III Public Pension Plans, Their Reform, and Saving
- Alfredo Cuevas, George Mackenzie, and Philip Gerson
- Published Date:
- September 1997
This section describes how public pension plans could be expected, on a priori grounds, to affect saving. It uses the LCH only as a starting point and acknowledges the implications of imperfect capital markets and imperfect information about the future. Three cases are presented: (1) Introduction of a PAYG plan where no plan existed previously. The literature invariably covers such cases, and the conclusions are relevant for the many countries that are contemplating a substantial expansion of an existing plan with narrow coverage. (2) Incremental reform to an established PAYG plan with universal coverage. (3) Introduction of a defined-contributions plan. Some of the main findings of this section are illustrated mathematically in Appendix I.
Introducing a Pay-as-You-Go Plan
A PAYG plan can conceivably redistribute wealth from generations with a high saving propensity to generations with a low saving propensity. The first retirees under the new scheme often enjoy a very high implicit rate of return to their contributions— and thus an increase in their wealth—because eligibility conditions are generous and contribution rates are low.8 In effect, the first retirees benefit from a windfall—they receive a quantity of SSW that may be very large compared with their existing wealth. Because retirees are at the stage where they are running down their asset holdings and because their wealth has increased, their consumption expenditure will also increase.
What of the members of the current working generation? The imposition of a payroll tax will reduce their disposable income, but offsetting this effect is the promise of a retirement pension. If the implicit rate of return to their payroll tax contributions is high enough, the people currently working will act as if their wealth has not declined. Hence, according to the strict version of the LCH, they will not reduce their consumption spending, but will instead reduce their voluntary saving. In these circumstances, introducing a PAYG plan will depress saving: both private sector and aggregate saving will fall.9 The capital stock is permanently reduced by the impact of the windfall received by the first generation of retirees, even though only the initial generation of retirees really benefits from the introduction of the PAYG scheme.10 Subsequent generations do not receive a windfall, but may be led to believe that they will not have to provide for themselves, or at least not to the same extent, in retirement.
The impact of the substantial SSW created for the first generation of retirees can, however, be countered, at least partially, through compensating transfers between the affected generations. For example, the current working generation may spend less of its own money on caring for its parents because their income from social security has risen, or the elderly may increase their bequests or inter vivos transfers to their children. There is evidence that a “bequest” motive underlies much saving (Kotlikoff and Summers, 1981). Nonetheless, these transfers do not extinguish SSW; they simply transfer it across generations.
The increase in the wealth of those plan participants nearing retirement can also, paradoxically, increase saving if the increase in wealth resulting from the introduction of the PAYG plan prompts a desire for early retirement, and thus a need for supplementary savings (Feldstein, 1974; Munnell, 1976). The increase in wealth enjoyed by persons close to retirement can mean, for example, that by saving a little more, they can retire early without either suffering an undue decline in retirement income or having to depress drastically their preretirement level of consumption.
The impact on saving of the introduction of a PAYG plan is also conditioned by the effects of capital market imperfections and the shortsightedness that often overcomes people making decisions in the face of an uncertain and distant future. The presence of myopia and capital market imperfections means both that people save too little and that they are income constrained. In consequence, their saving rate may be so low that they cannot maintain their level of consumption by lowering saving to pay for the payroll tax contributions that finance the PAYG scheme, and they will have difficulty borrowing to sustain their original level of consumption. They will also discount future pension benefits heavily. In these circumstances, the impact of the SSW windfall bestowed on early retirees is muted.11
Even if people are more farsighted and save more, social security “saving” will not be equivalent to voluntary private saving. Social security saving is “locked up” until retirement and cannot function as a vehicle for precautionary saving, as can short-term bank deposits or other financial investments. This too will attenuate the depressing effect of the SSW windfall on saving.
Apart from the difference in liquidity between voluntary saving and involuntary social security saving, there is a question as to how social security contributions are viewed by plan participants. If they are viewed as a form of saving—albeit involuntary and long term—that earns an adequate rate of return in the form of the expected future stream of pension benefits, then they are more likely to substitute for voluntary saving. Contributors are most likely to adopt this view of the payroll tax when contributions and benefits are tightly linked.
But the link can be weak when there is a redistributive element in the pension plan, like ceilings or floors. For example, with a flat-rate pension, there is little or no relationship between the value of contributions and the value of the pension.12 The increase in payroll tax contributions that results when wages and salaries increase does not lead to an automatic increase in future benefits. In this case, there is no reason to expect substitution of voluntary savings for increased payroll tax contributions.13
Even when the legal structure of the system entails a tight link, that link can be virtually destroyed by episodes of financial instability. In many countries, the social security system has been unable to maintain the replacement ratio stipulated by law, at least in real terms. Adverse demographic trends have contributed to this phenomenon, even when financial conditions have been stable. It is, in fact, common for the replacement ratio to be set at high levels at the outset of the retirement period, only to fall more or less continuously thereafter. It is not uncommon for the real value of a retirement pension to be highly variable, as it has been in countries experiencing hyperinflationary episodes. These elements of uncertainty can be expected to reduce the substitutability of social security for private saving.
It should not be surprising that the empirical evidence is hard to interpret, given the theoretical ambiguity of the impact of pension regimes on saving; the ways this impact can be conditioned by the economic, institutional, and social environment; and the substantial statistical problems researchers have stumbled across in testing for a relationship (see Appendix II for a fuller treatment of the empirical literature). In the United States, where most studies have been conducted, to the extent that a consensus exists, it is that social security has depressed private saving to some degree, although the offset effect is significantly less than 100 percent. In other countries, the results are mixed. SSW effects have been found in Italy and Japan, for example, but not in Canada, France, or the United Kingdom.
Reforming Existing Pay-as-You-Go Plans
When a mature PAYG system with broad coverage is already in place, the transitional impact, if any, has worked its way through. The impact of the SSW windfall on the first generation of retirees will be largely water over the dam (although the private sector saving rate will remain lower than it was previously), because subsequent generations will not enjoy an unusually high implicit rate of return. Nonetheless, changes to the structure of the existing system can in principle affect the saving rate, and the way they affect it will depend on the characteristics of financial markets and saving behavior already noted. Before considering the more radical reform represented by the adoption of a defined-contributions plan, it is important to consider how incremental or piecemeal reforms will affect an existing PAYG system.
Reform affects two separate groups of people: the current generation of pensioners and current working contributors.
A reduction in the real value of the average pension will probably increase the sum total of public and private sector saving. Pensioners’ income has to fall unless transfers from family members compensate for the reduction. This kind of approach to reform is extreme and may be seen as unfair because it breaks the implicit contract made with pensioners while they were working. Nonetheless, it can raise the saving rate. The increase in public sector saving that results from the reduction in transfer payments should not be fully offset by the fall in private saving that takes place as pensioners or their families attempt to sustain the pensioners’ level of consumption. A full offset is unlikely because the perceived wealth of the private sector as a whole (and, perhaps more important, its current income) will have fallen.14
Current Participants and Future Pensioners
An increase in saving that results from reform must derive from either a reduction in the average expected pension or an increase in contributions. Pensions may be reduced by reducing the accrual rate; lowering the base of pensionable income; and increasing the age or number of years of plan participation at which the maximum pension is achieved.
All of these measures can reduce the average expected pension of future retirees. Moreover, as these measures take effect, they reduce private sector disposable income.15 A reduction in the accrual rate, unless it is retroactive, will affect the flow of pensions only very gradually.16 A reduction in the pensionable base (e.g., making pensions a function of the average annual salary of, say, the last 15 years rather than the last 4) can have a much more draconian effect, but the sudden imposition of such a change can be seen as unfair. The same is true of the third option.
Nonetheless, any of these measures will reduce the expected future income of the current working generation unless it has been anticipated. If perceived wealth is reduced, and there is no change to the retirement age, the current working generation will have to save more to attain the same standard of living in retirement. In this case, both private and public saving can increase.17
An increase in the contribution rate will lower the disposable income of contributors because the effective incidence of the payroll tax, whatever its legal incidence, falls mostly on labor. This reduction in disposable income will presumably reduce both perceived wealth and consumption unless the increase in the rate is seen as financing a higher pension down the road. Private saving may fall, but the sum of public and private saving will rise. The high level of contributions in most countries where social security coverage is broad, and the need to avoid exacerbating labor market problems, however, argues against further rate increases in most countries.18
How much extra saving can be generated through incremental reform along these lines? It is generally accepted that measures to increase public sector saving will not be wholly offset by a decline in private saving (conceivably, private saving may even increase). By combining projections of the impact of reform on the pension plan’s accounts with an assumption about the impact of increased public saving on private saving, one can derive an illustrative range of estimates of the impact of incremental reform on saving.
Projections for the public pension systems of Canada, France, Germany, and Sweden show a clear tendency toward increasing cash-flow deficits in the absence of reform.19 These tendencies can be arrested through various changes to the structure of benefits (Table 2). The most potent change is a marked reduction in the replacement rate that affects existing and new pensioners alike. Thus, for example, the balance of Germany’s pension plan increases by 1.9 percent of GDP between 1995 and 2005, when the replacement rate is lowered by 5 percentage points for all pensioners (entailing a decline in the average pension of about 10 percent). Assuming that the offset factor for private saving is 50 percent, the impact on the saving rate over this ten-year period will be about 1 percent of GDP. In Canada, where the replacement rate is lower, the impact on the balance of the system of a similar adjustment will be 0.7 percent of GDP over the same period. With the same offset factor for private saving, the impact on aggregate saving will be 0.4 percent.
Scenarios include impact of pension reform on public sector interest payments and receipts.
Canada’s and Sweden’s baselines are already indexed to the CPI.
Scenarios include impact of pension reform on public sector interest payments and receipts.
Canada’s and Sweden’s baselines are already indexed to the CPI.
Reforms that affect the number of new pensioners in a given period and the average pension these pensioners receive may ultimately reduce the replacement ratio substantially. Initially, however, the reforms will have only a modest impact on aggregate saving. Thus, the same 5 percentage point reduction in replacement rates in Germany—but for new pensioners only—will initially increase public saving by only 0.2 percent of GDP. Its impact on the public sector deficit will increase to 1.1 percent by 2005 and reach 2.3 in 2015. The initial impact of an increase in the effective retirement age must be greater than that of a reduction in the replacement rate for new pensioners because, assuming no change in the retirement age, it will be tantamount to a 100 percent reduction in the replacement rate for new pensioners.20
These illustrative simulations of mature pension systems with aging populations also suggest lessons for countries contemplating a substantial increase in the coverage of an existing PAYG system. These countries are probably best advised to ensure that the terms of benefits are not excessively generous, so that politically unpalatable increases in contribution rates do not become necessary as the system matures further. A range of reforms may be required, including increases in the effective retirement age, a reduction in the accrual rate, and a redefinition of the base of pensionable income. In addition, the plan’s finances should be based on conservative demographic assumptions so that the risk of unforeseen increases in the dependency ratio is minimized. These procedures, together with administrative reform where necessary, will at least minimize the risk of reducing public sector saving. They may also entail forsaking a strictly PAYG system for a partially if not a fully funded system. The degree of funding achieved will depend on, among other factors, the impact of reform on average replacement ratios and contribution rates.
Introducing a Defined-Contributions Plan
The effects on saving arising from the introduction of a defined-contributions plan will be heavily dependent on whether the plan is intended to replace an existing PAYG system and on the modalities of its introduction. To begin with, it is useful to analyze the potential effects of the introduction of a compulsory defined-contributions plan, assuming for the sake of the exposition that no PAYG plan is already in place.
When No Pay-as-You-Go System Is in Place
A defined-contributions plan is essentially a compulsory saving plan, so that contributors, upon retirement, get back only what they put in, plus the accumulated return on their investment. Given that the present value of expected pension benefits should equal the present value of contributions at the market rate of interest, no net SSW is created, and there is thus no intergenerational wealth effect. An intergenerational wealth effect is created, however, if the return to contributions is artificially subsidized by the government, and may occur in the presence of imperfect capital markets and myopic plan contributors.
Consequently, the establishment of a defined-contributions plan should not reduce saving. At worst, it will simply result in a fully offsetting reduction in voluntary saving. It can increase saving, however, when the contribution rate is above the rate at which participants would save if left to their own devices and if capital market imperfections inhibit borrowing on the strength of future income. This increase is more likely to materialize when the conditions for withdrawing the funds are comparatively strict, as in Chile, where contributors may withdraw funds only upon retirement. Singapore’s CPF now allows contributors substantial withdrawals to finance the purchase of a house and home insurance, medical expenses, and certain types of savings accounts (Box 2).21 Thus, an account with the CPF is far more substitutable with liquid financial assets than an account with a Chilean pension fund. A number of studies of saving in Singapore and of the substitutability between voluntary saving and saving through the CPF have not as yet produced a consensus on this issue, however.22
When a Pay-as-You-Go Plan Is in Place
The state-regulated defined-contributions plan in Chile was introduced alongside a quite extensive PAYG system (Box 2). Participation in the new system was optional for persons already contributing to one of the state-run funds, but fresh entrants to the labor market (young workers), if not self-employed, were and are required to participate. The state-run system will be phased out automatically as the remaining contributors gradually retire and die. In this respect, the Chilean reform differs from the reforms implemented subsequently by Peru and Argentina. In Peru, participation in the new plan was optional even for new labor market participants. In Argentina, the defined-contributions plan was added as a second tier on top of the existing state-run system; simultaneously, a number of changes were made to the first tier with a view to improving its finances. (SeeAppendix III for a further discussion of the reforms in Peru and Argentina.)
Box 2.Two Pioneers in Pension Reform
Chile’s privately managed, publicly regulated plan was introduced in 1981. Participation was voluntary for persons already contributing to the state system, but obligatory for first-time employees. Nonetheless, the combined effect of participation in the new plan by new labor market entrants and substantial migration from the state system has raised the coverage of the new system to over 90 percent of the labor force, although only about 60 percent of these contribute regularly.
Participants pay a contribution that is a fixed percentage of their gross salary to an account registered in their name with the pension fund of their choice, as well as a management fee with a fixed percentage and a flat component. Some 21 funds were operating as of end-1994. Switches between funds cost the contributor nothing. The funds’ administrative costs, which include sales commissions, are high compared with large, well-administered public schemes. Average returns on contributors’ accounts have been high, reflecting, among other things, the government’s policy of indexing financial assets. With the development of Chile’s capital market and the advent of financial stability, the range of assets in which pension funds may invest has been broadened.
Upon retirement—the standard age is 65 for men and 60 for women—contributors may choose among several combinations of lump-sum payments (programmed withdrawals) and annuities (which are indexed); at present, the former is the preferred choice. Given the age of the system, comparatively few persons have retired.
Singapore’s Central Provident Fund (CPF) was established in 1955 and is a fully funded individual account system, with lump-sum and annuity benefits; there is no redistributive aspect. It currently covers about three-fourths of the population. Contribution rates in 1994 were 20 percent each for employers and employees and were capped at a monthly salary level of S$6,000 (US$4,000). The government does not contribute, except in its capacity as an employer. By law, virtually all the assets of the CPF are invested in government bonds, and the proceeds of the bond sales are mainly invested abroad.
After reaching the age of 55, members may withdraw as a lump sum all but a fixed amount (S$34,600 in 1993), which is used to finance a fixed monthly pension payable at age 60. Although the starting amount of this pension is indexed to the CPI, the pension for an individual is not indexed once it commences, and ceases once the individual exhausts the fixed sum plus accumulated interest (after about 14 years). A member’s account is fully portable. Since 1968, when accounts were just a source of pension income, withdrawal privileges have been steadily expanded. Now, withdrawals are allowed for housing, medical, education, and home insurance expenses and for investing in certain types of savings assets. Employee contributions and all withdrawals are excluded from taxable income, as are interest earnings on member balances.
One of the often-cited benefits of the Chilean reform is its apparently beneficial effect on capital market development. A defined-contributions plan can have an impact on capital markets because it generates a surplus that must be invested. Even if voluntary saving is reduced by exactly the amount of the saving channeled to the defined-contributions plan—as might occur if the defined-contributions plan were introduced as a supplementary second tier to the PAYG plan—its portfolio will differ from the portfolios of institutions through which the voluntary saving is channeled. For example, issues of stocks and bonds may increase at the expense of bank loans. If contributions substitute for payroll taxes, a new market for financial assets (including government paper) is created. If positive externalities are associated with the development of securities markets, a Chilean-style reform may boost economic growth and saving (Holzmann, 1996).
Of basic importance is how the funds are invested. A defined-contributions plan is unlikely to have an impact on saving different from that of a PAYG plan if the defined-contributions plan’s reserves are invested in government securities and the government increases its current expenditure. In such a case, there is no real counterpart to the system’s reserves. For Singapore’s CPF, the reserves must be invested in government securities, but they are reinvested—mostly abroad. In Chile, the funds of the private pension plans that manage contributors’ accounts are invested in a variety of instruments. A large share of the portfolio—about 40 percent in 1994—is invested in government paper, but the government’s efforts at fiscal consolidation have not been inhibited by what could have been, in a sense, a captive market.
One important consideration is the treatment given to the government’s implicit liability under the old system to those of its former participants who move to the new system. In Chile, the government issued “recognition bonds” (bonos de reconocimiento) to these participants. Their value was an approximation of the present value of the expected pension benefits participants had earned while contributing to the old system.23 In some systems, the records necessary for an accurate determination of the value of the bono have not been maintained, so that the formula determining the value is ad hoc. Clearly, the larger the calculated value of the average bono, the greater the potential for a wealth effect.
The Chilean reform and, to a much lesser extent, the reform in Peru, caused an increase in the public sector deficit because payroll tax contributions to the state-run system fell when contributors switched to the new, privately administered system. This loss was not fully offset by the contributions made to the new system, because the rate for the new system was less than the combined employer-employee rate for the old. Nonetheless, the impact on aggregate saving of the increase in the deficit was largely offset by the surplus of the new private financial institutions.24 The rest of the increase was reflected in an increase in take-home pay, which probably increased private sector consumption.25 Thus, the increase in the deficit—as conventionally measured—overstates the expansionary impact of the reform. Had the contribution rate to the new system been set to equal the combined payroll tax, the surplus of the pension plans would have fully offset the increase in the public sector deficit.26
The impact of a Chilean-style reform on national saving depends critically on the method chosen to finance the increase in the deficit that results from the reform. When the deficit is financed through borrowing and the contribution rate of the new system is set equal to the combined payroll tax rate of the old, the disposable income and wealth of the new plan’s contributors will not be affected and their saving will not change.27 The current generation of pensioners will also be unaffected unless the reform includes measures that affect the value of their pension. The government runs an increased deficit, as discussed above, but it is exactly offset by the increased private sector surplus in the form of pension plan surpluses (what were tax payments become involuntary private sector saving). In this case, it is not clear why national saving would be affected.
The result will be different if the government chooses to deal with its increased deficit by adopting a program of fiscal consolidation. The resulting increase in public sector saving should lead to an increase in aggregate saving, because, as discussed, the offset effect of private saving would be less than 100 percent. Thus, a Chilean-style reform cum fiscal consolidation should increase aggregate saving, although it would be the fiscal consolidation, not the pension reform, that increases saving. In fact, one tentative conclusion of research on the Chilean experiment is that the fiscal consolidation of the early to mid-1980s—which was prompted partly by the pension-reform-induced increase in the deficit-contributed to the increase in the national saving rate (Diamond and Valdés-Prieto, 1994; and Holzmann, 1995, 1996). The conversion of the state-run pension system’s implicit debt to explicit debt (the bonos) may also have helped prompt this consolidation by generating a recognition effect, thereby making consolidation more palatable politically. When the contribution rate is higher than the payroll tax rate, pension reform may increase saving even without fiscal consolidation, although fiscal consolidation should increase saving even more.
The implicit rate of return is the rate that equates the present expected value of contributions with the present expected value of pension benefits. The high implicit rate of return enjoyed by the early participants in the OASDI scheme in the United States is well documented (Duggan, Gillingham, and Greenlees, 1993). If the implicit rate of return equals the market rate of interest, no net SSW is created.
This view is associated with Feldstein (1974).
Subsequent generations also experience an increase in wealth from participation in the pension system to the extent that the implicit rate of return on their contributions exceeds the interest rate. The implicit rate of return may well become negative, however.
Another effect can be present when annuities are not available on reasonably favorable terms. Then, the introduction of a pension plan can be seen as reducing the risk of living so long after retirement that one’s savings are exhausted. In this case, the lessened risk can exercise a depressing effect on saving.
There may be some relationship, to the extent that to be eligible a potential beneficiary must work and contribute for a certain number of years. However, there will be no relationship between average salary during the qualifying period and the value of the pension, and extensions of the contributory period will not increase that value.
If a plan has a ceiling on contributions and a maximum pension benefit that are set deliberately to impart a redistributive component to the system, as with the U.S. OASDI program, there is a range of incomes over which a tight link is established. Once income exceeds a certain level, the link is broken. Contributions increase, but future benefits do not. Providing a flat-rate pension may lead to a reduction in saving, but the size of the decline will not vary with income.
This kind of policy, by weakening confidence in the public pension system, could stimulate private saving. Repeated recourse to it would, however, tend to defeat the purpose of a public pension scheme. The analysis assumes that the reduction in the public sector deficit is not used to finance tax cuts or expenditure increases.
The introduction of a PAYG system, in contrast, has no effect on private sector disposable income.
For example, suppose the accrual rate is 0.02, which means that for each extra year an individual works, the pension increases by 2 percent of the pensionable income base. If the rate is reduced to 0.015, it will take ten years for the replacement rate of new retirees to drop by 5 percentage points, and the average replacement rate will scarcely be affected. Nonetheless, this change in the accrual rate will entail a big decline in the steady-state replacement rate—from 70 percent for a 35-year working life to 52.5 percent.
This is not a foregone conclusion, given that saving behavior will depend on, among other considerations, how farsighted people are. Although perceived wealth is reduced, there is no substitution effect—the return to private saving is unaffected. Rather than let consumption in retirement fall by the full extent of the decline in the pension, current workers will reduce pre-retirement consumption (i.e., pre-retirement saving will be increased) if consumption both before and after retirement is positively related to wealth.
The argument that payroll taxation is distortional requires the assumption that at least part of the incidence of the payroll tax is on employers.
The projections are based on those described in Chand and Jaeger (1996).
The very large longer-term impact projected from an increase in the retirement age in France to 67 years results mainly from the current low effective age of retirement and from the impact of the deficit reduction on public sector interest payments (Table 2).
In comparison, Malaysia’s provident fund allows substantial withdrawals at age 50 and for critical illnesses.
A recent IMF publication found that there was strong substitutability (Husain, 1995). The imperfect substitutability of a Chilean-style pension account with liquid financial assets may explain why participation of the self-employed (which is voluntary) remains low in the Chilean plan. Saving for a house or for a commercial investment may seem far more attractive to young adults than participation in a retirement plan. See Gillion and Bonilla (1992).
The bonos contribute to the pool of savings that finance the pension plan participants elect upon retirement, but cannot be cashed.
In Peru, the increase in the deficit was smaller than in Chile because of the comparatively small size of the state system and the slower rate of migration to the new system.
In Chile, when contributors switched from the state system to the new plan, their employer no longer had to remit to the government either the employer’s or the employee’s portion of the pay-roll tax financing the state scheme. The employee’s gross pay rose by the full amount of the payroll tax. Because the rate of contribution to the new plan was set below the (combined) rate of the payroll tax, take-home pay increased.
In this latter case, the reform, while not reducing aggregate saving, could have entailed a financing problem for the public sector if it had not been able to borrow the full amount of the loss in contributions from the pension funds and if other sources had not been available. However, if the pension fund had been lending to the private sector, that lending presumably would have displaced bank or other lending, which could then have been directed to the government. Institutional constraints on the composition of financial institutions’ portfolios could nonetheless create problems.
If, however, the expected rate of return to contributions under the new system exceeds the expected implicit rate of return under the old, contributors’ wealth increases. This will depress their saving.