Policy Research Institute, Ministry of Finance, Japan, Public Policy Review, Vol.10, No.3, October 2014 481 The Impact of Demographic Changes on Social Security Payments and the Individual Income Tax Base Long-term Micro-simulation Approach * Kazuya Matsuda Visiting Researcher, Policy Research Institute, Ministry of Finance Yumiko Ozeki Director for Econometric Analysis, Policy Research Institute, Ministry of Finance Kazuaki Kikuta Researcher, Policy Research Institute, Ministry of Finance Junji Ueda Former Director for Econometric Analysis, Policy Research Institute, Ministry of Finance Abstract This paper analyses the quantitative impact of future demographic changes on Japan s fiscal situation, focusing on an inevitable increase in social security payments along with an increase in social security benefits, and its effect on individual income tax-base erosion through further social insurance deductions. The analysis is conducted using a micro-simulation methodology. In the analysis, we use the individual data contained in the Comprehensive Survey of Living Conditions (2010) in order to calculate theoretical values of individual income taxes at both the individual and the macro level from 2009, the base year, to 2060 in five-year intervals. We then compare the theoretical values with those for 2009 as the base year. We also compute theoretical values of the social security payments for each relevant year in the future, referring to several projections of social security expenditures under the current social security system. Major findings of this paper are as follows: (1) When individuals enrolled in the same social security program equally bear a future increase in social security payments due to demographic changes, the total amount of income tax revenue in 2060 will be approximately 3.7 trillion lower than in 2009. In this case, the * In preparing this paper, we received valuable comments at a meeting of the Financial Review paper panel. (Ministry of Finance, Policy Research Institute, November 2013). Regarding the provision of data obtained from the Comprehensive Survey of Living Conditions that were used in this paper, we obtained cooperation from relevant people at the Ministry of Health, Labour and Welfare. We hereby express our appreciation for their kind cooperation. Anything amiss you may find in this paper is entirely attributable to the authors. It should be noted that the contents of this paper represent the personal opinions of the authors, not the official opinions of the organizations to which they belong.
482 K Matsuda, Y Ozeki, K Kikuta, J Ueda / Public Policy Review ratio of income tax revenue to GDP will decline gradually from 1.842% in 2009 to 1.780% in 2060. We also find that the total amount of income tax revenue in 2060 will be reduced by around 0.8 trillion from 2009, when the effect of an increase in social security payments alone is taken into consideration while the other effects including demographic changes such as population decline are excluded. (2) When individuals bear a future increase in social security payments according to the ability to pay, the total amount of income tax revenue will decrease by around 4.0 trillion between 2009 and 2060. In this case, the ratio of the total individual income taxes to GDP will gradually drop to 1.674% by 2060. Total individual income taxes in 2060 will be reduced by around 1.4 trillion from 2009 when the effect of an increase in social security payments alone is taken into account. Keywords: individual income tax, taxation base, social insurance deductions, micro simulation, demographics JEL Classification: D31, D33, H24 I. Introduction In recent years, the aging of society, coupled with the low birthrate, has proceeded in Japan, and in the future, a further dramatic demographic change is expected to occur. In this situation, social security benefits, including pension, health and long-term care benefits, are expected to increase. To finance increased social security benefits, the burden of social security payments, mainly those paid by employed workers, is expected to grow under the existing social security system. For example, with regard to pension insurance payments, it is legally prescribed that the insurance premium rate be raised in stages in the period through fiscal 2017. Meanwhile, under each of health and long-term care insurance programs, there is a mechanism that secures a prescribed proportion of the cost of benefits through insurance revenue, which means that an increase in benefits will automatically lead to an increase in insurance premiums. On the other hand, these social security payments may be deducted from income in the calculation of taxable income under the existing income tax system. This means that an increase in social security payments will automatically result in the erosion of the income tax base. This paper conducts quantitative simulation as to how much an automatic increase in social security payments arising from the future aging of society will erode the individual income tax base and the total income tax revenue based on the assumption that the existing social security system and social insurance deductions will continue. To that end, with regard to the total amount of insurance payments for each of pension and health and long-term care programs, we use an estimated figure based on the projection for social security benefits under the existing social security system.
Policy Research Institute, Ministry of Finance, Japan, Public Policy Review, Vol.10, No.3, October 2014 483 If we are to conduct such simulation with a certain level of accuracy, it is necessary to take into consideration that the marginal tax rate under an excess progressive tax system varies according to the income level and that there are many people whose taxable income is smaller than the amount of social insurance deductions. This means that the impact of a change in the amount of social insurance deductions on individuals taxable income and tax payments varies according to their respective income level and attributes. Therefore, this paper conducts analysis through a micro-simulation methodology using data concerning individuals and households collected through questionnaires contained in the Comprehensive Survey of Living Conditions (hereinafter referred to as the individual data ). First, we estimate the amount of future social security payments for the social security programs in which each individual is enrolled and calculate the theoretical value of his/her income tax payments. Then, we estimate the theoretical value of future income tax payments at the macro level, namely, the total amount of future income taxes to be paid by individuals, and compare it with the value in the base year (2009). This paper is structured as follows. First, we discuss preceding studies and our analysis methodology in Section II and explain the data used in our analysis in Section III. In Sections IV and V, we conduct simulation under multiple scenarios and explain the results. Finally, we summarize our analysis and suggest how to improve the analysis in the future. II. Preceding studies and our analysis methodology The approach of applying a micro-simulation methodology taking into consideration individuals attributes to the estimate of the future income tax base and tax payments has already been adopted by fiscal authorities in the United States and European countries. In the United Kingdom, a micro-simulation methodology is used when estimating income tax revenue in the period through 2060. The results show how much the amount of macro-level income tax payments will change in accordance with various assumptions, including an increase in part-time workers (change in income distribution), a productivity improvement and a decline in the birthrate 1. The U.S. Congressional Budget Office (CBO) uses the CBOLT model (Congressional Budget Office Long-Term micro simulation model) based on a microsimulation methodology when making projections for future individual income tax revenue. Among studies on individual income tax in Japan that have adopted a micro-simulation methodology using individual data is one by Tajika and Yashio (2008). They identified the situation of the tax burden in Japan using individual data collected through questionnaires concerning income and savings in the Comprehensive Survey of Living Conditions (2004). Based on a simulation, they examined how the tax burden by household type will change if a certain tax system reform is implemented. Meanwhile, Yada (2011) presented an original model for calculating income and residential taxes and social security benefits using individual data from the Comprehensive Survey of Living Conditions in 2007 while using 1 See HM Treasury (2009), Long-term public finance report: an analysis of fiscal sustainability
484 K Matsuda, Y Ozeki, K Kikuta, J Ueda / Public Policy Review the approach of Tajika and Yashio (2008) as a reference. In our analysis, we calculate the theoretical value of income tax payments at the individual level according to the individual data based on the calculation model presented by Yada (2011) and estimate how the theoretical value of income tax payments at the macro level will change in accordance with changes in social security payments by aggregating the results for individuals through a method that will be explained later. With regard to social security payments, Ueda, Horiuchi and Morita (2010), Ueda, Horiuchi and Tsutsui (2011) and Ota and Nakazawa (2013) estimated the amount of future social security payments based on projections for the costs of health and long-term care. In our analysis, we set an assumed amount of increase from the level in 2009 in total social security payments at each future selected point of time spaced at 5-year intervals in the period through 2060. The increases in total security payments are allotted to individuals according to their attributes, including the type of social security program in which they are enrolled and the presence or absence of tax payment, based on the individual data. The theoretical value of income tax payments at the individual level is calculated, and the theoretical value of income tax revenue at the macro level obtained by aggregating the results for individuals is compared with the value of income tax revenue in 2009. In this way, we conduct quantitative analysis as to how much impact an increase in social security payments arising from demographic changes will have on the total income tax revenue. III. Data used in our analysis III-1. Comprehensive Survey of Living Conditions Now, we explain the data used in our analysis. Tajika and Yashio (2008) used individual data from the Comprehensive Survey of Living Conditions in 2004, compiled by the Ministry of Health, Labour and Welfare, while Yada (2011) used individual data from the same survey in 2007 2. As the Comprehensive Survey of Living Conditions is conducted through welfare offices across the nation, the collection rate from low-income households is high. Consequently, information concerning the tax burden, social security benefits and social security burden obtained through this survey is deemed to be accurate 3. Therefore, we use individual data from the Comprehensive Survey of Living Conditions in 2010. Questionnaires contained in the Comprehensive Survey of Living Conditions include the households, health, long-term care, income and savings questionnaires, and we use individual data from the household and income questionnaires. From the household questionnaire, 2 The Comprehensive Survey of Living Conditions, which is a major statistical survey based on the Statistics Act, is intended to investigate basic matters concerning the people s daily life, including health, medical care, welfare, pensions and income. It is conducted once every three years on an extensive basis. 3 See Yada (2011)
Policy Research Institute, Ministry of Finance, Japan, Public Policy Review, Vol.10, No.3, October 2014 485 information concerning individuals attributes such as age and employment status can be obtained. From the income questionnaire, information concerning items necessary for the calculation of the individual income tax amount, such as the income level and the amount of social security payment can be obtained. III-2. Data analyzed Survey subjects of the income questionnaire are randomly selected from the survey subjects of the household questionnaire. The number of survey subjects of the income questionnaire is 26,115 households (with a total of 70,175 household members), whereas the number of survey subjects of the household questionnaire is 228,864 households (with a total of 609,008 household members). The calculation of the income tax amount requires information obtained from both the household and income questionnaires, so the subjects of our analysis are households and household members that are the survey subjects of the income questionnaire. Regarding these data, data culling is implemented according to the procedures used by Yada (2011). First, we exclude from the sample group individuals whose birth date data, essential for calculating the amount of tax deduction from income, are unavailable, households to which those individuals belong and members of those households. Next, we exclude individuals who declare social security payments but who do not specify the payment amount, households to which those individuals belong and members of those households. A single-person household of a person living separately from his/her family for business reasons and the households of his/her family should be treated as one and the same household for the purpose of calculating spouse and dependent tax deductions. However, it is impossible to match these two categories of households based on information from the individual data alone and it is difficult to appropriately calculate income deductions for members of the households. Consequently, single-person households of persons living separately from their families and households of their families are excluded from the sample group. Households and household members excluded from the sample group are as shown in Table 1. The number of analyzed sample households is 21,829 and the number of analyzed sample household members is 57,857 4. 4 Yada (2011) excluded both individuals who declared payment of property tax but who did not specify the payment amount and members of the households to which those individuals belong. However, as the amount of property tax payment does not directly affect the calculation of the income amount in this paper, we do not exclude them.
486 K Matsuda, Y Ozeki, K Kikuta, J Ueda / Public Policy Review Table 1 Analyzed sample data Table 2 Household data Table 2 shows the average income and the average social security payment for each of the 10 household income brackets into which the sample households were divided based on the individual data concerning the subject group. The average household income for the sample households is 5,271,000 per year, while the average social security payment is 482,000 per year (including 200,000 in health insurance, 210,000 in pension insurance and 44,000 in long-term care insurance).
Policy Research Institute, Ministry of Finance, Japan, Public Policy Review, Vol.10, No.3, October 2014 487 IV. Simulation of social insurance deductions and income tax amount IV-1. Calculation of income tax amount (theoretical value) in 2009 IV-1-1. Calculation method of the income tax amount for households As the Comprehensive Survey of Living Conditions in 2010 included questions concerning income, tax, and social security payments in 2009, we adopt 2009 as the base year and calculate the theoretical value of the amount of tax paid by the sample households and the total tax amount in the year. The amount of tax paid by the sample households is calculated on the basis of the method used by Yada (2011). Individuals tax base (taxable income) related to the comprehensive taxation is calculated based on the individual data, and the income tax rate corresponding to the taxable income is applied in order to calculate the amount of income tax payment. The amount of a household s individual income tax payment is the sum of the income tax paid by all members of the household. The tax system in 2009 that was used for the calculation of the amount of income tax payment is as shown in Table 3. Table 3 Applicable tax system 380,000 380,000 380,000 Aged 16-22: \630,000 Aged 70 or older: \480,000 The deduction amount is Aged 70 or older: \480,000 Elderly parents living together: \580,000 Cannot be used at the same
488 K Matsuda, Y Ozeki, K Kikuta, J Ueda / Public Policy Review The specific calculation method of the amount of income tax payment is as follows: (i) Calculation of the total income First, in order to calculate individuals income amount, data concerning various incomes are obtained from the income questionnaire, and the total income amount based on the following formula (hereinafter referred to as the Total Income ) is calculated 5. Total Income = Salary income + pension income + business income + agricultural and livestock breeding income + home work income The salary income is calculated by subtracting from the figure indicated in the compensation of employees column the corresponding salary income deduction, while the pension income is calculated by subtracting from the figure indicated in the public pensions column the corresponding public pension deduction. With regard to business income, agricultural and livestock breeding income and home work income, the figures indicated in the relevant columns are used as they are. The income questionnaire of the Comprehensive Survey of Living Conditions also includes a property income column, and Yada (2011) included property income in the Total Income. However, in the survey, a very small group of people in a specific age group among the sample individuals declared extremely large amounts of property income. It is presumed that most of those incomes are interest and dividend incomes, which are subject to separate taxation, so we exclude property income from our analysis, focusing on income related to the comprehensive taxation as narrowly defined 6. (ii) Calculation of income deductions Next, with regard to individuals, we calculate the amount of income deductions from the Total Income calculated in (i) above. Income deductions applied in our analysis are the basic deduction, social insurance deductions, spouse deduction, special spouse deduction and dependent deduction. As for the basic deduction, 380,000 is universally deducted from the Total Income. Regarding social insurance deductions, the amount of social security payment obtained from the income questionnaire is deducted from the Total Income. People who have a spouse may be eligible for spouse deduction or special spouse deduction. In order to decide whether spouse deduction (special spouse deduction) should be applied to a husband or wife, the basic deduction and social insurance deductions are deducted from the Total Income of each of the two. Spouse deduction (special spouse deduction) is applied to whoever has the 5 Incomes covered by comprehensive taxation include temporary income and income from the sale of property. However, these incomes are not included in the Total Income because their amounts cannot be obtained from the Comprehensive Survey of Living Conditions. 6 The income questionnaire asks respondents to specify, in the asset incomes column, the total sum of the real estate income and the interest and dividend incomes, including those taxed separately at source, that are indicated in their tax declarations. However, all asset incomes are excluded because the breakdown of the total is unknown.
Policy Research Institute, Ministry of Finance, Japan, Public Policy Review, Vol.10, No.3, October 2014 489 higher income. Dependent deduction is applied to whoever has the highest income after spouse deduction (special spouse deduction) within the household 7. (iii) Application of excess progressive tax rates The amount of individuals income tax payment is calculated by multiplying the income after income deductions in (ii) above (taxable income) by the corresponding excess progressive tax rate. While the actual amount of income tax payment is arrived at by subtracting dividend deduction and housing loan deduction from this amount, information concerning those deductions cannot be obtained from the Comprehensive Survey of Living Conditions. Therefore, the amount of income tax payment before those deductions is deemed to be the final amount of income payment in our analysis. Finally, the amount of a household s income tax payment is arrived at by adding up the amounts of income tax payment by all members of the household. IV-1-2. Extension to the national scale Next, the amount of income tax payments by 1,829 households and 57,857 household members that was calculated in Section IV-1-1 is extended to the national scale to calculate the amount of macro-level income tax revenue. In the individual data from the Comprehensive Survey of Living Conditions, a multiplying factor is allotted to each region. A multiplying factor is a value that indicates how many households a sample household in the region represents. By adding up the figures obtained by multiplying the amounts of income tax payments by households by multiplying factors, we can estimate the total amount of income tax at the macro level. However, as was pointed out by Yada (2011), although it is possible to estimate nationwide figures by applying multiplying factors to the individual data if the questionnaire collection rate is 100%, this method cannot appropriately estimate nationwide figures because the collection rate is not 100% in reality 8. In addition, the collection rate varies according to the generation and structure of the sample household. As a result, there is a wide disparity between the individual data and the national census data in the distribution of individuals and households. 7 Other income deductions include disability deduction, widow (widower) deduction and medical expense deduction. However, this paper does not take into consideration these deductions because necessary data concerning them cannot be obtained from the Comprehensive Survey of Living Conditions. 8 If the multiplying factors are applied to the data set, the number of households comes to approximately 22.68 million, much fewer than the approximately 51.84 million in the national census (2010). When creating the data set in Section III-2, we excluded some data. Even if the data are not excluded, the number of households obtained by applying the multiplying factors comes to only 27.08 million. This is because the questionnaire collection rate is around 50 to 60%.
490 K Matsuda, Y Ozeki, K Kikuta, J Ueda / Public Policy Review To resolve these problems, Inagaki and Kaneko (2008) and Yada (2011) corrected the multiplying factors so as to narrow the disparity between the individual data and the national census data in the distribution of households characteristics. In our analysis, we also correct the multiplying factors through a similar method and calculate the amount of macro-level income tax revenue by multiplying the household number adjustment ratio (the ratio of the number of households obtained from the national census to the number of households arrived at by applying the multiplying factors to the individual data) by the amount of income tax revenue arrived at by applying the corrected multiplying factors 9. Table 4 shows the theoretical values of the amounts of macro-level social security payments and income tax revenue obtained by multiplying the household number adjustment ratio (2.350) by the amounts of social security payments and income tax revenue arrived at by applying the corrected multiplying factors. The amount of social security payments is approximately 24.7 trillion and the amount of income tax revenue is approximately 8.7 trillion. Let us compare these theoretical values with the actual figures for fiscal 2009. As for the amount of social security payments, the amount of employees mandatory social contributions (households income and outlay accounts) in the System of National Account (SNA), compiled by the Cabinet Office, was approximately 28.7 trillion in fiscal 2009, around 4 trillion higher than the abovementioned theoretical value. Now, we compare the theoretical value and the actual figure for fiscal 2009 (approximately 28.7 trillion) item by item by looking at the breakdown of the social security payments (Attached Table 10 of the SNA). While the theoretical values of the amounts of health and long-term care insurance premiums payments are almost equal to the actual values indicated in the SNA, the theoretical value of the amount of pension insurance payments is smaller than the actual figure indicated Table 4 Amounts of social security payments and income tax revenue (actual figures are on a fiscal year basis) (Sources) Social security payments, health insurance payment, long-term care insurance payment and employment and other insurance payment: FY2009 GDP (final), Cabinet Office Income tax revenue: FY2009 Settlement of general-account revenue and expenditures, Ministry of Finance 9 As for the correction method of the multiplying factors, see Supplementary Explanation A.
Policy Research Institute, Ministry of Finance, Japan, Public Policy Review, Vol.10, No.3, October 2014 491 in the SNA. In some cases, respondents may have neglected to declare or under-declared their pension insurance payments. Regarding other social security payments, the theoretical value of employment insurance payments is slightly larger than the actual figure indicated in the SNA. The amount of income tax revenue in fiscal 2009 was approximately 12.9 trillion, and the amount of income tax payments excluding taxes related to interest and dividend income and income from the sale of property that were withheld at source (approximately 2.3 trillion) was approximately 10.6 trillion, around 1.9 trillion higher than the theoretical value of income tax revenue. This is presumably because the calculation of income tax revenue in our analysis does not take into consideration property income, temporary income and income from the sale of property and also because it is difficult to identify the amount of income tax payments by some high income earners based on data from the Comprehensive Survey of Living Conditions. IV-2. Impact of the erosion of the tax base due to an increase in social security payments IV-2-1. Categorization by insurance program It is presumed that changes in the individual income tax base associated with demographic changes occur mainly because of two factors. One is the impact of an increase in social security payments arising from an increase in social security benefits caused by the aging of society. Even if the income level and distribution remain constant in the future, it is expected that an increase in social security payments will lead to expansion of social insurance deductions, which in turn will automatically result in the erosion of the income tax base. The other factor is the impact of a change in the income distribution caused by a shrinkage of the entire population and a decline in the proportion of the working-age population. In this section, in order to examine the impact of the former factor, we first calculate the amount of income tax revenue taking into consideration only an increase in the burden of social security payments, including payments for pension and health and long-term care insurance, with the demographics and the income level and distribution in 2009 as preconditions. Specifically, we estimate how much individuals social security payments will increase compared with 2009 in tandem with growth in benefits due to the aging of society based on the estimates of future payments obtained from the model used by Ueda, Horiuchi and Tsutsui (2011) and Ota and Nakazawa (2013). As for pension insurance payments, the amount of future payments is calculated based on the schedule of pension premium increases presented in the pension system reform of 2004 10. How individuals future payments for health insurance will change may vary depending on the insurance program in which they are enrolled. 10 The estimate is made for selected years at a five-year interval (2015, 2020, 2025, 2030, 2035, 2040, 2045, 2050, 2055 and 2060).
492 K Matsuda, Y Ozeki, K Kikuta, J Ueda / Public Policy Review That is because the projected increase in benefits differs from program to program due to differences in the attributes of members across programs. There are also other reasons such as that the ratio of payments covered by public funds differs from program to program and that the presence of the mechanism of providing contributions and grants in order to make fiscal adjustments between different insurance programs significantly affects the total amount of necessary insurance payments. As for future changes in long-term care insurance payments, it is necessary to separately estimate changes for Type 1 insured persons and Type 2 insured persons because the self-payment rate is different between these two types. In light of the abovementioned points, we first estimate the necessary increase in payments for each insurance program and then categorize the sample individuals by insurance program and look into how each individual s insurance payments will change in the future. However, while the household questionnaire of the Comprehensive Survey of Living Conditions includes questions concerning enrollment in health insurance and public pension programs, public pension programs are categorized only as the national health insurance program, employees insurance, or health insurance for people aged 75 or older (latter-stage Table 5 Social security programs and categorization criteria
Policy Research Institute, Ministry of Finance, Japan, Public Policy Review, Vol.10, No.3, October 2014 493 elderly health insurance). In the case of some individuals, it is not clear which insurance program they are enrolled in. Therefore, it is necessary to allocate members of the sample households to specific programs based on prescribed criteria (Table 5). Regarding health insurance, each individual is assumed to be a member of one of the latter-stage elderly health insurance, a mutual aid association, a health insurance society, the Japan Health Insurance Association or the national health insurance program. First, individuals aged 75 or older are categorized as members of the latter-stage elderly health insurance program 11. Next, individuals aged 74 or younger who are a dependent within a household are assumed to be a member of the insurance program in which the supporter (the highest income earner) of the household is enrolled. The highest income earners of a household are assumed to be a member of one of a mutual aid association, a health insurance society or the Japan Health Insurance Association based on the following criteria. First, civil servants are assumed to be a member of a mutual aid association. Permanent ordinary employees other than civil servants are categorized as a member of either a health insurance society or the Japan Health Insurance Association. In our analysis, individuals working for a company with a workforce of more than 100 employees are assumed to be a member of a health insurance society and those working for a company with a workforce of 99 or less are assumed to be a member of the Japan Health Insurance Association 12. People aged 74 or younger who are not enrolled in one of a mutual aid association, a health insurance society or the Japan Health Insurance Association are categorized as a member of the national health insurance program. As for long-term care insurance, individuals are categorized as either a Type 1 insured person or a Type 2 insured person. The categorization of the long-term insurance type is based on age: individuals aged 65 or older are categorized as a Type 1 insured person and individuals aged 40 to 64 are categorized as a Type 2 insured person. Regarding pension insurance, individuals are categorized as a member of either the national pension plan (Type 1 or Type 3 insured person) or an employee pension plan (Type 2 insured person) 13. Individuals who are a Type 1 or Type 3 insured person and individuals 11 People with a disability who are aged between 65 and 74 are also covered by the latter-stage elderly health insurance program. However, such people are categorized as a member of one of a mutual aid association, a health insurance society, the Japan Health Insurance Association or the national health insurance program because people with a disability cannot be identified on the basis of the individual data. 12 The establishment of a health insurance society by a single company requires a minimum of 700 insured persons. The establishment of a joint health insurance society by multiple companies requires a minimum of 3,000 insured persons. Allocating individuals to health insurance programs on the basis of the number of employees at their companies is not a perfect approach. However, it is difficult to find any other appropriate criterion, so the categorization is made universally on the basis of the number of employees. 13 Although civil servants belong to mutual aid associations, they are treated in the same way as members of employee pension programs for convenience s sake because the future rates of pension premiums for mutual aid associations have not yet been determined.
494 K Matsuda, Y Ozeki, K Kikuta, J Ueda / Public Policy Review aged 20 to 59 who did not fill in the pension plan enrollment column but who make pension insurance payment are categorized as a Type 1 insured person. Table 6 shows the number of individuals categorized as members of the various insurance programs based on the abovementioned criteria. Based on the number of individuals enrolled in health insurance programs that was calculated by applying the corrected multiplying factor, the enrollment rate for the national health insurance is 22%, lower than the actual rate of 31%. On the other hand, the enrollment rate for mutual aid associations is 14%, higher than the actual rate of 7% 14. A presumed factor behind these disparities is the possibility that the questionnaire collection rate is higher among members of mutual aid associations than Table 6 Number of members of social security programs (People) (Proportion) (10,000 people) (Proportion) (People) (Proportion) (10,000 people) (Proportion) (People) (Proportion) (10,000 people) (Proportion) (Sources) Number of members of health insurance programs: Ministry of Health, Labour and Welfare (2011), Basic Data on Health insurance Number of long-term care Type 1 insured persons: Ministry of Health, Labour and Welfare (2010), Report on Long-term Insurance Care. Number of long-term care Type 2 insured persons: National Institute of Population and Social Security Research (2012), an estimated total number of people aged 40 to 64 in the medium fertility, mediummortality projections of the Population Projections for Japan Number of members of pension insurance programs: Ministry of Health, Labour and Welfare (2011), Points of the Pension System. 14 The rate for the data excluded in Section III-2 is almost the same as the rate for the analyzed data.
Policy Research Institute, Ministry of Finance, Japan, Public Policy Review, Vol.10, No.3, October 2014 495 among members of the national health insurance program. Another factor is the possibility that dependents enrolled in the national health insurance program are categorized as members of other health insurance programs. Judgment as to whether or not supporters are enrolled in the national insurance program is made based on their age (whether or not they are aged 75 or older) as well as the presence or absence of membership in employees insurance programs, so it is unlikely that supporters enrolled in the national insurance program are erroneously categorized as members of other insurance programs. Meanwhile, dependents are categorized as members of the insurance program in which their supporters (the highest income earner in the household) are enrolled, so those dependents who are members of the insurance program of the supporters who are not the highest income earner may be erroneously categorized as members of other insurance programs. As for long-term care insurance, the enrollment rates for Type 1 and Type 2 insurance based on the number of insured persons calculated by applying the corrected multiplying factor are similar to the actual rates. In the case of pension insurance, the theoretical enrollment rate for the national pension plan is slightly lower than the actual rate, while the theoretical enrollment rate for employee pension plans is slightly higher. IV-2-2. Extension of social security payments to the national scale Next, we calculate the amount of individuals future insurance payments for their respective social security programs. In this section, we adopt a base scenario assuming that the rate of an increase (compared with fiscal 2009) in each of pension, health and long-term care insurance payments per capita will be linked to the necessary increase in insurance payments for each social security program and that all individuals enrolled in the same social security program will equally bear the burden of insurance payments (in other words, the payment will increase at the same rate for all individuals). Regarding health and long-term care insurance payments, we adopt the estimated amounts of future health and long-term care insurance payments (nominal), the number of members, the rate of increase in health and long-term care insurance payments per capita that were used by Ueda, Horiuchi and Morita (2010) and Ueda, Horiuchi and Tsutsui (2011) and Ota and Nakazawa (2013) and calculate the amount of future real insurance payments per capita for each social security program in each year in the future based on the following formulas. By multiplying the health and long-term care insurance payments in 2009 obtained from the individual data by the rates of increase in real insurance payments per capita, we calculate the estimated amount of future health and long-term care insurance payments to be made by each individual.
496 K Matsuda, Y Ozeki, K Kikuta, J Ueda / Public Policy Review Health insurance payment: Real insurance payment per capita = (Total insurance payments/number of members)/rate of increase in health care expenses per capita 15. Long-term care insurance payment: Real insurance payment per capita = (Total insurance payments/number of insured persons)/ rate of increase in long-term care expenses per capita *The total insurance payments for Type 2 insurance represent the portion of the payment amount that was paid by members themselves. As for pension insurance payments, we calculate the rate of increase based on the schedule of pension premium increases presented in the pension system reform of 2004. The insurance premium rate for employee pensions is scheduled to be gradually raised to 18.3% in 2018, and the insurance premium amount for the national pension program is scheduled to be raised to 16,900 per month in 2018. The payments will remain constant thereafter. As the pension premium rate for employee pension programs is increased in September and the pension premium amount is increased in April, we make adjustments to represent the payments on a calendar year basis by calculating the weighted average based on the number of months of payment and calculate the rate of increase compared with 2009. By multiplying the pension insurance payments in 2009 obtained from the individual data by the rate of increase thus calculated, we calculate the amount of future pension insurance payments to be made by each individual. Social security payments other than health and long-term care insurance payments and pension insurance payments, such as employment insurance payments, are assumed to remain constant. Tables 7-1 to 7-3 show the rate of increase in payments for each program of health and long-term care insurance and pension insurance (increase in real insurance payments per capita compared with 2009). Table 8 shows the results obtained by estimating individuals annual social security payments in each year based on the rate of increase in insurance payments calculated as above and calculating the amounts of macro-level income tax revenue according to the 15 When calculating the total amount of medical and long-term care payments, Ueda, Horiuchi and Morita (2010) and Ueda, Horiuchi and Tsutsui (2011) and Ota and Nakazawa (2013) assume that medical and long-term care expenses will be linked to the economic growth rate, etc. In other words, regarding the rate of increases in medical and long-term care insurance payments per capita in 2009 and 2010, they adopt figures arrived at in light of external factors such as the estimate of the medical care cost and the revisions of the medical and long-term care fees, while for the period from 2011 onward, they take into consideration the growth rates of nominal GDP and wages. However, in order to focus exclusively on the impact of an increase in social security payments arising from the aging of society, it is necessary to exclude the impact of an increase associated with the economic growth rate.
Policy Research Institute, Ministry of Finance, Japan, Public Policy Review, Vol.10, No.3, October 2014 497 Table 7-1 Rate of increase in real insurance payment (health insurance) per capita Table 7-2 Rate of increase in real insurance payment (long-term care insurance) per capita processes explained in Sections IV-1-1 and IV-1-2. According to Table 8, the income tax base will shrink in line with an increase in social security payments. The income tax revenue is estimated to be reduced by approximately 347.0 billion in 2020 and by approximately 758.0 billion in 2060 compared with 2009. In other words, the theoretical value of income tax revenue related to comprehensive taxation will be reduced by 8.7% in 2060 compared with 2009.
498 K Matsuda, Y Ozeki, K Kikuta, J Ueda / Public Policy Review Table 7-3 Increase in pension insurance payments Conversion into a calendaryear basis Table 8 Theoretical value of income tax revenue after extension to the national scale (base scenario) IV-3. Macro-demographic changes Following the analysis of the impact of an increase in social security payments based on the demographics of 2009 in the previous section, we analyze the impact of demographic changes, including a population decline, and associated changes in the income distribution
Policy Research Institute, Ministry of Finance, Japan, Public Policy Review, Vol.10, No.3, October 2014 499 on the amount of income tax revenue. As for the estimate of a future population, we use the medium fertility, medium-mortality projections of the Population Projections for Japan (an estimate as of January 2012) by the National Institute of Population and Social Security Research. We categorize the individual data by gender and age groups (increments of five years) and calculate the average amounts of social security payments and income for each group. By multiplying these figures by the respective population sizes of the groups as estimated by the Population Projections for Japan, we arrive at the amount of macro-level tax revenue in each selected year in the future. Table 9 shows the results of the calculation. Under this method, we assume that the real value of social security payments and income tax payments by each individual in 2009 will remain constant into the future, so it should be kept in mind that future changes in the economic environment and wage level as well as changes in the labor supply and income deductions are not taken into consideration. Table 9 shows that demographic changes will result in a significant decline in the amount of macro-level income tax revenue (income tax revenue in 2060 will be reduced by approximately 3.7 trillion compared with 2009). However, the pace of decline varies from period to period. In the 20 years between 2009 and 2030, the tax revenue is estimated to decline by approximately 1.2 trillion, followed by a further decline of 2.5 trillion in the 30 years between 2030 and 2060. Table 9 Theoretical value of income tax revenue assuming demographics changes (base scenario)
500 K Matsuda, Y Ozeki, K Kikuta, J Ueda / Public Policy Review Table 10 Comparison of decreases in income tax revenue (compared with 2009) (base scenario) changes (b) (b)/(a) Table 10 is a comparison between these results and the decline in the amount of income tax revenue calculated by taking into consideration an increase in social security payments while assuming constant demographics (Table 8). The ratio of the amount of income tax revenue reflecting demographic changes to the amount of income tax revenue not reflecting them consistently increases from 1.53 in 2015 to 4.92 in 2060. In other words, the further into the future we look, the greater the impact of changes in the income distribution due to demographic changes will be relative to the impact of an increase in social security payments. Next, in order to examine how much the capability to levy individual income tax will change, we consider how the ratio of income tax revenue to GDP will change. We assume that labor s share of income as well as the rates of future changes in employee compensation and GDP will stay constant. Based on these assumptions, we first calculate the average employee compensation in 2009 by gender and age groups based on the individual data and then estimate employee compensation in each selected year in the future by taking into consideration future demographic changes. By multiplying the value of GDP in 2009 by the growth rate of employee compensation obtained in this way, we arrive at a hypothetical value of GDP in each selected year in the future. Table 11 shows changes in the hypothetical value of GDP and the theoretical value of individual income tax payments. According to the table, the ratio of the theoretical value of income tax revenue to GDP will decline gradually from 1.842% in 2009 to 1.780% in 2060.
Policy Research Institute, Ministry of Finance, Japan, Public Policy Review, Vol.10, No.3, October 2014 501 Table 11 Changes in the ratio of income tax revenue to GDP (base scenario) Finally, we compare per-capita income tax payments under different assumptions. Table 12 shows per-capita income tax payment in each year based on the assumption that the demographics in 2009 will stay unchanged and the assumption that the demographics will change. For example, in 2060, the increase in social security payments will have the effect of reducing per-capita income tax payment by 5,922 compared with 2009, while the demographic change will have the effect of reducing it by 4,825. The latter figure presumably reflects the impact of an increase in the ratio of elderly people, who generally pay relatively less income tax. The increase in social security payments will have a greater impact in the earlier part of the period between 2009 and 2060. On the other hand, while the macro-demographic change will lead to an increase in total income tax revenue between 2020 and 2030, it will result in a very steep decline in the revenue later, particularly between 2030 and 2040.
502 K Matsuda, Y Ozeki, K Kikuta, J Ueda / Public Policy Review Table 12 Comparison of Income tax payment per capita (base scenario) V. Calculation under an Alternative Scenario In Sections IV-2 and IV-3, we conducted simulation based on the base scenario assuming that the burden of a future increase in social security payments will be equally borne by individuals enrolled in the same social security program. In this section we conduct simulation based on an alternative scenario that takes into consideration the fiscal situation of each social security program and members payment ability when allocating the necessary increase in health and long-term care insurance payments to individuals. Specifically, the alternative scenario assumes that for individuals enrolled in the latterstage elderly health insurance program and the national health insurance program, insurance payments will stay constant into the future, with the ensuing shortfall in insurance revenue to be covered by individuals enrolled in mutual aid associations and health insurance societies. This assumption is based on the thinking that it may be realistically difficult to increase the insurance premium rates for individuals enrolled in the latter-stage elderly health insurance program and the national health insurance program, whose insurance payment ability is deemed to be relatively low. As for long-term care insurance, the alternative scenario assumes that insurance payments by Type 1 insured persons (people aged 65 or older) will stay constant, with the ensuing shortfall in insurance revenue to be covered by Type 2 insured
Policy Research Institute, Ministry of Finance, Japan, Public Policy Review, Vol.10, No.3, October 2014 503 persons, who are working-age people. In addition to categorizing individuals by type of insurance program, we also pay attention to the presence or absence of income tax payment as a benchmark of individuals payment ability. The alternative scenario assumes that social security payments by individuals who did not pay income tax will stay constant into the future, with the ensuing shortfall in social security revenue to be covered by individuals who paid income tax in the same year. Based on the above assumptions, we conduct simulation similar to the one conducted in Sections IV-2 and IV-3. When calculating the amount of income tax revenue that reflects macro-demographic changes, we adjust the rate of increase in each of health care, long-term care and pension insurance payments so as to match the total amount of macro-level insurance payments with the total amount under the base scenario 16. Tables 13-1 and 13-2 show the rates of increase in health care, long-term care and pension insurance payments under the alternative scenario. Tables 14 to 17 show the results of the calculation of the amounts of macro-level income tax revenue in each selected year in the future based on the rates of increase in insurance payments under the alternative scenario. Table 13-1 Rate of increase in real insurance payment (health insurance) per capita (alternative scenario) Table 13-2 Rate of increase in real insurance payment (long-term care insurance) per capita (alternative scenario) 16 As for the specific adjustment method, see Supplementary Explanation B.