Health Policy and Planning Advance Access originally published online on December 19, 2006
Health Policy and Planning 2007 22(1):49-59; doi:10.1093/heapol/czl037
The distribution of net benefits under the National Health Insurance programme in Taiwan
1 Institute of Public Health, National Yang Ming University, Taipei, Taiwan.
2Harvard School of Public Health, Boston, MA, USA.
3Bureau of National Health Insurance, Taipei, Taiwan.
*Corresponding author. Department of Population and International Health, Harvard School of Public Health, 124 Mount Auburn Street, 410S Cambridge, MA 02138, USA. E-mail: wyip{at}hsph.harvard.edu
 |
Abstract |
|---|
 Top Abstract IntroductionMethodsResultsDiscussionConclusionReferences |
|---|
The redistributive effects of a social insurance programme are determined by how the programme is paid for—who pays and how much do they pay?—and how the benefits are distributed. As a result, the redistributive effects of a social health insurance programme should be evaluated on the basis of its net benefit—the difference between benefits and payment. Among the rich body of empirical analysis on equity in health care financing, however, most studies have relied on partial analysis, assessing equity by source of financing while ignoring the benefit side, or looking at equity in benefits but ignoring the funding side. Either approach risks misleading findings. In this study, therefore, the primary objective was to assess the distribution of net benefits across income groups under Taiwan's National Health Insurance (NHI) programme.
This study observed a nationally representative sample of 74 012 NHI enrolees from 1996 to 2000. The unique NHI databases in Taiwan provide comprehensive enrolment and utilization information, and allowed linkage to each enrolee's income tax files. In addition to crude estimates, two-part models and ordinary least-square models were used to adjust inpatient and outpatient benefits for health care needs (age, sex, major disease status and physical disability).
After adjusting for health care needs, the distribution of net benefits showed an apparent pro-poor pattern, with the lowest income group receiving the highest net benefits (NT$3353) and the top income group receiving the lowest net benefits (–NT$3072) in 1996. Although a clear pro-poor pattern was observed among those enrolees who paid wage-based premiums, this vertically equitable pattern was less evident among the enrolees who paid fixed premiums. Overall, a trend of increasing net benefits was observed in all income groups between 1996 and 2000, and all the NHI enrolees can be considered better off over time.
In addition to contributing to the limited literature on equity in net benefits, the study provides an important policy reference to developing countries with large underground economies and relatively small populations of regular wage-earners as it indicates that using fixed premiums as a major financing scheme may pose a serious equity concern and policy challenge.
Key Words: Taiwan, national health insurance, net benefits
KEY MESSAGES
|
|
Introduction |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionConclusionReferences |
|---|
In addition to providing financial risk protection, social health insurance is often promoted for equity reasons. The redistributive effects of a social insurance programme are determined by how the programme is paid for—who pays and how much do they pay?—and how the benefits are distributed. As a result, the redistributive effects of a social health insurance programme should be evaluated on the basis of its net benefit—the difference between benefits and payment (Zschock 1989
; Culyer and Wagstaff 1993). Among the rich body of empirical analysis on equity in health care financing, however, most studies have relied on partial analysis, assessing equity by source of financing while ignoring the benefit side, or looking at equity in benefits but ignoring the funding side (Culyer et al. 1992; Wagstaff et al. 1992; van Doorslaer et al. 1999; Wagstaff et al. 1999; van Doorslaer et al. 2000). Either approach risks misleading findings.
Limited research exists on the distribution of the net benefits of social health insurance programmes. Conducting such studies faces three major challenges: (1) the lack of complete information on sources of both funding and benefits at the household/individual level; (2) the difficulty in linking insurance-related information to income-tax data along with information on household structure; (3) the complexity involved in constructing appropriate algorithms for allocating benefits and payments. A study that tried to overcome these obstacles and estimate the redistributive consequences of public health care programmes in the province of Manitoba, Canada (Mustard et al. 1998) showed a progressive redistributive effect at the household level in the distribution of non-cash health care benefits financed purely by taxation sources. While the results provide a lesson for developed countries where their health insurance programmes are financed through general taxation, they may not be generalizable to systems not financed entirely by general tax revenue, such as a social health insurance system.
The primary objective of this study was to assess the distribution of net benefits across income groups under Taiwan's National Health Insurance (NHI) programme. Taiwan established its NHI in 1995 with the objective of providing equal access to health care for all. It is financed by a combination of premium and tax revenue through government subsidies. As a single insurer, the NHI has also maintained detailed individual-level data on enrolment, health care utilization, and income, which allows us to analyze the distribution of net benefits. In addition to filling a knowledge gap in the redistributive effects of social health insurance, our results will also provide valuable lessons for economies that have a relatively large non-wage sector which renders wage-based premium financing infeasible.
Health care financing in Taiwan
The NHI is a social insurance system financed by premiums and taxes. In 2000, premiums from the insured and employers constituted 72% of NHI revenues, and the remaining 28% was subsidized by the government through taxes. Enrolment is mainly through employment, and enrolees are classified into one of the subcategories of six main insurance categories, according to their occupation (Table 1). Premium contributions are collected in two ways: (1) waged-based premiums paid by regular wage earners, and (2) fixed premiums (varied according to job category) paid by those without a well-defined monthly wage. For those individuals in category I, such as the employees of public or private enterprises, and employers, whose monthly wage information can be easily obtained, premiums are based on wages. In 2000, there were eight wage brackets with a cap of NT$87 600 (2000 annual average: 1 US$ = NT$31.225). The contribution rate was 4.25% for all wage brackets. Thus, premiums are higher for those in higher wage brackets. Those in categories II-VI, including seasonal or temporary workers without a fixed employer, farmers and fishermen, whose monthly wages are not well defined and are more difficult to assess, are enrolled in NHI through various workers’ associations or the local government. They pay fixed premiums, pre-determined by the Bureau of National Health Insurance (BNHI) according to occupation.
|
The wage-based premium is shared by an individual, his or her employer and the government. The premium contribution of each source varies by insurance category. For example, in private enterprises, employees pay 30%, the employer pays 60% and government pays the remaining 10%. Employers and the self-employed pay 100% of their premiums. The government subsidizes 100% of the premiums for low-income individuals, veterans and aborigines. Premiums for dependants are levied on a per capita basis; an individual pays for a maximum of three dependants. See Table 1 for a more detailed description of the insurance categories and shares of premium contribution by category.
Although Taiwan's NHI was designed to be pro-poor, the dual financing schemes may not necessarily lead to pro-poor redistribution. The fixed premium and the cap set at NT$87 600 for wage-based premiums may limit the system's redistributive function.
Benefits coverage
The NHI programme offers comprehensive and equal benefit coverage to all its enrolees. The benefits package includes outpatient services in clinics and hospitals, inpatient hospital services, dental services, Chinese medicine services, diagnostic tests and examinations, prescription drugs and certain over-the-counter drugs, preventive services, day care for the mentally ill and home care. Cosmetic surgeries, long-term care, dentures, hearing aids and prosthetics are not covered.
In order to minimize the moral hazard inherent in a comprehensive universal health insurance programme, NHI beneficiaries bear some cost-sharing obligations. For outpatient care, from 1996 to 2000, beneficiaries paid co-payments of NT$50 for clinic visits or outpatient visits to district hospitals, NT$100 for an outpatient visit to regional hospitals and NT$150 for an outpatient visit to academic medical centres. For inpatient services, beneficiaries are required to pay co-insurance for medical services as well as for the cost of beds and meals. The payment varies with the length of stay: 10% for the first 30 days, 20% for days 30 through to 60, and 30% for 61 days and beyond (Bureau of National Health Insurance [Taiwan] 2005). On average, the co-payment rate is relatively lower in Taiwan than in other countries.
Although the NHI benefit coverage was designed to be comprehensive and equal for all enrolees, due to differences in disease patterns and health care utilization, the distribution of benefits to rich and poor may not necessarily be equal. Hence, it is important to examine whether NHI's equal and comprehensive package has achieved equity in meeting health needs.
|
Methods |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionConclusionReferences |
|---|
Study design and sample
The NHI sample files, constructed and managed by the National Health Research Institute (NHRI), consist of comprehensive utilization and enrolment information for a nationally representative sample of 100 000 NHI beneficiaries out of a population of 21 400 826 enrolees throughout Taiwan in 2000. A random sampling design was used by the NHRI to select a representative sample (National Health Research Institute [Taiwan] 2005a, b). After excluding those individuals whose files lacked complete information, household registration record or NHI enrolment record, or who died during the study period, the final sample contained 74 012 individuals. This study followed the study sample for the fiscal years 1996, 1998 and 2000 to estimate the net benefits of the NHI programme by income quintiles.
Data
Seven electronic databases were linked to construct a database for the study. The NHI enrolment files provide information on enrolment status, premiums paid and physical disability status. The NHI ambulatory care claims files contain records of the use of outpatient services at the individual level, and the NHI inpatient files contain records of each admission case. Records include information such as gender, date of birth, date of service and reimbursements of all insured physician services, procedures, laboratory tests, diagnostic imaging and prescription drugs for each outpatient care visit or hospital admission. The NHI major-disease file identifies individuals in the sample with serious diseases such as cancer and major mental illnesses. In addition, the household registry was used to provide information on the aboriginal status and residential location of each individual in the sample. Death certificates were used to track those who died during the study period from 1996 to 2000.
The income tax files covering the years 1998–2000, managed by the Ministry of Finance, were also linked to provide gross income information on average individual and household incomes over the 3-year period, number of household members within the economic family unit, number of household members aged 70 or over, and number of household members aged 20 or under. Unfortunately, the Ministry of Finance did not provide information on actual tax paid by individuals or households. Of the sample subjects, about 19% did not file a tax return for all 3 years. The majority of these individuals were exempt from income tax mainly because their annual incomes were below the level at which tax returns are required. Hence, these 19% of 74 012 individuals were included in the lowest income quintile (Q0). The distribution of income tax data in our sample is similar to that in the general population. Unique, anonymous individual identification numbers and dates of birth were used to link all seven datasets.
Variables
Average per capita household income
A measure of total household income was derived from the income tax files. Each individual's household income was defined as total household gross income and calculated by summing the gross incomes from the income tax returns of all members of the same economic household unit. Household income was then adjusted for household structure (number of individuals aged 70 or above and number of individuals aged 20 or under) according to the equivalence scale proposed by Aronson et al. (1994) and Buhmann et al. (1988):
|
|
and 
, we followed Wagstaff et al. (1999) to set the two parameters at a value of 0.5. Then we divided the household-structure-adjusted household income by the number of household members to get a per capita household income. We then constructed the average per capita household income by averaging the per capita household incomes over the years 1998–2000. Individuals were grouped into quintiles of approximately 15 000 individuals each according to average per capita household income. The lowest quintile group (Q0) contained the poorest individuals including those who did not file any income tax return. Premium
A measure of premiums was derived from the enrolment file. Premiums are collected from three sources: individual contribution, employer contribution and government subsidy. Share of premium contribution by the insured, employer and government for each insurance category is specified in Table 1. The formulae used by the Bureau of National Health Insurance to calculate individual, employer and government contribution of premium were as follows:
Total employer contribution for a household = insurable wage x premium rate (4.25%) x employer's share of premium x (1 + national average number of dependents per insurance household)
Total employer contribution for a household = insurable wage x premium rate (4.25%) x government's share of premium x (1 + national average number of dependents per insurance household)
Premium per subject = premium ÷ actual household size
In order to relieve possible overwhelming financial pressure on large families, the government set the maximum number of payable dependants at three. Furthermore, in order to prevent employers from discriminating against employees of larger families, instead of being based on number of payable dependants, the calculation of employer contribution is based on a national average number of dependants per household. The national average numbers of dependants per household were 1.10, 0.95 and 0.88 in 1996, 1998 and 2000, respectively (Bureau of National Health Insurance [Taiwan] 2001).
On the other hand, due to data unavailability, the number of payable dependants and actual household size of each subject in our study sample were not available. So a national average number of payable dependants and a national average household size under each insurance category were used as proxies, respectively.
Net payment
Net payment at the individual level includes individual premium contribution, employer premium contribution and co-payments for using the NHI health services. Government subsidies to premium contribution from general taxation were not included in the calculation of net payment because we were unable to precisely allocate the tax burdens to each individual due to the lack of data on actual taxes paid by each individual. This limitation will be discussed in more detail below. The data on co-payments comes from the NHI claims files. However, claims files do not record direct out-of-pocket payments for uninsured services, registration fees, fees for meal and room upgrades, or payments for special nursing-aid services in hospital. In this study, the distributive effects of the three sources of net payment were investigated separately.
Total health care benefit
In this study, total health care benefit is defined as the consumption of insured health care services, as measured in NT-dollars. Detailed financial information regarding provider payments can be found in the computerized NHI medical services claims files. Total health care benefits included consumption of all services offered in ambulatory care visits and hospital admissions. The distributive effects of ambulatory care, hospital care and total benefits were analysed separately.
Net benefits
Net benefits were calculated by subtracting net payments (individual contribution, employer contribution and co-payments) from the total health care benefit.
Statistical analyses
The unit of observation of this study was each individual in the sample. Both benefits and payment measures were adjusted for the Consumer Price Index and defined in real terms. As health care needs are a likely confounder, all benefits measures were adjusted for this factor. In this study, a set of four variables (age, sex, major disease status and physical disability) was used to adjust for differences in health care needs across each income quintile. The distribution of each variable across income quintile is presented in Table 2. Inpatient expenditure was modelled using the two-part models (Duan et al. 1983; Cretin et al. 1990; Jones 2000). In the first part, the probability of hospitalization was estimated using logistic regression. The inpatient expenditure, conditional on a positive inpatient use, was then estimated using ordinary least square (OLS) linear regression. Since the distribution of medial expenditure was highly skewed, the natural logarithm of inpatient expenditure was used in the models. Predicted log medical expenditures were re-transformed to a raw scale in order to calculate the predicted total medical expenditure using the smearing technique (Duan 1983). Outpatient expenditure was modelled using OLS linear regression since over 80% of individuals incurred outpatient expenses. SAS and STATA were the statistical packages used. The data linkage process was conducted within the Bureau of National Health Insurance and followed the government's confidentiality regulations during the linkage and analysis processes.
|
|
Results |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionConclusionReferences |
|---|
Table 2 presents some descriptive statistics, demographic characteristics, physical disability, major diseases, and utilization of outpatient and inpatient services across income quintiles. The results indicate that some differences were observed in the distribution of these variables across each quintile. The lowest income quintile (Q0) had slightly more males than females, more individuals with physical disability and major diseases. In terms of health care utilization, those individuals of the Q0 group had a lower probability of using any outpatient services, but higher probability of any inpatient service use than those in other income quintiles. Both their outpatient and inpatient expenses were higher than those of people in higher income quintiles.
Table 3 shows the distribution of the study sample by insurance category in 1996, 1998 and 2000, and the distribution of the full NHI population by insurance category in 2000. The sample distribution was similar to the full NHI population distribution by insurance category. The three largest insurance groups were private-enterprise employees, workers enrolled through worker's associations, and farmers and fishermen. These three groups comprised almost 80% of the NHI enrolee population.
|
Table 4 presents the distribution of the sample population by premium types and income. Those who pay fixed premiums dominated the bottom two income groups (78.2% in Q0 and 66.4% in Q1) in 1996 but were less dominant in higher income groups. Wage-based enrolees dominated the top two income groups (66.4% in Q3 and 80.1% in Q4). Although the proportion of those in the wage-based group increased from 51.3% in 1996 to 52.6% in 2000 while the proportion of those in the fixed-premium group decreased from 48.7% in 1996 to 47.4% in 2000, the income distributions of these two groups stayed the same.
|
Tables 5a–c
report the distributions of crude measures of total benefits, net payments and net benefits by income quintiles in 1996, 1998 and 2000. While in 1996, no substantial difference in total benefits was observed among income quintiles, we can note a clear progressive pattern for inpatient care and an evident regressive pattern for outpatient services. The lowest quintile group incurred higher inpatient expenditures than the higher income quintiles, but had lower outpatient expenditures than the top three quintiles.
|
|
|
In 1996, a clear progressive distribution of net payments was observed among income quintiles, with the lowest quintile (Q0) paying the lowest amount (NT$6952) while the top quintile (Q4) contributed the highest amount (NT$13 973). More specifically, the most sizable differences in premium contributions across income quintiles were observed for employer contributions, ranging from NT$1528 in the lowest quintile to NT$8195 for the top quintile (Q4). Employer contribution for the top income quintile was almost five times that of the lowest quintile group. Overall, employer contribution seems to show the strongest income redistributive effects. Further, we note that the co-payment rate was low relative to the size of benefits received, and the distribution of co-payments was slightly progressive, which may be the result of greater outpatient use at academic medical centres (which require higher co-payment) among higher income groups.
Most importantly, our results show that net benefits decreased as income increased. The distribution of net benefits was pro-poor, from NT$3608 for the lowest income quintile to –NT$2895 for the top income quintile. The relationship between net benefits and income became stronger over time; larger net benefit differentials across the income groups were observed in 1998 and 2000. Overall, the patterns reported in 1998 and 2000 were similar to those observed in 1996.
Tables 6a–c report the distribution of total benefits, net payments and net benefits in 1996, 1998 and 2000, adjusting for health care needs (age, sex, major disease status and physical disability). The patterns were similar to the crude measures and the distribution of adjusted net benefits remained pro-poor; as income increased, net benefits decreased. In 1996, the lowest quintile received the highest net benefits (NT$3353) while the top income group received the lowest net benefits (–NT$3072) after adjusting for health care needs. Similar findings were reported for 1998 and 2000.
|
|
|
Figures 1a–c
show the trends of adjusted total benefits, net payments and net benefits from 1996 to 2000. The pattern in Figure 1a indicates that after adjustment, an increasing trend in total benefits was observed for all income groups. The magnitude of increase was particularly large for the lowest income quintile (Q0). On the other hand, a much flatter increasing trend was observed for net payment among all income quintiles (Figure 1b). As the increases in adjusted total benefits over time were greater than increases in net payments in all income groups, an increasing trend in net benefits was observed for all income groups (Figure 1c). More specifically, the lowest income quintile had the largest increase in net benefit over time. In addition, although the top income group had negative net benefits for all 3 years, the negative value decreased over time, along with the trend of increasing net benefits observed in all other groups. Thus every group fared better at the end of the period than at the beginning.
|
|
|
Since the NHI has two different methods of calculating contributions, wage-based and fixed-premium, it is useful to conduct separate net-benefit analyses for the wage-based and fixed-premium enrolees. After adjusting for health care needs, Tables 7a and 7b present the distributions of total benefits, net payments and net benefits for those who pay wage-based premiums and those who pay fixed premiums, respectively.
|
|
Individual contribution and employer contribution were two major sources of net payments for those who pay wage-based premiums, and strong progressive patterns were observed in these two categories. Therefore, the distribution of net payments shows a strong progressive relationship with income. On the other hand, for those paying fixed premiums, none of the three net payments sources (individual contribution, employer contribution and co-payments) had a strong progressive pattern. Hence, the distribution of net payments in the fixed-premium group showed a much weaker income-redistribution effect than that in the wage-based group. Furthermore, net payments for wage-based individuals were substantially higher than for those paying fixed premiums at every income level.
A clear pro-rich pattern was observed in total benefits for both the wage-based and fixed-premium groups at all income levels except the lowest quintile. Among the wage-based individuals, the lowest quintile received the highest total benefits. Although the lowest quintile of the fixed-premium population (NT$16 777) did not receive the highest total benefits, its benefits were still higher than those of quintiles Q1 and Q2. Furthermore, those paying fixed premiums tended to receive substantially higher benefits than those paying wage-based premiums at every income level except for the lowest quintile. One plausible explanation is that the fixed-premium group had a higher proportion of the elderly than did the wage-based premium group.
A strong pro-poor effect was observed in net benefits among the wage-based population, whereas a clear pro-rich effect was observed in net benefits among the fixed-premium population. Furthermore, while individuals paying fixed premiums received large positive net benefits at every income level, only the lowest two quintiles of the wage-based population received positive net benefits, with the remaining income groups receiving negative net benefits. Wage-based enrolees received much lower net benefits than fixed-premium enrolees at all income levels. Wage-based enrolees were subsidizing fixed-premium enrolees.
|
Discussion |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionConclusionReferences |
|---|
We can identify at least four significant findings. First, the results suggest that determining the equity of the NHI programme based on either health care utilization or financial burdens only may be misleading. When both factors are taken into consideration, the distribution of net benefits of the NHI programme is clearly pro-poor and indicates that the NHI has successfully served its social mission by providing sound coverage to the poor.
Secondly, the substantially larger differences observed in net payments made by the rich and the poor are likely to be the major contributing factor to the pro-poor distribution of net benefits. Of all three net payment sources, employer contribution was most progressive and contributed significantly to the equity of the NHI financing schemes. This may be attributable to the strong redistributive function of employer contribution and the concentration of fixed-premium enrolees in lower income quintiles. When contemplating financing the NHI through income tax or general taxation, policy-makers should not overlook the strong income-redistribution effects of the existing wage-based premium system.
Thirdly, given the low co-payment levels in Taiwan, total health care utilization by rich and poor is similar. That no discernible difference can be found across income quintiles may suggest that everyone's health care needs have been satisfied under the National Health Insurance with its low cost-sharing obligation. However, the rich and the poor have different utilization patterns. The rich tend to use more outpatient care while the poor tend to use more inpatient care. The distribution of inpatient care is more regressive than that of outpatient care, and the results remain robust after adjusting for individual medical needs. To avoid unfairly impacting the poor, these differential usage patterns should be kept in mind when designing schemes to streamline health care utilization.
Fourthly, and most important, the separate analyses for wage-based and fixed-premium populations indicate that the NHI programme has very poor horizontal equity due to the substantial net benefit differences observed between wage-based and fixed-premium individuals at the same income level. Moreover, the strong regressive distribution of net benefits observed among the wage-based population, and the apparent progressive distribution of net benefits observed among the fixed-premium population, suggest that wage-based premium schemes can lead to much better vertical equity than fixed-premium schemes. Since many developing countries are characterized by a sizable underground economy and a relatively small regular wage-earner population, a wage-based premium scheme may not be feasible. On the other hand, relying heavily on fixed-premium schemes for financing may pose serious concerns regarding equity. The trade-off between feasibility and equity becomes a policy dilemma for developing countries in devising financing schemes for social health insurance.
An additional encouraging observation regarding the NHI programme is that it maintains a satisfaction rate of almost 80%. However, its financial sustainability is questionable. This study offers a plausible explanation of the relationship between a high satisfaction rate and vulnerable financial sustainability. All income groups, except for the top quintile, enjoyed positive net benefits between 1996 and 2000. Moreover, everyone, even those in the top income group, is experiencing increasing net benefits. Since everyone benefits more over time under the NHI programme, it is not surprising that it has a high satisfaction rate. However, the weakening financial sustainability of the NHI programme may be the hidden cost of this high satisfaction rate. The long-term failure to raise premiums in line with actuarial cycles and estimates and the rise in health care expenditures have resulted in the increasing trend of net benefits across all groups. This may be the root of the current financial crisis of the NHI. In addition, although government subsidies provide positive net benefits to the majority of the population, the insured tend to forget that these subsidies come from their own pockets. In the face of financial crisis, premiums must be raised or other financing schemes (income tax or general taxation) employed in order to keep the NHI financially sound. Without a rise in premiums, it is doubtful whether the net benefits can remain positive and continue to increase.
In interpreting our results the following qualifications should be kept in mind. The distribution of net benefits in favour of the poor may be underestimated for the following reasons. First, since the income tax information provided by the Ministry of Finance does not include data on actual tax payments and we do not have an appropriate allocation algorithm, we are not able to allocate individual tax contribution to government subsidies. Hence, we did not include government subsidies as a part of net payments or in the analyses of net benefits. Since the rich tend to carry a larger tax burden than the poor, the distribution of net benefits in favour of the poor was probably underestimated in this study. Secondly, although risk reduction is the main goal of social health insurance, this study does not measure the benefits from risk protection, benefits that tend to be larger for the poor than the rich. Due to lack of information on out-of-pocket payments other than co-payments, the interpretation of the results is limited to the net benefits of the NHI programme. The results are not sufficient to describe the equity of total health care expenditure in Taiwan. Thirdly, since we only used four observable variables (age, sex, major disease status and physical disability) to adjust for health care need, there may be other unobservable health statuses. If these unobservable variables are correlated with income, the coefficient estimates for income can be biased upwards if lower income is correlated with poorer health. Fourthly, excluding individuals with incomplete data in our analyses may be a source of bias. If poor people are more likely to have incomplete data, the pro-poor pattern of net benefit of the NHI programme could be stronger than we observed. If incomplete data occurs at random, then the omission of incomplete data is not a problem. Finally, several data limitations should be noted: (1) the household economic unit for tax purposes is not necessarily the same as the co-resident household unit; (2) individuals belonging to the lowest quintile may still have income, even though they do not file an income tax return; (3) our estimate of medical needs using age, sex, major disease status and physical disability may not adequately reflect individual medical needs.
|
Conclusion |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionConclusionReferences |
|---|
This study is a first step towards evaluating the equity of the NHI programme by taking both health care benefits and payments into consideration. Taking these two factors into account, the distribution of net NHI benefits is pro-poor. Future research evaluating the equity of a health care system should take into account benefits as well as payments. Focusing on only one of these factors may produce misleading results. We also found that using two different financing schemes leads to an inequitable distribution of net benefits between the wage-based and fixed-premium enrolees at the same income level, and the distribution of benefits in the fixed-premium scheme shows worse vertical equity than in the wage-based premium scheme. The study provides an important policy reference to developing countries with large underground economies and relatively small populations of regular wage-earners as it indicates that using fixed premiums for financing health care may pose a serious equity concern and policy challenge.
|
Acknowledgement |
|---|
The study is supported by a grant from Taiwan's Ministry of Education, Aim for the Top University Plan, and Taiwan's National Health Research Institute grant 94A1-HPSP01-01.
|
References |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionConclusionReferences |
|---|
Aronson JR, Johnson P, Lambert PJ. 1994. Redistributive effect and unequal tax treatment. Economic Journal 104:262–70.[CrossRef]
Buhmann B, Rainwater L, Schmaus G, Smeeding TM. 1988. Equivalence scales, well-being, inequality, and poverty: sensitivity estimates across ten countries using the Luxembourg Income Study (LIS) database. Review of Income and Wealth 34:115–42.[CrossRef]
Bureau of National Health Insurance [Taiwan]. 2001. An overview of National Health Insurance. Online at: [http://www.nhi.gov.tw/english/e_05pro_103.asp].
Bureau of National Health Insurance [Taiwan]. 2005. National Health Insurance Annual Statistical Report 2004Taipei, Taiwan.
Cretin S, Duan NH, Williams AP Jr, Gu XY, Shi YQ. 1990. Modeling the effect of insurance on health expenditures in the People's Republic of China. Health Services Research 25:667–89.[Web of Science][Medline]
Culyer AJ and Wagstaff A. 1993. Equity and equality in health and health care. Journal of Health Economics 12:431–57.[CrossRef][Web of Science][Medline]
Culyer AJ, van Doorslaer E, Wagstaff A. 1992. Utilisation as a measure of equity by Mooney, Hall, Donaldson and Gerard. Journal of Health Economics 11:93–8.[CrossRef][Web of Science][Medline]
Duan NH. 1983. Smearing estimate: a nonparametric retransformation method. Journal of the American Statistical Association 78:605–10.[CrossRef]
Duan N, Manning WG, Morris CN, Newhouse JP. 1983. A comparison of alternative models for the demand for medical care. Journal of Business and Economics Statistics 1:115–26.
Jones AM. 2000. Health econometrics. In Culyer AJ and Newhouse JP (Eds.). Handbook of Health EconomicsAmsterdam Elsevier BV volume 1A pp. 285–9.
Mustard C, Barer M, Evans RG, et al. 1998. Paying taxes and using health care services: the distributional consequences of tax financed universal health insurance in a Canadian province. Conference on the State of Living Standard and the Quality of Life in Canada, Ottawa.
National Health Research Institute [Taiwan]. 2005a. Overview of random samples of NHI beneficiaries. Taipei, Taiwan Online at: [http://www.nhri.org.tw/nhird/date_cohort.htm].
National Health Research Institute [Taiwan]. 2005b. Comparison of distribution of basic characteristics between four waves of random samples of NHI beneficiaries and the entire population of NHI beneficiaries. Taipei, Taiwan Online at: [http://www.nhri.org.tw/nhird/talk_01_8.htm#1].
Van Doorslaer E, Wagstaff A, van der Burg H, et al. 1999. Equity in the finance of health care: some further international comparisons. Journal of Health Economics 18:263–90.[CrossRef][Web of Science][Medline]
Van Doorslaer E, Wagstaff A, van der Burg H, et al. 2000. Equity in the delivery of health care in Europe and the US. Journal of Health Economics 19:553–83.[CrossRef][Web of Science][Medline]
Wagstaff A, van Doorslaer E, Calonge S, et al. 1992. Equity in the finance of health care: some international comparisons. Journal of Health Economics 11:361–87.[CrossRef][Web of Science][Medline]
Wagstaff A, van Doorslaer E, van der Burg H, et al. 1999. The redistributive effect of health care finance in twelve OECD countries. Journal of Health Economics 18:291–313.[CrossRef][Web of Science][Medline]
Zschock DK. 1989. Health sector disparities in Peru. Bulletin of the Pan American Health Organization 23:323–36.[Medline]
Accepted for publication 20 October 2006.
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||