Financing National Health Insurance and Challenge of Fast Population Aging: The Case of Taiwan

Financing National Health Insurance and Challenge of Fast Population Aging: The Case of Taiwan Minchung Hsu Pei-Ju Liao GRIPS Academia Sinica October 15, 2010 Abstract This paper aims to discover the impacts of fast population aging, which is happening in many developing countries, on financing a national health insurance (NHI) program that is a current trend and encouraged by WHO. We use Taiwan, which has implemented NHI since 1995 and experienced sharp fertility decline and fast population aging, as an example. We undertake a stochastic equilibrium OLG model with household heterogeneity, market incompleteness and endogenous working/saving decisions to provide a quantitative investigation. We calibrate the model to match the macro and micro data of Taiwan in 2000s as a benchmark. With taking into account population aging in 2030 and 2050, we find that the labor tax burden has to increase from 14.3% in 2009 to 19.3% in 2030 and to 24.1% in 2050 to balance the government budget. If we consider the real medical cost growth, the situation will be even worse. Assuming the real cost growing at the same rate as in 2000s until 2030, labor tax rate will increase to 31.6% in 2030, and reach 45% in 2050. Related policy issues will also be discussed. National Graduate Institute for Policy Studies, Tokyo, Japan. Email: minchunghsu@grips.ac.jp. Institute of Economics, Academia Sinica, Taipei, Taiwan. Email: pjliao@econ.sinica.edu.tw. All errors are ours. Acknowledgement will be added. 1

1 Introduction Most of developed countries have provided universal health insurance (UHI) to their citizens. 1 Some middle income countries also achieved universal coverage, for example Korea, Singapore and Taiwan, and the WHO actually encourages countries to pursue it (WHO 2008 annual report) for health care equality. Public provision of UHI is widely adopted, particularly for those developing countries pursuing UHI. However, given the global trend of population aging and fertility declining, financing the UHI is expected to be a real challenge in the near future: a smaller working population has to shoulder a higher average cost of UHI due to a larger old population, who need more health care. This paper uses Taiwan, which has implemented a public UHI program since 1995 and experienced sharp fertility decline and fast population aging, as an example to understand the possible problems on UHI financing and related policy issues on UHI and population. This analysis also provider important policy implications for those countries has recently achieved or are currently pursuing UHI. Taiwan, in 1995, implemented a public UHI program (named as National Health Insurance, NHI, in Taiwan) to replace the original employment-based health insurance system. Even though the cost of the NHI is low currently compared with many European countries, its financing problem has been observed. Furthermore, with the fast aging population in Taiwan, it is important to understand the potential impacts of the NHI financing on the economy when taking into account the expected future demographic changes. In this paper we focus on the impacts on tax burden, capital accumulation, labor supply, output level and welfare. Related policy issues will be also discussed. We undertake a equilibrium a dynamic equilibrium over-lapping-generations model with household heterogeneity, financial market incompleteness and endogenous working/saving decisions to provide a quantitative investigation. In the model, households face idiosyncratic income shocks and medical expenditure shocks. Because of the market incompleteness, households can not fully insure the uncertainties. Precautionary assets will be accumulated for the sake of self insurance. The NHI also provide partial insurance against the medical uncertainty. We calibrate the model to match the macro and mirco data of Taiwan in 2000s as a benchmark. The crucial part is to ensure that the medical cost, NHI cost and the tax burden in the model can match the data in 2000s as a reasonable starting point. We also make sure that the capital output ratio and average labor supply in the model can match the aggregate data so that the saving and working decisions in the model can be consistent with the household behaviors in the real economy. 1 The US is an exception, but thinking of a reform. 2

With taking into account population aging in 2030 and 2050, we find that the labor tax burden will increase from 14.3% in 2009 to 19.3% in 2030 and to 24.1% in 2050 by assuming real medical cost unchanged. Average NHI cost per capita will increase more than 60% in 2050. Labor supply is 33% lower in 2050 because of population aging and increased tax distortion. If we consider the real medical cost growth, the situation will be even worse. Assuming the real cost growing at the same rate as in 2000s until 2030, the medical service will become 161% more costly than in 2009. In this case, average NHI cost will increase by 132% and labor tax rate will increase to 31.6% in 2030. Moreover, in 2050 with the same real cost as in 2030, the labor tax rate will reach 45% because of the worse population structure. A similar model has been adopted by Jeske and Kitao (2009) and Hsu and Lee (2010) to study health insurance related issues. Jeske and Kitao (2009) investigate the effects of the US tax policy on private health insurance take-ups. Since labor supply is not their focus, they treat it inelastic. We allow endogenous labor decisions to study the impact on labor supply and its welfare implication. Hsu and Lee (2010) study the impacts of UHI provision on private health insurance markets, asset holdings and welfare. Population aging is not taken into account in their analysis, but is crucial in this paper. 2 Taiwan s Health Insurance System and Demographic Features 2.1 NHI The NHI in Taiwan was expanded from three major health insurance programs provided by the government: Labor Insurance, Government Employees Insurance, and Farmers Health Insurance. By 1994, 57% of total population was covered by the existing insurance programs; the remaining 43% of the uninsured population includes mainly those who are under age, students, elderly, and non-working adults in the households whose heads are not in the public sector (Cheng and Chiang, 1997). The population coverage rate of the NHI soon boosts to 97% in 1998 (Chou et al. 2003). In 1994, around 85% of hospitals and 70% of clinics were contracted with those insurance programs. After 1995, the Bureau of National Health Insurance (BNHI) becomes the monopsony of medical care services, and the contracted medical care institutions increase to around 96.5% of total hospitals and 89.5% of total clinics in 1997 (Chou et al. 2003). The premium basically is depending on labor income 5.17% of registered labor income. It can be viewed as a labor income tax for financing the NHI. In practice, there are seven categories for insurance premium payment, which define how to share the insurance premium by employees, 3

2009 2011 2013 2015 2017 2019 2021 2023 2025 2027 2029 2031 2033 2035 2037 2039 2041 2043 2045 2047 2049 2051 2053 2055 Life Expectancy 90 85 80 75 70 65 Mael Female Figure 1: Forecasted life expectancy employers and the government in each category. In general, both private and public employers are required to share 60% of the premium for their employees and the government share 10%; see BNHI (2010) for more details. However, employers, especially private ones, can always transfer the additional cost to employees in order to keep their total labor cost unchanged. Those without labor income pay 60% of the average premium and the rest is subsidized by the government. 2.2 Demographics Taiwan has experienced a fast demographic change that includes a sharp fertility decline, an increase of children survival rate and a rise of life expectancy. The fertility has sharply declined since 1970. Fertility rate is 1.1 in 2009, which is unable to maintain a zero population growth, and expected to stay low in the future if there is no major policy/environment change. Life expectancy is expected to increase from 79.3 in 2009 to 85.2 in 2050 (figure 1). All the demographic trends implies a fast aging population. The population above age 65 is currently 14.6% as of the population between age 15 to 64 (i.e. the old-age dependency ratio), but is expected to become 68.6% in 2050. Figure 3 presents the forecasted trend. 4

2009 2011 2013 2015 2017 2019 2021 2023 2025 2027 2029 2031 2033 2035 2037 2039 2041 2043 2045 2047 2049 2051 2053 2055 80 70 60 50 40 30 20 10 0 Forecasted old-age dependency ratio (%) Figure 2: Forecasted old-age dependency ratio (age 65+/age 15-64) 34,000 Medical Expenditure per capita (in 2000 price) 32,000 30,000 28,000 26,000 24,000 2000 2001 2002 2003 2004 2005 2006 2007 2008 Figure 3: Growing medical cost per capita (in 2000 price) 5

3 The Model We undertake a theoretical approach to understanding the impact of population aging on the financing of NHI. In the model economy, there is no aggregate uncertainty, but households face an idiosyncratic labor productivity shocks and a medical expenditure shock. Financial markets are incomplete as in the real world that means full contingent claims against the risks are unavailable. Instead, first, households can partially self-insure by accumulating precautionary asset holdings. Second, with the NHI coverage, they can partially insure against medical expenditure shocks. 3.1 Demographics The economy is populated by a continuum of finitely-lived households (measure one) and they maximize expected discounted lifetime utility from consumption and leisure. The population consists of two generations - the young and the old. Young agents supply labor and earn wage income and old agents are retired from market work. Young agents become retired with probability ρ o every period and the old die and leave the economy with probability ρ d every period. On average, the young work for (1/ρ o ) years, and the old live for (1/δ d ) years before they die. In each period, the economy has new-born young households which replace the old households who die such that measure of total population stays constant. A similar setting, the stochastic aging and death, is also used in Jeske and Kitao (2009) to capture the features of retirement and death, which clearly have effects on agents saving decisions, in an Aiyagari-Bewley type model. The demographic setting with the probabilities described above implies that every period there ρ is o ρ o +ρ fraction of old people and ρ d d ρ o +ρ fraction of young people. A population aging reflects in d a higher old-dependence ratio that implies a smaller working/tax-paying population. 3.2 Labor and Medical Expenditure Shocks Young household s effective labor supply depends on the hours worked and idiosyncratic labor productivity shock z, which is stochastic. In each period t, an idiosyncratic labor productivity shock takes one of l < values in a finite set Z = {z 1, z 2,..., z l }. Each household s productivity shock evolves independently according to a first-order Markov process with transition probability matrix π z, which is l l and an invariant distribution π z. Both young and old households faces medical expenditure shocks x, which is also stochastic. In each period t, each household s medical expenditure shock takes one of m < values in a finite set X i = {x 1,i, x 2,i,..., x m,i } for i {old, young}. Each household s medical expenditure 6

shock also evolves independently according to a first-order Markov process with transition probability matrix π x,i, which is m i m i for i {old, young} and an invariant distribution π x,i for i {old, young}. 3.3 Asset and Health Insurance Market Structures 3.3.1 Asset market There is a non-state contingent claim which is an asset that households can purchase at one unit of consumption good and pays off (1 + r) 1 units of consumption good next period. With trading this non-state contingent claim, households can partially insure themselves against any combination of idiosyncratic productivity shocks and medical expenditure shocks by accumulating precautionary asset holdings. One assumption that we made to present market incompleteness is that households are subject to a borrowing constraint. This borrowing limit on households asset holdings specially affects the asset holding decision of low-wealth households since they cannot smooth their consumption over time when they are hit by falls in their disposable incomes. 3.3.2 National Health Insurance Program The NHI mandatorily covers a constant fraction ω of household s medical expenditure x. Households pay (1 ω) x units of consumption good when the medical expenditure x is realized under the NHI coverage. This NHI program is financed by a income-contingent premium (P NHI, equivalently a labor tax) and government general revenues. According to the current rule of the NHI premium, the premium tax rate τ NHI is 5.17% per person for households with labor income in general. For those without labor income, they pay 60% of the average NHI premium. 3.4 Government Government s revenue consists of revenues from different tax instruments, labor income tax τ n, capital income tax τ k, consumption tax τ c, and newly issued government debt D. Government runs a social insurance (safety net) program and an universal health insurance program, which is described above. Government provides a social insurance that guarantees a minimum level of consumption c for every households by supplementing the income in case the household s disposable income 7

plus assets (net after medical expenditure) falls below c. 2 We consider a simple transfer rule proposed by Hubbard et al. (1995). The transfer T will be made if the household s disposable income plus assets (net after medical expenditure) is smaller than a minimum level of consumption. The transfer amount will be exactly equal to the difference. There is other government expenditure G, which is constant. Social insurance (safety net) program, universal health insurance program and other government expenditure are financed by the revenues from consumption tax and income tax. government budget constraint: G + [T + ωx]dφ + (1 + r)d = [τ n (wzn) + τ k (ra) + τ c c + P NHI ]dφ + τ k (rb) + D (1) where Φ is the distribution of the whole population over state variables, T is a transfer to the individual made for social insurance, x is individual medical expenditure, and a is an individual asset holding. 3.5 Production Technology On the production side, we assume that there is a continuum of competitive firms operating a technology with constant returns to scale. Aggregate output Y is given by Y = F (K, L) = AK θ L 1 θ, where K and L are the aggregate capital and effective labor employed by the firm s sector and A is the total factor productivity which we assume to be constant. Capital depreciates at rate of δ every period. θ denotes the capital income share. 3.6 Household 3.6.1 Preference We adopt a standard utility function u(c, n), which is consistent with balance growth path and widely used in the growth literature, as below: [ c φ (1 n) 1 φ] 1 µ u (c, n) =, (2) 1 µ where µ is the relative risk aversion coefficient. 2 In a model with expenditure shocks, it is possible that an unlucky household encounters an expenditure shock larger than its income and so can not sustain a positive consumption. This will create a problem in the model. To prevent this undesired situation, a mechanism providing minimum consumption support is necessary. See similar settings, for example, in Jeske and Kitao (2009) and Attanasio et al. (2009). 8

labor Supply The utility function given by equation (2) implies that labor supply can be expressed as a function of consumption and effective wage rate: 3.6.2 Young household s problem n = 1 (1 φ)c φ(1 τ n )wz. (3) The state of an agent is summarized by a vector s = (a, z, x), where a denotes asset holdings brought into the period, z the idiosyncratic shock to labor productivity, x the idiosyncratic health expenditure shock that has to be paid. subject to V (s) = max c,n,a { u (c, n) + β (1 ρo ) E [ V ( s )] + βρ o E [ W ( s )]} (1 + τ c )c + a + q (x) i HI = Wel y + T Wel y (1 τ n ) wzn + [1 + (1 τ k ) r] (a + b) [1 ω] x P NHI P NHI = τ NHI (wzn); T = max{0, (1 + τ c )c Wel y } a 0; 1 > n 0; where W is the value when the agent becomes old, and T is the transfer made by the meanstested social insurance system. Accidental bequests left by old agents died at the beginning of the period, b, are equally distributed to all survived agents. 3.6.3 Old household For the retired, they do not supply labor and have no labor income. Their labor productivity z is fixed at 0. Therefore they only face medical shocks without income shocks. An old agent s problem is: W (s) = max c,a { u (c, 0) + β (1 ρd ) E [ W ( s )]} 9

subject to (1 + τ c )c + a = Wel o P NHI + T; P NHI = 0.6 τ NHI (wzn)dφ; Wel o [1 + (1 τ k ) r] (a + b) [1 ω] x; T = max{0, (1 + τ c )c Wel o }; a 0. 3.6.4 Recursive Competitive Equilibrium A stationary recursive competitive equilibrium consists of household decision rules of asset holding a, labor supply n, and consumption c, a set of firm decision rules of capital rented K and effective labor employed L, a price system of w and r, a government policy of tax rates τ n, τ k, τ c, a government debt D, a policy of NHI coverage ω, minimum consumption floor c, and a distribution of households over the state variables Φ(s), such that: a) given the price system, the decision rules of K and L solve the firm s problem; b) given the price system, the insurance premium and the policy of tax rates, the decision rules of (a, n, c) solve household s problem; c) government policies (τ k, τ n, τ c, P NHI, D, c) satisfy the government s budget constraints; d) Φ(s) is stationary; e) all markets clear: L = (zn)dφ(s) and K + D = (a + b)dφ(s); f) resource feasibility condition is satisfied Y = C + G + K δk + X; where C is the aggregate consumption, and X is the aggregate medical expenditure. 4 Calibration for Benchmark We calibrate the model to match the Taiwanese economy in 2000s as a benchmark. To calibrate medical expenditure and income shocks, we use micro data from Panel Study of Family Dynamics (PSFD) which provides panel survey data on Taiwanese households. The baseline of 10

our calibration is that the medical cost, NHI cost and the tax burden in the model can match the data in 2000s as a benchmark for comparison. We also make sure that the capital output ratio and average labor supply in the model can match the aggregate data so that the saving and working decisions in the model can be consistent with the household behaviors in the real economy. 4.1 Demographic features According to the labor statistics, average age of entering labor market is 21 and average retirement age is 55 in Taiwan. So we set model age 1 as 21 years old in the data and on average an agent work for 34 years that implies ρ o = 1/34. The fraction of working-age population (age 21 54) is 72% as of the all population of age 21 and above. 4.2 Utility and Production Functions The model period is set to be one year. The risk aversion parameter µ is set at 2. The utility discount factor (β) is chosen so that capital-output ratio is equal to 2, which is suggested by Chow and Lin (2002). The leisure utility parameter φ is chosen so that aggregate labor hours is equal to 0.3, which is the average per capita labor hours. In the production function, the capital income share (θ) is 0.544, and follow Chow and Lin (2002) to set the depreciation rate of capital (δ) at 0.04. The scaling production parameter A is calibrated to normalize the average wage income in the benchmark into unity. 4.3 Labor Productivity and Medical Expenditure Shocks We use household income data in 2004 and 2005 from PSFD to calculate the shock states of labor productivity and the transition probabilities for characterizing income uncertainty. We equally divide the whole individual income distribution into 4 groups from bottom 25% to top 25% in 2004. The proportional deviations from the sample mean are defined as the 4 states of labor productivity in the model. See table 1 for the definition. We use the transition of income from 2003 to 2004 to determine the transition probabilities given each initial state. Table 2 presents the result. To characterize medical expenditure shocks, We calibrate a Markov process by using the data from PSFD. We define four medical expenditure states as low, fair, high, and very high, which represent medical expenditure in the bottom 60%, from 60 to 85%, from 85 to 95% and in the top 5%, respectively because of the skewness of medical expenditure. Because the average medical expenditure from the survey data is lower than the national average medical cost (from 11

Table 1: States of labor productivity shocks Income Average As of sample State range (NT$ in 2004) mean (2004) 1. Bottom 25% 0 25 % 12,399 30.1% 2. Next 25% 25 50 % 26,772 64.9% 3. Next 25% 50 75 % 40,300 97.8% 4. Top 25% 75 100 % 91,427 221.8% Source: PSFD Table 2: Transition probabilities of labor shocks State 1 State 2 State 3 State 4 State 1 0.7108 0.2304 0.0392 0.0196 State 2 0.1345 0.5965 0.2515 0.0175 State 3 0.0455 0.1717 0.6161 0.1667 State 4 0.0118 0.0588 0.1588 0.7706 Source: PSFD 2008 National Health Expenditures, Department of Health, Taiwan), we adjust the average medical expenditures of the four groups by the proportional difference so that the aggregate medical cost to output ratio in our model can match the national medical cost to GDP ratio in 2004, which is 6%. These expenditures, therefore from low to very high, were 1857, 9234, 29509 and 131434 Taiwan dollars in 2004, and also were 0.37%, 1.84%, 5.88% and 26.19%, respectively, as of the GDP per capita in 2004 for the working-age population. For retired people, the average medical expenditures of the four groups were 8883, 58365, 206561, and 681059 Taiwan dollars in 2004, and were 1.77%, 11.63%, 41.16%, and 135.71% as of the GDP per capita in 2004. We set the fourstate medical expenditure shocks, X y and X o for the young and the old respectively, as the above percentages of average labor income in the model (see table 3 and 4). The transition probabilities for the Markov chain of medical expenditures are calibrated based on PSFD in 2003 and 2004. The results are reported in Table 5 and Table 6. 12

Table 3: States of medical expenditure the young (X y ) Expenditure Average As of average State range ($ in 2004) income (2004) Low bottom 61% 1,857 0.4% Fair 61 86% 9,234 1.8% High 86 95% 29,509 5.9% Very High 95 100% 131,434 26.2% Source: PSFD, DGBAS and author s calculation. Table 4: States of medical expenditure the old (X o ) Expenditure Average As of average State range ($ in 2004) income (2004) Low bottom 61% 8,883 1.8% Fair 61 85% 58,365 11.6% High 85 95% 206,561 41.2% Very High 95 100% 681,059 135.7% Source: PSFD, DGBAS and author s calculation. 4.4 NHI, Safety Net, Government Taxes The expenditure coverage rate of NHI ω is calculated by the ratio of total NHI payment to total medical expenditure. We find the coverage rate is 65%. According to the current rule of the NHI premium, the premium tax rate of NHI is 0.9 5.17% per person for households with labor income in general. For those without labor income, they pay 60% of the average NHI premium. The minimum consumption floor provided by the safety net is set to 5% average earning based on the average government subsidy per low-income individual. Consumption tax rate is set at 5%, capital income tax is 0% and labor income lax rate is 14.3%. Government debt to output ratio is 32.5%. 13

Table 5: Transition probabilities of X y Low Fair High Very High Low 0.7140 0.1983 0.0647 0.0230 Fair 0.4059 0.4307 0.1238 0.0396 High 0.3698 0.3151 0.2466 0.0685 Very High 0.2857 0.3714 0.1429 0.2000 Source: PSFD Table 6: Transition probabilities of X o Low Fair High Very High Low 0.7140 0.2092 0.0698 0.0070 Fair 0.4393 0.3873 0.1156 0.0578 High 0.2714 0.2714 0.3572 0.1000 Very High 0.3055 0.1667 0.2222 0.3056 Source: PSFD 5 Quantitative Analysis The benchmark economy is calibrated to match the average Taiwanese economy in 2000s and the young/old ratio in 2009. Because the average age of entering labor market is 21 and average age of retirement is 55, the old-age dependency ratio defined as (age 55+/age 21-54) is 38.9% in 2009. Medical cost to output ratio is 6%. Average labor income tax rate is 14.3%. 5.1 Population aging and NHI bruden If the average age of entering labor market and average retirement age keep the same, the old-age dependency ratio is forecasted to become 85.2% in 2030 and 132.5% in 2050 (calculated from Population Projections for Taiwan, the Council for Economic Planning and Development, CEPD, 2008). The population aging will increase cost of NHI and the burden of working population. To evaluate the impacts quantitatively, we simulate the new steady state equilibria given the new young-old ratios. We keep the medical cost (X and Xo) and government expenditure G, the same 14

Table 7: Summary of Parameter Values Name of the parameters Notations Values Risk Aversion µ 2.00 Depreciation Rate δ 0.04 Capital Income Share θ 0.544 Prob. of being retired ρ o 1/34 Fraction of the Young ρ d ρ o +ρ d 0.72 Table 8: Other Parameters Name of the parameters Notations Values Min. consumption level as of average labor income c 5% NHI premium tax rate τ NHI 0.9 0.0517 Consumption tax rate τ c 0.05 Capital tax rate τ k 0.0 Labor tax rate τ n 0.143 Debt/GDP ratio 0.325 15

Table 9: Impacts of Population Aging old/young τ n burden labor x/y NHI cost 2009 38.9% 14.3% 0.30 6.0% 0.0% 2030 85.2% 19.3% 0.24 10.0% 44.3% 2050 132.5% 24.1% 0.20 14.0% 62.3% Note: τ n : labor income tax rate excluding NHI tax; x/y: average medical cost to output ratio. as in the benchmark, and assume the government uses additional labor tax to balance its budget with all other tax rates, policies and Debt-GDP ratio the same as in the benchmark. The result is presented in table 9. We can see that the labor income tax burden increases from 14.3% to 24.1% in 2050. On the other hand, labor supply (hours) decreases from 0.3 to 0.2 in 2050, a 33% drop. This decline in labor partial because of the population aging and partially because of the increased tax rate discourages labor supply. The fraction of old people is higher in 2030 and 2050, and so average NHI cost becomes higher too. It increases by 44.3% in 2030 and by 62.3% in 2050 compared with year 2009. Because average NHI cost becomes higher and labor supply becomes lower, the government has increase tax rate to satisfy its budget constraint. 5.2 Increasing medical cost From 2001 to 2009, we find that medical cost per capital increase constantly every year. The real growth rate is 2.3% per year. If this real growth trend lasts for 40 years, in 2003 the real medical cost per person will be 161% as of in 2009, and in 2050 it will be 254% as of the cost in 2009. If this medical cost increase is taken into account with population aging, the situation will be much worse. To show it quantitatively, we assume the real cost increases 2.3% annually from 2009 to 2030, and then becomes stable to 2050. 3 In this case, average NHI cost will increase by 132% and labor tax rate will increase to 31.6% in 2030. Moreover, in 2050 with the same real cost as in 2030, the labor tax rate will reach 45% because of the worse population structure. Labor supply largely is reduced from 0.3 to 0.18 in 2050. 3 Given the population structure in 2050, real medical cost growth from 2030 to 2050 would lead to a 100% labor tax rate and zero labor supply. 16

Table 10: Impacts of Population Aging and Real Medical Cost Rising old/young τ n burden labor x/y NHI cost 2009 38.9% 14.3% 0.30 6.0% 0% 2030 85.2% 31.6% 0.23 17% 132% 2050 132.5% 45.0% 0.18 26% 161% Note: τ n : labor income tax rate excluding NHI tax; x/y: average medical cost to output ratio. 5.3 Policy discussion (To be added) 5.4 Concluding Remarks This paper uses a calibrated stochastic OLG model to discover the impacts of population aging on Financing NHI in the future 20 and 40 years. We calibrate the model to match the macro and mirco data of Taiwan in 2000s as a benchmark. The crucial part is to ensure that the medical cost, NHI cost and the tax burden in the model can match the data in 2000s as a reasonable starting point. We also make sure that the capital output ratio and average labor supply in the model can match the aggregate data so that the saving and working decisions in the model can be consistent with the household behaviors in the real economy. With taking into account population aging in 2030 and 2050, we find that the labor tax burden will increase from 14.3% in 2009 to 19.3% in 2030 and to 24.1% in 2050 by assuming real medical cost unchanged. Average NHI cost per capita will increase more than 60% in 2050. Labor supply is 33% lower in 2050 because of population aging and increased tax distortion. If we consider the real medical cost growth, the situation will be even worse. Assuming the real cost growing at the same rate as in 2000s until 2030, the medical service will become 161% more costly than in 2009. In this case, average NHI cost will increase by 132% and labor tax rate will increase to 31.6% in 2030. Moreover, in 2050 with the same real cost as in 2030, the labor tax rate will reach 45% because of the worse population structure. This paper is still an on-going project. Related policy issues on NHI benefits, retirement age and government subsidy will be future discussed later. 17