policy options for financing the future health and … › p › seido › output › ...government...

Policy Options for Financing the Future Health and Long-term Care Costs in

Japan*

Tadashi Fukui** and Yasushi Iwamoto***

October 2005

Abstract

As the Japanese population structure changes, health care and long-term care costs will steadily increase. The current style of financing (pay-as-you-go) will create a large increase in future burden and intergenerational inequity of burdens. This paper studies an alternative policy that pre-funds the benefits for the elderly.

During a transition process, the scheme keeps a higher contribution rate to accumulate enough funds. Under our baseline scenario, the sum of contribution rates of health insurance and long-term care insurance increases from 5.06 percent of earnings to 12.41 percent. The rate of increase in total burdens including taxes devoted for subsidies is 63 percent.

Our sensitivity analysis has shown that quantitative implications depend on the settings of social costs, the labor force, and the interest rate. The setting of labor force does not affect very much. The assumption of social costs has a significant impact.

Even under the most optimistic scenario, the rate of increase in total burdens is 34 percent. Although we cannot tell the exact number of necessary contribution rate that is able to transfer the funded system, what is sure is that the significant increase in contribution rate is needed. JEL classification numbers: H55, I10 Keywords: social insurance; pre-funding; health insurance; long-term care; long-term care

* An earlier version of this paper was prepared for the meeting of the ESRI International Collaboration Projects, February 15 and 16, Tokyo, and the 16th East Asian Seminar on Economics, June 23-25, Manila. We would like to thank Nadarajan Chetty, Olivia Mitchell and Epictetus Patalinghug and participants for helpful comments at the meetings. ** Associate Professor, Kyoto Sangyo University *** Corresponding author: Professor, University of Tokyo. [email protected].

1

1. Introduction

The 2003 Annual Report on Japanese Economy and Public Finance (the Japanese Cabinet

Office) wrote “the sustainability of fiscal and the social security systems goes unsure and the

early accomplishment of bold institutional reforms is requested,” and listed for the first time the

social security reforms preparing for aging population not as a problem in the future but as an

urgent issue.

In the public pension reform of 2004, Japan has cut the future pension benefit substantially.

Since the aggregate pension benefit per national income is designed to be stable in the future, the

financial problem of the public pension is resolved to some extent. A remaining problem in the

public pension program is that the intergenerational inequality is not reduced, because the

primary effect of the reform was to cut the future generation’s benefit and contribution.

The health insurance and long-term care insurance are expected to increase their spending,

too. It is not as easy to cut these benefits as pension benefits, however, because health care and

long-term care services are basic services for people’s life. As shown in Bradford and Max

(1997) and Cutler and Sheiner (2000), an unfunded health insurance creates non-negligible

intergenerational income redistribution. When these costs are financed with a current scheme

(essentially, pay-as-you-go system), the large increase in future burden will be necessary, thus

creating intergenerational imbalance of burden. A way of restoring the equity is for the social

insurance program to pre-fund the future rising costs. In this paper, we consider the effect of

introducing the pre-funding scheme on the intergenerational distribution of burdens. From the

following three points, this paper gives new insights on the future financial problems of social

insurance programs.

First, this paper treats both health care and long-term care. Feldstein (1999) proposed a

pre-funding of Medicare program, and Suzuki (2000) conducted a simulation study of

pre-funding Japanese health insurance program. This paper also considers long-term care

insurance, which was introduced in 2000. While annual spending of long-term care is smaller

than health expenditure, long-term care costs concentrate heavily on the older period. Therefore,

pre-funding this cost requires a large amount of saving. Taking account of both health and

long-term care insurance is a necessary task to grasp the future financial problem of social

insurance program.

2

Our analysis of intergenerational inequity of burdens has a similar spirit with generational

accounting pioneered by Auerbach, Gokhale and Kotlikoff (1991). Since we focus on the way of

financing health and long-term care costs, however, the analytical framework is somewhat

different from the generational accounting. We specify the future policy variables and calculate

burdens under the specified policy scenario. Our focus is confined in health and long-term care

insurances, while the generational accounting considers a broader range of governmental

programs. We instead provide a detailed discussion of those programs.

Second, our simulation has a long time horizon until the 2100 Fiscal Year, while the

government projects the social security costs for health care and long-term care until the FY2025.

The tradition of governmental projections was established, when the aging process was expected

to reach the peak at that time because most of baby boomers will end their lifetime. However,

due to the decline of fertility rate after late 1970s, the aging is now expected to continue after

FY2025.

Third, the robustness of future projection is examined. The main alternative scenario here

assumes that all the present factors but the population structure will sustain in the future, and

projects the labor force. This mechanical projection procedure is applied to economic growth, the

health care expenditure, and the long-term care costs mechanically, instead of constructing a

sophisticated model for forecasting. A virtue of the mechanical projection is its easiness to grasp

the relations among variables. Comparing the projected values with the government’s forecasts

can clarify how projections are influenced by assumptions that the government employs.

The organization of the paper is as follows. Section 2 describes the current problem for the

budget of Japanese social security programs. We point out the growth of health and long-term

care costs is the most serious problem for the sustainability of the government budget.

Section 3 focuses on the projections of labor force participation. The government

projection is optimistic, because it assumes that more elderly and female will participate in the

labor market. We provide an alternative projection that does not expect the increasing labor force

participation. Although the estimated labor force depends on the assumption of labor supply

behavior, the trend decline of total labor force will be significant in any scenarios.

Section 4 is concerned with the health care expenditure and the long-term care service

costs. We project these costs until FY2100 under three scenarios. Even under the most optimistic

scenario that assumes the per capita cost will not grow, the costs per income will steadily

3

increase until 2060s.

Section 5 conducts some policy simulations that consider how to finance the future health

and long-term care costs. We project these costs from FY2004 to FY2100. It is shown that the

balanced budget operation of health and long-term care insurances will create a large inequity of

burdens among generations. Raising the premium immediately and pre-funding for the future

rising costs will help to equate the burdens of generations. However, in our baseline scenario, the

total burden has to be raised by around 60 percent immediately.

We give some concluding remarks in Section 6.

4

2. Long-term Projection of Social Security Benefits

The Ministry of Health, Labor and Welfare (MHLW) occasionally publishes “The

Perspective of the Benefit and Burden of the Social Security System” (Shakai-hoshou no Kyufu

to Futan no Mitooshi), which projects the benefit and burden of public pension, health care, and

social works including long-term care. According to the projection made in May 2004 (we call it

the MHLW 2004 projection hereafter), total social security benefits are expected to grow from

23.5 percent of National Income at factor cost to 29 percent between FY2004 and FY2025 (see

Table 1).1

Table 1: Projection of Social Security Benefits and Burdens

Fiscal Year2004 2010 2015 2025

Social security benefits 23.5 25.4 27.0 29.0 Public Pension 12.6 12.8 12.9 12.2 Health care 7.1 8.2 9.2 11.2 Welfare etc. 3.8 4.3 4.7 5.7 Long-term care 1.4 2.2 2.7 3.6Social security burdens 21.3 24.2 26.6 29.5 Social insurance premium 14.2 15.5 16.7 18.3 Government subsidies 7.1 8.7 9.6 11.2

Note) Numbers are a percent of National Income at factor cost.Source) The Perspective of Benefit and Burden of Social Security System ,Ministry of Health, Labor and Welfare, May 2004.

The future trend varies with programs. Public pension benefit will not grow and actually

become slightly lower in FY2025 than the current level. This stabilization of future pension

benefit was achieved by the pension reform of 2004. Before this reform, the contribution rate of

the pension program for private employees was planned to gradually increase from 13.58 percent

1 The ratio to NI at factor cost shows a meaningless fluctuation when the VAT tax rate changes. The Japanese government had started to use the ratio to NI long before the VAT was introduced. In this paper, we will focus on the ratio to GDP.

5

of earnings to 23.1 percent in FY2020. Since this number seemed unacceptable, controlling the

contribution rate became a major target of the reform. The final contribution rate was reduced to

18.3 percent, and the significant cut of benefits in the future was implemented. However, this

reform does not solve all of major problems. As shown in Maekawa (2004), since the benefit will

be cut gradually, it fails to remedy the intergenerational imbalance of net benefits. The immediate

cut of benefit and increase in contribution should have been necessary to reduce the

intergenerational inequity.

The growing component of social costs is now benefits of health insurance and long-term

care insurance. Health care benefit as a ratio to NI will increase by 4.1 percentage points until

FY2025, and the long-term care benefit will grow by about 2.2 percentage points. How to

finance these costs is now a tough issue for policy makers.

It would be necessary for arguing the sustainability of social security system to make

long-term forecasts of the economy and the population during 50 years or longer. However, it is

impossible to predict such a long term exactly, because of uncertainty over future circumstances.

For instance, the government population projections have not been able to foresee the decline of

fertility that have sustained in the last three decades. The actual decline has been more drastic

than the projected. The unexpected deviation affects the design of public pension system heavily.

Therefore, an examination of the detail of assumptions underlying the projection is in order.

When the health and long-term care insurances are operated as the pay-as-you-go system,

the burden of workers will steadily increase. It will create intergenerational imbalance of burdens.

Pre-funding future social security costs will help to reduce this imbalance by forcing the current

generation to pay more and reducing the the extra burden of future generations. The financial

condition of pre-funding system relies on future economic variables like the interest rate, the

wage growth rate, and the technological change of medical treatments. The reliability of a future

projection is a crucial factor of the success of pre-funding scheme. We will examine how the

pre-funding system is affected by the future conditions of the Japanese economy.

6

3. Economic Growth and Labor Force

This section and next section will discuss the properties of governmental projections, and

provide our alternative projections on labor force, economic growth, health care cost and

long-term care cost. This section focuses on variables that determine income.

“The Perspective of the Benefit and Burden of Social Security” does not rely on a general

equilibrium model. It assumes the growth rate of real national income is the sum of the growth

rates of wage and labor force. This setting can be justified in the following situation. A

production function is homogeneous of degree one with capital K and labor L, and has a

labor-augmenting technological progress. It is written as

( )ALKFY ,= . (1) Here, Y and A represent the output and the efficiency of labor respectively. Differentiating (1)

with respect to time yields

LL

AA

LL

AA

KK

YKF

YY K

&&&&&&++

−−= . (2)

When the growth rates of capital and efficiency unit of labor are the same, the first term of

the RHS of (2) becomes zero. The economic growth rate is then the sum of the growth rate of

labor-augmenting technological change (the wage growth rate) and the growth rate of labor

input.

In the MHLW 2004 projection, the growth rate of labor input is assumed to be 0.1 percent

until FY2008, -0.2 percent in FY2009 and FY2010, and -0.5 percent after FY2011. The labor

force in FY2025 will be 61.58 million, which is smaller by 7.24 percent than the 2004 level. This

setting is based on a detailed projection of labor force, “A Forecast of Labor Force Participation

Rates” (Roudouryoku-ritsu no Mitooshi), which was compiled by the Employment Security

Bureau of MHLW in May 2002 (We call it the MHLW 2002 labor force projection). The lower

part of Table 2 shows that the MHLW 2002 labor force projection expects an increase in the

labor force participation rates of the elderly and women. The upper part of Table 2 shows labor

forces, which we calculate by multiplying the labor force participation rates reported by the

Employment Security Bureau and the official population projection (Medium variant) for each

7

age group. The calculated labor forces in 2025 are 62.97 million people, which is slightly larger

than the MHLW 2004 projection. The initial point of MHLW 2002 labor force projection is 2000,

when the total labor force is 67.66 million. Since the labor force dropped to 66.42 million in

2004, the MHLW 2004 projection starts a smaller initial point and adopts a similar growth rate,

thus reaching a smaller labor force in the future.

Table 2: The Projection of the Labor Force by MHLW

Male Female

Year 2000 2025 2000 2025

Labor force 40,140 36,310 27,520 26,655(thousands)

(percent)15-19 18.4 20.1 16.6 17.820-24 72.7 77.6 72.7 73.725-29 95.8 95.9 69.9 75.330-34 97.7 97.6 57.1 65.035-39 97.8 97.8 61.4 67.440-44 97.7 97.8 69.3 75.245-49 97.3 97.5 71.8 77.050-54 96.7 96.9 68.2 73.555-59 94.2 94.4 58.7 67.560-64 72.6 85.0 39.5 60.565- 34.1 29.5 14.4 13.0

Labor force participationRates by age group

Note: The 2025 labor force is calculated by multiplying the projected population(Medium variant) reported in the Population Projections for Japan (January2002, National Institute of Population and Social Security Research) by the laborforce participation rates by age groups.Source: The 2025 labor force are reported in the Perspective of the Labor ForceParticipation Rate (July 2002, the Employment Security Bureau, Ministry ofHealth, Labor and Welfare) and the 2000 labor force and labor forceparticipation rate reported in the Labor Force Survey (the Statistics Bureau,Ministry of Internal Affairs and Communications.)

While the government projection assumes the future increase in the labor force

8

participation, we may worry about what will happen if it does not realize. To examine the impact

of projecting the future labor supply optimistically, we calculate an alternative scenario.2 We

used the labor force participation rates by age from the Population Census 2000 and the future

population from “the Population Projections for Japan: 2001-2050” (January 2001, National

Institute of Population and Social Security Research). The Population Census reports the labor

force participation rates of every single age to 84 year-old (85 year-old and over are en bloc). We

rescaled the labor force participation rates by age groups so that our estimates of the total labor

force in 2000 matches with the values reported in the 2004 Labor Force Survey.3

Table 3 shows our mechanical projection. The labor force will decrease by 7.99 million

from 2000 to 2025 to be 58.39 million persons. A decrease in the labor force will continue after

2025 and every decade will lose more than five million labor forces. In 2050, the total labor

forces will be 44.69 million people, which is less by 21.7 million and the rate of change from

2004 is 32.7 percent. When the population projection is based on the Low Variant, the projected

number becomes more pessimistic. In 2050, the rate of change from 2004 becomes 38 percent.

Since the effect of low birth rate appears after newborn generations start to work, the labor force

in 2025 does not change very much.

2 Iwamoto (1998) surveyed the existing studies on the labor force, and also projected the labor force until 2020. His mechanical projection assumes that the present labor force participation rates by age group will sustain in the future and that the future population structure will change. Iwamoto’s (1998) estimates are not quite different from preceding estimates. One reason is that the labor force participation rates of working-age population do not have a room for changing drastically, and another reason is that the future population of now existing generations can be predicted considerably surely, although the trend of the birth rate in the future is uncertain. 3 The Population Census reports the labor force in 2000 to be 66.1 million, while the Labor Force Survey reports it to be 67.66 million.

9

Table 3: Mechanical Projection of the Labor Force

（thousand person)

Labor Force the Rate ofChange

the Rate ofChange

Year 2004 2010 2020 2025 2030 2040 2050 (percent) (percent)

Medium variant (baseline)

Both sexes 66,390 64,432 60,353 58,393 56,000 50,128 44,687 -7,997 -12.0 -21,703 -32.7

Males 39,040 38,059 35,580 34,449 33,134 29,783 26,550 -4,591 -11.8 -12,490 -32.0

Females 27,350 26,373 24,774 23,945 22,866 20,345 18,138 -3,405 -12.5 -9,212 -33.7

Low variant

Both sexes 66,390 64,432 60,331 58,172 55,306 48,080 41,190 -8,218 -12.4 -25,200 -38.0

Males 39,040 38,059 35,568 34,336 32,769 28,626 24,520 -4,704 -12.1 -14,520 -37.2

Females 27,350 26,373 24,763 23,837 22,537 19,454 16,670 -3,513 -12.8 -10,680 -39.0

the Changefrom 2004

to 2025

the Changefrom 2004

to 2050

Note: Author's calculation using the labor force participation rates by age groups reported in the 2000 Labor Force Survey and the 2000Population Census (Ministry of Internal Affairs and Communiations) and the Population Projections for Japan: 2001-2050, January 2001(National Institute of Population and Social Security Research) on the assumption that 2000 labor force participation rates will sustain.

10

The MHLW 2004 projection regarded the labor force as labor input. Since the efficiency

varies with ages, however, the labor force and labor input may move differently. The procedure

of the MHLW 2004 projection can be refined. We estimated an efficiency unit of labor implied

by the two MHLW projections, assuming that the efficiency is proportional to wage. The

age-wage profile by age group and sex was taken from published cross-tables of the 2003 Basic

Survey on Wage Structure (Chingin Kozo Kihon Chosa, MHLW). For each age, the efficiency

unit of labor input is calculated as a product of the total wage per worker and projected labor

forces. We rescaled the aggregated efficiency unit in 2004 so that it equals the total labor forces

in that year. We considered three scenarios; our mechanical projection (pessimistic), MHLW

2004 projection (optimistic), and MHLW 2002 projection (more optimistic).

Figure 1 depicts the three scenarios of labor input. In the MHLW 2002 labor force

projection,. the labor input in 2025 will be 63.02 million, according to our normalization

described above. The counterpart of our mechanical projection is 59.03 million. The MHLW

2004 projection underlying the social security cost projection lies between them. While the

difference among projection is significant, all of them project a further decline after 2025. In

2100, the labor input will be 30.08 million in the most optimistic case and 28.08 million in the

most pessimistic case. The bottom line is that even in the most optimistic scenario, the large

decline of labor force will be unavoidable.

11

Figure 1: Comparison of Three Labor Input Projections

20,000

30,000

40,000

50,000

60,000

70,000

2004 2014 2024 2034 2044 2054 2064 2074 2084 2094

Year

Effic

ienc

y un

it of

labo

r

very optimistic optimistic pessimistic

12

4. Long-Term Projections of Health and Long-term Care Benefits

The MHLW projection assumes health care and long-term care will grow more rapidly

than income. If our mechanical projection procedure is applied both costs, the estimates will be

smaller than the official projection. To clarify the stance of the official projection, this section

conducts a sensitivity analysis employing alternative settings of key variables. Combining two

scenarios of health and long-term care costs with three scenarios of labor forces, we add six

cases of projections. The following two subsections describe a procedure of projecting the two

costs. We then present results of sensitivity analysis.

4.1 Health care cost

Table 4 compares five recent projections about health care and long-term care. Since

projections in different years assume different inflation rates, comparing nominal variables is

misleading. When we look at the ratios to National Income, projected health expenditures differ

mildly.

13

Table 4: The Perspectives of Health Care and Long-term Care Costs Reported by Ministry of Health, Labor and Welfare (trillion yen)

Projecteddate

Projected inMarch 1994

Projected in November 1996 Projected inSeptember

1997

Projected inOctober 2000

Projected inMay 2002

Projected inMay 2004

Fiscal Year

Introduction ofthe Long-TermCare Insurance

Introduction ofthe Long-TermCare Insurance

Health Care Costs1993 241995 24 24 242000 38 262010 68 35 342025 141 107 - 108 96 90 71 60 59

(11 - 19） (11.5 - 18) (10 - 16) (10 - 15) (11) (11) (11)Long-Term Care Service Costs

2005 62010 8 92015 122025 13 - 20 14 - 21 21 20 19

(2) (2.5) (3) (3.5) (3.5)

Note: Numbers are benefits of each social insurance. The Numbers in parentheses are the ratio to National Income. Source: Welfare Vision for the 21st Century, Ministry of Health and Welfare, May 1994 and others.

- 14 -

The future health expenditures are projected by extrapolating the most recent actual values

of nominal health expenditure, which do not synchronize with the setting of the inflation rate and

the economic growth rate. The extrapolation of nominal health expenditure without any

reasonable link to inflation can be problematic from the viewpoint of economic theory.

Economists usually make projections based on real values, and their methodologies belong to a

family of mechanical projection. Iwamoto (2004) surveyed the several projections made by the

existing studies such as Ogura and Irifune (1990), Ogura (1995), Niki (1995), and Iwamoto,

Takeshita and Bessho (1997), Nishimura (1997), and Tokita et al. (1997), and concluded that the

national health expenditure would become about 1.4 times as big as the level of 2000 in the

following 30 years. Our mechanical projection follows the method established by existing

studies and assumes that the per capita health expenditure by age group reported in “National

Medical Expenditure (FY2003)” of MHLW (shown in Table 5) will sustain in the future.

- 15 -

Table 5: The Expenditures for Health Care and Long-Term Careper Capita by Age Group (2004 Fiscal Year)

(yen)

Age Group Health careLong-term

care

0-4 158,9005-9 92,10010-14 66,70015-19 58,10020-24 71,20025-29 85,60030-34 97,50035-39 103,00040-44 119,100 5,700 (40-64)45-49 149,90050-54 199,20055-59 245,00060-64 323,10065-69 433,400 43,80070-74 564,800 97,00075-79 749,400 (75-) 203,20080-84 429,40085-89 799,90090-94 1,236,10095- 1,786,500

Note: The values are the sum of benefit from insurer and co-payment of patients.Health care expenditure is calculated by proportionally adjustingthe FY2002 value reported in National Health Expenditure(MHLW) so that the national aggregate matches the FY healthcare expenditure reported by MEDIAS, which totals payments ofthe public health insurance. National Health Expenditurecategorizes those who are 75 year-old and over as one age group.Long-term care expenditure: the values are calculated bymultiplying 12 by the actual benefit for service in October 2004reported in Monthly Report of Long-Term Care Benefit(MHLW). Those who are 40 year-old and over can be therecipient of long-term care insurance. The Report categorizesthose who are from 40 to 64 year-old as one age group.

Since we want to make our mechanical projection comparable with the MHLW 2004

- 16 -

projection, the data by age group were converted to the FY2004 income level. First, per capita

health care cost by age group was proportionally adjusted so that the national aggregate matches

the numbers reported in MEDIAS (Medical Information Analysis System), which reports health

care costs paid by public health insurance. The national aggregate of health care costs was

calculated as a product of population and per capita cost by age group. The adjusted age-cost

profile was used to project the future health care cost.

We also decomposed the total cost to social security payment and out-of-pocket expenses.

Unfortunately, MEDIAS does not provide this decomposition. We obtained the social security

payment from National Medical Expenditure (Kokumin Iryohi).4 In allocating social insurance

payment among generations, we use statutory coinsurance rates in 2004 and assume that they

will sustain in the future. Since April 2003, the coinsurance rate has been 20 percent for 0-2

year-old, 30 percent for 3-69 year-old. For 70 year-old and more, the rate is 10 percent in

principle (20 percent is applied to high-income earners). Due to the lack of data, we assume that

10 percent is applied to all of those aged 70 and over. Actual out-of-pocket payments are less

than statutory coinsurance rate, because there is a stop-loss rule and other schemes to reduce

out-of-pocket expenses. We proportionally scaled the statutory coinsurance rates so that the sum

of estimated social insurance payments match the aggregate value reported by MEDIAS in 2004.

Since available health expenditure data do not separately report the spending of age 0-2

and age 3-5, we assume health expenditure per capita will be uniform in this age group, and then

calculate the average coinsurance rate. Since actual costs concentrate on newborn babies, this

procedure may slightly overestimate the true social security payment. At the same time, many

municipal governments offer extra benefits to health expenditure of infants from their general

budget. Since we do not incorporate these subsidies into our estimation, it results in

underestimation of social security benefits. The overall impact on an estimation bias is

ambiguous.

We then reproduced the MHLW 2004 projection of the future health insurance benefits.

The MHLW 2004 projection assumes that per capita health care cost for the non-elderly (under

4 The release of National Medical Expenditure is roughly one year behind of MEDIAS. We first estimated 2004 data of NME by multiplying the growth rate of comparable variable of MEDIAS between 2003 and 2004 to the 2003 data of National Medical Expenditure. Since out-of-pocket expenses of National Medical Expenditure include payments of medical treatments not covered by social security programs, we estimated the out-of-pocket payments as a residual.

- 17 -

65) will grow at the wage growth rate and that per capita cost for the elderly aged 65 and over

will grow faster by 1.1 percentage points than wage. The projection on the non-elderly is the

same as our mechanical projection. For the costs beyond 2025, we need to specify its growth rate.

If health care cost continuously grows faster than wage, it will ultimately exhaust all of resources

available to consumers. We thus need to assume the growth will stop sometime in the future. We

consider two cases in which the faster growth will stop in FY2025 or FY2100. In the former case,

per capita health care cost is 5.56 times as big as that of non-elderly. A comparable ratio in 2004

is 4.3. In the latter case, the growth of health care cost for the elderly is extreme; the ratio in

FY2100 will be 13.09.

As a measure of financial burden, we focus on the ratio of social security benefits to

compensation of employees and mixed income in terms of national accounts. The denominator

of the ratio is assumed to be proportional to the amount of labor input. There are three settings of

health care cost; (1) the per-capita cost of the elderly grows at the wage growth rate (optimistic),

(2) it will grow faster by 1.1 percentage point than wage until FY2025 (pessimistic), and (3) it

will grow faster by 1.1 percentage point than wage until FY2100 (very pessimistic). Combining

them with three scenarios of labor input yields nine cases of future projection. An upper part of

Table 6 shows these ratios.

The MHLW 2004 projection shows that the ratio of health expenditure to the labor input in

FY2025 will become 1.58 times as big as the level of FY2004. This scenario was reproduced in

our social cost-pessimistic labor force-optimistic scenario, which indicates the ratio will grow by

a factor of 1.56 between FY2004 and FY2025. On the other hand, the ratio under the mechanical

projection will grow by a factor of 1.38. Therefore, the MHLW 2004 projection stands on more

pessimistic prospects of a future increase in health expenditure than our mechanical projection.

The most pessimistic scenario is a combination of very pessimistic projection of health care cost

and mechanical projection of labor force. In this case, the ratio in FY2025 will be 1.63 times of

FY2004 level.

A noticeable point is that the ratio will keep increasing afterwards, and the ratio to the

FY2004 level under the setting employed by the governmental projection will be 1.98 times in

FY2050. In other words, FY2025 is not the terminal point for thinking about the sustainability of

- 18 -

the health insurance system.5

5 We should note that our projections may be biased upward, because they do not take into account the effect of aging on the terminal care expense. In Japan, Suzuki and Suzuki (2001) and Ohkusa (2002a) provided a modified projection that took account of this point. Since the terminal care expense for more aged becomes lower, a longer longevity will shift the average age of death to further elder so that the future health expenditure by age group is expected to be lower than the present level. Taking this effect into consideration, the above-mentioned studies pointed out that the MHLW 2004 projection overestimated the future health expenditure. Suzuki and Suzuki (2001) reported it amounted to about 4.4 percent of the health expenditures for the elderly. On the other hand, Ohkusa (2002a) concluded that the overestimation reached about from 15 to 30 percent of the total health expenditures. The divergence in their estimates is quite large, because the data limitation made it difficult for them to separate medical expenditures of survivors and those who died.

- 19 -

Table 6: Projection of the ratio of Health and Long-term Care Costs to Labor Input

2010 2015 2025 2030 2040 2050 2100

Health caresocial cost: optimisticlabor force: very optimistic 1.08 1.16 1.29 1.35 1.47 1.62 1.55

optimistic 1.09 1.18 1.32 1.38 1.51 1.66 1.59pessimistic1) 1.12 1.22 1.38 1.44 1.59 1.74 1.66

social cost: pessimisticlabor force: very optimistic 1.12 1.26 1.52 1.59 1.75 1.93 1.84

optimistic2) 1.13 1.28 1.56 1.62 1.79 1.98 1.89pessimistic 1.16 1.32 1.63 1.70 1.88 2.07 1.98

social cost: very pessimisticlabor force: very optimistic 1.12 1.26 1.52 1.65 1.99 2.41 3.66

optimistic 1.13 1.28 1.56 1.69 2.04 2.47 3.75pessimistic 1.16 1.32 1.63 1.77 2.14 2.59 3.92

Long-term caresocial cost: optimisticlabor force: very optimistic 1.25 1.52 2.03 2.31 2.81 3.14 3.31

optimistic 1.26 1.54 2.08 2.36 2.88 3.21 3.39pessimistic1) 1.29 1.59 2.17 2.47 3.02 3.37 3.55

social cost: pessimisticlabor force: very optimistic 1.34 1.73 2.61 2.97 3.61 4.03 4.25

optimistic2) 1.35 1.76 2.67 3.03 3.70 4.13 4.35pessimistic 1.39 1.81 2.79 3.18 3.88 4.33 4.56

social cost: very pessimisticlabor force: very optimistic 1.34 1.73 2.61 3.15 4.32 5.43 10.41

optimistic 1.35 1.76 2.67 3.22 4.42 5.56 10.65pessimistic 1.39 1.81 2.79 3.37 4.64 5.83 11.15

Notes: Numbers are the ratio of health care cost and long-term care cost to labor input (FY 2004 = 1).1) Mechanical projection. The furure labor force participation rate and per capita social cost by age groupwill be constant.2) A projection that replicate the MHLW 2004 projection, which is reported by Table 1.Source: Authors' calculation.

4.2 The long-term care cost

The government expects long-term care cost will grow more rapidly than health care cost.

According to the MHLW 2004 projection in Table 4, the long-term care insurance benefit in

FY2025 (the ratio to National Income) will be 2.7 times as much as that in FY2004. The growth

rate of per capita long-term care cost is set to be higher than the wage growth rate, although the

detailed assumption was not explicitly documented.

The municipal governments prepare a three-year business plan for the long-term care

insurance. Before the first revision was made in FY2003, the Projection of Demand for

- 20 -

Long-term Care (Kaigo-Sahbisu-Ryou-tou no Mitooshi, June 2002) were provided by MHLW,

and expected the increased care in the following five years. For example, the provisional

nationwide sum of usage of home-visit service will increase by 39.3 percent from FY2003

(142,194 visits) to FY2007 (198,033 visits).6 The projection did not provide the total amount of

long-term care costs. We calculate the sum of the provisions weighted by the actual costs in

FY2003, and the resulting growth during FY2003-2007 amounted to 26.4 percent.

We projected the future long-term care insurance benefits with a similar methodology as

health insurance. Since the most recent data for annual spending by age group is for FY2002 at

the writing of this paper, we estimated annual spending from monthly data. The long-term care

expenditure by age group in October 2004 was calculated from “Monthly Report of Long-Term

Care Benefits Survey (Kaigo Kyufuhi Jittai Chosa Geppo)” (MHLW).7 We obtained annual

spending by multiplying it by 12, and report it in the right column in Table 5. The future

long-term care expenditure was then projected under three settings. The first one (optimistic)

assumed that the age-expenditure profile would not change and that only the population structure

would change. We then decomposed total spending into social security benefit and out-of-pocket

expense by assuming the ratio of out-of-pocket payment to total costs in FY2002 (8.99 percent8)

will sustain.

We calculated three scenarios of long-term care cost; (1) the per-capita cost of each age

group would grow at the wage growth rate (optimistic), (2) it would grow faster by 1.2

percentage points than wage until 2025 so that the total cost per income well approximated the

MHLW 2004 projection (pessimistic), and (3) it would grow faster by 1.2 percentage points than

wage until 2100 (very pessimistic). Properties of these settings parallel to those of health care

costs. Combining them with three projections of labor force yields nine scenarios. The resulting

ratios are reported in a lower part of Table 6.

The growth factor from FY2004 to FY2025 ranges between 2.03 and 2.79. Under the

6 This projection presumed that the number of the certified as the care required in FY2003 was 3,279,000, whereas the actual number of the certified was 2,983,000 at the end of FY2003. 7 The data source does not directly report cost by age group. The available cross tables are (1) cost by severeness of disability and (2) the numbers of beneficiaries by severeness of disability and age group. Assuming long-term care cost of each severeness of disability was the same across age groups, we estimated cost by age group with these two cross tables. 8 It is slightly lower than the statutory coinsurance rate (10 percent), because there are some measures that lightens burden of out-of-pocket payments.

- 21 -

governmental projection, it is 2.67, which is relatively pessimistic. If the per long-term care

grows at the wage growth rate, the growth factor is between 2.03 and 2.17 depending on the

setting of labor force. The future size depends on whether the demand for service quantity

increases or not.

- 22 -

5. Simulation of Health care and Long-Term Care Insurance Policy

5.1 Procedures of the simulation

This section simulates policies that finance future health and long-term care insurance

benefits projected in the previous section. The effect on fiscal balances and burdens on

generations are focused on.

Since we regard benefits of these insurances as necessary services when a person becomes

sick or disabled, it is not very meaningful to discuss the intergenerational distribution of benefits.

Our simulation focuses only on the financing side.

From October 2007, half of benefits of the health insurance for those aged 75 and over will

be financed with government subsidies, which is in turn financed by general tax revenues. Until

October 2002, 30 percent of benefits for those aged 70 and over was financed with government

subsidies. From 2002 to 2007, a transition process gradually raises the eligible age and the share

of government subsidies. Half of long-term care insurance benefit is financed with government

subsidies. Other major subsidies are 50 percent of benefits of National Health Insurance

(Kokuho) and 13 percent of the Government-managed Health Insurance for Employees (Seikan

Kenpo).

We assume that the social insurance premium and taxes devoted for social insurance

benefits are paid from the same income base, which is compensation of employees and mixed

income in terms of national accounts. For simplicity, we also assume that these incomes will

grow at the same rate as GDP (and labor input) after FY2004 and that there is no administrative

cost in social insurance programs.9

The initial year and the terminal year of the simulation were set FY2004 and FY2100

respectively, because the population projection by the National Institute of Population and Social

Security Research is available within these periods.

We consider two policies:

Policy A: Balanced budget operation in which the benefit of each year is financed by taxes

and insurance premiums of each year

9 For instance, the administrative cost of Society-Managed Health Insurance for Employees is actually about 4 percent of total benefits in FY2001.

- 23 -

Policy B: Attempt to reduce the intergenerational inequity of burdens by pre-funding the

future social insurance payments for the elderly (details will be described in Section 5.3)

5.2 Balanced budget

We define a burden rate as the ratio of burdens (the sum of insurance premiums and

government subsidies financed by taxes) to 90 percent of the sum of compensation of employees

and mixed income. Rescaling the sum of compensation of employees and mixed income aims to

calibrate the statutory premium rate. The health insurance premium (including government

subsidies for the non-elderly and excluding those for the elderly) in the initial year, FY2004, is

calculated as 8.21 percent. The actual health insurance premium for the enrollees of

Government-managed Health Insurance for Employees is 8.2 percent. Since enrollees of the

Government-managed Health Insurance for Employees work for small companies, their average

salary is lower than that of the whole workers. Since the government subsidy aims to offset the

earnings difference, we tried to calibrate the statutory insurance premium. For the long-term care

insurance, the simulated premium rate in FY2004 is 1.11 percent, and it matches the actual

insurance premium paid by the enrollees of Government-managed Health Insurance for

Employees. Under the balanced budget, the burden is equal to social insurance benefits

(excluding out-of-pocket payment). Therefore, we actually calculated the ratio of benefit to

incomes.

Figure 2 shows the burden rates for health care, long-term care, and the total of them when

the Policy A is carried out under the governmental projection (social cost-pessimistic and

labor-force optimistic). The burden rate for health care keeps increasing until FY2059 when it

amounts to 19.94 percent. The burden rate for long-term care will amount to 10.97 percent in

FY2066. Although the paths of these two burden rates appear parallel in Figure 2, we should note

that the absolute level of the long-term care cost is low. The burden rate for long-term care grows

much more rapidly than that for health care, because the long-term care benefit concentrates on

more aged population than the health care benefit. For the same reason, a peak of the burden rate

for long-term care will follow a peak of that for health care. A total burden rate of both

insurances reaches a peak that is 30.65 percent in FY2064. The ratio of total burden to GDP then

amounts to 15.63 percent.

- 24 -

Figure 2: The Burden Rates under Policy A (Balanced Budget)

0%

5%

10%

15%

20%

25%

30%

35%

2004 2014 2024 2034 2044 2054 2064 2074 2084 2094

Total costs

Health care costs

Long-term care cost

- 25 -

Figure 3 shows the lifetime burden rate of each generation under the governmental

projection. The interest rate is assumed to be 1 percentage point higher than the wage growth rate.

It is defined as the ratio of lifetime burden to lifetime income. The lifetime variable is the sum of

present discounted value of annual numbers during the period between the initial point of the

simulation and the terminal point. Therefore, the rate is a prospective burden during lifetime for

the generation born in 2001 or earlier. To estimate lifetime income, we used the age-wage profile

which was used to calculate labor input in Section 2.1. The annual discount rate is set 1 percent.

The horizontal axis of Figure 3 represents the birth year of each generation. The lifetime

burden rates are not smooth in early generations, perhaps because our prospective calculation

covers only a short period for them. Figure 3 indicates the balanced budget system will pose

heavier burdens on younger generations. The inequality of burdens is particularly severe for the

current working-age generation. The lifetime burden rate for those who were born in 1945 is 14.8

percent. That for those born in 2001 is 26.2 percent.

- 26 -

Figure 3: Lifetime Burden Rates by Generation under Policy A (Balanced Budget)

10%

12%

14%

16%

18%

20%

22%

24%

26%

28%

1904 1914 1924 1934 1944 1954 1964 1974 1984 1994

birth year

- 27 -

5.3 Equalizing the burden by pre-funding policy

One possible way of avoiding the increasing burden of future generations depicted in

Figure 3 is a policy that levies a constant burden rate over time. That policy intends to charge a

high burden in advance so that it can accumulate enough funds to prepare for increasing costs in

the future. Feldstein (1999) advocated the idea of prefunding the Medicare, which is the US

public health insurance for the elderly.

As a scenario of pre-funding health and long-term care costs, we consider the following

policy. For health insurance, a pre-funding part finances insurance-payments of health care costs

for the elderly (age 65 and over). Workers who are age 15 and over pay premiums. The health

care costs for those aged 64 and under and government subsidies to benefits for the elderly are

financed as a pay-as-you-go system. Long-term care insurance employs a pre-funding scheme,

while government subsidies are financed as a pay-as-you-go system. Since enrollees of the

current system are age 40 and over, we assume workers of this age group pay premiums.

The setting of the interest rate is a key factor in determining the performance of a

pre-funding scheme. When we focus on the proportion of the burdens to income, not the absolute

level but the difference between the interest rate and the wage growth rate matters. The MHLW

projection on public pension finance in May 2004 assumed that the nominal interest rate would

be 3.2 percent and that the growth rate of nominal wage would be 2.1 percent. Our baseline case

sets 1 percentage point difference between the interest rate and the growth rate. As alternative

scenario, the difference is set 0 percent or 2 percent.

Since the current health and long-term care insurance is a pay-as-you-go scheme, a

transition to a funded system should be designed. For the health care costs, we first calculate the

contribution rate that is sufficient enough for the cohort born in FY2001 to finance the expected

value of their lifetime health insurance benefits after 65 years old. This rate is 4.96 percent under

the governmental projection and the baseline scenario of the interest rate. If the cohorts born in

FY2001 and later pay this premium rate, the total accumulated funds in FY2100 amount to

111.11 percent of GDP. The transition process is designed to achieve this level of funds with a

constant premium rate during the transition. Since the existing generations did not pre-fund their

health care costs, 4.96 percent of the premium rate is not enough to hit the target in FY2100. It

turns out that when the contribution rate of 8.52 percent will successfully accumulate the

required funds.

- 28 -

The evolution of the funded system will be achieved in the following way (Table 7

summarizes the numbers that represents the burdens under Policies A and B). When health care

costs (excluding government subsidies) is financed as a pay-as-you-go scheme in FY2004, the

contribution rate for the health care costs for the age 64 and under is 4.32 percent, and that for

the age 65 and over is 3.89 percent. When the transition to a pre-funding scheme starts in

FY2005, the contribution rate for the elderly jumps up to 8.52 percent under the governmental

ptojection, and keeps its level until FY2100. After then, the contribution rate for the elderly will

shrink to 4.96 percent. The contribution rate for health care costs of age 64 and under is almost

stable. It ranges between 3.97 percent and 4.3 percent.

Table 7: Contribution Rates under the PAYG and Pre-funding Schemes

(A) (B)Pay-as-you-go Pre-funding

2004 2005-2100 2005-2100 2101-

Total 12.01 12.28 - 28.12 19.62 - 26.70Health insurance non-elderly 4.32 3.97 - 4.30 elderly 3.89 3.98 - 9.08 8.52 4.96 government subsidies 2.69 2.84 - 10.07Long-term care insurance 1.11 1.16 - 4.83 3.95 2.17

Note) Numbers are percent of earnings (90 percent of compensation ofemployees and mixed income).The interest rate is based on the baseline case,and other parameters are based on the governmental projection.

The transition process of long-term care insurance is designed parallel to the health

insurance. The contribution rate that is sufficient enough for the cohort born in FY2001 to

finance the expected value of their lifetime long-term care insurance benefits is 2.17 percent

under the governmental projection. If the cohorts born in FY2001 and later pay this premium rate,

the total accumulated funds in FY2100 amount to 68.35 percent of GDP. This amount of funds

will be accumulated by the contribution rate of 3.95 percent during the transition. When

long-term care costs (excluding government subsidies) is financed as a pay-as-you-go scheme in

- 29 -

FY2004, the contribution rate is 1.11 percent. When the transition to a pre-funding scheme starts

in FY2005, the contribution rate jumps up to 3.95 percent, and keeps its level until FY2100.

After then, the contribution rate will be 2.17 percent.

For the calibrated value in FY2004, the health insurance payments for the non-elderly is

4.32 percent of earnings, those for the elderly is 3.89 percent, and the long-term care costs is 1.11

percent. When social insurance premiums and government subsidies are combined, the burden

rate becomes 12.01 percent. Under the pay-as-you-go system, the social costs for the elderly will

grow steadily. The total burden ratio will go up to 28.12 percent of earnings. The burden rate of

health insurance payments will go up to 9.08 percent. The growth of government subsidies is

more rapid. Its peak is 10.07 percent.

Under the pre-funding scheme, the burden rate should increase from 12.01 percent in

FY2004 to 19.62 percent in FY2005. The rate of change of total burdens at the initial point is 63

percent. Since the government subsidies will grow as in the case of pay-as-you-go system. The

peak of total burden rate is 26.70 percent in FY2064.

5.4 Sensitivity Analysis

Figure 4 compares the lifetime burden rates between the balanced budget and the

pre-funding scheme. We should note that the burden rates for those born after 2001 does not

cover their whole lifetime, because the simulation terminates in FY2100. When the system is

changed from the balanced budget to the pre-funding, generations who are born before 1997 will

face a higher lifetime burden, and younger generations will benefit from a lower lifetime burden.

The curve of lifetime burden in Figure 4 becomes flatter. The pre-funding scheme helps to

reduce the inequality of burdens. This will be made possible by the currently living generation’s

acceptance of sharing burdens of future generations.

- 30 -

Figure 4: Lifetime Burden Rates under Policy A (Balanced Budget) and Policy B (Pre-funding)

7%

12%

17%

22%

27%

32%

1908 1918 1928 1938 1948 1958 1968 1978 1988 1998 2008 2018 2028 2038 2048 2058 2068 2078

birth year

Balanced budget Pre-funding

- 31 -

As the results of sensitivity analysis, Table 8 reports contribution rates of pre-funding

scheme in FY2005-2100 and after FY2100 under 27 scenarios. The assumption of labor force

participation does not affect the contribution rate very much. Between the pessimistic scenario

and the very optimistic scenario, the difference of the health insurance contribution rate is 0.36 to

0.92 percentage points. The gap is large when the projection of social costs is pessimistic. This

reflects the fact that a funding scheme is not influenced by a demographic change. On the other

hand, the interest rate affects the contribution rate. Under the optimistic scenario (the interest rate

is 2 percentage points larger than the wage growth rate), the contribution rate is 1.5 to 4.68

percentage points lower than the pessimistic scenario. The setting of social costs has a larger

impact on the contribution rate. The difference of a contribution rate in a transition period

between a pessimistic scenario and an optimistic scenario is 3.13 to 6.33 percentage points.

The above findings are applied to the case of long-term care insurance. The effect of the

interest rate is relatively large compared with the size of benefit. Since the long-term care cost

concentrates on the late elderly, the difference of contribution rate is 1.08 to 3.41 percentage

points between the most optimistic scenario and the most pessimistic one.

Even under the most optimistic scenario, the burden rate combined with health care and

long-term care should increase 16.05 percent as the starting point. The rate of change is 34

percent, which is smaller than the governmental projection, but is still non-negligible. Under the

most pessimistic scenario, the necessary rate of increase becomes 85 percent. Although

quantitative implications depend on the settings, a significant increase in contribution rate is

needed to implement a pre-funding scheme.

- 32 -

Table 8: Sensitivity Analysis of Contribution Rates in a Pre-funding Social Insurance Scheme

(Percent)

Health care for the elderly Long-term care

2005-2100 2101- 2005-2100 2101-

interest rate - growth rate = 2 percentsocial cost: optimisticlabor force: very optimistic 6.25 2.74 2.66 1.08

optimistic 6.36 2.81 2.71 1.10pessimistic 6.61 2.90 2.82 1.14

social cost: pessimisticlabor force: very optimistic 7.56 3.45 3.31 1.38


social cost: very pessimisticlabor force: very optimistic 9.38 6.06 4.47 2.81


interest rate - growth rate = 1 percentsocial cost: optimisticlabor force: very optimistic 6.84 3.85 3.08 1.65






interest rate = growth ratesocial cost: optimisticlabor force: very optimistic 7.75 5.36 3.74 2.51






Note) Numbers are percent of earnings (90 percent of compensation of employees andmixed income). In FY2004, the contribution rate is 3.89 percent for health insurance forthe elderly and 1.11 percent for long-term care insurance.

- 33 -

Suzuki (2000) conducted a similar calculation of the transition to the fully-funded health

insurance system. While our calculation unites whole health insurance, his calculation was

decomposed into insurance subsystems. The transition was assumed to start FY1995 and reach a

fully-funded scheme in FY2100. For the Society-Managed Health Insurance for Employees

(Kumiai Kenpo), the contribution rate will increase from 7.8 percent to 9.8 percent.

Under Suziki’s specification, a fully-funded scheme finances individual’s lifetime health

care costs. Before individuals begin to work, their pre-funding account has to borrow money. The

resulting aggregate funds will be lower than in our scheme. This is one reason that we obtained a

much larger hike of the contribution rate during the transition to the pre-funding scheme.

Figure 5 draws total burden rates under the baseline scenario and six alternative scenarios

which take an alternative scenario of one variable among three. Since the baseline scenario takes

a medium assumption about the three variables, it takes the middle of seven lines. The highest

line is the case of the most pessimistic social cost scenario. Since the growth of social costs is

assumed to continue at the end of this century, the burden rate of the future generation is far

larger than in other scenarios. The lowest line is the case of the optimistic social cost scenario.

The social costs have the most significant impact on the burden rate. The impact of the labor

force on the burden rate is the least among the three variables. The higher interest rate can reduce

the burden rates considerably. For the generation born in 2001, the lifetime burden rate is 25.62

percent with the baseline scenario and 20.15 percent with the 1 percentage point higher interest

rate.10

10 Feldstein and Samwick’s (1997) simulation of pre-funding the US social security program assumed much higher interest rate, thus resulting in a smaller contribution rate for pre-funding. Given the recent poor performance of the Japanese asset prices, following their assumption does not look plausible.

- 34 -

Figure 5: Total Burden Rates under Alternative Scenarios

15.00%

20.00%

25.00%

30.00%

35.00%

40.00%

45.00%

50.00%

1909 1919 1929 1939 1949 1959 1969 1979 1989 1999 2009 2019 2029 2039 2049 2059 2069 2079

birth year

baseline social cost: optimistic social cost: very pessimistic interest rate: 2 percentinterest rate: 0 percent labor force: very optimistic labor force: pessimistic

- 35 -

6. Conclusion

As the Japanese population structure changes, the current style of financing

(pay-as-you-go) will create a large increase in future burden. It will also create intergenerational

inequity of burdens. We studied an alternative policy that pre-funds the benefits for the elderly

aged 65 and over. Pre-funding is not very familiar in Japan. Our analysis aims at providing

realistic scenarios of this policy option, to stimulate policy discussions.

During a transition process until FY2100, the scheme keeps a higher contribution rate to

accumulate enough funds. Under the parameters implied by the governmental projection, the

sum of contribution rates of health insurance and long-term care insurance increases from 5.00

percent of earnings to 12.47 percent. The rate of increase in total burdens including taxes devoted

for subsidies is 63 percent.

Our sensitivity analysis has shown that quantitative implications depend on the settings of

social costs, the labor force, and the interest rate. The setting of labor force does not affect very

much. The assumption of social costs has a significant impact. Although we cannot tell the exact

number of necessary contribution rate that is able to transfer the funded system, what is sure is

that the significant increase in contribution rate is needed. Even under the most optimistic

scenario among 27 scenarios, the necessary increase in contribution rates of the two social

insurances is 3.91 percentage points. The rate of increase in total burdens is 34 percent.

Implementing a pre-funding may be a challenge, because the initial increase of burden is

politically unpopular. However, the cost of not introducing this scheme is heavy burden of future

generations. Although we cannot implement an adequate contribution rate, raising the

contribution rate aggressively will help to reduce the intergenerational imbalance of burdens.

- 36 -

References

Auerbach, Alan J., Jagadeesh Gokhale, and Laurence J. Kotlikoff (1991), “Generational

Accounts: A Meaningful Alternative to Deficit Accounting,” in David Bradford ed., Tax

Policy and the Economy, Vol. 5, Cambridge, MA: MIT Press, pp. 55-110.

Bradford, David F., and Derrick A. Max (1997), “Implicit Budget Deficits: The Case of a

mandated Shift to Community-Rated Health Insurance,” in James M. Poterba ed., Tax

Policy and the Economy, Vol. 11, Cambridge, MA: MIT Press, pp. 129-167.

Culter, David M., and Louise Sheiner (2000), “Generational Aspects of Medicare,” American

Economic Review Papers and Proceedings, Vol. 90, No. 2, May, pp. 303-307.

Feldstein, Martin (1999), “Prefunding Medicare,” American Economic Review Papers and

Proceedings, Vol. 89, No. 2, May, pp. 222-227.

, and Andrew Samwick (1997), “The Economics of Prefunding Social Security and

Medicare Benefits,” in Ben S. Bernanke and Julio Rotemberg eds., NBER

Macroeconomics Annual 1997, Cambridge, MA: The MIT Press, pp. 115-148.

Iwamoto, Yasushi (1998), “Japan’s Labor Force in the year 2020,’ Keizai Kenkyu, Vol. 49, N0.4,

October, pp. 297-307 (in Japanese).

, Satoshi Takeshita, and Masashi Bessho (1997), “Iryo Hoken Zaisei to Kouhi Hutan,”

Financial Review, No.43, November, pp. 174-201 (in Japanese).

Maekawa, Satoko (2004), “Changes in Benefits and Burdens, by Generation, Arising from Social

Security Reform,” Financial Review, No. 72, August, pp. 5-19.

Niki, Ryu (1995), Nihon no Iryouhi: Kokusai Hikaku no Kanten Kara, Igaku Syoin (in

Japanese).

Nishimura, Syuzo (1997), “Chouki Tsumitategata Iryou Hoken Seido no Kanousei Nitsuite,”

Iryo Keizai Kenkyu, Vol.4, December, pp.13-34 (in Japanese).

Ogura, Seiritsu (1995), “Koureika no Kosuto: Nippon ni Okeru Kouteki Shikin Kakuho no

Tenbou,” in Noguchi, Yukio and Devid A.Wise eds., Koureika no Nitibei Hikaku, Nihon

Keizai Shinbun, pp.161-205 (in Japanese).

, and - Irifune (1990), “Waga Kuni no Jinkou no Koureika to Kaku Kouteki Iryo Hoken

no Syushi Nituite,” Financial Review, No. 17, August, pp.51-77 (in Japanese).

Ohkusa, Yasushi (2002a), “Koureika no Iryohi heno Eikyou oyobi Nyuin Kikan no Bunseki,”

- 37 -

Kikan Shakai Hosho Kenkyu, Vol. 38, No.1, June, pp.52-66 (in Japanese).

Suzuki, Wataru (2000), “Iryo Hoken ni Okeru Sedaikan Fukohei to Tsumitatekin wo Motsu Fea

na Zaisei Hosiki heno Iko,” Nihon Keizai Kenkyu, No. 40, March, pp. 88-104 (in

Japanese).

, and Suzuki Reiko (2001), “Jumyou no Choukika wa Roujin Iryouhi Zouka no Youinka?,”

JCER Discussion Paper, No.70 (in Japanese).

Tokita, Tadahiko et al. (1997), “The Present and Future National Medical Expenditure in Japan,”

Economic Analysis (Keizai Bunseki in Japanese), Vol. 152, September.

policy options for financing the future health and … › p › seido › output › ...government...

Documents