The Effects of Reform in the National Health Insurance Contribution Scheme:

Transcription

1 Research Paper The Effects of Reform in the National Health Insurance Contribution Scheme: On Equity in Health Insurance Contribution Burden December 2017 Eunkyeong Lee John M. Kim Myung-jae Sung Chang-woo Lee

2 Korea Institute of Public Finance 336, Sicheong-daero, Sejong-si, Korea Tel: Fax: URL: C 2017 KIPF

3 The Effects of Reform in the National Health Insurance Contribution Scheme: On Equity in Health Insurance Contribution Burden December 2017 Eunkyeong Lee John M. Kim Myung-jae Sung Chang-woo Lee

4 The Effects of Reform in the National Health Insurance Contribution Scheme: On Equity in Health Insurance Contribution Burden 2 Contents Ⅰ. Introduction 7 Ⅱ. Current State and Issues of the National Health Insurance Contribution Scheme Current State of the Contribution Scheme Issues of the Current Contribution Scheme 17 Ⅲ. Previous Studies and Introduction of Scenarios Previous Studies Introduction of Scenarios 25 Ⅳ. Effect of Reform in the National Health Insurance Contribution Scheme: Cross-Sectional Analysis Background Analysis of Income Redistributive Effects Expected Effects of Improvement in Insurance Contribution Assessment 43 Ⅴ. Effect of Reform in National Health Insurance Contribution Scheme: Longitudinal Analysis Current State of Contribution Burden on Households 60

5 Contents 3 The Effects of Reform in the National Health Insurance Contribution Scheme: On Equity in Health Insurance Contribution Burden 2. Change in Insurance Contribution Burden by Scenario Analysis of Redistributive Effect 81 Ⅵ. Conclusion and Policy Implications 89 References 94

6 The Effects of Reform in the National Health Insurance Contribution Scheme: On Equity in Health Insurance Contribution Burden 4 List of Tables <Table II-1> Changes in the Rate of Health Insurance Contribution for the Employee Insured and the Amount borne by the Self-Employed Insured per Point 14 <Table II-2> Current State of the Employee Insured and the Self-Employed Insured 15 <Table II-3> Insurance Contribution Revenue by Year 16 <Table IV-1> Descriptive Statistics of the Household Income and Expenditure Survey (Annualized Basis) by Statistics Korea (As of 2014) 29 <Table IV-2> Distributions of Various (Cash) Transfer Incomes by Income 40 <Table IV-3> Joint Distribution of Average Property by Income and Age (Based on the 2013 NaSTaB Data) 47 <Table IV-4> Post-Reform Distribution of Health Insurance Contribution Burden by Income 48 <Table V-1> <Table V-2> <Table V-3> <Table V-4> <Table V-5> <Table V-6> <Table V-7> Current State of Household Income and Health Insurance Contribution Burden 62 Longitudinal Analysis ( ): Gross Household Income, Health Insurance Contribution and Contribution Burden Rate 64 Average Income, Average Insurance Contribution and Contribution Burden to Income Ratio by Income Quintile: by Household 65 Households with the Employee Insured: Average Income, Average Insurance Contribution and Contribution Burden to Income Ratio by Income Quintile 67 Households with the Self-Employed Insured: Average Income, Average Insurance Contribution and Contribution Burden to Income Ratio by Income Quintile 68 Health Insurance Contribution and Contribution Burden Ratio by Household in Standard A 70 Longitudinal Analysis of Standard A: Incomes Subject to Contribution Assessment, Household Insurance Contribution and Contribution Burden Ratio 72

7 Contents 5 <Table V-8> <Table V-9> Average Income, Average Insurance Contribution and Contribution Burden to Income Ratio by Income Quintile in Standard A 73 Health Insurance Contribution and Contribution Burden Ratio by Household in Standard B 75 <Table V-10> Longitudinal Analysis of Standard B: Incomes Subject to Contribution Assessment, Household Insurance Contribution and Contribution Burden Ratio 77 <Table V-11> Income, Insurance Contribution and Contribution Burden to Income Ratio by Income Quintile in Standard B 78 <Table V-12> Income-Based Contribution Rate and Property-Based Contribution Rate By Household in Standard C 80 <Table V-13> Income, Insurance Contribution and Contribution Burden to Income Ratio by Income Quintile in Standard C 81 <Table V-14> Redistributive effect of Health Insurance Contribution ( ) 83 <Table V-15> Longitudinal Analysis: Redistributive Effect of Health Insurance Contribution 83 <Table V-16> Redistributive Effect of Health Insurance Contribution in Standard A ( ) 84 <Table V-17> Longitudinal Analysis: Redistributive Effect of Health Insurance Contribution in Standard A 85 <Table V-18> Redistributive Effect of Health Insurance Contribution in Standard B ( ) 86 <Table V-19> Longitudinal Analysis: Redistributive Effect of Health Insurance Contribution in Standard B ( ) 87 <Table V-20> Redistributive Effect of Health Insurance Contribution in Standard C ( ) 87 <Table V-21> Longitudinal Analysis: Redistributive Effect of Health Insurance Contribution in Standard C ( ) 88

8 The Effects of Reform in the National Health Insurance Contribution Scheme: On Equity in Health Insurance Contribution Burden 6 List of Figures [Figure IV-1] [Figure IV-2] [Figure IV-3] [Figure IV-4] [Figure IV-5] [Figure IV-6] [Figure IV-7] [Figure IV-8] Gini Coefficient by Income Stage (Estimation Results of Raw Data from the HIES) 34 Change Rate of Gini Coefficient by Income Item (Estimation Results of Raw Data from the HIES) 39 Change Rate of Gini Coefficient per One Million KRW of Burden/Benefit per Household by Income Item (Estimation Results of the HIES) 41 Effect of Health Insurance Contribution Reform on Contribution Amount by Income (Change in Amount) 53 Effect of Health Insurance Contribution Reform on Contribution Amount by Income (Change Rate) 54 Distributions of Income Share and Gross Property Share by Income (Based on Estimation Results of the 2009 NaSTaB Data) 54 Gini Coefficient by Stage: Effect of Reform in Health Insurance Contribution Scheme 57 Income Redistributive Effect of Reform in Health Insurance Contribution Scheme 58 [Figure V-1] Household Contribution Burden Ratio by Income Quintile 66

9 Ⅰ Introduction Generally, the healthcare system for public health mobilizes its financial resources from the public sector and operates in the form of social insurance or taxation. In Korea, health insurance is operated as social insurance. There are three main reasons for justifying such governmental intervention (i.e. social insurance) in goods that the private sector can provide like private insurance (Boadway et al., 2006). First, the administrative expense of social insurance is lower than that of private insurance. Another reason concerns market failure due to information asymmetry. In private insurance, information asymmetry occurs between the insurer and the insured as a result of adverse selection and moral hazard. The same applies to the supplierinsurer relationship as well. Lastly, social insurance serves as a redistributive device. While private insurance is characterized by voluntary purchase, risk-rated insurance premiums and differentiated benefits depending on contract contents, the national health insurance as social insurance is, in principle, characterized by compulsory purchase, contributions based on the ability to pay (usually measured by income level) and equal insurance benefits (i.e. the use of healthcare services). Due to these characteristics of social insurance, national health insurance plays a redistributive role (Hindriks & De Donder, 2003). First, the group with high health risk pays a lower contribution than the expected risk whereas the group with low risk pays a higher contribution than the expected risk. As a result, the redistribution occurs from the low-risk group to the high-risk group. Second, as the health insurance contribution increases along with income,

10 The Effects of Reform in the National Health Insurance Contribution Scheme: On Equity in Health Insurance Contribution Burden 8 redistribution takes place from the high-income group to low-income group, even when they face the same risk. In this sense, the national health insurance is functioning as a redistributive device while pooling the health risk of individuals to spread out healthcare expenses. According to Boadway et al. (2006), there is no need to resort to social insurance for redistribution purposes because income tax has a redistributive effect in the first-best situation with perfect information However, social insurance is used as an effective tool for redistribution policy in the second-best situation where there are tax distortions under imperfect information. This is because the redistributive effect of social insurance does not cause the same type of distortion as income tax (Rochet, 1989; Cremer & Pestieau, 1996). In particular, social health insurance exerts a high redistributive impact when risk is negatively correlated with income and low-income class is more likely to be exposed to diseases (Boadway et al., 2001). Among various roles of health insurance as social insurance, this study focuses on its redistributive effects in light of contribution burden. The level of individual contribution burden is determined based on the insurance contribution scheme. Despite the consolidation of the national health insurance finance in the early 2000s, the contribution scheme has been dualized into one for the employee and the other for the self-employed. This has caused a persistent controversy over the issue of equity. It is because insurance contribution is imposed only on earned income in the case of the employee while it is determined by various factors, such as global income, property, automobile, gender and age, in the case of the self-employed. This equity controversy not only hinders social integration but also hampers the financial stability of the health insurance. This study aims to analyze the effects of the reform of the health insurance contribution scheme on the improvement of equity (redistribution) and the finance of the health insurance through the change in contributions paid by households or individuals. The focus of this study does not lie in proposing a yet another alternative in the midst of the controversy over the reform. Rather, its objective is to analyze the effects of a reformed contribution scheme based on the existing alternatives. In doing so, the analysis will be based on the potential level of contribution burden, not limited to the current level of burden. In other words, when we assume that the contribution scheme is reorganized to be single

11 Introduction 9 income-based, the reform effect will be analyzed based on all incomes that can be achieved potentially including under-reported and unreported incomes not just on incomes that National Tax Service or National Health Insurance Service (NHIS) is aware of. Generally, when the contribution scheme reform is discussed, revenue-neutral reform is the usual basis. However, this study is not limited to a revenue-neutral reform because it aims to analyze potential effects. To be more specific in terms of methodology, it is because the ultimate purpose of the reform is to increase the range of objects of contribution i.e. to find and expand the sources of tax revenue. In this respect, the analysis will be conducted based on the expected effects of the reform that includes unreported incomes. Therefore, in our analysis, the contribution level will be slightly higher than the current level. In summary, broadening the scope of incomes subject to insurance contribution assessment means both the diversification of the type of income subject to contribution assessment and an increase in the rate of income reporting. This study uses three standards. Standards A and B are simplifications of the contribution scheme into a completely income-based one. Standard B is a hypothesis that applies minimum contribution while Standard C is a hypothesis in which the contribution assessment base is increased to include all incomes and property regardless of employment status in contrast to the current scheme that imposes property-based contribution only on the self-employed by adding a 20% property-based contribution to an 80% income-based contribution. When we assume that the existing contribution scheme is converted to an income-based one according to these three standards, we will look mainly at the redistributive effects of Standard A, which is the most simplified version, and use the effects of Standard C, which is the least feasible yet includes property-based contributions, for comparative purposes. In examining the distribution (incidence) and redistribution effects of the national health insurance system, this study combines the short-term effects from the cross-sectional viewpoint and long-term effects from the life-cycle viewpoint, using two data sources: one from the Household Income and Expenditure Survey (hereinafter the HIES) published from 2006 to 2014 by the Statistics Korea, and the other from the National Survey of Tax and Benefit (hereinafter NaSTaB) published from 2008 to 2014 by the Korea Institute of Public Finance. These

12 The Effects of Reform in the National Health Insurance Contribution Scheme: On Equity in Health Insurance Contribution Burden 10 data are complementary to each other in the analysis of short and long-term effects. Unlike the public data (sample cohort data provided by the NHIS), they also include detailed information on income and health insurance contribution as well as various demographic and economic variables, which helps analyze the effect of the contribution scheme reform from various angles. The NaSTaB data is longitudinal, and this has several advantages: it distinguishes individual health insurance contributions made by the employee from those made by the self-employed; it allows an accurate look at individual contributions since they are recorded by the units of won; and incomes can be accurately identified since the proof of income tax payment is collected for the NaSTaB data set. However, as only about half of the income tax reporters submit the document of proof, large numbers of income information are omitted in the samples, and there is not sufficient detailed information with which to make an estimate of the omitted income variable. As for the HIES s cross-sectional data, it is impossible to distinguish the employee insured from the self-employed insured, and many items related to income tax burden are omitted in the data set. However, it has abundant information on the amount of income by individual and income type as well as detailed information needed for the estimation of the income tax burden, which makes it possible to estimate household incomes at a reliable level. What we mean by income here is not taxable income used in the NHIS (i.e. income that the taxation authorities use for tax administration purposes), but a comprehensive concept of income that encompasses unreported nontaxable incomes (transfer income, other incomes subject to separate withholding taxation, and unreported business income, etc.). Since this study aims to analyze incidence and redistributive effects when the object of contribution assessment is expanded to all incomes, we consider it more desirable to base the analysis on the broader and more comprehensive conception of household income. This study consists of the followings. Chapter II presents the current state and problems of the national health insurance contribution scheme. Chapter III first discusses scenario analyses conducted in previous studies and reform scenarios prepared by the aforementioned planning task force, the party-government consultative body and individual parties, and then introduce

13 Introduction 11 a set of scenarios to be used in this study. Chapter IV conducts a cross-sectional analysis on the effect of the reform in the health insurance contribution scheme using the HIES data. Chapter V conducts a panel analysis on the effect of the reform in the health insurance contribution scheme using the NaSTaB data. And lastly, Chapter VI presents the conclusion and policy implications.

14 Ⅱ Current State and Issues of the National Health Insurance Contribution Scheme 1 Current State of the Contribution Scheme In Korea, the national health insurance system was introduced in 1977, and universal healthcare was achieved in At the time of introduction, insurance contribution was imposed and levied on employees by each workplace association and on the self-employed by separate associations organized at Si/Gun/Gu level. However, as the NHIS became a single insurer in July 2000, a single contribution assessment standard was needed. Initially, the government intended to consolidate the contribution scheme with an income-based standard, but the preexisting separate contribution schemes by employment status were maintained after all because a proper system was not in place to track down individual incomes. Currently, the health insurance contribution scheme 1) is largely divided into two parts: the employee insured and the self-employed insured. Looking at each part, however, the scheme can be divided, again, into several groups based on income. In the case of the employee insured, the contribution is imposed per each employee and calculated by multiplying the monthly salary by the rate of contribution (6.12% as of 2016), half of which (3.06%) is paid by the employer. Here, the calculation is based on the monthly salary reported in the 1) Refer to the Introduction to the System section in the NHIS homepage. (Accessed )

15 Current State and Issues of the National Health Insurance Contribution Scheme 13 previous year (business income in case of the employer), and the amount is settled based on the total amount of wages of the pertinent year reported after contribution is imposed. Contribution by the employee insured has upper and lower limits. If the monthly salary exceeds 78.1 million KRW, 78.1 million KRW (contribution: 4,779,729 KRW) is applied as the upper limit, and if the monthly salary is less than 0.28 million KRW, 0.28 million KRW (contribution: 17,136 KRW) is applied as the lower limit. Starting from September 2012, if the income other than the wage of a wage earner exceeds 72 million KRW per year, the additional contribution has been imposed on the monthly income based on the income other than wage. The monthly income refers to the income of the employee insured excluding the wage included in the monthly wage, and it is calculated by dividing the sum of the interest income, dividend income, business income, earned income, annuity income and other incomes by 12. Here, the sum reflects 100% of the interest, dividend, business and other incomes and 20% of earned and annuity incomes. For the rate of contribution, 3.06%, which is the rate borne by the employee, is applied on wage income. The rate of contribution for the monthly income does not have a lower limit, but the upper limit is the same as that of the monthly wage of 78.1 million KRW. Contribution for the self-employed insured is imposed per household. The amount of contribution is assessed by calculating a household s score based on the self-employed subscriber s income, property, automobile, the standard of living, and labor force participation rate and multiplying the total score by the amount per point. Here, different contribution standards apply based on an annual income of 5 million KRW. For low-income class with an annual income of 5 million KRW or less, income assessment is based on the standard of living and labor force participation rate, and for the households with an annual income of more than 5 million KRW, taxable income is calculated based on global income (20% of wage and annuity incomes and 100% of interest, dividend, business and other incomes). The score for the standard of living and labor force participation rate are divided into 30 grades and taxable income into 75 grades. As for property, which is included in the contribution assessment base regardless of annual income, the score is divided into 50 grades, and the object of contribution assessment includes housing, buildings, lands, ships, airplanes (100%

16 The Effects of Reform in the National Health Insurance Contribution Scheme: On Equity in Health Insurance Contribution Burden 14 of the assessment standard property tax) and key-money lease/monthly rent (30% of key-money lease/monthly rent). In the case of automobiles, the score is divided into 7 grades considering type, displacement volume and the years of use. As the self-employed insured have no employer who pays half of the contribution, unlike the employee insured, the whole contribution should be paid by households. In the case of the self-employed insured, the minimum contribution is a monthly payment of 3,590 KRW with a score of 20 assessment points and the maximum contribution is a monthly payment of 2,277,330 KRW with 12,680 assessment points. <Table II-1> presents changes in the rate of contribution for the monthly wage and monthly income of the employee insured and the amount of contribution paid by the self-employed insured per assessment point. As of 2016, the rate of contribution for the monthly wage of the employee insured is 6.12% and that for the monthly income is 3.06%, half of the former, and the amount borne by the self-employed insured per point is KRW. Table II-1 Changes in the Rate of Health Insurance Contribution for the Employee Insured and the Amount borne by the Self-Employed Insured per Point (Unit: %, KRW) Rate of contribution for the employee insured Rate of contribution for monthly salary Rate of contribution for monthly income Amount borne by the self-employed insured per point Source: NHIS, (Accessed ).

17 Current State and Issues of the National Health Insurance Contribution Scheme 15 <Table II-2> presents the current state of health insurance subscribers by employment status. The number of the employee insured is steadily increasing: in 2005, the number of the employee insured was 27 million (57%) and the self-employed insured 20 million (43%); in 2014, the figures posted 36 million (71%) and 15 million (29%), respectively. This is because employees working at workplaces with fewer than five employees became eligible for the employee insured status as the criteria was eased in 2003; and currently, employees working at workplaces have just one or more full-time employees are classified as the employee insured. On the contrary, the rate of dependents of the employee insured decreased steadily from 64% in 2005 to 57% in 2014, which is can be attributed to the strengthened eligibility criteria for the dependent status during this period. Table II-2 Current State of the Employee Insured and the Self-Employed Insured (Unit: 1 mil. people, 1 mil. households, %) Health Insurance Subscribers (A) 4,739 4,741 4,782 4,816 4,861 4,891 4,930 4,966 4,999 5,032 The employee insured The selfemployed insured Sub-total (B) 2,723 2,845 2,942 3,042 3,141 3,238 3,326 3,411 3,501 3,560 Subscribers 975 1,042 1,117 1,162 1,215 1,276 1,340 1,399 1,461 1,514 Dependents (C) 1,749 1,803 1,825 1,880 1,927 1,962 1,986 2,012 2,040 2,046 Subscribers (D) (Number of Households) Percentage of the employee insured (B/A) Percentage of the self-employed insured (D/A) Percentage of dependents (C/B) 2,016 1,896 1,840 1,774 1,720 1,652 1,604 1,556 1,498 1, Source: NHIS, National Health Insurance Statistical Yearbook, Each year

18 The Effects of Reform in the National Health Insurance Contribution Scheme: On Equity in Health Insurance Contribution Burden 16 <Table II-3> presents the contribution revenue by employment status. As of 2002, of the total contribution revenue of 11 trillion KRW, the contribution paid by the employee insured and the self-employed insured was 7 trillion KRW (63%) and 4 trillion KRW (37%), respectively. In 2014, however, the total revenue was 42 trillion KRW 34 trillion KRW (83%) paid by the employee insured and 7 trillion KRW (17%) by the self-employed insured, which indicates that the contribution paid by the former group represented an absolute share. In other words, this shows that the finance of Korea s national health insurance is highly dependent on contributions imposed on earned income. Table II-3 Insurance Contribution Revenue by Year Total (A) Insurance Contribution The employee insured (B) The self-employed insured (C) Percentage of contribution by the employee insured (B/A) (Unit: KRW 100 million, %) Percentage of contribution by the self-employed insured (C/A) ,277 68,719 40, ,409 91,684 45, , ,283 47, , ,209 48, , ,965 49, , ,485 53, , ,297 59, , ,377 59, , ,831 63, , ,416 67, , ,796 70, , ,751 71, , ,865 72, Note: 1. Based on settled amounts 2. Contribution paid by the employee insured and that by the self-employed insured are based on the assessed amount of contribution and the imposed amount of contribution, respectively. 3. National subsidy is excluded, and contribution paid by the employee insured includes voluntarily and continuously insured persons contribution and the monthly income-based contribution 4. Based on actual residence on resident registration Source: NHIS, National Health Insurance Statistical Yearbook, Each year

19 Current State and Issues of the National Health Insurance Contribution Scheme 17 2 Issues of the Current Contribution Scheme Possible problems regarding the health insurance contribution scheme are divided into three main areas. The first area includes problems related to whether the total amount and fiscal balance of the health insurance finance are adequate from the macro-fiscal perspective. In other words, some questions can be raised if fiscal resources provided by a certain contribution scheme are adequate enough to ensure sound and sustainable finance. The second area is related to a set of questions concerning equity from the micro-fiscal perspective: i.e. equity in contributions among individuals or classes defined by certain characteristics. The third area concerns how a contribution scheme is executed in the operational process, rather than the design of the scheme itself. This is related to errors, corruption or other work inefficiencies that can occur in the process of executing a given contribution scheme. Most of the problems raised about the health insurance contribution scheme are related to the issue of equity. As discussed earlier, equity issues arise from ambiguousness concerning whether the key criteria in contribution assessment are based mainly on income or the standard of living. Equity here refers to whether contribution is made fairly among two or more groups. Thus, according to which groups are in comparison, the equity issue can be divided again into the one between the employee insured and the self-employed insured and the one within each group with the same employment status. The latter s case, comes down to a problem in which the upper class bears relatively less burden than the lower class as the comparison is based on income or the standard of living within each group with the same employment status. In the former s case i.e. equity between the employee insured and the self-employed insured, the problem of measurement and different standards of contribution assessment come in to play as complex factors, on top of the above problem. To look at representative problems in which these equity issues are pointed out to be present in specific ways and, thus, need improvement, they include differences by employment status in such aspects as contribution assessment factors, the eligibility of dependents, income calculation criteria and the subject of contribution payment.

20 The Effects of Reform in the National Health Insurance Contribution Scheme: On Equity in Health Insurance Contribution Burden 18 A. Difference in Contribution Assessment Factors by Employment Status As for most of the employee insured (those with global income of 72 million or less KRW, contribution is assessed based on earned income only. In contrast, it is difficult to track down the income of the self-employed. For this reason, contribution paid by the self-employed insured is assessed based on every item, such as property and automobile, through which the ability to pay can be determined. Included in property are real estate, key-money lease/monthly rent, etc., and most of the property owned by the self-employed insured are housing for residential purposes. Contribution is assessed based on automobile and property although they do not generate actual income. Also, the object of contribution assessment in the case of the self-employed insured includes interest, dividend, business, annuity and other incomes, from which inequity arises. For the employee insured, contribution is imposed only on earned income even when their global or financial income is the same as or more than that of the self-employed insured. As for the self-employed insured, however, contribution assessment includes every single item of income. Therefore, the burden of health insurance contribution can be higher on the self-employed insured even when the employee insured and the self-employed insured might have the same amount of income. Another issue of concern on inequity between the employee insured and the self-employed insured is the share of contribution. Currently, the employee insured pay 50% of the contribution and the rest is paid by the employer whereas the self-employed insured pay 100% of the contribution, which has caused a controversy over inequity. However, from the economic perspective, the 50% contribution borne by the employer is actually passed on to employees in the form of the labor costs, and employees get to pay the whole or part of it. Also, the claim about inequity is untenable since it overlooks the fact that the government takes up the role of the employer in bearing financial support that corresponds to contributions paid by the self-employed insured. The misguided perception seems to be due to the legal provision on the governmental support, from which it is difficult to recognize that such governmental support corresponds to the contribution borne by the employer in the case of the self-employed insured.

21 Current State and Issues of the National Health Insurance Contribution Scheme 19 Yet, inequity occurs not only between the employee insured and the self-employed insured, but also within the self-employed. The employee insured should pay an additional contribution of 180,000 KRW if their global income other than earned income is million KRW. But if their global income is million KRW, they do not have to pay any additional contribution. That is, the difference of only 20,000 KRW in annual incomes results in an additional contribution as much as 180,000 KRW, and this has led to complaints. Within the self-employed insured, different grades are assigned according to contribution assessment factors, such as income, property, labor force participation and automobile. However, the grading system is not standardized; neither does the score for each item. Still, the amount of contribution is assessed by tallying up the scores, assigning one of 105 grades to the sum of the scores and, then, multiplying it by the amount per point. B. Eligibility of Dependents Depending on employment status, the eligibility criteria for dependents differ as well. While there is no limit in the number of dependents to be covered for the employee insured, all household members of the self-employed insured should share the payment of contribution as the insured since such factors as gender, age, labor participation rate. In the case of the self-employed insured, all household members of the self-employed insured should share the payment of contribution because the amount of contribution is assessed by household based on gender, age and labor force participation rate, etc. In practice, therefore, dependent coverage applies to the employee insured only. To take a look at the eligibility of dependents for the employee insured, eligibility is approved if each item of global income is less than 40 million KRW, In other words, dependents are considered eligible not when the total global income does not exceed 40 million KRW, but so long as all of the following requirements are met at once: interest and dividend incomes of less than 40 million KRW; earned and other incomes of less than 40 million KRW, annuity income of less than 40 million KRW and business income of 5 million KRW or less. As such, controversies over inequity can ensue since it amounts to a free ride in comparison to employees or self-employed persons who should

22 The Effects of Reform in the National Health Insurance Contribution Scheme: On Equity in Health Insurance Contribution Burden 20 pay insurance contributions if they have earned income as few as 5 million KRW. C. Regressivity in Assessment of Contribution by the Self-Employed Insured In the case of the self-employed insured, the contribution scheme is regressive since contribution burden is likely to be less for those with more income and/or property. Of course, regressivity exists to a certain degree also in the case of the employee insured due to the upper and lower limits of contribution, but it is not as serious as in the case of the self-employed insured where the contribution rate drops drastically as income and property increase. The reason for the regressivity in the latter s case is that there are grades according to the level of income or property with points assigned to each grade yet the points do not go up in direct proportion to the income level. Let us examine the rate of contribution (monthly contribution/the median value of annual income divided by 12 months) based on the monthly contribution calculated by multiplying points assigned to a given grade by the amount per point. As for the Grade 1 with the low income, the contribution rate is 14.9%, which is more than twice that of the employee insured (6.12%). Meanwhile, Grade 50 s contribution rate stands at 2.8%, which is a significant decrease to as few as about a fifth of that of Grade 1. This clearly reveals regressivity in contributions made by the self-employed as the more income one has, the lower the ratio of the contribution rate to income becomes. Property-based contribution assessment shows a much more serious degree of regressivity. While the amount of property soars by 3,000 times from Grade 1 to Grade 50, the difference in points between the two grades is merely 67 times (1,475 points/22 points). As such, while the property-based contribution is to the tune of 58% in Grade 1 (the poor class with almost no property) the figure posts just 3.5% in Grade 50 (the wealthy class with a large amount of property). As the property-based grade goes down, the property-based contribution rate increases at a more abrupt pace. This shows how regressive property is as a contribution assessment base for the self-employed insured.

23 Ⅲ Previous Studies and Introduction of Scenarios 1 Previous Studies Previous studies on the national health insurance contribution scheme have focused on reform scenarios i.e. policy reforms about how the current contribution scheme is to be changed. However, we aim to focus on the effects of changes in a reform scenario on redistribution and the finance of the health insurance, rather than the scenario itself. However, the effects can differ greatly depending on how a scenario is designed, which leads us to find it necessary to analyze scenarios first. Thus, we begin this chapter by reviewing previous studies on reform plans. In response to the consolidation of the finance of the National Health Insurance in 2002, Choe Byeong-ho et al. (2001) raise the equity problem in the current contribution scheme and discuss improvement measures with the scope of their study limited to the self-employed insured. They suggest two alternatives. One way (Alternative 1) is to grant points to each factor like gender, age, income, property and automobile and, then, add all the granted points to assess the amount of contribution. And the other way (Alternative 2) is to assess the amount of contribution only based on income, property and automobile for those with a certain level of income or more and based on gender, age, property and automobile, instead of income, for those with less than a certain level of income. In their study, a comparison is made between the two alternatives and the current scheme, and they present the attendant changes in contribution. In a comparison between Alternative 1 and the current scheme, 82.48% of the

24 The Effects of Reform in the National Health Insurance Contribution Scheme: On Equity in Health Insurance Contribution Burden 22 households with no income data is expected to pay less while 72.73% of the households with income data is expected to pay more. Also, to take a further look at the households with income data by income level, those with lower income are expected to experience the lowest increase in contribution. In their research for the NHIS, Lee Yong-gap et al. (2006) suggest an income-centered, equitable contribution scheme plan. While income is selected as the contribution assessment factor, they include capital gain to reflect the payment ability based on property. Therefore, a person who has income (earned income and global income) pays the contribution in proportion to his or her income whereas for a person who has property yet no income, the amount of contribution is assessed in proportion to capital gain. And a person who has no income pays just the basic contribution (4,000~7,900 KRW.) As for minors under 19 years old, the authors claim that they are to be exempt from contribution because they are dependents. As for people who need social protection (seniors aged 65 or older or the disabled, etc.), they argue that a reduction in contribution or exemption should be considered. In Lee Yong-gap et al. (2006), a transitional model is designed in consideration of the current conditions, and four scenarios are presented as per the rate of contribution and basic contribution. According to their simulation analysis, the contribution burden decreases for the households of the self-employed insured and the burden increases for the households of the employee insured by 17-19% while the neutrality of the entire finance is maintained. Their study marks the first to propose a specific idea on basic contribution. Overall, the proposed alternative is income-centered yet partially considers property in that property-based contribution is imposed on the wealthy with a large amount of property while contribution assessment is based on income. Of course, there remain concerns about a significant rise in contribution for the employee insured and the public s acceptance of contribution assessment based on capital gain. In Shin Young-seok el al. (2007) and Shin Young-seok (2011), a reform plan is divided into short-term and mid- to long-term tracks as part of a measure to secure the finance of the national health insurance. For the short-term, the research points out areas in which improvement can be made quickly under the current contribution scheme; for the mid- to long-term, it argued for an

25 Previous Studies and Introduction of Scenarios 23 income-based, single contribution scheme. As for the short-term plan, the research proposes the strengthening of the criteria for dependent eligibility, the disqualification of a person with annuity or business/lease income of one million KRW or more as a dependent, the reorganization of the scheme of double contribution assessment regarding property and automobile, the abolishment of the 5 million KRW income criterion for the self-employed insured, the introduction of basic contribution, and a shift in assessment base from evaluated income to taxable income. And the mid- to long-term plan argues for an incomecentered reorganization by adopting such measures as the change of employment status for those working at a workplace with one or more employees from the self-employed insured to the employee insured, the introduction of basic contribution (for 20% of the whole nation including farmers and fishermen, the low-income vulnerable class and self-employed small business owners) and the expansion of contribution assessment base to include incomes other than earned income. In their research for the NHIS, Shin Hyeong-jun et al. (2008) present a follow-up study on policy measures to improvement equity in the contribution scheme. Instead of a complete reform, the study focuses on a partial reform to address certain unreasonable elements in the existing system, which includes the abolishment of the 5 million KRW income criterion for the self-employed insured, the introduction of basic contribution, and improvement of the employee insured dependent scheme. Shin et al. takes a reserved position about a complete shift toward an income-centered contribution scheme based on their analysis, which shows that the income estimation rate is only about 40% in the case of the self-employed insured (24% in the case of households with the income of 5 million KRW or less.) For this reason, they propose a scenario that abolishes the 5 million KRW income criterion for the self-employed insured (less than 5 million KRW for 84.2% of the self-employed insured and 5 million KRW or more for 15.8% of the self-employed insured), unifies the income grade system, replaces the points assigned according to gender and age in the contribution assessment based on the standard of living with the introduction of basic contribution based on age and assesses the amount of contribution by abolishing the existing redundancies, such as income, automobile and property, in the contribution assessment based on the standard of living. According to

26 The Effects of Reform in the National Health Insurance Contribution Scheme: On Equity in Health Insurance Contribution Burden 24 the scenario, the contribution is assessed, instead, by multiplying total points (i.e. basic points + income points + property points + automobile points) by the amount per point. In this study, Shin et al. conduct a simulation analysis on two models in which points are differently distributed among contribution assessment factors. Thus, Model 1 is characterized by a high concentration of points in the property factor and Model 2 in the income factor. The result of analysis of Model 1 (a larger share of the property factor) indicates that 24% of the households would see an increase in their insurance contributions while 76% would pay less contribution. In the case of Model 2 (a larger share of the income factor), the result shows that 25% is expected to pay more and 75% to pay less, which indicates no significant difference between the two models. As for research by Korea Employers Federation (2012), it is pointed out that the establishment of an equitable contribution scheme is a more urgent task rather than the consolidation of the different contribution schemes applied to the employee insured and the self-employed insured. Regarding a change to an income-centered contribution assessment scheme, they also argue there will be much weightier burden on the national treasury due to a loss arising from excluding property and automobile from the contribution assessment base for the self-employed insured as well as an already low rate of income estimation within the group. Still, their research stops short of coming up with a specific reform plan while proposing only the improvement of the unreasonable dependent scheme. Ryu Geon-sik (2011) introduces 5 brief plans regarding the reform of the contribution scheme. In the first plan, he calls for the strengthening of income estimation measures on the self-employed insured and the reduction of contribution assessed based on housing and automobile. The second and third plans are related to the strengthening of dependent eligibility criteria (the exclusion of siblings and the disqualification of person with a certain amount of property.) The fourth one includes a phased expansion of the contribution assessment base for the employee insured from earned income to non-earned income (starting with annuity income.) And the fifth plan concerns a shift toward an income-centered contribution scheme regardless of employment status in the mid to long term.

27 Previous Studies and Introduction of Scenarios 25 Shin Hyeon-woong (2015) narrows the problems of the current contribution scheme down to three: the inclusion of property and automobile in the contribution assessment on the self-employed; contribution assessment based only on earned income in the case of the employee insured; and lax eligibility criteria on the dependents of the employee insured. In that regard, he suggests limiting the assessment base to income while extending the object of contribution assessment to include non-earned income, and strengthening the dependent eligibility criteria. 2 Introduction of Scenarios The purpose of this study is to establish simple and reasonable standards for the health insurance contribution scheme and focus on the analysis of its effect, not to present a new reform scenario. The basic principle for the reform that this study has looked at is to eliminate a distinction between the employee insured and the self-employed insured and assess the amount of contribution in proportion to the ability to pay. Here, the ability to pay is to be based on global income, and all Korean people should pay contributions in principle. Income subject to contribution assessment is global income (earned income, business income, financial income, annuity and other incomes) that is generated regularly. All incomes should be reflected 100% and the contribution rate is 3.06%, which is the rate borne by the employee insured, out of the contribution rate of the pertinent year, 6.12% (as of 2016). In other words, the total contribution rate is 6.12% in the case of earned income, with half paid by the employee and the other half paid by employers. As for the rest of the income, however, the rate starts from 3.06% because there is no employer. Under the standard, the contribution is assessed in direct proportion to income. Thus, it can be imposed on the low-income class with a unit of 100 KRW or 10. To resolve this inconvenience, we present the second standard in which the minimum fixed amount of contribution is imposed if the annual global income falls in the range of less than 0.28 million KRW x 12 months (3.36 million KRW), which is the minimum contribution range for the employee

28 The Effects of Reform in the National Health Insurance Contribution Scheme: On Equity in Health Insurance Contribution Burden 26 insured, and a fixed-rate contribution for those with the annual global income exceeding the range. For the employee insured, the monthly minimum contribution will be 17,136 KRW (0.28 million KRW x 6.12% (as of 2016)), individuals will pay the half of it (i.e. 8,568 KRW.) Therefore, individuals with a monthly global income of less than 0.28 million KRW pay the minimum fixed amount of 8,568 KRW, and individuals with a monthly global income of 0.28 million KRW or more pay their contributions at a fixed rate of the monthly global income x 6.12%/2. Only in the case of earned income, their employers will pay the other half (3.06 %.) The contribution burden presented under this simplified standard is the result obtained when the contribution is assessed only based on income. From the life-cycle viewpoint, however, the retired, elderly group live on property like real estate or financial assets as their income base weakens. In this regard, it seems reasonable to consider property, rather than income, when measuring their ability to pay insurance contribution. Of course, property-based contribution assessment is very unfeasible, but we include in our analysis the complementation of income-based contribution with property-based contribution to compare how the amount of contribution changes from an academic perspective. Here, we will assess the rate of contribution based respectively on income and property with 80% of the contribution from income and 20% from property under the premise that the total contribution revenue is equal to the first standard. The range of property here is the same as that is currently applied to the self-employed insured by the NHIS, such as housing, real estate, financial assets, automobile and key-money lease/monthly rent. However, as there are many unresolved issues e.g. inaccuracy of property reporting in the survey data and whether debt should be added or deducted, we limit the share of property in contribution assessment to 20%. In including property in insurance contribution assessment, the third standard can be an attractive alternative in that individuals with more property are likely to have the ability to pay more contributions. However, much caution should be paid to a property-based health insurance contribution assessment. Also, it should be noted that it is very unfeasible in practice since there are many demerits in terms of transparency, objectivity and comprehensiveness in contribution assessment despite the merit of satisfying the ability principle.

29 Ⅳ Effect of Reform in the National Health Insurance Contribution Scheme: Cross-Sectional Analysis 1 Background Recently, shifting the existing health insurance contribution scheme to a single, income-based standard is being actively discussed. The reform plans suggested from various circles are different in their details, but are similar in that they are focused on changing the current standard to a single, income-base one. The main object of this chapter is to evaluate the expected policy effects of a shift toward an income-based contribution scheme from the equity perspective. Most previous analyses of the equity effect of the health insurance system focus on the joint distribution between the burden of insurance contribution and benefits. If the research method employed in the previous research is adopted, income groups are classified based on taxable income estimated by the taxing authority or the NHIS. This method is effective in analyzing the distribution, redistribution and cross-subsidization effects of the health insurance system itself, such as insurance benefits relative to contribution burden within the frame of the health insurance scheme, and it (the former) is employed in. most of the recent studies that analyzes the expected effects of a change to an income-based contribution assessment standard. However, these analyses are limited in that they discuss the equity effect within the given framework of the health insurance system while failing to provide information on the equity effects on the

30 The Effects of Reform in the National Health Insurance Contribution Scheme: On Equity in Health Insurance Contribution Burden 28 distribution and redistribution of income, including all types of incomes in a comprehensive sense, beyond the frame of a direct comparison between contribution burden and insurance benefits. Complementing previous studies and expanding the prospect of equity effects relating to the national health insurance, we aim to analyze and evaluate income distribution and redistribution effects based on a general sense of income encompassing all other types of income than taxable income, which is the current object of contribution assessment and independent households so as to objectively evaluate the level of the understanding of the health insurance system, policy effects and its merits and demerits. This chapter analyzes and evaluates effects on equity using the HIES data published by Statistics Korea. 2) In the following Chapter V, we estimates and evaluates equity effects using the NaSTaB data. The reason for using these two different data sets is that relying on only one of them cannot cover the comprehensive effects of reform that this study aims to probe. In other words, we use the two structurally different data sets in terms of the method and form of data accumulation basically because we aim to examine both the short-term and long-term effects of reform on equity. 2 Analysis of Income Redistributive Effects A. Scope of Analysis To analyze the policy effects of the national health insurance system, this chapter uses raw data (based on annual data) from the HIES published by Statistics Korea. The policy effects of the national health insurance system will be examined in two main aspects: the distribution of contribution by income 2) In addition to the HIES data, data from the Survey of Household Finances and Living Conditions is also cross-sectional. Although the latter contains information on property, measurement errors are much fewer in the former because it uses the method of keeping a household ledger every month for data investigation and accumulation. In the follow-up research, it still deserves to conduct an analysis including the latter data set.

31 Effect of Reform in the National Health Insurance Contribution Scheme: Cross-Sectional Analysis 29 level and the income redistributive effects. To look into changes in policy effects based on a time-series analysis, we set the period of analysis to be between 2006 and 2014 in that data from the period includes information on single-person households as well. <Table IV-1> presents the descriptive statistics (as of 2014) on household characteristics contained in the HIES. Table IV-1 Descriptive Statistics of the Household Income and Expenditure Survey (Annualized Basis) by Statistics Korea (As of 2014) (Unit: Age, Number of People, KRW) Average Standard Deviation Maximum Minimum Age of householder Number of members of household Number of the employed Market income 43,304,592 31,489, ,867,825 0 Private income 45,725,985 31,170, ,867,825 0 Gross Income 48,087,628 30,254, ,353,880 0 Disposable income 43,858,486 25,996, ,417,972 0 Post-tax income 41,433,144 24,734, ,305,601 0 Final income 46,734,821 26,523, ,077,856 0 Consumption expenditure 26,533,888 16,895, ,638,775 0 Income Tax 1,774,693 3,358,160 55,606,979 0 Property Tax 203, ,330 5,685,972 0 Social security contributions 2,432,345 2,411,277 43,561,419 0 Private transfer income 2,421,393 8,025, ,824,041 0 Public transfer income 2,361,643 6,311,668 86,652,948 0 Note: Estimates made by the authors using the data annualized based on raw data from the Household Income and Expenditure Survey The policy effects of the health insurance system are divided into two main parts: the collection (or payment) of contribution (burden) and insurance benefits (coverage). Therefore, we include both insurance contribution and insurance benefits in probing policy effects and income redistributive effects by income level.

32 The Effects of Reform in the National Health Insurance Contribution Scheme: On Equity in Health Insurance Contribution Burden 30 As for changes in distribution by income level, we make estimation about and analyze the distribution of incidence 3) by income level based on market income, the amount of insurance contribution and insurance benefits to compare the effective rate of contribution burden, etc. Also, to understand the relative size of the policy effect of the national health insurance system, we evaluate the size of the policy effect of the insurance system through a relative comparison with income tax. Because it can be considered that the calculation of health insurance contribution is primarily based on market income, earned income and business income (including lease income), which takes up the majority of market income, will be the object of income tax. For convenience and within the range in which confusion does not arise, we will refer to them 3) The recent studies respectively by Sung Myung-jae (2016) and by Cho Yoon-je, Kim Jong-il, Park Jong-kyu, Sung Myung-jae, Hee-suk Yun and Lee Chang-won (2016, Chapter 5) are representative ones that make estimation about the distribution of various tax burdens and fiscal expenditure benefits. In order to estimate the distribution of tax burdens including various in-kind benefits, income tax and excise tax, we obtained the same values, in principle, by adopting the same research data and methodology employed in the above studies, and used the values in our analysis. Therefore, this study and the previous studies mentioned above use the same values of tax burden and fiscal expenditure benefits by household. However, this study is differentiated from the previous ones in terms of the scope of research topic and the object of research. The previous studies estimate overall income redistribution effects by showing the overall joint distribution of tax burden and fiscal expenditure benefits, but they do not discuss incidence distribution, equity and income redistribution effect regarding specific individual items. On the contrary, this study is differentiated in that it focuses on conducting a comparative analysis on the incidence distribution and income redistribution effects of individual items-especially the national health insurance contribution and coverage item-and those of other items. Meanwhile, there are some cases in the previous studies in which the tax parameters of the past were used for estimating the amount of each health insurance coverage item because statistical data on the performance value of the health insurance benefits of the analysis year had not been announced yet at the time of analysis or had not been updated due to a lack of information. In this study, we have updated health insurance benefit parameters based on performance value and re-estimated the amount of insurance benefits per household amount by households for the corresponding items. The health insurance benefits in , 2011 and 2014 have been newly re-estimated in comparison with Sung Myung-jae (2016). The distribution of final income has been revised to the same degree as revisions in the estimation results of in-kind benefits. Attention should be given to the fact that those parts show different estimation results from the previous studies. In the case of Cho Yoon-je et al. (2016) and Sung Myung-jae (2016), there are some parts where they conduct an analysis on the same object (incidence analysis of various tax burdens and fiscal expenditure benefits in 2013), but the estimation results are more or less different from each other. This is because Cho Yoon-je et al. (2016) and Sung Myung-jae (2016) adopt different research methods in some items. The former uses estimated values of excise tax burden from the survey data without adjustment while the latter applies an adjustment coefficient per tax item so as to equalize the average burden on each household calculated based on revenue per tax item and the average burden per household calculated as a result of estimating the survey data. Among these two, our analysis adopts the latter s method of applying an adjustment coefficient.

33 Effect of Reform in the National Health Insurance Contribution Scheme: Cross-Sectional Analysis 31 as earned global income tax. The main object of analysis is policy effects of the national health insurance system in light of income distribution and redistribution, and we take a look at those effects by household since the data under consideration i.e. the statistics of the health insurance contribution is reported by household. Also, from the welfare economic perspective, the household is the smallest unit of consumption to study income redistributive effects and, thus, considering the household as the basic unit of analysis is a widely used research method in the academia. However, as health insurance contribution is imposed on individual incomeearners, the expected incidence and policy effects of the reform will be calculated in the form of burden borne by individual while taking the household, which is the basic unit of welfare, as a reference point for analysis of income redistributive effects. B. Analysis Method For the incidence analysis of health insurance contribution and the health insurance coverage (benefits), we divide income level into 10 income deciles based on gross income. 4) The reason for using gross income for analysis is to minimize the arbitrariness of analysis by tallying up all types of incomes available. The measurement of income redistribution effects is based on the fall rate (or reduction rate) of the Gini coefficients of both pre-tax income and post-tax income. Since the lower absolute value of the Gini coefficient means less income inequality, it indicates a positive effect on income redistribution. In contrast, the higher absolute value of the Gini coefficient indicates a negative effect on income redistribution because it means more income inequality. 4) In this chapter, the classification of income level is based on gross income. In the previous studies considered in Section 2 above, the classification is based on taxable income related to health insurance contribution assessment. As taxable income is limited to a narrow sense of income that the taxing authorities grasps for the purpose of taxation (or health insurance contribution assessment), it should be noted that gross income-based income deciles and contribution assessment-based income deciles are completely different.

34 The Effects of Reform in the National Health Insurance Contribution Scheme: On Equity in Health Insurance Contribution Burden 32 Health insurance contribution per household was investigated directly from the HIES. However, the survey does not have information on the amount of health insurance benefits. In particular, it has some information on medical expenses, but it is difficult to determine whether health insurance benefits have been dispensed (or applied) and how much expenditure has been caused by the application from that information alone. Therefore, this study has adopted the research data and methodology for analysis employed in Sung Myung-jae (2016) to replicate and report the distribution of burden and benefits per item in relation to various tax burdens and cash and in-kind benefits by income level including health insurance benefits between 2007 and Meanwhile, the estimated values of health insurance benefits per household are needed to estimate the effect of health insurance benefits on income redistribution. To this end, the amounts of contribution and benefits by household have been separately estimated and stored for use in the process of recalculating the distribution of contribution and benefits by income level for the purpose of estimating the income redistribution effect of individual health insurance coverage items. To resolve an estimation bias arising from not using statistics on the performance figure of the corresponding year, this study made an estimation by revising the amount of insurance benefits for the corresponding year based on the insurance benefit performance statistics of the same year. While basically using the same methodology in the previous research, revision was also made to the amount of insurance benefits by household for estimation by updating statistical data on the total amount of health insurance benefit performance and the average amount of benefits per capita by gender and age (data from the National Health Insurance Statistical Yearbook of each year) and modifying the parameters used in the estimation model. In addition, this chapter estimates and analyzes the expected policy effect of a change to an income-based contribution assessment. For that purpose, we adopt the method of microsimulation analysis.

35 Effect of Reform in the National Health Insurance Contribution Scheme: Cross-Sectional Analysis 33 C. Changes in Income Redistributive Effect As in other studies, this study also measures the income redistribution effects of various tax and fiscal expenditure benefit items based on the fall rate of the Gini coefficient of pre-tax/post-tax income or pre-benefit/post-benefit income. [Figure IV-1] through [Figure IV-3] presents an comparison of the estimated rates of change for individual items by comparing the fall rates of the Gini coefficient of individual tax and fiscal expenditure benefit items from the market income stage to final income stage in 2006, 2010 and [Figure IV-1] shows the Gini coefficient by stage while [Figure IV-2] and [Figure IV-3] presents the fall rate of the (marginal) Gini coefficient by corresponding item and the fall rate of the Gini coefficient per unit based on an assumption that an average contribution or benefit of one million KRW per household for each item constitutes one unit, respectively. In calculating the fall rate of the Gini coefficient by stage, the estimation value varies depending on which Gini coefficient was chosen as the denominator for the calculation of change rate. To maintain the objectivity of comparison, we chose the private income Gini coefficient, which serves as a starting point for income redistribution in the public sector, as the denominator and the extent of a fall in the Gini coefficient by stage as the numerator. Then, we converted them into a percentage and measured the value as the income redistribution effect of each item, such as tax burden and fiscal expenditure benefits by stage.

36 The Effects of Reform in the National Health Insurance Contribution Scheme: On Equity in Health Insurance Contribution Burden 34 [Figure IV-1] Gini Coefficient by Income Stage(Estimation Results of Raw Data from the HIES) 년 Market Income 시장소득 Private Income 민간소득 Public Pension + 공적연금 Gross Income 총소득 Earned Global Income Tax - 근로. 종합소득세 Interest Dividend Income Tax - 이자. 배당소득세 Property Tax - 재산세 Public pension Contribution - 공적연금기여금 Health Insurance Contribution - 건강보험료 Disposable Income 가처분소득 Value Added Tax - 부가가치세 Individual Consumption Tax - 개별소비세 Liquor Tax - 주세 Traffic Energy Environmental Tax - 교통. 에너지. 환경세 Post-Tax Income 세후소득 Basic Living Security (In-Kind) + 기초생보 ( 현물 ) Health Insurance Benefit + 건강보험수혜 Education + 교육 Childcare + 보육 Final Income 최종소득 년 Market Income 시장소득 Private Income 민간소득 Public Pension + 공적연금 Gross Income 총소득 Earned Global Income Tax - 근로. 종합소득세 Interest Dividend Income Tax - 이자. 배당소득세 Property Tax - 재산세 Public pension Contribution - 공적연금기여금 Health Insurance Contribution - 건강보험료 Disposable Income 가처분소득 Value Added Tax - 부가가치세 Individual Consumption Tax - 개별소비세 Liquor Tax - 주세 Traffic Energy Environmental Tax - 교통. 에너지. 환경세 Post-Tax Income 세후소득 Basic Living Security (In-Kind) + 기초생보 ( 현물 ) Health Insurance Benefit + 건강보험수혜 Education + 교육 Childcare + 보육 Final Income 최종소득 Private Income Public Pension Gross Income Earned Global Income Tax Interest Dividend Income Tax Property Tax Public pension Contribution Health Insurance Contribution Disposable Income Value Added Tax Individual Consumption Tax Liquor Tax Traffic Energy Environmental Tax Post-Tax Income Basic Living Security (In-Kind) Health Insurance Benefit Education Childcare Final Income Source: Calculated by the authors based on the HIES

37 Effect of Reform in the National Health Insurance Contribution Scheme: Cross-Sectional Analysis 35 As shown in [Figure IV-1], the market income Gini coefficient as of 2006 was and the final income Gini coefficient was , a decrease by 17.8%. In relation to the decrease, it is estimated that the fall rate of the Gini coefficient of private transfer income and the fall rate of the Gini coefficient due to the rest of the public sector made up 5.6% points and 12.2% points, respectively. In 2010, the Gini coefficients of market income and final income posted and , respectively a drop by 17%. The income redistribution effect of the private and public sectors are estimated to have been 3.6% points and 13.5% points, respectively. While there was a decrease in, income redistribution effects through private transfer income, a slight increase was observed in income redistribution through the public sector (income tax, property tax and various social security contributions, etc.) and fiscal expenditure (expenditure on education and social affairs) increased slightly. In particular, there was an increase in the gross income redistribution effect because the latter was larger than the former. In 2014, the Gini coefficients of market income and final income went down to and , respectively, an 18.4% decline. The income redistribution effects of private transfer income and the public sector are estimated to have been 3.4% points and 15.0% points, respectively. The income redistribution effect through private transfer income is estimated to have decreased while income redistribution through the public sector is estimated to have increased following the period between 2006 and [Figure IV-2] shows the estimation and comparison of income redistribution effects of individual items related to private transfer income and taxation and fiscal expenditure benefits in the public. First, as presented above, the income redistribution effect through private transfer income continuously declined from 5.59% (2006) through 3.59% (2010) to 3.37% (2014). On the contrary, the income redistribution effect through public pension of cash transfer income continuously increased from 1.15% through 3.03% to 4.53%. Although it is not a precise expression, the rapid increase in the income redistribution effect of public pension is estimated to have canceled out the income redistribution effect of private transfer income. As shown in <Table IV-2>, a 27.7% increase is observed in private transfer income and other social security benefits. During

38 The Effects of Reform in the National Health Insurance Contribution Scheme: On Equity in Health Insurance Contribution Burden 36 the same period, public pension saw an increase by 167%. To cite an empirical verification based on regression analysis in Sung Myung-jae and Park Gi-baek (2009), it could be interpreted, with a caution, that there was a crowding-out effect where the increase in public transfer income reduced the increase effect of private transfer income to a certain degree. From a time-series perspective, this implies a certain type of crowding-out effect where a more increase in the income redistribution effect of public pension was accompanied by a more decline in the income redistribution of private transfer income. 5) Meanwhile, it is found there was no negative correlation between the income redistribution effect of other social security benefits among pubic transfer income (cash) and that of private transfer income. 6) It can be interpreted that as the number of elderly household rapidly increased in the low-income deciles (mainly the income deciles 1 through 3) due to population aging, both the rate of the households that received public pension benefits and the average amount of public pension benefits increased at the same time and private transfer income was reduced due to the crowding-out effect. In the case of other social security contributions like unemployment benefits, we estimate that they are mainly attributable to a decline in other social security benefits, such as unemployment benefits, due to a decreasing share of the population in an economically active age and an increasing share of elderly/retirement households among the low-income deciles. 5) Originally, crowding-out effect means that there is a negative correlation between public transfer income and private transfer income. It requires caution to stretch the meaning by interpreting that there is always a negative correlation between the income redistribution effects of the two as well. 6) A slight reduction was observed in the income redistribution effect of other social security benefits including unemployment benefits and national basic living security cash: from 2.05% (2006) through 1.70% (2010) to 1.55% (2014). Due to the negative correlation between private transfer income and public transfer income, as public transfer income increases, private transfer income or the proportion of private transfer income of the total transfer income decreased. Generally, if the crowding-out effect between the two is significant, it is highly likely that there is also a negative correlation between the income redistribution effects of the two. However, if the crowding-out effect between the two is not sizable, the correlation between the income redistribution effects of the two can be uncertain. Unlike public pension, we also have found the income redistribution effect of other social security benefits shows a positive correlation with private transfer income. The size of expenditure on other social security benefits is greatly influenced by economic fluctuations. Therefore, although other social security benefits might have the crowding-out effect, this could remain unidentified if the impact of economic fluctuations dominates the crowding-out effect.

39 Effect of Reform in the National Health Insurance Contribution Scheme: Cross-Sectional Analysis 37 According to [Figure IV-2], the income redistribution effect of earned global income tax was 3.23%, 2.98% and 3.32% in 2006, 2010 and 2014, respectively, showing fluctuation in the lower 3% range in general. Taxation on income in Korea does not incorporate inflation indexation. Therefore, unless tax laws are revised, a rise in prices leads to an increase in nominal income over time. This eventually results in increased effective tax rates i.e. tax creeping effect. In this case, the income redistribution effect of (earned global) income tax can increase. Nevertheless, the reason that the income redistribution effect of earned global income fluctuates repeatedly in the lower 3% range is assumed to be because the effective tax burden and income redistribution effects tend to maintain a similar level due to intermittent adjustments made to the level of income deduction and that of tax exemption limit. The income redistribution effect of health insurance contribution was 0.14% in 2006, 0.15% in 2010 and 0.39% in 2014, showing a gradual increase as suggested by the very small absolute values. As briefly explained above, because the health insurance contribution scheme has basically a proportional tax rate scheme, it is institutionally difficult to exhibit progressivity. On the contrary, because the lower-income class has a smaller percentage of market income, it can be said that health insurance contribution reduces the relative income distribution gap to a slight degree due to the nature of distribution. In addition, the reduction of market income percentage in the low-income class due to an aging population seems to increase the income redistribution effect of health insurance contribution little by little. The income redistribution effect of national health insurance benefits was 3.40% in 2006, 4.26% in 2010 and 4.67% in 2014, which shows a rapidly rising trend as well as high absolute value. What should be noted is that the income redistribution effect of national health insurance benefits is higher than that of (earned global) income tax and that the gap between the two is rapidly increasing. Generally, it is believed that income redistribution through taxation is mainly achieved through income tax, and even tax and financial experts tend to think so. However, it is notable that health insurance benefits show a much higher income redistribution effect than that of income tax. In terms of life cycle, medical expenses are concentrated in old age. Also, old age can be seen as the age group that belongs to low-income class since

40 The Effects of Reform in the National Health Insurance Contribution Scheme: On Equity in Health Insurance Contribution Burden 38 it overlaps with retirement age. Therefore, it can be said that health insurance benefits faithfully performs the function of income redistribution during a lifetime through consumption smoothing over the life cycle as well as income redistribution between the high-income class and the low-income class. Of course, the expense burden structure of the health insurance system is the pay-as-you-go scheme, instead of a funded system. In this respect, it is more reasonable to regard the result as redistribution due to inter-generational (income) transfers, rather than due to consumption smoothing over the life cycle. That said, we do not find it too invalid to interpret that, functionally speaking, the health insurance system performs the role of redistribution through consumption smoothing. In summary, the income redistribution effect of the health insurance system is slight in the stage of contribution payment, but the effect becomes very strong in the benefit stage.

41 Effect of Reform in the National Health Insurance Contribution Scheme: Cross-Sectional Analysis 39 [Figure IV-2] Change Rate of Gini Coefficient by Income Item(Estimation Results of Raw Data from the HIES) (Unit: %) 년 사적이전소득 (B) +Public Pension(C) + 공적연금 (C) +Other Social Transfer Incomes(D) + 기타사회이전소득 (D) -Earned Global Income Tax(E) - 근로. 종합소득세 (E) -Interest Dividend Income Tax(F) - 이자. 배당소득세 (F) -Property Tax(G) - 재산세 (G) -Public Pension Contribution(H) - 공적연금기여금 (H) -Health Insurance Contribution(I) - 건강보험료 (I) - 기타사회보장기여금 (J) -Other Social Security Contributions(J) -Value Added Tax(K) - 부가가치세 (K) - 개별소비세 (L) -Individual Consumption Tax(L) -Liquor Tax(M) - 주세 (M) - 교통. 에너지. 환경세 -Traffic Energy En vironmental Tax(N) (N) - 담배세 (O) -Tobacco Tax(O) + 기초생보 ( 현물 ) (P) +Basic Living Security (In-Kind)(P) + 건강보험수혜 (Q) +Health Insurance Benefit(Q) + 교육 (R) +Education(R) + 보육 (S) +Childcare(S) + 주거수혜 (T) +Housing Benefit(T) 년 사적이전소득 (B) +Private Transfer Income(B) + 공적연금 (C) +Public Pension(C) + 기타사회이전소득 (D) +Other Social Transfer Incomes(D) - 근로. 종합소득세 (E) -Earned Global Income Tax(E) - 이자. 배당소득세 (F) -Interest Dividend Income Tax(F) - 재산세 (G) -Property Tax(G) - 공적연금기여금 (H) -Public Pension Contribution(H) - 건강보험료 (I) -Health Insurance Contribution(I) - 기타사회보장기여금 -Other Social Security Contributions(J) (J) - 부가가치세 (K) -Value Added Tax(K) - 개별소비세 (L) -Individual Consumption Tax(L) - 주세 (M) -Liquor Tax(M) - 교통. 에너지. 환경세 -Traffic Energy En vironmental Tax(N) (N) - 담배세 (O) -Tobacco Tax(O) + 기초생보 ( 현물 ) (P) +Basic Living Security (In-Kind)(P) + 건강보험수혜 (Q) +Health Insurance Benefit(Q) + 교육 (R) +Education(R) + 보육 (S) +Childcare(S) + 주거수혜 (T) +Housing Benefit(T) 년 사적이전소득 (B) +Private Transfer Income(B) + 공적연금 (C) +Public Pension(C) + 기타사회이전소득 (D) +Other Social Transfer Incomes(D) - 근로. 종합소득세 (E) -Earned Global Income Tax(E) - 이자. 배당소득세 (F) -Interest Dividend Income Tax(F) 재산세 (G) -Property Tax(G) - 공적연금기여금 (H) -Public Pension Contribution(H) - 건강보험료 (I) -Health Insurance Contribution(I) - 기타사회보장기여금 (J) -Other Social Security Contributions(J) 부가가치세 (K) -Value Added Tax(K) - 개별소비세 (L) -Individual Consumption Tax(L) - 주세 (M) -Liquor Tax(M) - 교통. 에너지. 환경세 (N) -Traffic Energy ironmental Tax(N) Env 담배세 (O) -Tobacco Tax(O) + 기초생보 ( 현물 ) (P) +Basic Living Security (In-Kind)(P) + 건강보험수혜 (Q) +Health Insurance Benefit(Q) + 교육 (R) +Education(R) 보육 (S) +Childcare(S) + 주거수혜 (T) +Housing Benefit(T) Note: 1. Income redistribution effect was measured using the fall rate of the Gini coefficient. The Gini coefficient fall rate is calculated by measuring the rate of change in the Gini coefficient of income before/after adding up/subtracting each income item (measured based on a marginal change of the Gini coefficient divided by the private income Gini coefficient) and changing the sign of the measured value. 2. If the Gini coefficient fall rate is a positive number, it means that the income redistribution effect is positive; if it is a negative number, the income redistribution effect is negative (-) because the Gin coefficient increases. Source: Calculated by the authors based on the data from the HIES

42 The Effects of Reform in the National Health Insurance Contribution Scheme: On Equity in Health Insurance Contribution Burden 40 Table IV-2 Distributions of Various (Cash) Transfer Incomes by Income (Unit: 1,000 KRW, %) Private Transfer Income / Average ,809 2,217 2,033 2,091 1,707 1,602 1,636 1,576 1,876 2,413 1, ,982 2,228 2,013 1,786 2,014 1,467 1,764 1,712 2,029 2,437 1, ,848 2,464 2,350 1,854 1,888 2,302 1,828 2,119 1,644 2,468 2, ,453 1,871 2,164 1,680 1,361 1,534 1,799 1,610 2,108 2,126 1, ,208 1,898 1,712 2,009 1,479 1,981 1,425 1,467 2,102 3,214 1, ,600 1,787 1,802 2,175 2,122 2,608 2,331 2,012 1,633 2,733 2, ,480 2,252 1,704 2,135 2,571 1,825 2,165 2,837 3,243 3,654 2, ,434 2,055 2,079 1,635 2,293 2,782 2,547 2,563 3,293 4,075 2, ,386 1,888 1,862 2,147 2,015 3,019 2,714 2,648 2,892 3,644 2, Public Pension / Average ,092 1, , , ,123 1, ,115 1,106 1, , ,181 1, ,064 1,391 1,111 1,306 1,213 1,315 1,080 1,040 1,280 1,091 1, ,237 1,278 1,383 1,215 1,033 1,374 1,262 1,020 1,292 1,368 1, ,354 1,381 1,769 1,356 1,454 1,526 1,268 1, ,742 1, ,453 1,768 1,724 1,747 1,739 1,556 1,553 1,462 1,456 1,352 1, ,050 2,100 2,267 1,877 1,841 2,358 1,433 1,765 1,605 2,043 1, Other Social Security Benefits / Average , ,404 1, Change Rate Private Transfer Income Average Public Pension Other Social Security Benefits Note: 1. Based the results estimated by applying the estimation method employed in Sung (2016) to raw data from the HIES published from 2006 to 2014 by Statistics Korea 2. Based on all households including single-person households Source: Calculated by the authors based on the data from the HIES

43 Effect of Reform in the National Health Insurance Contribution Scheme: Cross-Sectional Analysis 41 [Figure IV-3] Change Rate of Gini Coefficient per One Million KRW of Burden/Benefit per Household by Income Item (Estimation Results of the HIES) (Unit: %p/ 1 mil. KRW per household) 년 사적이전소득 (B) +Private Transfer Income(B) 공적연금 (C) +Public Pension(C) 기타사회이전소득 (D) +Other Social Transfer Incomes(D) 근로. 종합소득세 (E) -Earned Global Income Tax(E) 이자. 배당소득세 (F) -Interest Dividend Income Tax(F) 재산세 (G) -Property Tax(G) 공적연금기여금 (H) -Public Pension Contribution(H) 건강보험료 (I) -Health Insurance Contribution(I) 기타사회보장기여금 (J) -Other Social Security Contributions(J) 부가가치세 (K) -Value Added Tax(K) 개별소비세 (L) -Individual Consumption Tax(L) - 주세 (M) -Liquor Tax(M) - 교통. 에너지. 환경세 -Traffic Energy onmental Tax(N) (N) Envir - 담배세 (O) -Tobacco Tax(O) 기초생보 ( 현물 ) (P) +Basic Living Security (In-Kind)(P) 건강보험수혜 (Q) +Health Insurance Benefit(Q) 교육 (R) +Education(R) 보육 (S) +Childcare(S) + 주거수혜 (T) +Housing Benefit(T) 년 사적이전소득 (B) +Private Transfer Income(B) + 공적연금 (C) +Public Pension(C) + 기타사회이전소득 (D) +Other Social Transfer Incomes(D) - 근로. 종합소득세 (E) -Earned Global Income Tax(E) 이자. 배당소득세 (F) -Interest Dividend Income Tax(F) - 재산세 (G) -Property Tax(G) - 공적연금기여금 (H) -Public Pension Contribution(H) - 건강보험료 (I) -Health Insurance Contribution(I) -Other Social Security Contributions(J) - 기타사회보장기여금 (J) 부가가치세 (K) -Value Added Tax(K) - 개별소비세 (L) -Individual Consumption Tax(L) - 주세 (M) -Liquor Tax(M) -Traffic Energy onmental Tax(N) - 교통. 에너지. 환경세 (N) Envir - 담배세 (O) -Tobacco Tax(O) 기초생보 ( 현물 ) (P) +Basic Living Security (In-Kind)(P) + 건강보험수혜 (Q) +Health Insurance Benefit(Q) + 교육 (R) +Education(R) + 보육 (S) +Childcare(S) + 주거수혜 (T) +Housing Benefit(T) 년 사적이전소득 (B) +Private Transfer Income(B) + 공적연금 (C) +Public Pension(C) + 기타사회이전소득 (D) +Other Social Transfer Incomes(D) - 근로. 종합소득세 (E) -Earned Global Income Tax(E) 이자. 배당소득세 (F) -Interest Dividend Income Tax(F) - 재산세 (G) -Property Tax(G) - 공적연금기여금 (H) -Public Pension Contribution(H) - 건강보험료 (I) -Health Insurance Contribution(I) - 기타사회보장기여금 (J) -Other Social Security Contributions(J) - 부가가치세 (K) -Value Added Tax(K) - 개별소비세 (L) -Individual Consumption Tax(L) - 주세 (M) -Liquor Tax(M) - 교통. 에너지. 환경세 -Traffic Energy onmental Tax(N) (N) Envir - 담배세 (O) -Tobacco Tax(O) 기초생보 ( 현물 ) (P) +Basic Living Security (In-Kind)(P) + 건강보험수혜 (Q) +Health Insurance Benefit(Q) 교육 (R) +Education(R) + 보육 (S) +Childcare(S) + 주거수혜 (T) +Housing Benefit(T) Source: Calculated by the authors based on the data from the HIES

44 The Effects of Reform in the National Health Insurance Contribution Scheme: On Equity in Health Insurance Contribution Burden 42 As shown in [Figure IV-3], if the burden/benefit by item is the same for each with one million KRW, the estimated income redistribution effect of in-kind basic living security benefits (healthcare and education benefits) is highest. Next come housing benefits, health insurance benefits, traffic energy environmental tax, public pension and other social security benefits, followed by private transfer income and earned global income tax. 7) However, in the case of in-kind basic living security benefits, the average burden falls far short of one million KRW in terms of all households as the absolute size of the benefits is small and in relative terms, the benefits are so small that recipients are limited to 1. Therefore, although the income redistribution effect by unit is highest, the figure stands at as small as 0.25%p (2006)-0.75%p (2014) with its size effect taken into consideration. The size of the total income redistribution effect of housing benefit is also merely about 0.06%p because the total benefit size is small although the income redistribution effect by unit is high just like the result of basic livelihood security (In-kind) (Refer to [Figure IV-2]). In the case of health insurance contributions, the income redistribution effect based on one million KRW of the average burden by household is very low with 0.17%p (2010)-0.34%p (2014). Because the health insurance rate of contribution is not assessed based on a single tax rate scheme and, thus, it lacks an institutionally progressive structure, the result suggests that the burden of health insurance contribution has a progressive structure and a positive effect on income redistribution due to the progressivity of market income distribution alone (compared to gross income), On the contrary, in the case of health insurance benefits, the income redistribution effect based on one million KRW of the average benefit by household for all households is quite high: 3.09% (2006); 2.51% (2010); and 2.20% (2014). However, the scale of the income redistribution effect by unit is in decline. From 2006 to 2014, the health insurance benefits for the 7) When based on 2014, the income redistribution effect is high in order of 8.24%p for in-kind basic living security benefits (healthcare and education benefits), 3.92%p, for housing benefit, 2.97%p for traffic energy environmental tax, 2.34%p for public pension, 2.26%p for other social security benefits, 2.20%p for health insurance benefits, 1.87%p for earned global income tax, and 1.39%p for private transfer income.

45 Effect of Reform in the National Health Insurance Contribution Scheme: Cross-Sectional Analysis 43 low-income deciles increase much more rapidly. However, this implies that the marginal income redistribution effect is decreasing while the overall income redistribution effect of health insurance benefits is on a rapid rise in aggregate terms due to continuous population aging. One of the factors leading to this phenomenon seems to lie in the fact that the marginal speed at which elderly households flows into low-income deciles due to the aging population is gradually diminishing. For instance, in the case of Income 1, the share of elderly households exceeds 80% significantly, which means the income redistribution effect is already nearing saturation. Income redistribution is also gradually reaching saturation among elderly households in s 2 and 3, the share of which is rapidly increasing to the mid-60% and the mid-30%, respectively. 3 Expected Effects of Improvement in Insurance Contribution Assessment A. Reform Scenarios The health insurance system has life-cycle characteristics in that contributions are mainly borne by age groups who earn income actively i.e. the young and the middle-aged while benefits are received primarily in the old age are received. The most noticeable mismatch between the main period of contribution payment and the main period of benefit occurs in this case. If the life-cycle characteristics are taken into consideration, the mismatch problem can be eased to a certain degree. Thus, this study also analyzes expected policy effects of a case in which the property-based contribution method is added to reflect the life-cycle effect partially. Of course, regarding a property-based health insurance contribution scheme is rare even in developed countries. Also, given the situation of Korea, it is highly unlikely to adopt property-based standards for insurance contribution assessment; there is little prospect that they would be introduced. However, it is worthwhile to compare the equity effect of the two policy alternatives: i.e.

46 The Effects of Reform in the National Health Insurance Contribution Scheme: On Equity in Health Insurance Contribution Burden 44 a single income-based scheme and an incorporation of the life-cycle effect (at least partially). Therefore, this section, regardless of feasibility, analyzes and discusses the equity effect of reform in the health insurance contribution scheme including the case of a partial incorporation of a property-based assessment scheme to improve the readers understanding by broadening the horizon of discussion on the subject matter. The comparison and analysis of the equity effect are conducted based on the following three standards: a case where contribution is assessed in proportion with all global incomes in principle (Standard A); a case where the lower limit of contribution is set when the income does not reach a certain level (Standard B); and a case where a combination of income-base and property-based assessment schemes is introduced with the respective proportion being 80% of the amount of contribution assessed based on Standard A and 20% of the amount of contribution assessed based on property (Standard C). The policy effect analysis based on the below three standards will be equally applied in this chapter and in Chapter V (panel analysis). Standard A: Earned income, business income, financial income (when it is more than 20 million KRW), annuity income (when it is more than 12 million KRW), other incomes (when they are more than 3 million KRW) 8) - However, when the income is more than the standard amount, contribution is assessed based on the whole amount, not just on the excess. Standard B: Standard A + Minimum insurance rate applied Minimum insurance rate: applied when it is annual 3.36 million KRW or less (0.28 million KRW 12 months) Minimum contribution: 0.28 million KRW 6.12% 12 months = 17,136 KRW 12 months = 205,632 KRW (Payment by the employee insured: 8,568 KRW per month, 102,816 KRW per year) 8) It should be noted that other incomes are excluded from analysis since the HIES by Statistics Korea does not include the income item set forth as other incomes under the Income Tax Act.

47 Effect of Reform in the National Health Insurance Contribution Scheme: Cross-Sectional Analysis 45 Standard C: Health insurance contribution rate is assessed based on both income and property However, income-based : property-based = 80% : 20% Income-based contribution: 80% of the health insurance contribution assessed based on Standard A (i.e. income-based assessment) Property-based contribution: property-based contribution rate should be 1/4 of income-based contribution on average. Scope of property: financial assets + real estate assets (the rate of market-value adjustment of assessment standard is assumed to be 80%.) Microsimulation was conducted to estimate and analyze the expected effect of contribution scheme reform scenarios introduced above. The raw data (based on annualized data) included in the HIES 2014 by Statistics Korea was used for the simulation. This study analyzes the expected effect of a reform in the contribution assessment scheme that covers a more comprehensive scope of income reported in the HIES, beyond the narrowly defined scope of taxable income, which is currently used as the assessment standard. Therefore, the measured effect of change in the amount of contribution is much higher in this study, compared to the existing contribution scheme, which is based on the scope of taxable income that the health insurance contribution assessment authority (i.e. the NHIS) or the taxing authority can track down in practice. In other words, since the HIES contains unreported business incomes that are not off the radar, it should be noted that reform-driven changes in contribution burden are the maximum effect that is expected when all incomes are subject to contribution assessment i.e. the potential maximum effect, rather than the actual effect. B. Distribution of Estimated Property in the HIES The HIES does not provide property-related information. However, when the property-based standard is added to the health insurance contribution scheme as in Standard C, information on property distribution by household is needed.

48 The Effects of Reform in the National Health Insurance Contribution Scheme: On Equity in Health Insurance Contribution Burden 46 In this section, we came up with a total of 100 cells by combining current income-based 10 deciles included in the 7th-year NaSTaB data (those included in the 2013 data set) and 10 age groups classified according to the householder s age with a 5-year interval (a total of 10 age groups divided with a 5-year interval from less than 25 to 65 years old or older). Then, the mean and standard deviation of property were estimated for each cell. Under the assumption that property is normally distributed in each cell and the same applies to the data from the HIES, we estimated the distribution of property and used it for analysis. <Table IV-3> presents the distribution of the average value of property for each cell according to the NaSTaB data and the distribution of estimated property in the HIES based on the former. According to the NaSTaB data (as of 2013), the extent of relative disparity is significant by income decile in the distribution of financial assets. Yet, the disparity is slightly smaller in the distribution of real estate assets in relative terms. As for financial assets, the decile ratio is similar to that of market income in <Table IV-3>. As shown in <Table IV-3>, the ratio of 10 to 1 is much less than 10, which implies that disparity in the distribution of real estate assets by income decile is less than relative income disparity. This is predictable in the sense that the structure of property held by households in Korea is mainly composed of real estate assets and that a considerable number of low-income, post-retirement elderly households hold property like housing based on the life-cycle hypothesis.

49 Effect of Reform in the National Health Insurance Contribution Scheme: Cross-Sectional Analysis 47 Table IV-3 Joint Distribution of Average Property by Income and Age (Based on the 2013 NaSTaB Data) (Unit: 1,000 KRW) Financial Assets Average ~24 1, ,501 7,300 9, ,000 18, ,466 25~29. 9,995 5,666 9,664 10,552 17,182 44,329 14,401 24,200 43,619 14,928 30~34 47,531 1,946 10,114 4,837 18,374 16,086 13,982 36,546 43,167 45,174 20,082 35~39 1,626 4,161 15,096 19,085 10,723 25,945 35,717 43,527 43, ,509 41,484 40~44 2,451 3,251 8,905 14,704 14,529 22,284 22,865 72,751 42,239 52,398 32,907 45~49 5,316 5,791 8,433 7,601 12,962 18,449 34,047 40,395 66,229 79,347 33,088 50~54 6,792 8,690 8,589 9,787 23,112 17,713 14,438 42,562 38, ,300 43,795 55~59 2,354 7,507 18,404 32,594 32,758 24,899 42,736 35,409 66,844 84,634 38,483 60~64 5,982 16,858 21,532 17,772 47,511 36,429 51,326 78,629 43, ,557 35,864 65~ 5,707 10,491 18,421 48,240 35,795 50,660 84,801 67,631 49,739 91,875 19,697 Average 5,601 9,707 13,451 18,743 21,191 23,492 33,338 49,963 49,388 90,995 29,897 Real Estate Assets Average ~ ~29. 4, ,116 16,069 7,644 53, , ,015 16,547 30~34 61, ,466 11,557 16,122 79, , , ,535 40,989 63,311 35~39 5, ,271 53,234 53,119 69, ,938 66, , , ,168 40~44 24,814 46,183 30,768 36, , , , , , , ,225 45~49 104,033 37,246 25,573 82,831 37,295 97, , , , , ,451 50~54 82, ,511 43,352 47,658 94, , , , , , ,048 55~59 3,368 30,016 84, , , , , , , , ,568 60~64 23, , , , , , , , , , ,400 65~ 72, , , , , , , , ,440 1,035, ,770 Average 65,962 95, ,834 92, , , , , , , ,064 Note: Based on estimations made by the authors Source: Calculated by the authors based on the 2013 NaSTaB data C. Incidence by Income <Table IV-4> shows the results of the health insurance contribution distribution by income decile estimated through the simulation of reform scenarios (Refer to Subsection A above.). Here, it needs to be noted that the

50 The Effects of Reform in the National Health Insurance Contribution Scheme: On Equity in Health Insurance Contribution Burden 48 income information contained in the HIES was investigated more comprehensively than incomes reported to the taxing authorities during tax return. As for the reformed standard for contribution assessment, we applied the consolidation of the current scheme into a single income-based one (Standards A and B) and the combination of income-based and property-based schemes (Standard C). The key of all three standards is to expand significantly the scope of income subject to contribution assessment. When the rate of insurance contribution is maintained at the current level yet the scope of income that is subject to contribution assessment is expanded and includes unreported incomes exceeding the current size of taxable income tracked down by the taxing authorities, the amount of contribution assessed based on reformed scenarios is expected to be the maximum possible level and, thus, increase greatly, compared to the current level. Table IV-4 Post-Reform Distribution of Health Insurance Contribution Burden by Income (Unit: 1,000 KRW, times, %) Average 10/ 1 Private Income 6,414 13,525 20,939 28,187 34,951 40,516 49,349 58,379 71, ,336 43, Private Transfer 1,386 Income 1,888 1,862 2,147 2,015 3,019 2,714 2,648 2,892 3,644 2, Public Pension 2,050 2,100 2,267 1,877 1,841 2,358 1,433 1,765 1,605 2,043 1, Other Social Security Benefits Public Transfer Income 2,644 2,970 2,959 2,533 2,791 3,035 2,046 2,507 2,219 2,499 2, Gross Transfer Income 4,042 4,858 4,830 4,679 4,801 6,041 4,733 5,129 5,087 6,119 5, Gross Income 10,456 18,383 25,769 32,866 39,752 46,557 54,083 63,508 77, ,455 48, Income Tax ,007 1,447 2,295 3,499 8,304 1, Property Tax Direct Tax ,216 1,651 2,608 3,757 8,709 2, Public Pension Contribution ,192 1,440 1,747 2,219 3,070 1, Health Insurance Contribution ,000 1,122 1,298 1,578 1,954 2,724 1,

51 Effect of Reform in the National Health Insurance Contribution Scheme: Cross-Sectional Analysis 49 Table IV-4 Continue Average 10/ Standard A ,158 1,391 1,687 2,007 2,446 3,671 1, Standard B ,167 1,400 1,694 2,015 2,453 3,677 1, Standard C ,213 1,400 1,636 1,893 2,243 3,223 1, Other Social Security Contributions Total Social Security Contributions ,386 2,057 2,425 2,876 3,499 4,384 6,045 2, Total Direct Tax ,295 1,948 2,940 3,641 4,527 6,107 8,141 14,754 4, Disposable Income 10,177 17,677 24,473 30,918 36,812 42,916 49,556 57,400 68,946 97,701 43, Value Added Tax ,084 1,377 1,701 1,959 2,238 2,554 2,844 3,868 1, Individual Consumption Tax TrafficEnergyEn vironmental Tax ,007 1,224 1, Liquor Tax Tobacco Tax Total Excise Tax 607 1,172 1,906 2,396 2,965 3,381 3,880 4,357 4,929 6,595 3, Post-Tax Income 9,570 16,505 22,567 28,522 33,847 39,535 45,676 53,043 64,017 91,106 40, In-Kind Basic Living Security Benefits Health Insurance 2,420 Benefits 2,275 2,167 2,059 1,999 2,078 2,008 2,067 2,063 2,091 2, Education ,096 2,083 2,555 3,191 3,531 4,236 4,610 5,578 2, Childcare Housing In-Kind Benefits 3,418 2,887 3,475 4,544 5,076 5,757 6,014 6,750 7,139 7,958 5, Final Income 12,988 19,392 26,042 33,066 38,923 45,292 51,690 59,793 71,157 99,064 45, EarnedBusiness LeaseAnnuity Income 5,635 11,850 19,349 26,093 33,746 40,048 47,897 56,654 69, ,644 41,

52 The Effects of Reform in the National Health Insurance Contribution Scheme: On Equity in Health Insurance Contribution Burden 50 Table IV-4 Continue Rate of Change from Current Level Standard A Standard B Standard C Ratio to Gross Income Current Level Standard A Standard B Standard C Ratio to Taxable Income Current Level Standard A Standard B Standard C Average 10/ 1 Average 10/ 1 Average 10/ 1 Note: 1. Except health insurance benefits and health insurance contribution, the rest of the analysis above cites <Appendix Table 2> in Sung Myung-jae (2016). For the distribution of health insurance benefits, the parameters used for the year 2013 were revised and estimated using performance values from the 2014 data because there was no information available on average health insurance benefits per capita by gender and age when Sung (2016) performed the estimation. In addition, the distribution according to the reform of contribution assessment is based on estimations made by the authors using the raw data of the HIES by Statistics Korea. Source: Calculated by the authors based on the HIES data The amount of health insurance contribution according to each standard (1.433 million KRW per household [Standards A and C] to million KRW [Standard B]) is estimated to greatly increase from the current level (1.136 million KRW). The increase rate of contribution is 26.1% in Standards A and C and 26.9% in Standard B, showing that the latter is a little higher than the former. Standard A and Standard B have a quite similar character because both of them are income-based with the only difference being whether or not the lower limit is set on insurance contribution. In other words, Standard B imposes the minimum amount of contribution on those who earn the lowest income and

53 Effect of Reform in the National Health Insurance Contribution Scheme: Cross-Sectional Analysis 51 Standard A does not while both of them assess contribution based on income. It is a natural outcome that Standard B with the lower limit on contribution shows a slightly higher level of the total contribution burden than Standard A across all income deciles. In the case of a shift toward a single income base, it is estimated that a change in the level of contribution burden becomes quite asymmetrical by income decile. In both Standards A and B, the average contribution burden on Income 1, which refers to the lowest 10%, decreases by 33,000 KRW (Standard A) or 20,000 KRW (Standard B) while the contribution level increases in the remaining income deciles. In particular, the higher the income decile is, the level of absolute burden goes higher (Refer to [Figure IV-4].) As for s 3 to 10, the increase rate of contribution burden fluctuates in the range between 16% and 35%, but the deviations of the contribution increase rate are not significant by income decile (Refer to [Figure IV-5].) The average health insurance contribution declines only in 1, unlike other income deciles. Since the share of income is smaller than that of property in 1, the amount of contribution gets smaller in that group if the contribution scheme changes to a single, income basis from the current system in which contribution assessment for the self-employed insured is based on other components than income (e.g. property, automobile ownership, etc.) Recently, elderly households makes up a near majority of the low-income class especially 1. From the life-cycle perspective, elderly households belong to the retirement period. Also, the level of average property accumulation is higher than other age groups due to the characteristics of the old age. As shown in [Figure IV-6], the fact that the share of property is higher than that of income in 1 proves this point. In the case of the self-employed insured, this leads to a relatively larger amount of contribution compared to income level since the level of property they hold is also subject to contribution assessment under the current system. Therefore, those in the lowest-income decile should pay a relatively higher level of health insurance contribution according to the value of property they have under the current system because the share of property is relatively larger among those with the lowest income. If the insurance contribution scheme is converted to a single income basis, the absolute burden will be reduced.

54 The Effects of Reform in the National Health Insurance Contribution Scheme: On Equity in Health Insurance Contribution Burden 52 As explained briefly above, the difference between Standards A and B consists in whether or not the lower limit is applied to insurance contribution. Generally, those who earn income less than the minimum standard are more likely to be the low-income class. Also, the number will decrease as the level of income increases. This expectation holds true in the result of our estimation as well. The differential in insurance contribution between Standards A and B is an average of 13,000 KRW in 1, followed by 8,000 KRW in 3, 9,000 KRW in 5, 8,000 KRW in 8 and 6,000 KRW in 10, which shows that the differential decreases more gradually in higher-income deciles. In Standard C, the average value of the total contribution burden is the same as Standard A except that 20/100 of the average amount of the total contribution is subject to a single contribution rate in proportion to the total value of property (80% for real estate assets). If the rate of contribution to the value of property is inversely calculated for the average burden per household of property-based contribution to be 1/4 of income-based contribution based on the distribution of estimated property of the HIES estimated in Subsection B above, the value is estimated to be 0.164%. If Standard C is applied (with the rate of contribution to the value of property being 0.614%), the average amount of the total contribution is the same as Standard A, but it is estimated that a change in the amount of burden exhibits a very different pattern from Standard A. In Standards A and B where contribution is assessed solely based on income, the average amount of contribution burden decreases in 1, the figure increases greatly in Standard C. On the contrary, the rate of change in health insurance contribution is estimated to decrease in higher-income deciles, compared to Standards A and B. As such, the decile ratio (=10 decile 1 decile), which measures the progressivity of health insurance contribution, is estimated to decrease 8.5 times much less than Standard A (31.4 times) and Standard B (28.3 times) not to mention the current level (18.2 times). Since this is smaller than the decile ratio of market income or gross income, the income redistribution effect of Standard C is estimated to have a negative value. Based on our cross-sectional analysis, Standard C s effect on income redistribution is expected to have a negative value. This phenomenon seems to result from a discrepancy between

55 Effect of Reform in the National Health Insurance Contribution Scheme: Cross-Sectional Analysis 53 the distribution of income and that of property. A negative income redistributive effect of a property basis incorporated into the contribution scheme is largely due to the fact that the analysis focuses on a short-term structure of income redistribution structure from a cross-sectional perspective while ignoring the life-cycle effect. Thus, caution should be paid that the effect of the incorporation of a property-based assessment on the redistribution of lifetime income does not always have a negative value when reevaluated based on lifetime income in accordance with the life-cycle hypothesis. [Figure IV-4] Effect of Health Insurance Contribution Reform on Contribution Amount by Income (Change in Amount) (Unit: 1,000 KRW) 1,200 1, (Unit: 1,000 KRW) Standard A Standard B Standard C Average Source: Based on estimations made by the authors through an analysis of the raw data in the HIES 2014 by Statistics Korea

56 The Effects of Reform in the National Health Insurance Contribution Scheme: On Equity in Health Insurance Contribution Burden 54 [Figure IV-5] Effect of Health Insurance Contribution Reform on Contribution Amount by Income (Change Rate) (Unit: %) Standard A Standard B Standard C Average Source: Based on estimations made by the authors through an analysis of the raw data in the HIES 2014 by Statistics Korea [Figure IV-6] Distributions of Income Share and Gross Property Share by Income (Based on Estimation Results of the 2009 NaSTaB Data) (Unit: %, times) Income Share Gross Pr operty Share(%) 4.13 Income Share Gross Property Share Relative Ratio of Gross Property Share to Income Share Gross Property Share/Income Share(times) Dec ile Source: Illustrated based on the contents of <Table III-2> in Sung Myung-jae (2011)

57 Effect of Reform in the National Health Insurance Contribution Scheme: Cross-Sectional Analysis 55 D. Income Redistributive Effect Compared to their income level, some of the self-employed insured should pay a relatively higher contribution despite a low level of income under the current contribution scheme because of a disparate assessment standard including property (e.g. automobile, etc.). This adversely affects the progressivity of the distribution structure of health insurance contribution burden. First, if the health insurance contribution scheme is consolidated to a single, income-based assessment as in Standards A and B, the progressivity of health insurance improves due to the asymmetrical distribution of the income that is subject to contribution assessment by gross income decile i.e. the distribution structure where a higher-income decile has a higher percentage of the income subject to assessment out of gross income. Additionally, since the scope of income that is subject to contribution becomes more comprehensive when the health insurance contribution scheme is reformed, health insurance contribution also increases by 26 to 27% from the current level. In other words, the reform will result in an increase in both the progressivity of health insurance contribution burden and the size of the total health insurance. If other conditions are the same, the income redistributive effect gets larger as progressivity in the structure of contribution burden increases. Meanwhile, an increase in the size of the total contribution burden, the tax burden structure of which is progressive vis-à-vis income, also means an increase in the income redistribution effect of health insurance contribution when other conditions are the same. Therefore, the contribution scheme is converted to an income-based one will lead to an improvement in the income redistribution effect of health insurance contribution because both the progressivity of the contribution burden structure and the scale of total health insurance contribution burden will increase. [Figure IV-7] presents changes in the Gini coefficient in each major item stage, from the market income stage to the final income stage, as of 2014 under the current health insurance system. In the case of health insurance contribution, the current level of Gini coefficient and the expected levels of the post-reform Gini coefficient are listed side by side. [Figure IV-8] shows the income redistribution effect by item, which was measured based on the relative ratio

58 The Effects of Reform in the National Health Insurance Contribution Scheme: On Equity in Health Insurance Contribution Burden 56 by calculating the ratio of reduction in the Gini coefficient to the Gini coefficient of the previous stage with the private income Gini coefficient as the denominator. Also, it shows the income redistribution effect expressed as the post-reform Gini coefficient reduction rate. As a result, when the contribution scheme is converted into an income-based assessment (Standards A and B), the post-tax income-based Gini coefficient (income after health insurance contribution) was estimated to be slightly smaller than in the present. This implies, as examined above, that when the contribution scheme is reformed, both enhanced progressivity and an increase in the size of the total burden reduced disparity in the relative income after health insurance contribution, compared to the current system, by increasing the income redistribution effect simultaneously. As such, the Gini coefficient reduction rate was estimated to be 0.71%p (Standard A) or 0.69%p (Standard B), which is a slightly rise from the current 0.39%p. In the case of Standard C, in which income-base and property-base contributions are combined, the income redistribution effect showed a negative value. As shown in [Figure IV-7], the Gini coefficient of income after the deduction of health insurance contribution was , which was up Gini points from the current , in Standard C. As a result, the income redistribution effect, which was measured by the Gini coefficient fall rate, was estimated at 0.02% unlike the current 0.39%, showing a negative effect on income redistribution. (Refer to [Figure IV-8]).

59 Effect of Reform in the National Health Insurance Contribution Scheme: Cross-Sectional Analysis 57 [Figure IV-7] Gini Coefficient by Stage: Effect of Reform in Health Insurance Contribution Scheme Note: 1. Based on estimations made by the authors through an analysis of the raw data in the HIES 2014 by Statistics Korea 2. PY, GY, DY, PTY and FY refer to private income, gross income, disposable income, post-tax income and final income, respectively. Source: Calculated by the authors based on the HIES data

60 The Effects of Reform in the National Health Insurance Contribution Scheme: On Equity in Health Insurance Contribution Burden 58 [Figure IV-8] Income Redistributive Effect of Reform in Health Insurance Contribution Scheme Note: 1. Based on estimations made by the authors through an analysis of the raw data in the HIES 2014 by Statistics Korea 2. The income redistributive effect was measured by calculating the Gini coefficient fall rate compared to the Gini coefficient of the previous stage with the private income Gini coefficient as the denominator Source: Calculated by the authors based on the HIES data

61 Ⅴ Effect of Reform in National Health Insurance Contribution Scheme: Longitudinal Analysis This chapter examines the effect of the reform in the health insurance contribution scheme on redistribution and fiscal revenue from the health insurance using the NaSTaB data. Using panel data investigated from 2008 to date, the NaSTaB contains detailed information on various variables related to individual income and household expenditure. In particular, it serves as a very useful source for understanding the contribution revenue size of the national health insurance and the contribution burden on a national scale because it includes contribution payment by individual and the health insurance subscription type and whether or not there are dependents. In the NaSTaB, the payment amount of the health insurance contribution is investigated by individual. Since 2009, individual contributions have been investigated separately on the employee insured and on the self-employed insured, which makes it possible to analyze the national health insurance subscription type by individual or household. Data on insurance contribution is very rare in the domestically investigated survey data, and it is hard to obtain even in the sample cohort data provided by the NHIS. Also, since the NaSTaB is panel-type data, changes in employment status can be identified by individual or household. Thanks to the merits of the NaSTaB data including variables related to insurance contribution, a cross-sectional analysis using the data is expected to provide an opportunity to compare the effects of health insurance contribution on redistribution and fiscal revenue, which are difficult to be confirmed by the HIES alone. This would also contribute to confirming the robustness of the

62 The Effects of Reform in the National Health Insurance Contribution Scheme: On Equity in Health Insurance Contribution Burden 60 analysis results. Furthermore, we expect that a longitudinal analysis using the characteristics of panel data would be differentiated from the analysis presented in Chapter IV in that the former can reveal the long-term effect of health insurance contribution on redistribution. 1 Current State of Contribution Burden on Households Data from the 1st to 7th NaSTaB ( ) was used to examine the contribution burden of the national health insurance on households. In the NaSTaB data, income and health insurance contribution are investigated by individual and other survey questions related to expenditure are investigated by household. Since insurance contribution is paid by household with a distinction made between payers and dependents, this study focuses on the burden of insurance contribution on households. First, to examine the current state of health insurance contribution burden by household, we define health insurance contribution as the ratio of contribution to income. Therefore, the contribution burden rate of a household is refers to the amount of contribution payment compared to household income. To calculate this directly, data collected at both individual and household levels was used in measuring contribution burden per household. The individual-level data contains detailed information on individual incomes, such as earned income and business income, and the payment amount of individual health insurance contribution is listed by employment status. For the calculation of contribution burden, individual gross income was calculated by adding up all kinds of income data covered in the NaSTaB by individual. As for health insurance contribution, the amount of contribution payment investigated in the data set was used. What should be noted here is that although the half of the contribution paid by employers is ultimately incident on individuals, leading to a reduction in real wages, contribution burden is calculated in this study based on the nominal amount of contribution paid by individual or household. <Table V-1> presents health insurance contribution burden by year using household data. The household data was created by adding up individual incomes

63 Effect of Reform in National Health Insurance Contribution Scheme: Longitudinal Analysis 61 from the individual-level data and the payment amount of individual insurance contribution per household and linking the sum with the household-level data. Meanwhile, the health insurance subscription type can be identified only in the individual-level data. So, we used this in classifying data into households with the employee insured, households with the self-employed insured and households with both the employee insured and the self-employed insured. Since the distinction by subscription type started from 2009, the current state can be identified by employment status from For this reason, we generated the household-level data using the 2nd (2009) to 7th (2014) NaSTaB data. As for household income, health insurance contribution by household, and contribution burden rate by household, all of them are weighted averages. First, household income has steadily been increasing since 2009, and it reaches about 40 million KRW as of In 2014, household income posts million KRW in the case of the employee insured and million KRW in the case of the self-employed insured. Household income is estimated at a relatively low level of about million KRW among households with both the employee insured and the self-employed insured or those that have changed the subscription type. Health insurance contribution increased from a yearly average of 0.83 million KRW in 2009 to a yearly average of about 1.17 million KRW in In the households with the employee insured, the average annual amount of contribution payment was either similar to or slightly higher than the average annual amount paid by the households with the self-employed insured. However, for the households with the self-employed, the rate of contribution burden was two times as high as the contribution burden rate for the households with the employee insured.

64 The Effects of Reform in the National Health Insurance Contribution Scheme: On Equity in Health Insurance Contribution Burden 62 Table V-1 Current State of Household Income and Health Insurance Contribution Burden (Unit: 10,000 KRW, Ratio) 2009~ Household Income , (46.75) Employee insured household Self-employed insured household Employee insured household + Self-employed insured household Health insurance contribution by household Employee insured household Self-employed insured household Employee insured household + Self-employed insured household Contribution compared to household income Employee insured household Self-employed insured household Employee insured household + Self-employed insured household , (59.30) 2, (95.53) 2, (84.22) (1.40) (1.73) (3.03) (2.33) (.001) (.0004) (.004) (.0005) 3, (48.11) 4, (62.17) 3, (116.35) 2, (82.88) (1.44) (1.74) (3.25) (2.83) (.001) (.0005) (0.006) (.0006) 3, (53.81) 4, (68.52) 3, (143.31) 2, (87.29) (1.65) (1.93) (4.26) (2.99) (.001) (.0003) (0.009) (.0006) Note: ( ) shows ratio. Source: Calculated by the authors using the 2 nd to 7 th NaSTaB data 3, (58.10) 4, (83.70) 3, (96.30) 2, (110.93) (1.74) (2.29) (3.28) (3.42) (0.001) (.0002) (0.006) (.0005) 3, (79.62) 5, (84.59) 3, (145.42) 2, (193.75) (1.94) (2.71) (3.65) (2.92) (0.004) (.0003) (0.019) (.0009) 4, (62.76) 5, (89.41) 3, (127.29) 2, (96.09) (2.02) (2.68) (4.14) (3.26) (0.001) (.0002) (0.006) (.0007)

65 Effect of Reform in National Health Insurance Contribution Scheme: Longitudinal Analysis 63 <Table V-2> presents gross household income, health insurance contribution by household, and the contribution burden by household, which were calculated using longitudinal data. The sum of the average household income is the average of yearly household incomes added up per household; the sum of the average household insurance contribution is the average of yearly household insurance contributions added up per household. Household income and health insurance contribution of each year were converted to the level of 2014 using the consumer price index. The sum of the average household income for the six years between 2009 and 2014 was about 200 million KRW, and an average of 5.54 million KRW of health insurance contribution was paid. The contribution burden rate represents the average ratio of the sum of yearly health insurance contributions to the sum of yearly household incomes per household. Average household income and average household contribution are the average values calculated by regarding the data between 2009 and 2014 one pool of cross-sectional data. As such, the contribution burden rate is estimated at , which is slightly low when compared to the cross-sectional data. The bottom of <Table V-2> lists the sum of household incomes and that of household contributions with only those households investigated four times or more in the NaSTaB as subjects. It is found that the sum of average household incomes is higher by 30 million KRW than the value calculated with all households as subjects, and this is because the number of investigations was high (four times or more) for the former. The sum of average household contributions stands at million KRW, which is also higher than the value calculated with all households as subjects. However, the contribution burden rate is slightly lower yet quite similar in the case in which all households are included, compared to those households investigated four times or more in the NaSTaB as subjects. Neither does the average household contribution show much difference between the two.

66 The Effects of Reform in the National Health Insurance Contribution Scheme: On Equity in Health Insurance Contribution Burden 64 Table V-2 Longitudinal Analysis ( ): Gross Household Income, Health Insurance Contribution and Contribution Burden Rate (Unit: 10,000 KRW, Ratio) Sum of Average Household Incomes Sum of Average Household Contributions Contribution Burden Rate Average Household Income Average Household Contribution All Households 20, , Households Investigated 4 Times or More 23, , Note: Converted to the price level of 2014 using each year s consumer price index Source: Calculated by the authors using the 2 nd to 7 th NaSTaB data <Table V-3> shows the result of weighted calculations where average household income, average household insurance contribution and the contribution burden rate are classified into 5 income quintiles. The respective distributions of average income, average insurance contribution, and the contribution burden rate by household are not very different from the distributions of the same variables by individual. Average income and average contribution increase as the income quintile increases, but the contribution burden rate remains fairly consistent across all quintiles. However, the contribution burden rate in Income Quintile 1 is twice that of other income quintiles, showing the possible income regressivity of the national health insurance contribution scheme.

67 Effect of Reform in National Health Insurance Contribution Scheme: Longitudinal Analysis 65 Table V-3 Average Income, Average Insurance Contribution and Contribution Burden to Income Ratio by Income Quintile: by Household (Unit: 10,000 KRW) Income Quintile Average Income Quintile 1 Quintile 2 Quintile 3 Quintile 4 Quintile 5 Average Contribution Contribution Burden Average Income Average Contribution Contribution Burden Average Income Average Contribution Contribution Burden Average Income Average Contribution Contribution Burden Average Income Average Contribution Contribution Burden , , , , , , , , , , , , , , , , , , , , , , , Source: Calculated by the authors using the 2 nd to 7 th NaSTaB data

68 The Effects of Reform in the National Health Insurance Contribution Scheme: On Equity in Health Insurance Contribution Burden 66 [Figure V-1] Household Contribution Burden Ratio by Income Quintile Quintile 분위 1 2Quintile 분위 2 Quintile 3분위 3 4Quintile 분위 4 Quintile 5분위 5 Source: Calculated by the authors using the 2 nd to 7 th NaSTaB data <Table V-4> and <Table V-5> present average income, average insurance contribution and the contribution burden rate by income quintile within households with the employee insured and within households with the self-employed insured, respectively. The average income of households with the employee insured in Quintile 1, which the low-income class, is slightly higher than that of households with the self-employed insured in Quintile 1; on the contrary, the average income of households with the self-employed insured in Quintile 5, which is the high-income class, is higher than that of households with the employee insured in Quintile 5.

69 Effect of Reform in National Health Insurance Contribution Scheme: Longitudinal Analysis 67 Table V-4 Households with the Employee Insured: Average Income, Average Insurance Contribution and Contribution Burden to Income Ratio by Income Quintile (Unit: 10,000 KRW) Income Quintile Average Income Quintile 1 Quintile 2 Quintile 3 Quintile 4 Quintile 5 Average Contribution Contribution Burden Average Income Average Contribution Contribution Burden Average Income Average Contribution Contribution Burden Average Income Average Contribution Contribution Burden Average Income Average Contribution Contribution Burden , , , , , , , , , , , , , , , , , , , , , , , , Source: Calculated by the authors using the 2 nd to 7 th NaSTaB data

70 The Effects of Reform in the National Health Insurance Contribution Scheme: On Equity in Health Insurance Contribution Burden 68 Table V-5 Households with the Self-Employed Insured: Average Income, Average Insurance Contribution and Contribution Burden to Income Ratio by Income Quintile (Unit: 10,000 KRW) Income Quintile Average Income Quintile 1 Quintile 2 Quintile 3 Quintile 4 Quintile 5 Average Contribution Contribution Burden Average Income Average Contribution Contribution Burden Average Income Average Contribution Contribution Burden Average Income Average Contribution Contribution Burden Average Income Average Contribution Contribution Burden , , , , , , , , , , , , , , , , , , , , , , , , Source: Calculated by the authors using the 2 nd to 7 th NaSTaB data

71 Effect of Reform in National Health Insurance Contribution Scheme: Longitudinal Analysis 69 2 Change in Insurance Contribution Burden by Scenario A. Insurance Contribution Burden in Standard A The income that is subject to insurance contribution in Standard A was calculated by adding up earned income, business income, financial income, annuity income and other incomes included in the individual-level data. In the case of households, the income was calculated by reorganizing the aforementioned individual-level income data by household and adding up the values. Insurance contribution was calculated by multiplying the household-level income mentioned above by half of each year s contribution rate of the national health insurance. <Table V-6> presents health insurance contribution and the contribution burden rate in Standard A, which were calculated using a series of data mentioned above. Health insurance contribution by household in Standard A shows a slight increase across all years under consideration from the current level. A slight increase is observed in the amount of contribution among households with the employee insured and among those with both the employee insured and the self-employed insured; the amount of contribution declines among households with the self-employed insured. However, the ratio of contribution to household income i.e. the contribution burden rate is lowered in Standard A. This can be because the scope of income covered in Standard A is larger than the current scope. The rate of contribution burden is similar or increases among households with the employee insured, but the rate decreases significantly among households with the self-employed insured when Standard A applies. On the contrary, the contribution burden rate rises greatly among households with both the employee insured and the self-employed insured.

72 The Effects of Reform in the National Health Insurance Contribution Scheme: On Equity in Health Insurance Contribution Burden 70 Table V-6 Health Insurance Contribution and Contribution Burden Ratio by Household in Standard A (Unit: 10,000 KRW, Ratio) Health Insurance Contribution by Household (Current) Health Insurance Contribution by Household (Standard A) Households with the Employee Insured (Current) Households with the Employee Insured (Standard A) Households with the Self-Employed Insured (Current) Households with the Self-Employed Insured (Standard A) Households with Both the employee insured and the self-employed insured (Current) Households with Both the employee insured and the self-employed insured (Standard A) Contribution to Household Income Ratio (Current) Contribution to Household Income Ratio (Standard A) Households with the Employee Insured (Current) Households with the Employee Insured (Standard A) (1.40) (1.14) (1.73) (1.46) (3.03) (2.37) (2.33) (2.00) (.001) (1.44) (1.24) (1.74) (1.59) (3.25) (3.09) (2.83) (2.07) (.001) (1.65) (1.46) (1.93) (1.84) (4.26) (4.02) (2.99) (2.39) (.001) (1.74) (1.62) (2.29) (2.31) (3.28) (2.67) (3.42) (3.17) (0.001) (1.94) (2.18) (2.71) (2.37) (3.65) (3.93) (2.92) (5.28) (0.004) (2.02) (1.79) (2.68) (2.58) (4.14) (3.59) (3.26) (2.67) (0.001) (.0004) (.0005) (.0003) (.0002) (.0003) (.0002)

73 Effect of Reform in National Health Insurance Contribution Scheme: Longitudinal Analysis 71 Table V-6 Continue Households with the Self-Employed Insured (Current) Households with the Self-Employed Insured (Standard A) Households with Both the Employee Insured and the Self-Employed Insured (Current) Households with Both the Employee Insured and the Self-Employed Insured (Standard A) (.004) (0.006) (0.009) (0.006) (0.019) (0.006) (.0005) (.0006) (.0006) Source: Calculated by the authors using the 2 nd to 7 th NaSTaB data (.0005) (.0009) (.0007) <Table V-7> presents the sum of incomes that are subject to contribution assessment, the sum of contributions, and the ratio of contribution burden to household income calculated based on longitudinal data in Standard A. Each year s household incomes that are subject to contribution assessment and household contributions were converted to the price level of 2014 using the consumer price index. When the contribution scheme is changed to Standard A, the average of the sum of all incomes that are subject to contribution assessment for the 6 years between 2009 and 2014 stands at about 140 million KRW and average contribution is 5.85 million KRW. A slight increase is observed in the amount of insurance contribution in Standard A, compared to the current level. The contribution burden ratio refers to the average ratio of the sum of each year s household contributions to the sum of the same year s household incomes. Average household income and average household contribution refer to average values calculated by regarding the data between 2009 and 2014 as one pool of cross-sectional data. The contribution burden rate in Standard A calculated as above is , which is a slight increase compared to the current level

74 The Effects of Reform in the National Health Insurance Contribution Scheme: On Equity in Health Insurance Contribution Burden 72 (0.0263). The bottom of <Table V-7> shows the sum of household incomes that are subject to contribution assessment, the sum of household contributions, etc. with only the households investigated four times or more in the NaSTaB as subjects; the results indicate that there is not so much change in the contribution burden rate. Table V-7 Longitudinal Analysis of Standard A: Incomes Subject to Contribution Assessment, Household Insurance Contribution and Contribution Burden Ratio Average sum of incomes subject to contribution assessment Average sum of household contributions Contribution burden ratio (Unit: 10,000 KRW, Ratio) Average of perhousehold incomes subject to contribution assessment Average household contribution All households Households investigated 4 times or more in the NaSTaB 14, , , , Note: Converted to the price level of 2014 using each year s consumer price index Source: Calculated by the authors using the 2 nd to 7 th NaSTaB data <Table V-8> shows the result of weighted calculations where average household insurance contribution and the contribution burden rate are classified into 5 income quintiles according to Standard A. Average contribution increases as the income quintile increases, and the contribution burden rate is the same across all quintiles in the same year. This is because the amount of contribution was calculated based on the income that had been calculated under the same condition. In other words, the contribution burden rate stays the same across all income quintiles because different contribution rates were not applied according to income level.

75 Effect of Reform in National Health Insurance Contribution Scheme: Longitudinal Analysis 73 Table V-8 Average Income, Average Insurance Contribution and Contribution Burden to Income Ratio by Income Quintile in Standard A (Unit: 10,000 KRW, Ratio) Income Quintile Average Contribution (Current) Quintile 1 Quintile 2 Quintile 3 Quintile 4 Quintile 5 Average Contribution (Standard A) Contribution Burden (Standard A) Average Contribution (Current) Average Contribution (Standard A) Contribution Burden (Standard A) Average Contribution (Current) Average Contribution (Standard A) Contribution Burden (Standard A) Average Contribution (Current) Average Contribution (Standard A) Contribution Burden (Standard A) Average Contribution (Current) Average Contribution (Standard A) Contribution Burden (Standard A) Source: Calculated by the authors using the 2 nd to 7 th NaSTaB data

76 The Effects of Reform in the National Health Insurance Contribution Scheme: On Equity in Health Insurance Contribution Burden 74 B. Insurance Contribution Burden in Standard B The income that is subject to insurance contribution in Standard B is calculated by adding up earned income, business income, financial income, annuity income and other incomes at the individual level and reorganizing the individual-level gross income into a house-level income. However, when the annual household income subject to contribution assessment was 3.36 million KRW or less, the minimum contribution was applied; if it exceeded 3.36 million KRW, the fixed-rate contribution was applied. <Table V-9> presents health insurance contribution by household and the contribution burden rate calculated per household according to Standard B. In Standard B, household health insurance contribution increases in each year under consideration, compared to the current level. However, the ratio of household contribution to household income i.e. the contribution burden rate does not show much change although there exists a slight variation across the years.

77 Effect of Reform in National Health Insurance Contribution Scheme: Longitudinal Analysis 75 Table V-9 Health Insurance Contribution and Contribution Burden Ratio by Household in Standard B (Unit: 10,000 KRW, Ratio) Health Insurance Contribution by Household (Current) Health Insurance Contribution by Household (Standard B) Households with the Employee Insured (Current) Households with the Employee Insured (Standard B) Households with the Self-Employed Insured (Current) Households with the Self- Employed Insured (Standard B) Households with Both the employee insured and the self-employed insured (Current) Households with Both the employee insured and the selfemployed insured (Standard B) Contribution to Household Income Ratio (Current) Contribution to Household Income Ratio (Standard B) Households with the Employee Insured (Current) Households with the Employee Insured (Standard B) Households with the Self-Employed Insured (Current) Households with the Self- Employed Insured (Standard B) Households with Both the Employee Insured and the Self-Employed Insured (Current) Households with Both the Employee Insured and the Self- Employed Insured (Standard B) (1.40) (1.14) (1.73) (1.46) (3.03) (2.36) (2.33) (1.98) (.001) (1.44) (1.23) (1.74) (1.59) (3.25) (3.08) (2.83) (2.06) (.001) (1.65) (1.45) (1.93) (1.82) (4.26) (4.01) (2.99) (2.35) (.001) (1.74) (1.61) (2.29) (2.31) (3.28) (2.66) (3.42) (3.14) (0.001) (1.94) (2.17) (2.71) (2.37) (3.65) (3.93) (2.92) (5.27) (0.004) (2.02) (1.78) (2.68) (2.58) (4.14) (3.58) (3.26) (2.66) (0.001) (.0004) (.0005) (.0003) (.0002) (.0003) (.0002) (.004) (0.006) (0.009) (0.006) (0.019) (0.006) (.0005) (.0006) Note: ( ) means ratio. Source: Calculated by the authors using the 2 nd to 7 th NaSTaB data (.0006) (.0005) (.0009) (.0007)

78 The Effects of Reform in the National Health Insurance Contribution Scheme: On Equity in Health Insurance Contribution Burden 76 When Standard B is applied, average contribution increases and the contribution burden rate is similar or increases, compared to the current level, among households with the employee insured. As for households with the self-employed insured, both of the figures decrease. <Table V-10> presents the sum of incomes that are subject to contribution assessment, the sum of contributions, and the ratio of contribution burden to household income calculated based on longitudinal data in Standard B. Each year s household incomes that are subject to contribution assessment and household contributions were converted to the price level of 2014 using the consumer price index. When the contribution scheme is changed to Standard B, the average of the sum of all incomes that are subject to contribution assessment for the 6 years between 2009 and 2014 stands at about 140 million KRW and average contribution is 5.93 million KRW. A slight increase is observed in the amount of insurance contribution in Standard A, compared to the current level. A slight increase is observed in the amount of insurance contribution, compared to the current level. The contribution burden ratio refers to the average ratio of the sum of each year s household contributions to the sum of the same year s household incomes. Average household income and average household contribution refer to average values calculated by regarding the data between 2009 and 2014 as one pool of cross-sectional data. The contribution burden rate in Standard B calculated as above is , which shows an increase from the current level (0.0263). The bottom of <Table V-10> shows the sum of household incomes that are subject to contribution assessment, the sum of household contributions, etc. with only the households investigated four times or more in the NaSTaB as subjects; the results indicate that there is no significant change in both the contribution burden rate and average household contribution.

79 Effect of Reform in National Health Insurance Contribution Scheme: Longitudinal Analysis 77 Table V-10 Longitudinal Analysis of Standard B: Incomes Subject to Contribution Assessment, Household Insurance Contribution and Contribution Burden Ratio Average sum of incomes subject to contribution assessment Average sum of household contributions Contribution burden ratio (Unit: 10,000 KRW, Ratio) Average of perhousehold incomes subject to contribution assessment Average household contribution All households Households investigated 4 times or more in the NaSTaB 14, , , Note: Converted to the price level of 2014 using each year s consumer price index Source: Calculated by the authors using the 2 nd to 7 th NaSTaB data While <Table V-9> presents average household income and average insurance contribution and the contribution burden rate that are calculated according to Standard B, <Table V-11> shows the result of weighted calculations where average household insurance contribution and the contribution burden rate are classified into 5 income quintiles according to Standard B. Average contribution increases as the income quintile increases, but the contribution burden rate remains nearly the same across all quintiles except for Income Quintile 1. This is because the minimum contribution was applied to households whose incomes subject to contribution assessment are less than 3.36 million KRW while contribution assessment was based on the income calculated under the same condition. The contribution burden rate as of 2014 stands at 0.30 in Quintiles 2 to 5, but Quintile 1 shows a high level of Since the fixed-amount contribution was applied to only the low-income class, the contribution burden rate is higher in Quintile 1, compared to other quintiles.

80 The Effects of Reform in the National Health Insurance Contribution Scheme: On Equity in Health Insurance Contribution Burden 78 Table V-11 Income, Insurance Contribution and Contribution Burden to Income Ratio by Income Quintile in Standard B (Unit: 10,000 KRW, Ratio) Income Quintile Average Contribution (Current) Quintile 1 Quintile 2 Quintile 3 Quintile 4 Quintile 5 Average Contribution (Standard B) Contribution Burden (Standard B) Average Contribution (Standard B) Contribution Burden (Standard B) Average Contribution (Standard B) Contribution Burden (Standard B) Average Contribution (Standard B) Contribution Burden (Standard B) Average Contribution (Standard B) Contribution Burden (Standard B) Source: Calculated by the authors using the 2 nd to 7 th NaSTaB data C. Insurance Contribution Burden in Standard C Standard C combines income-based and property-based contribution assessment, through which this study aims to deduce the respective rates (%) of income-based contribution and property-based contribution to be applied under the premise that the amount of contribution revenue estimated

81 Effect of Reform in National Health Insurance Contribution Scheme: Longitudinal Analysis 79 in Standard A remains secured and that the respective shares of income-based and property-based contributions in the revenue are set at 80% and 20%. First, to deduce the rate of income-based contribution, the income subject to contribution was calculated by adding up earned income, business income, and financial income exceeding 20 million KRW, 50% of the public annuity income, and other incomes exceeding 3 million KRW per household. Then, we came up with the amount of income-based contribution by calculating incomes subject to contribution assessment per household and 80% of the total contribution revenue deduced according to Standard A. The scope of property was limited to the data available in the NaSTaB, such as housing, buildings, land, ship and airplane, and the total amount of property was calculated using this data. The scope of property that is subject to the insurance contribution assessment can be adjusted when the scope of assets, such as key-money lease/the monthly rent and debts, is determined in the future. In that respect, we used the sum of the property data available in the NaSTaB for the present in Standard C of this study. Based on this, property-based contribution, which is to cover 20% of the contribution revenue in Standard A, was deduced. As shown in the <Table V-12>, the rate of income-based insurance contribution to cover the contribution revenue in Standard A is about 2%; the rate of property-based insurance contribution stands at about 0.15%. <Table V-12> presents changes in the rate of household insurance contribution burden when Standard C is applied. The contribution health insurance burden rate by household in Standard C is similar to those in Standards A and B. The contribution burden rate decreases for households with the employee insured and households with the self-employed insured whereas the burden rate increases for the households with both the employee insured and the self-employed insured. When compared with Standard A, which collects the same amount of insurance contribution revenue, the insurance contribution burden rate decreases relatively for households with the employee insured; the figure increases slightly for households with the self-employed insured and households with both the employee insured and the self-employed insured.

82 The Effects of Reform in the National Health Insurance Contribution Scheme: On Equity in Health Insurance Contribution Burden 80 Table V-12 Income-Based Contribution Rate and Property-Based Contribution Rate By Household in Standard C (Unit: Ratio) Income-based insurance contribution rate (Standard C) Property-based insurance contribution rate (Standard C) Insurance contribution compared to household income (Current) Households with the Employee Insured (Current) Households with the Employee Insured (Standard C) Households with the Self-Employed Insured (Current) Households with the Self-Employed Insured (C) Households with both the employee insured and the self-employed insured (Current) Households with both the employee insured and the self- employed insured (Standard C) (.001) (.0004) (.001) (.0005) (.001) (.0003) (0.001) (.0002) (0.004) (.0003) (0.001) (.0002) (.004) (0.006) (0.009) (0.006) (0.019) (0.006) (.0005) (.0006) Source: Calculated by the authors using the 2 nd to 7 th NaSTaB data (.0006) (.0005) (.0009) (.0007) <Table V-13> shows the result of weighted calculations where average household insurance contribution and the contribution burden rate are classified into 5 income quintiles according to Standard C. Average contribution increases as the income quintile increases, but the contribution burden rate decreases as the income quintile increases. When compared to Standard A in <Table V-6>, the contribution burden rate of the low-income class increases while that of the high-income class decreases in relative terms. When Standard A is applied, the difference in the contribution burden rate is not great between Income Quintile 1 and Income Quintile 5. When Standard C is applied, however, the contribution burden rate of Quintile 1 increases greatly while that of Quintile 5 decreases significantly. Therefore, it can be expected that the income redistributive effect of Standard C is negative, compared to Standard A even

83 Effect of Reform in National Health Insurance Contribution Scheme: Longitudinal Analysis 81 when the same amount of contribution revenue is assumed to be secured in both scenarios. Table V-13 Income, Insurance Contribution and Contribution Burden to Income Ratio by Income Quintile in Standard C (Unit: 10,000 KRW, Ratio) Income Quintile Quintile 1 Quintile 2 Quintile 3 Quintile 4 Quintile 5 Average Contribution Contribution Burden Average Contribution Contribution Burden Average Contribution Contribution Burden Average Contribution Contribution Burden Average Contribution Contribution Burden Source: Calculated by the authors using the 2 nd to 7 th NaSTaB data 3 Analysis of Redistributive Effect Generally, the analysis of redistributive effects is designed to measure changes in income distribution with regard to mandatory expenditure, such as tax, and the result is commonly determined with the difference between the Gini coefficient of pre-tax income and that of post-tax income. Based on the result, vertical equity, horizontal equity and the degree of re-ranking can be identified. Regarding the redistributive effect and equity issues related to the financial sources of health insurance, previous studies have been conducted within the same frame. However, since WHO report suggested equity issues relating to the financial sources of health insurance should be measured based on contribution burden rather than income in 2000, the WHO developed a

84 The Effects of Reform in the National Health Insurance Contribution Scheme: On Equity in Health Insurance Contribution Burden 82 contribution burden-based index called Fairness in Financial Contribution (FFC) and has been using the index in its cross-country analysis on equity related to the financial sources of health insurance. The FFC index measures changes in burden after the payment of health insurance contribution, as opposed to the conventional, income-based redistributive effect (RE) that measures changes in the distribution of income after the payment of health insurance contribution. In this study, we examined equity in health insurance contribution through the income redistributive effect after the payment of health insurance contribution for the convenience of comparison with previous studies. First, the redistributive effect of the current health insurance contribution scheme was analyzed. The Gini coefficient of the income before the payment of the health insurance contribution was calculated using household income, for which individual income was added up per household and household weight. The Gini coefficient of the income after the payment of health insurance contribution was calculated using the household income the sum of individual income per household, after the payment of household health insurance contribution the sum of individual payment of health insurance contribution per household, and household weight. The redistributive effect is measured by subtracting the Gini coefficients of the post-contribution payment income from the Gini coefficient of the pre-contribution payment income. The first row of <Table V-14> shows the results of the redistributive effect calculated for each year. The difference between the Gini coefficient of the pre-contribution payment income and that of post-contribution payment income has a negative value every year, which means the income inequity after the payment of the health insurance contribution becomes more serious than before the payment of the health insurance contribution. The redistributive effect can be expressed as the sum of vertical redistributive effect, horizontal redistributive effect, and the degree of re-ranking. Looking at the vertical redistribution rates presented in the third row of <Table V-14>, most of them are above 100%, which shows a structure where more income is transferred from the poor to the rich than actual in the absence of horizontal redistributive effect. The Kakwani index an index that measures the progressivity of insurance contribution also shows a negative value every year, suggesting that health insurance contribution is regressive vis-a-vis income.

85 Effect of Reform in National Health Insurance Contribution Scheme: Longitudinal Analysis 83 Table V-14 Redistributive effect of Health Insurance Contribution ( ) Redistributive Effect (RE) Vertical Redistributive Effect Vertical Redistribution Rate Contribution Rate Kakwani Index RE=GX-GX_P V=[g/(1-g)]* KE (V/RE)* g KE Source: Calculated by the authors using the 2 nd to 7 th NaSTaB data <Table V-15> presents the calculated values of redistributive effect from the longitudinal perspective using the sum of average household incomes and the sum of health insurance contributions between 2009 and 2014 listed in <Table V-2>. As in the cross-sectional analysis, the longitudinal analysis also shows that the redistributive effect of the current contribution scheme has negative values. The last row of <Table V-15> concerns only the households investigated four times or more in the NaSTaB as subjects between 2009 and 2014, and the redistributive effect has a negative in this group as well. Table V-15 Longitudinal Analysis: Redistributive Effect of Health Insurance Contribution Redistributive effect (RE) RE=GX-GX_P Gini coefficient before payment of health insurance contribution GX Gini coefficient after payment of health insurance contribution GX_P Note: The last row shows the values obtained by calculations based only on those households that were investigated four times or more in the NaSTaB. Source: Calculated by the authors using the 2 nd to 7 th NaSTaB data

86 The Effects of Reform in the National Health Insurance Contribution Scheme: On Equity in Health Insurance Contribution Burden 84 A. Redistributive Effect of Standard A The insurance contribution of Standard A was deduced as explained above, and the redistributive effect was measured using this. <Table V-16> presents redistributive effects that were calculated by comparing the current precontribution assessment income i.e. current household income and the postcontribution assessment income, which was calculated by excluding insurance contribution calculated according to Standard A from the current household income. The results of <Table V-16> show that insurance contribution in Standard A has almost no redistributive effect. However, Standard A seems to be more positive than the current contribution scheme in that the current health insurance contribution scheme shows negative redistributive effects across all years in <Table V-14>. Table V-16 Redistributive Effect of Health Insurance Contribution in Standard A ( ) Redistributive Effect (RE) Vertical Redistributive Effect Vertical Redistribution Rate Contribution Rate Kakwani Index RE=GX-GX_P V=[g/(1-g)]* KE (V/RE)* g KE Source: Calculated by the authors using the 2 nd to 7 th NaSTaB data <Table V-17> presents the calculated values of redistributive effect using the sum of average household incomes and the sum of health insurance contributions between 2009 and 2014 listed in <Table V-7>. Unlike the

87 Effect of Reform in National Health Insurance Contribution Scheme: Longitudinal Analysis 85 cross-sectional analysis, the longitudinal analysis shows that the redistributive effect of the current contribution scheme has positive values. This indicates that the contribution scheme of Standard A can be positive for income redistribution. The last row of <Table V-17> concerns only the households investigated four times or more in the NaSTaB as subjects between 2009 and 2014, and the redistributive effect is positive. Table V-17 Longitudinal Analysis: Redistributive Effect of Health Insurance Contribution in Standard A Redistributive effect (RE) RE=GX-GX_P Gini coefficient before payment of health insurance contribution GX Gini coefficient after payment of health insurance contribution GX_P Note: The last row shows the values obtained by calculations based only on those households that were investigated four times or more between 2009 and 2014 Source: Calculated by the authors using the 2 nd to 7 th NaSTaB data B. Redistributive Effect of Standard B The insurance contribution of Standard B was deduced as explained above and the redistributive effect was measured using this. According to <Table V-18>, the insurance contribution of Standard B negatively works on the redistributive effect. In other words, when judged based on income, income inequality can be worsened under the Standard B contribution scheme. However, when compared with <Table V-14>, the Standard B contribution scheme has a less impact on income inequality than the current contribution scheme. It is because that while the RE value under the current contribution scheme is about , under the Standard B contribution scheme the figure is about Income inequality is worsened under both the current contribution scheme and the Standard B contribution scheme because both of them schemes show negative redistributive effects. However, the extent of the Standard B contribution scheme s impact on the redistributive effect is less than the current contribution

88 The Effects of Reform in the National Health Insurance Contribution Scheme: On Equity in Health Insurance Contribution Burden 86 scheme. In this respect, Standard B seems more positive for redistribution, compared to the current contribution scheme. Table V-18 Redistributive Effect of Health Insurance Contribution in Standard B ( ) Redistributive effect (RE) Vertical Redistributive effect Vertical Redistribution Rate Contribution Rate Kakwani Index RE=GX-GX_P V=[g/(1-g)]* KE (V/RE)* g KE Source: Calculated by the authors using the 2 nd to 7 th NaSTaB data <Table V-19> presents the calculated values of redistributive effect using the sum of average household incomes and the sum of health insurance contributions between 2009 and 2014 listed in <Table V-10>. As in the cross-sectional analysis, the longitudinal analysis shows that the redistributive effect of the current contribution scheme is negative. This indicates that the contribution scheme of Standard B can be negative for income redistribution. The last row of <Table V-19> concerns only the households investigated four times or more in the NaSTaB as subjects between 2009 and 2014, and the redistributive effect is negative.

89 Effect of Reform in National Health Insurance Contribution Scheme: Longitudinal Analysis 87 Table V-19 Longitudinal Analysis: Redistributive Effect of Health Insurance Contribution in Standard B ( ) (Unit: 10,000 KRW, Ratio) Redistributive effect (RE) RE=GX-GX_P Gini coefficient before payment of health insurance contribution GX Gini coefficient after payment of health insurance contribution GX_P Note: The last row shows the values obtained by calculations based only on those households that were investigated four times or more between 2009 and 2014 Source: Calculated by the authors using the 2 nd to 7 th NaSTaB data C. Redistributive Effect of Standard C The insurance contribution of Standard C was deduced as explained above and the redistributive effect was measured using this. <Table V-20> presents redistributive effects based on the post-contribution assessment income that was calculated using the current household income and insurance contribution calculated according to Standard C. According to the results, the Gini coefficient increases, and this indicates that the Standard C contribution scheme negatively impacts the redistributive effect. Table V-20 Redistributive Effect of Health Insurance Contribution in Standard C ( ) Redistributive effect (RE) RE=GX-GX_P Source: Calculated by the authors using the 2 nd to 7 th NaSTaB data <Table V-21> presents the calculated values of redistributive effect using the sum of average household incomes and the sum of health insurance contributions between 2009 and As in the cross-sectional analysis, the longitudinal analysis shows that the redistributive effect is negative. This

90 The Effects of Reform in the National Health Insurance Contribution Scheme: On Equity in Health Insurance Contribution Burden 88 indicates that the Standard C contribution scheme can be negative for income redistribution. To compare with Standard A, the NHIS can collect the same amount of contribution revenue in Standard C. Yet, the redistributive effect can be more negative than with Standard A. This seems to be because the amount of contribution is assessed based not only on income but on property as well. The last row of <Table V-21> concerns only the households investigated four times or more in the NaSTaB as subjects between 2009 and 2014, and the redistributive effect is negative. Table V-21 Longitudinal Analysis: Redistributive Effect of Health Insurance Contribution in Standard C ( ) Redistributive effect (RE) RE=GX-GX_P Gini coefficient before payment of health insurance contribution GX Gini coefficient after payment of health insurance contribution GX_P Note: The last row shows the values obtained by calculations based only on those households that were investigated four times or more between 2009 and 2014 Source: Calculated by the authors using the 2 nd to 7 th NaSTaB data

91 Ⅵ Conclusion and Policy Implications Due to its various assessment standards, the current health insurance contribution scheme generates equity problems between the employee insured and the self-employed insured, as well as within each of them. According to the principle of social insurance, the amount of contribution should be determined in proportion with the ability to pay. However, a regressive structure is observed in some parts of the current contribution scheme. To solve such inequity problems, many reform plans have been discussed to consolidate the current scheme into a single, income-based assessment standard. This means that the distinction between the employee insured and the self-employed insured is removed and that the object of contribution assessment is expanded from earned income to include other types of income as well. If such reform is made, those who have been paying less than their ability to pay will have increased contributions while there will be a decrease in the amount of contribution for the low-income group who have been overpaying. Therefore, the reform will improve overall equity and help secure the revenue of the health insurance system. When the economically active population is reduced due to population aging, the base of health insurance revenue will be threatened under the current scheme since it depends on earned income. In that respect, expanding the contribution assessment base to encompass various types of income is expected to have a positive effect on stabilizing the finance of the national health insurance system. By applying the three most simplified standards among diverse proposals to reform the existing scheme to a solely income-based one, this study analyzed redistributive effects of the reform on equity in contribution burden, and its