Japan's Pension System -Evaluating the 2004 Reform and Establishing Clear Principles for Further Reforms-

After opening remarks by YOSHITOMI Masaru, President and CRO, RIETI, ASO Yoshifumi (Professor, Faculty of Law, Keio University) gave the following presentation titled "Pension Debt and Implicit Tax."

One characteristic of pay-as-you-go pension systems is that while income may be transferred between generations, such systems are essentially zero-sum in nature. This means that some generations will benefit more than others. The second characteristic is that if government bonds are issued, it is possible to conduct an income transfer policy that is equivalent to a pay-as-you-go pension scheme. The third point is that the burden created by issuing government bonds is called an implicit tax, which is a major characteristic of the pay-as-you-go approach.

Let us consider a simplified scenario for a pay-as-you-go pension system in which interest rates, the population growth rate and per-capita wage growth are fixed and people's lives are divided into two periods. The younger generation pays the pension of the older generation. Considering this income transfer from the point of view of lifetime income, the generation of people who have already retired at the time the plan is launched benefits because it receives a pension without having to pay contributions, while all succeeding generations do not benefit in this way. Over the long term, the positive transfer to the initial generation that benefited is balanced out by the negative transfer from succeeding generations, so the scheme is zero-sum in nature.

A policy equivalent to this pay-as-you-go approach is feasible, and it is called the income transfer policy. Specifically, government bonds are issued at the starting point and income is transferred to senior citizens. In order to prevent the bond liability growing without limit thereafter, a tax is levied to ensure the bond balance is maintained within a fixed range compared to the GDP. This means that the pay-as-you-go approach and the income transfer policy are essentially equivalent. If a fully-funded pension scheme is integrated into this transfer policy, the transfer is zero, so it would be exactly the same as a pay-as-you-go system.

In a pay-as-you-go pension scheme, there is a net pension debt that is covered by the contributions of succeeding generations. When shifting from a pay-as-you-go system to a fully-funded system, that net pension debt must be made zero, so a heavy burden is temporarily placed on the generation working during the transition period. Capital accumulation, lifetime income and contribution rates, etc. are affected by the transition. If the transition period is short, lifetime income will decrease and pension contribution rates will increase. At the same time, when shifting to a fully-funded system, capital accumulation increases.

Measuring net pension debt with a simple model shows that an amount equivalent to approximately 30 times annual benefits is required. In a realistic model, it is around 25 times. Employees' Pension System assets currently stand at around five times benefits, but this debt is likely to increase, increasing the burden on future generations.

In order to make the pension system sustainable, the current excess debt must be eliminated by future surplus assets, but it is a mistake to discuss the pension system only from the point of view of sustainability. Which generations are going to bear the burden and how much of it are they going to bear? It is necessary to think about the issue from the perspective of generational equity.

Dr. Yoshitomi made the following comment.

In sum, the equivalence of the impact between issuing government bonds and having net pension debt means that even in the case of regular government bonds, the government's income only rises in proportion to the GDP growth rate. However, if interest rates are higher than the GDP growth rate, the bond balance will balloon to a bursting point unless taxes are continually increased by the amount of the difference. What is important is the difference between interest rates and the economy's growth rate. Whether this can be determined by any mechanisms will likely be debated. Another problem is how to manage the rate of return when there are substantial assets and inherent risks.

RIETI Research Counselor and Faculty Fellow TACHIBANAKI Toshiaki then gave a presentation titled "The Merits of Fully Tax-Funded Basic Pension Schemes and Counter Arguments."

I am presenting a proposal for a fully tax-funded basic pension program in my role as a Professor of Kyoto University. The merits of a pension system operated completely on tax revenues are that it:
(1) Delivers security because it ensures fixed pension benefits for all citizens.
(2) Eliminates the current problem of unpaid pension contributions owing to the relative certainty of collecting consumption taxes.
(3) Is capable of responding to the problem of category 3 insured persons and diversification in work formats.
(4) Is capable of eliminating the problem of differences among generations.
(5) Helps revitalize companies.
(6) Maximizes the economic growth rate.
(7) Puts current assets at zero and assures benefits for retirees.

There is a counter-argument that equates fully securing the basic pension component via tax revenues with public welfare assistance, but nobody is opposed to public outlays funded entirely from taxes for mandatory education and the dependent child allowance, etc., so a basic pension program could be regarded as a public expenditure in the same way. With respect to the criticism that people will lose their spirit of self-reliance, I think that spirit can be encouraged by privatizing and fully funding the second tier of the program. Against the argument that uncollected amounts are a problem even with a tax-based pension scheme, I think the problem can be solved by introducing an invoice system or taxpayer IDs. With respect to the opinion that consumption taxes should be allocated to eliminate the fiscal deficit, currently a large portion of public expenditure is taken up by pension benefit payments, so having the source of pension benefits depend on consumption taxes will lead to reduced public expenditure. This means that a fully tax-funded approach will likely help reduce the fiscal deficit. On the point that contribution-based pension systems are mainstream among developed countries, it is certainly true that the G5 countries use contribution-based pension systems, but there is no theoretical basis for Japan to imitate this. Japan should adopt the method that best improves the welfare of its citizens.

Next, RIETI Consulting Fellow KANEKO Yoshihiro gave a presentation titled "Enhancing the Financial Sustainability on a Basis of Clear Principles."

I will discuss the characteristics and feasibility of integrating the National Pension and Employees' Pension Systems. A new integrated pension system would basically be for the earnings-based component. The government would pay a guaranteed minimum pension to beneficiaries who would otherwise receive low benefits. The merits of an integrated system would be a low pension benefit burden and neutrality with respect to choice of employment. Its drawbacks include a high system transfer cost and the potential for inequity among participants because of differences in the amount of income captured.

Survivors' and disability pension schemes would be separated from a new integrated pension system. The guaranteed minimum pension amount and half of National Pension System benefits during transition would be covered by the government. With application of the macroeconomic slide, it is estimated that at most a 3.8% consumption tax would be required. Without the macroeconomic slide, a 5.8% consumption tax rate would be needed.

With regard to changes in the pension reserve, if participants in the National Pension System switched to a new integrated system based on an earnings-related pension scheme, the proportion of the economy accounted for by the pension system would rise, and it is estimated that the reserve would increase to 500 trillion yen. This would have a substantial impact on capital markets.

Switching over to a new integrated system would significantly change the relationship between benefits and contributions. Under the present system, because of the fixed amount component there is not a one-to-one relationship with the level of contribution, but under a new integrated system such a relationship would exist. Specifically, if the standard monthly earnings-related amount is at least 274,000 yen, participants would receive a larger pension than current participants in the Employees' Pension System. If the amount is 145,400 yen or less, they would receive the minimum guaranteed pension.

With regard to how to switch to this system, I think it would be possible by expanding the criteria for application of the Employees' Pension System and by encompassing part-time and other irregular workers.

RIETI Consulting Fellow YAMAZAKI Nobuhiko made the following comments about these presentations.
To Prof. Aso:

When debating fully-funded systems and pay-as-you-go schemes, it is beneficial to use the criteria of equity among generations and efficiency, but when the debate includes value judgments on intergenerational equity, it is insufficient to consider the issue from only an economic perspective. Ethical and governmental perspectives should also be considered based on the fact that the preceding generation educates the succeeding generation and hands over social capital free of charge.

To Prof. Tachibanaki:

You explained that everyone uniformly receiving 170,000 yen a month would be fair, but if people who succeeded in society also received that payment I don't think it would be politically possible given the nature of Japanese people. If the proposal is for means testing, the distinction between the pension and public welfare assistance would disappear and I don't think it would be consistent with the spirit of self-reliance. Even if the spirit of self-reliance was encouraged by privatizing the second-tier component while the first-tier component was means tested, I'm fairly skeptical about this proposal working out.

If we assume that uncollected consumption tax is a technical problem that can be solved, then uncollected social insurance premiums are also a technical problem that can be solved. It follows then that the social insurance system has not broken down completely.

To Prof. Kaneko:

If we switch to a new integrated pension system, pension benefits for middle and lower income earners would fall and benefits for high income earners would increase, so it is hard to see the significance in making this transition.

Next, Olivia S. MITCHELL (Professor of Insurance & Risk Management, The Wharton School of the University of Pennsylvania) provided the following comments, titled "Concluding Comments on the Future of Japanese Pension Reform."

With respect to the reform proposals, it is first necessary to construct good macroeconomic models in order to measure any effects of the reforms. It is also necessary to pay heed to the response from a behavioral science perspective. It is further necessary to think closely about distributional effects. Means testing is one method of considering distributional problems, but if the test only considers income while ignoring capital, I would be skeptical of the results, and distributional inequities would likely arise. To avoid these inequities microdata are required encompassing not only income, but also capital. With high-quality microdata, it is possible to observe the effects of means testing and changes in post-retirement consumption.

The pension system is modeled on a husband and wife participating for forty years, but isn't this an unnatural hypothesis that does not reflect reality? Microdata would likely help in this research as well.

Prof. Tachibanaki's proposal is extremely attractive in terms of the fact that taxes would be collected from people who hadn't paid before, but there is the problem that a consumption tax is very regressive. It may be possible to solve this problem, but first it is necessary to reconstruct the current system; one suggestion for doing this would be to introduce taxpayer IDs.

With regard to the impact of the reform proposals from a behavioral science perspective, it is not clear how employment and retirement behavior of senior citizens will be affected if taxes are increased and how employment formats of part-time and fulltime employees will change; research is necessary. It is also necessary to think about the possible effects on marriage if the category 3 insured person classification is eliminated.

This debate over pension system reforms is likely also applicable to health insurance and elderly care insurance. In my evaluation, Japan has taken an important first step toward reform.

Ole SETTERGREN (Director of the Department of Pensions, Swedish Social Insurance Agency) made the following comment about Prof. Aso's presentation.

Your presentation on the pay-as-you-go system was excellent, but the problem is that many points are unclear to a lot of people. When debating a fully-funded scheme to reform the pension system, there is a danger of losing focus if the discussion drifts to whether a larger pension reserve is required.

Dr. Yoshitomi added the following comments.

Sweden has extricated itself from both pay-as-you-go and fully-funded approaches and its system continues to be successful. Considering this as a third category, more discussion is necessary. Unlike the pay-as-you-go or fully-funded approaches, the Swedish system only uses the growth rate in wages and does not consider the relationship between interest rates and economic growth. I think this is the merit of their system.

When funds are accumulated, there is the problem of how to select the investment portfolio. In particular, what to do when the fund becomes enormous would be an issue for the future.

Prof. Kaneko responded to these comments.

To Mr. Yamazaki:

The specific direction of the RIETI model presented here concerning the proposal to integrate the National Pension System and the Employees' Pension System can be regarded as the Employees' Pension System absorbing the National Pension System. If there is gradual expansion in the number of people subject to the Employees' Pension System and companies are willing to bear equal shares between employer and employees, we envisage that the Employees' Pension System could eventually apply even to people making 10,000 yen per month.

Under the RIETI model, a guaranteed minimum pension scheme would be introduced to take the place of the basic pension component, and the earnings-related component would be strengthened. Accordingly, pension benefits for high income earners would increase proportionally and middle and low income earners would benefit less, but as in the U.S., by structuring the system so that the income replacement rate for high income earners is lowered and the rate for low income earners is raised, it would be possible to ensure the system had income redistributional effects.

To Prof. Mitchell:

Regarding whether to conduct means testing for the guaranteed minimum pension, it would perhaps not be necessary if social security numbers and taxpayer IDs were required in various social contexts. But if tax records were made a requirement, it would be necessary to take heed of the point that the insurance principle is related to guaranteed minimum pension benefits based on the assistance principle. However, if one considers resistance to means testing, then this method is also possible. In order to maintain public trust while implementing these systems, it would be necessary to provide pension information on a one-to-one basis, as in Germany and Sweden.

Next, Prof. Aso responded to comments by Mr. Yamazaki and Dr. Yoshitomi.

To Mr. Yamazaki:

I want to emphasize that my basic perspective is that income transfer in a pension system is zero sum in nature and the existence of excessive debt in the present puts a burden on the future. In addition, net debt already exists and how this debt will be borne is an important question. In order to promote rational debate, that debt should not be processed surreptitiously within the structure of the pension system; we should clarify the excess portion of the debt and logically discuss how it should be shouldered. A pay-as-you-go pension system causes an income transfer to senior citizens at the time the system commences and succeeding generations bear this debt. In general this transfer cannot be justified. The extent of this excess debt on succeeding generations can be simulated in various ways, but basically income transfer is zero sum in nature and we must think about who is going to bear its burden.

To Dr. Yoshitomi:

Among pay-as-you-go systems, the Swedish system is essentially a mechanism for transferring net pension debt to the future in a form that doesn't diffuse it. The pension rate of return is equivalent to the economic growth rate. However, if capital accumulation is held back by net pension debt, over the long term it will be better to revert to a fully-funded system. If the system reverts to being fully funded after a certain period of time, pension contributions will go up temporarily, but theoretically capital accumulation will recover, yields will increase, and the effects of contribution hikes will be curtailed, so even the generation in the transfer period will profit.

The fund will amount to perhaps two to three times the GDP. In the case of a closed economy, if interest rates are low and the fund is large, profitable investment becomes extremely problematic, but in economic terms it is more desirable to have a large amount of capital.

Finally, Prof. Tachibanaki responded to comments by Mr. Yamazaki and Prof. Mitchell.

To Mr. Yamazaki:

On the point of whether benefits would also be provided to high income earners under a fully tax-funded basic pension scheme, the original principle of the system holds that benefits would not be lowered for high income earners through means testing, but benefits could be reduced if called for by public sentiment. With respect to the argument that some consumption tax would not be collected, the rate of uncollected consumption tax is within 5%, so cannot be compared with the rate of uncollected social insurance premiums, which is around 40%.

To Prof. Mitchell:

Japan's added value tax is a proportional tax, but what I am thinking of is a progressive consumption tax and a progressive added value tax, so the regressive critique does not apply. In addition, I also think that it makes no sense for pension amounts to change depending on marital status. My proposal tries to solve this kind of problem by paying a fixed amount to each individual.

Q: What type of means testing is used in Australia?

A (Prof. Piggott): Means testing has been conducted in Australia since 1908 when old age pensions were introduced, so it has a long history. Eligibility is based on the value of reported assets, but it is not strict like Japan and it is very widespread. However, if the value of your assets at retirement rises or falls thereafter, you could lose or gain eligibility for benefits.

Q (Mr. Settergren): Is real estate included in means testing?

A (Prof. Piggott): Yes. However, the criterion is not the value of your house, but rather whether real estate is in your name and whether you live there.

Q (For Prof. Aso): Your presentation distinguished three generational periods: - 1, 0, and 1. If the pension system starts in period 0, generation 0 lightens the burden on generation - 1, but how is this accounted for? Also, who bears net debt occurring with a pay-as-you-go system when the population is constant?

A: I want the focus to be on the extent of the public transfer when a pension system starts. The structure of that transfer is such that the burden with respect to the initial generation is borne by succeeding generations. What happens to private support when this public transfer occurs depends on the response -- it differs depending on the conditions. Regarding who shoulders the burden in a constant situation, all generations do. To state it mathematically, a relationship is established in which the transfer to the initial generation is equivalent to the total burden of future generations.

Q (From Mr. Yamazaki): In your earlier response, you explained that under a pay-as-you-go system if there is a particular need for the initial generation to profit, then this income transfer is justified. But there is no particular difference for a generation that supported its parent's generation in a situation where there was no pension system. Also, considering only income transfer via a public system, the market in the world of economics can be thought of as fair. Including private transfer outside of this world constitutes fairness from a wider perspective. In particular, when taking a broad view of this transfer, including both education and intellectual property, I think such a gift to the preceding generation under a pay-as-you-go system can be adequately justified.

A (Dr. Yoshitomi): A more important point than justifying the gift to generation - 1 is that when implementing a pay-as-you-go system a gift is provided to generation - 1; that is, generation - 1 receives benefits without any contribution to the system, or the proportion of their contribution is extremely advantageous. In the case of introducing a fully-funded system for generation 0, we must compare by making the same assumptions as when implementing a pay-as-you-go system. Discussions over differences between pay-as-you-go and fully-funded boil down to whether or not to give a gift to generation - 1; the latter approach does not. Given that theoretically starting under the same conditions is easier to understand, arguing that the fully-funded approach should change in this way could increase its fairness compared to a pay-as-you-go system.

A (Prof. Aso): I want to reemphasize that the net debt initially incurred must be borne and that among pension systems the pay-as-you-go approach is a mechanism for bearing this debt. With a fully-funded system, generation - 1 is abandoned, but an extremely high rate of return is possible; this is not the case with pay-as-you-go. That difference is the financial resources transferred to the initial generation.
The point is how to think about the burden of current net debt when the system starts. There are two ways of asking people to bear the burden: (1) reduce benefits or (2) put the burden on future generations by increasing pension contributions. Fundamentally, it would be desirable to have a debate about which generations should bear how much of the burden. At the same time, intellectual property has public property value and I don't think education should be debated in the context of the pension system.

Q (For Prof. Tachibanaki): (1) How would means testing be carried out?
(2) Wouldn't a consumption tax be highly regressive with respect to income?
(3) If consumption tax maximizes the economic growth rate, which helps vitalize companies, more than social insurance premiums, shouldn't we totally abolish the tax system and consolidate it into indirect taxes and consumption taxes? What about taxes on assets?
(4) Even if there are merits for employees and employers on the supply side, won't consumption on the demand side be curtailed?
(5) With an indirect tax, won't the disposable income of low and medium income earners decrease?
(6) Won't indirect tax revenues be inevitably affected by demographic trends?

A: Regarding means testing and regressiveness, the consumption tax I am advocating is a progressive tax; because consumption taxes are regressive I have proposed a concrete measure to eliminate this, and I won't go into further detail here. Accordingly, my proposal is not regressive.

On the point of whether an indirect tax based system would truly boost economic growth, I have written a book that simulates concrete policies and even provides specific figures. I encourage you to read it.

My ideal is an expenditure tax; at present, I am advocating VAT, but I think in 10-20 years a progressive expenditure tax will become the mainstream. However, it will only be implemented if the technology advances and the authorities gain a good understanding of it.

Regarding this, Dr. Yoshitomi made the following comments.

One of the major conclusions of yesterday's session was that in order to look at social security overall, discussion was needed on the problem of how representative averages derived from models are. For this reason, microdata are inevitably necessary. The results outlined by Prof. Tachibanaki should not be treated as the end of the discussion; more research is needed on assumptions and mechanisms relating to anticipated behavior.