Division
data), whereas NESDB did not report capital formation in
health from other sectors.
• However, as the NHA bases itself on the SES, households
tend to under-report expenditure due to recall problems.
There are also sampling biases whereby urban households
tend to be non-cooperative samples.
As elaborated in Table 5, we are confident that NESDB is
highly over-estimated, while NHA is slightly underestimated.
National health spending should then fall in-between, but
rather shifts to the left. Two possible solutions to solve household under-reporting are proposed: firstly, the survey of
revenue generated by public and private providers from outof-pocket spending as suggested by Berman (1997); and secondly, the fine tuning of the SES questionnaire to avoid
non-sampling bias and a special survey of urban household
health expenditure to solve sampling bias (see discussion
below).
In a pluralistic health care system like that of Thailand and
other developing countries, the breakdown of expenditure
into public and private health institutions serves the national
policy interests. However, the 1994 version could not achieve
the breakdown of expenditure for ambulatory care from
inpatient care. The monitoring of outpatient and inpatient
expenditures would prompt policy-makers to introduce
different policy instruments to properly manage the inflation
of outpatient or inpatient expenditure in the future.
In order to sustain the NHA report, major development work
is needed, such as the development and amendment of existing financial information systems and the breakdown of health
expenditure by each financing agency to suit the NHA analysis. This must gain policy support, especially from affected
government departments led by NESDB. We propose the
Table 5. Comparison of NHA and NESDB estimates
Type
NHA
NESDB
Public consumption
Adequate: comprehensive coverage, actual
expenditure assessed.
Adequate: comprehensive coverage, take into
account unspent budget
Lend heavily on SES, likely to underestimate
due to sampling and non-sampling bias.
Missing some financing agencies, then underestimate
Public capital
Private consumption
Private capital
Total expenditure
Adequate coverage of new established
private clinics, hospitals, major capacity
expansion
Slightly underestimate due to private
consumption.
No estimation
Lend heavily on very inaccurate drug consumption,
tends to overestimate. Missing other financing agencies,
then underestimate this part.
No estimation
Highly overestimate, due to overestimation of the
largest proportion on private consumption.
05 Tangcharoensathien (jl/d)
20/10/99 8:29 am
Page 351
National Health Accounts in Thailand
351
Table 6. Future NHA structure
Use of funds
Public sources
Private sources
–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
1
2
3
4
5
6
7
8
9
10
11
12
Total
%
I. Personal health care
1. Inpatient care
• Acute public hospital care
• Acute private hospital care
• Psychiatric hospitals
• Other specialized hospital care
2. Ambulatory care
• Physician services in public sector
• Physician services private sector
• Dental services
• Other professional health services
• Other non-professional services
3. Medical goods
• Pharmaceuticals
• Therapeutic appliances
II. Other health expenditures
4. Investment on facility: Public
5. Investment on facility: Private
6. Administration
7. Safe Water Programme
8. Sanitation Programme
9. Communicable diseases control
10. Immunization
11. Rehabilitation services
12. Health education
13. Bio-medical research
14. Health system research
15. Food and Drug Administration
16. Other services
Total Health Expenditure
future NHA format in Table 6, not only to serve the national
policy interest but also to facilitate international comparisons.
As household expenditure takes a lion’s share of total health
expenditure, more accurate household expenditure information would significantly improve overall NHA validity. As
some National Statistical Office technical staff are closely
involved in NHA development, the questionnaire amendment
of each biennial SES is easily accomplished. The 1998 SES
questionnaire distinguishes expenditure on self-prescribed
drugs, on ambulatory care in public and private facilities, and
other expenses (dental, optical) in the month prior to the
interview date. Household expenditure for inpatient care in
public and private hospitals will be asked about for the
previous year. This helps the breakdown but may not solve
errors from sampling bias, especially urban household nonrespondents. We are thinking about introducing a small-scale
survey of urban household health expenditure, where there is
a higher access to health care. This may provide some correcting factors for the SES.
Future NHA development aims at providing a biennial
report, starting with 1994, then 1996, 1998 and so on. As we
use the SES for household health-expenditure estimation, the
05 Tangcharoensathien (jl/d)
352
20/10/99 8:29 am
Page 352
Viroj Tangcharoensathien et al.
proposed time line is in association with the biennial NSOSES. In March 1998, the NESDB Secretary General agreed
to be fully involved in subsequent NHA developments in collaboration with major stakeholders, users of NHA and
financing agencies. It is hoped that subsequent NHAs will
provide more accurate figures and breakdown and could
eventually replace the current NESDB series.
Conclusions
We have demonstrated the integration of research and
development work in a prototype NHA, in consultation with
stakeholders over definitions and methodologies. The exercise
showed discrepancies and challenges for further research and
development work. Further development of NHA is not possible without partnership-building with financing intermediaries, both public and private. The questionnaire amendment
of the national household SES is most important when the
household takes a major share of total health expenditure.
Public sources of finance are not difficult to access but criteria
for breakdown are needed. In a pluralistic health care system,
the breakdown of spending on public and private providers is
important to support the development of proper policy instruments. Sustainable NHA development and utilization of NHA
for monitoring, planning and policy development must be the
primary objective. International comparisons through standardized definitions cannot be realized without national
capacity and sustainable development. The 1996 and 1998
NHA exercises are currently well under way, supported by the
Health Systems Research Institute in collaboration with all
concerned financing agencies and NESDB, with the aim of
achieving the outlook proposed in Table 6.
Endnotes
1
The 1994 exchange rate was 25 Baht to 1 US$.
This is an official reconciliation at technical staff level for the
purpose of comparison with NHA. This figure was for internal use
and comparison with NHA only, not an official record.
3 Note that we reclassified here for the purpose of comparison.
We insist that CSMBS and state enterprise medical fringe benefit are
regarded as social health insurance arrangement and categorized as
public sources of finance, not income supports to beneficiaries.
2
References
Bennett S, Tangcharoensathien V. 1994. A shrinking state? Politics,
economics and private health care in Thailand. Public Administration and Development 14: 1–17.
Berman P. 1996. National Health Accounts in Developing Countries:
Appropriate Methods and Recent Applications. Data for Decision Making Project. Boston: Harvard University.
Berman P. 1997. National Health Accounts in developing countries:
appropriate methods and recent applications. Health Economics 6: 11–30.
Griffiths A, Mills M. 1982. Money for Health: A Manual for Surveys
in Developing Countries. Sandoz Institute for Health and SocioEconomics Studies and the Ministry of Health of the Republic
of Botswana. Geneva.
Griffiths A, Mills M. 1983. Health Sector Financing and Expenditure
Surveys. In: Lee K, Mills A (eds). The Economics of Health in
Developing Countries. Oxford: Oxford University Press.
Jaidee S. 1987. Report on policy and strategy for drug price control.
Bangkok: Chulalongkorn University, Drug Study Group.
Lertiendumrong J. 1998. Economic crisis and impact on private
health sector in Thailand. Paper presented at the Second European Congress on Tropical Medicine, 14–18 September 1998,
Liverpool University.
Mach EP, Abel-Smith B. 1983. Planning the Finances of the Health
Sector: A Manual for Developing Countries. Geneva: World
Health Organization.
Ministry of Finance. 1997. Comptroller General Department, Budget
execution report. Bangkok: Ministry of Finance.
Myer CN, Mongkolsmai D, Causino N. 1985. Financing health services and medical care in Thailand. Bangkok: USAID.
National Account Division. 1988. National Income of Thailand, 1988
edition. Bangkok: National Economic and Social Development
Board.
National Statistics Office. 1986. Report on household Socio-Economic Survey 1986. Bangkok: Veteran Administration Press.
Newbrander W, Carrin G, Le Touze D. 1994. Developing countries’
health expenditure information: what exists and what is
needed? Health Policy and Planning 9(4): 396–408.
Tangcharoensathien V. 1991. Health Expenditure in Thailand:
Review of methodologies. Bangkok: Health Planning Division,
Ministry of Public Health.
Tangcharoensathien V, Nittayaramphong S. 1994. Thailand: a private
sector out of control? Health Policy and Planning 9(1): 31–40.
Wibulpolprasert S, Buapradubkul P, Chantarasathit N. 1987. Health
care financing, Thailand. Bangkok: Ministry of Public Health.
Wibulpolprasert S (ed). 1994. Drug system in Thailand. Bangkok:
Health Systems Research Institute, Thailand Health Research
Institute and Food and Drug Administration.
World Bank. 1993. World Development Report 1993: Investing in
Health. Oxford: Oxford University Press.
World Bank. 1995. China National Health Accounts Assessment.
Washington DC: Human Development Department.
Acknowledgements
The authors wish to thank technical staff in various organizations for
providing useful data for the development of the first version of
NHA, especially NESDB, Ministry of Finance Comptroller General
Department, Ministry of Labour and Social Welfare Social Security
Office, Ministry of Interior Local Administration Department, Ministry of Commerce Insurance Department, National Statistical
Office. We are grateful to the Health Systems Research Institute who
funded this work.
We also express our sincere thanks to Professor Anne Mills of the
London School of Hygiene and Tropical Medicine, and Joseph
Kutzin of the World Health Organization for their comments on the
earlier draft and anonymous peers for their invaluable suggestions.
Support from the Health Systems Research Institute and Thailand
Research Fund Senior Research Scholar Programme are highly
appreciated.
Biographies
Viroj Tangcharoensathien is a full-time researcher in health financing and health economics at the Health Systems Research Institute
and also a Senior Research Scholar supported by Thailand Research
Fund. His current work is involved in government employee health
insurance reform in Thailand, the development of National Health
Accounts, caesarean section studies and drug management reform.
Adit Laixuthai is an economist. He formerly worked with Chulalongkorn College of Public Health, and is now working with the Thai
Farmer Bank Research Office.
Jitpranee Vasavit is a statistician at the National Statistical Office who
is responsible for the National Household Socio-Economic Survey.
Nuan-anan Tantigate is a research manager at the Health Systems
Research Institute. She coordinates several research programmes.
05 Tangcharoensathien (jl/d)
20/10/99 8:29 am
Page 353
National Health Accounts in Thailand
Wiphant Prajuabmoh-Ruffolo is a demographer working at Chulalongkorn Institute of Population.
Duangkamol Vimolkit is a health policy analyst working at the
National Economic and Social Development Board.
Jongkol Lertiendumrong is a researcher attached to the Senior
Research Scholar Programme jointly funded by Thailand Research
353
Fund and the Health Systems Research Institute. She is currently
doing her PhD in Health Care Financing at the London School of
Hygiene and Tropical Medicine.
Correspondence: Viroj Tangcharoensathien, Health Systems
Research Institute, 5th floor, Mental Health Department Bldg,
Tivanond Road, Nonthaburi 11000, Thailand. Email: viroj@hsrint.
hsri.or.th
Appendix
Summary table of methodological approaches
Financing agencies
Existing health expenditure data system
NHA development approach
1. MOPH and other related
ministries
Good computerized system of aggregate
health expenditure – easy to retrieve
expenditure by programmes (due to
programme budgeting system in place) but
no breakdown of expenditure as desired.
Criteria developed to breakdown expenditure
by type according to dummy table.
Difficulties for future breakdown of
ambulatory and inpatient care anticipated.
2. MOF / CSMBS
Good computerized system of actual
expenditure breakdown by outpatient (at
public facilities) and inpatient (at public
and private hospitals).
3. Local Government and
Municipalities
No systematic national data compilation.
Health is a small item.
Survey to all 140 municipalities, estimated
figures for non-respondents based on size of
municipality (number of population served).
Data from Bangkok Metropolitan Authority
is quite complete. Local government
expenditure becomes more important as
policy emphasises decentralization.
4. State Enterprises
No systematic national data compilation.
Health expenditure was reported as
personnel related expenditure.
Survey to all 62 enterprises, estimated figures
for non-respondents by size of enterprise
(number of employees).
5. Social Security Scheme and
Workmen Compensation Scheme
Good computerized database in the
Social Security Office. Care needed to
exclude cash benefit as part of sickness
and maternity benefit.
Need further breakdown of outpatient and
inpatient expenditure.
6. Private Insurance Plan
and Traffic Accident Insurance
Ministry of Commerce Insurance
Department has neither aggregate nor
disaggregated data.
Access to health expenditure data in private
insurance companies is problematic. Good
collaboration by top three (more than 70% of
market share), estimated for non-respondents.
Future partnership with Insurance
Department to compile routine health
expenditure data from all companies.
7. Employer-Provided Benefit
No responsible organization for data
compilation.
Estimation was made from the NSO biennial
surveys of industries and business enterprises
expenditure pattern.
8. Households
Estimation of consumption expenditure
from NSO biennial SES. There is no
information on capital formation.
Further development of questionnaire on
expenditure of ambulatory and inpatient care
at public and private facilities, in addition to
self-prescribed drugs and others. Capital
formation was estimated based on the number
of new establishments and expansion of
private medical institutions (clinics, hospitals,
pharmacies, dental clinics) and unit cost of
investment. This data comes from MOPH
Medical Registration Division.
9. Non-Profit Institution
Ministry of Interior has only directories
of non-profit institutions, foundations,
associations. No other information.
Survey was done with low response rate.
Several large NGOs active in health
programmes were interviewed and
expenditure estimated.