The existing pay-as-you-go government pension scheme (the New Zealand superannuation) faces a serious challenge in the years ahead: because of adverse demographic dynamics, it is likely to pose a growing burden on the fiscal and external positions. In addition, the system has other important drawbacks. For example, because benefits are not linked to contributions, it causes disincentives for work; and because old-age pensions are publicly provided, private saving is discouraged.


The existing pay-as-you-go government pension scheme (the New Zealand superannuation) faces a serious challenge in the years ahead: because of adverse demographic dynamics, it is likely to pose a growing burden on the fiscal and external positions. In addition, the system has other important drawbacks. For example, because benefits are not linked to contributions, it causes disincentives for work; and because old-age pensions are publicly provided, private saving is discouraged.

II. Pension Reform in New Zealand1


The existing pay-as-you-go government pension scheme (the New Zealand superannuation) faces a serious challenge in the years ahead: because of adverse demographic dynamics, it is likely to pose a growing burden on the fiscal and external positions. In addition, the system has other important drawbacks. For example, because benefits are not linked to contributions, it causes disincentives for work; and because old-age pensions are publicly provided, private saving is discouraged.

During 1997, the government proposed replacing the existing scheme with a new private scheme (the retirement savings scheme). The proposed scheme was a combination of: (i) a pure “defined-contribution” compulsory scheme with private accounts being managed by competing funds; and (ii) a public top-up scheme with additional funds provided by the government to guarantee a minimum pension. However, this scheme was decisively rejected in September 1997 in a national referendum.

This paper first examines the challenges facing the existing pension system. Although the focus of the analysis is on fiscal and saving effects of the long-term aging, other key issues such as distortions to work are also discussed.

The paper then discusses the referendum proposal, explaining how it would have addressed these challenges. Although the scheme would have been superior to New Zealand superannuation in a number of ways, it would have done little to increase saving, would have imposed a larger initial fiscal burden, and would have also entailed higher administrative costs and greater uncertainty.

The paper also examines further reforms to the existing system. The rejection of the proposed reform in the referendum implies that, for the foreseeable future, New Zealand will continue to rely on its public pension system. Consequently, the reform of the current superannuation scheme becomes the top priority, in order to ensure that the scheme is fiscally sustainable, its pressure on macroeconomic performance is moderate, and intergenerational transfers are not excessive. The reform should occur as early as possible. Parametric adjustments to the current scheme can be achieved by reducing the level of benefits (possibly through raising the retirement age or through changes in the indexation system), by building up financial reserves (through channeling short- and medium-term government surpluses into a public fund in order to smooth future pensions expenditure), or by some combination of these approaches.

The proposed pension reform in New Zealand carries some wider lessons for other countries that consider reforming their public pension systems. The paper discusses some of these lessons.

A. Introduction and Conclusions

1. The aging of the New Zealand population in the decades ahead has profound implications for the current public pension scheme, the New Zealand superannuation (NZS). The two main implications are:

  • The increase in the number of retirees will lead to a substantial increase in government pension expenditure. Such expenditure is expected to more than double (as a percent of GDP) during the first half of the 21st century, and to further increase during the second half.

  • This increase comes on top of other increases in projected fiscal costs related to demographic dynamics—such as health expenditure—putting additional pressure on the fiscal position and possibly also on the already large external current account deficit.

2. Besides the projected increase in the fiscal burden, there are also other problems inherent in a pay-as-you-go pension scheme, such as NZS (in which current expenditure on pensions is financed by taxes on current income). These problems include: disincentives to work and to save; diminished development of private capital markets and financial instruments; intergenerational redistribution; risk of political interference; and risk limitations, which severely restrict the rate of return which can be earned by individuals on their savings (see Appendix II.1).

3. Given these substantial problems with the pay-as-you-go system, many industrial countries, which currently use this system, are starting to examine possible ways to reform their pension schemes. However, a system based on private defined-contribution pension schemes (the main alternative to the pay-as-you-go system) also faces serious problems and challenges. These problems, which are discussed in more detail in Appendix II.1, include: high administrative costs; increased private and fiscal uncertainty; reliance on the development of well functioning annuity markets; complex regulatory mechanisms; and a transition period which might impose significant costs on the current generation (and might also lead to a deterioration in the fiscal position and a decline in saving). Given these problems with the main alternative to the pay-as-you-go system, and especially the last one, it is understandable that the introduction of such an alternative system may prove to be very unpopular.

4. New Zealand’s public pension system and the problems it is facing—including the rapid aging of its population—are not much different than those in many other industrial countries. In fact, the aging problem in New Zealand is less severe than in many other OECD countries. The main difference is the willingness of New Zealand to acknowledge the importance and the severity of these problems and to seriously consider possible steps to address them, in the context of an open public debate.2 In this respect, New Zealand’s experience can provide an important lesson for the rest of the world.

5. Probably the most important problem facing the pension systems in New Zealand, and in other countries, is the fiscal sustainability of the system, given current levels of benefits and projected demographic trends. In general, countries with public pension arrangements facing rapidly aging populations have four possible ways to address the resulting fiscal stresses:3

  • Through developing a significant fully funded defined-contribution scheme.

  • Through parametric adjustments of the current pension system, possibly combined with building up financial reserves.

  • By undertaking broader fiscal adjustments not related to public pensions.

  • By encouraging greater labor force participation or immigration.

6. In the case of New Zealand, the initial conditions prescribe that the main choice is between the first two channels, although some improvements can also be achieved through the latter two channels.

7. Recognizing these options, New Zealand’s government recently decided to present a detailed proposal aimed at implementing the first alternative. This proposed retirement savings scheme (RSS) was publicly debated and, in a national referendum in September 1997, was rejected by a large majority of voters.

8. The RSS proposal had the following key characteristics: it was a fully-funded defined-contribution compulsory private scheme; funds were managed by the private sector; savings had to be locked away until retirement age and had to be taken out as an annuity; and the government was to provide a top-up to private savings in order to guarantee a minimum pension.

9. The main purpose of this paper is to examine the topic of pension reform in New Zealand, both in the context of the RSS scheme and in the context of future possible pension reforms. An additional objective of the paper is to derive some general lessons from the experience of New Zealand, that may benefit other countries considering pension reform. After discussing the key problems with the current public scheme (the NZS), and especially the negative effect of demographic dynamics on its fiscal implications, this paper examines the following two key questions:

  • Would the proposed scheme have solved these problems?

  • Given that the scheme has been rejected, what are the remaining policy options to solve the problems?

10. This paper uses long-term simulations of the New Zealand economy in order to compare alternative scenarios: a continuation of the existing scheme, the adoption of the RSS, and alternative reforms to the existing NZS.

11. The main conclusions of the paper are the following:

  • The RSS would have been superior to NZS in a number of ways. However, it would have done little to increase saving and work incentives, would have imposed a larger initial fiscal burden, and would have also entailed higher administrative costs and greater uncertainty. Many of these problems could have been addressed by an improved design of the top-up part of the RSS.

  • Following the rejection of the RSS, the reform of the current superannuation scheme is now the top priority, in order to ensure that the scheme is fiscally sustainable, its pressure on macroeconomic performance is moderate, and intergenerational transfers are not excessive. The reform should occur as early as possible. Parametric adjustments to the current scheme can be achieved by reducing levels of benefits (possibly through raising the retirement age or through changes in the indexation system), by building up financial reserves (through channeling short- and medium-term government surpluses into a public fund in order to smooth future pension expenditure), or by some combination of these approaches.

12. The structure of the paper is as follows: Section B presents the long-term aging problem facing New Zealand and examines its effects on the existing superannuation scheme. Section C presents some theory and evidence on privatizing pension schemes, including lessons from the experience of three countries: Australia, Chile, and Singapore. Section D inspects the proposed scheme and its ability to solve the key problems. Section E examines further reforms. Section F discusses possible lessons that other countries can draw from New Zealand’s experience.

B. The Long-Term Aging Problem (and its Effects on the Current Superannuation Scheme)

The demographic outlook

13. The old-age dependency ratio (defined as the ratio of people at age 65 and older to people at ages 15-64) is forecast to increase sharply during the first half of the next century, and to more than double during 2005-40. The increase in the old-age dependency ratio is projected to continue during the second half of the 21st century, but at a slower pace.4

Fiscal and external sustainability

14. Demographic changes will put significant pressure in the years ahead on the existing pension scheme—the New Zealand Superannuation (NZS).5 A key issue regarding NZS is its fiscal sustainability. Assuming no change in current policies, simulation results show the following:6

  • Gross fiscal costs of NZS will increase from 4.9 percent of GDP in 2000 to 9.9 percent of GDP in 2050 (and to 11.5 percent of GDP in 2100).7

  • The increase in costs occurs mainly during the period 2005-40, as a result of doubling of the old-age dependency ratio (from 17.5 percent to 35.3 percent). After 2040, fiscal costs would continue to increase, but at a much slower rate.

  • The sharpest increase occurs during 2010-30 (from 5.1 to 8.8 percent of GDP).

15. The expected increase in government expenditure on pensions is significant for at least two reasons:

  • It comes on top of other increases in projected fiscal costs related to demographic dynamics. For example, Polackova (1997) projects that health expenditure will increase from 5.5 percent of GDP in 2000 to 11.8 percent of GDP in 2050.8 According to her projections, other major expenditure items, including education, will remain roughly constant or will slightly increase (as a percent of GDP), and the total public expenditure on other items, besides pensions and health care, is projected to increase from 15.2 percent of GDP in 2000 to 15.7 percent of GDP in 2050. As a result, the total projected increase in fiscal expenditure during 2000-50 is almost 12 percentage points of GDP (5.0 percentage points increase in pensions, 6.3 percentage points increase in health care, and 0.5 percentage points increase in other expenditure categories).

  • It would put considerable pressure on New Zealand’s external position. The New Zealand economy has large net external liabilities (currently 81 percent of GDP) and a large current account deficit. The public saving-investment balance, which now is in surplus, partly offsets the private deficit. Assuming no perfect Ricardian equivalence, a significant deterioration of the public balance would put considerable pressure on the external current account deficit.

16. Some have pointed out that this situation is no worse than in other industrial countries.9 Although this argument may be technically correct, its relevance for the future of New Zealand and its policy implications are not apparent. In any case, the external position faced by New Zealand provides less room for fiscal deterioration, compared with other countries.

Intergenerational transfers and welfare implications

17. The changing demographic structure—caused by the sharp increase of the old-age dependency ratio after 2000—amplifies the general problems inherent in a pay-as-you-go scheme. Equally important, it challenges the equity and impartiality on which the pay-as-you-go system rests. In a changing demographic environment, the intergenerational transfers are strongly dependent on the relative size of each specific generation, because each generation has to contribute (through taxes) to finance the retirement of the previous generation. In particular, people in New Zealand who are now in the labor force need to pay only modest taxes, compared with taxes facing people that will be in the labor force around the middle of the 21st century.10

C. Theory and Evidence on Privatizing Pension Schemes


18. Most industrial countries, and a substantial number of developing countries, are now using defined-benefit public pension schemes as their main instrument of providing income during retirement. In principle, such schemes can be fully funded. In practice, most schemes are approximating a “pay-as-you-go” system. Such a system has serious problems (as discussed in Appendix II.1). Because of these problems (but mainly because of the problem of fiscal sustainability in face of adverse demographic dynamics), many of the countries that currently use a pay-as-you-go public scheme are considering a shift to a different pension system.

19. The main alternative to the public pension scheme is a system based on a defined-contribution private scheme, in which individuals contribute a specific fraction of their income or wages to private savings accounts, and can withdraw their savings, including interest and profits, during their retirement. This system appears to offer several important advantages over a public scheme: a significantly higher rate of return to participants; a more efficient labor market (with higher participation rates); more competition in the financial system (potentially leading to more efficiency, innovation, and growth); and, possibly, a higher saving rate in the economy (meaning a stronger external position and perhaps a higher growth rate). It is quite clear, however, that a system based on a defined-contribution private scheme has its own potential problems and drawbacks (as discussed in Appendix II.1).

20. This section surveys some of the recent theoretical papers that examine switching from a public to a private scheme, and discusses some lessons that can be drawn from selected country experiences.11

Theoretical literature

21. The switch from a pay-as-you-go defined-benefit public scheme to a defined-contribution private scheme (also called a “privatization” of the pension system) has been examined in many theoretical papers. Recent papers include Feldstein (1995), Feldstein and Samwick (1997), Kotlikoff (1995), and Chand and Jaeger (1996).

22. Feldstein (1995) derives the theoretical conditions under which a debt-financed privatization of a pay-as-you-go pension system results in a higher level of welfare. The idea is that there are two offsetting effects on welfare: (i) the future rate of return in a private scheme (which depends on the marginal product of capital) is usually much higher than the rate of return in a public pay-as-you-go scheme (which depends on the growth rates of the labor force and of wages); and (ii) the privatization increases the stock of government debt (through the issue of “recognition bonds” to current participants), and the service of this additional debt requires additional resources in the future. Feldstein shows that the mathematical conditions for a welfare-enhancing privatization are such that they are easily satisfied under most possible parameters. He estimates the net present value of the net social welfare gain as a result of privatization using parameter estimates for the United States. Feldstein finds that the potential gain is very large and can reach (in present value terms) 2.5 times current GDP. When using alternative assumptions, which he views as very conservative, he estimates a gain slightly greater than GDP.

23. Feldstein and Samwick (1997) examine this argument in a simulation framework. Using detailed census and social security information for the United States, they model a transition from the current pension system to a fully funded system based on mandatory contributions to individual accounts. They estimate that the transition could raise wages (net of income and payroll taxes) by more than 35 percent in the long run.12

24. Kotlikoff (1995) simulates the welfare gain from privatization of the pension system in the United States. However, unlike Feldstein and Samwick (1997), which emphasize the potentially higher rate of return in a privatized system, Kotlikoff stresses the endogeneity of labor supply and concentrates on potential efficiency gains as a result of reducing distortions associated with labor taxation in the pay-as-you-go system. He finds that, even after the current generation is fully compensated for the transition costs it incurs, future generations will experience a significant increase in welfare. The respective long-run increases in capital stock, output, and wage are 8.5 percent, 8.1 percent, and 0.1 percent.13

25. Chand and Jaeger (1996) examine the fiscal aspects of pension reform in selected industrial countries (the G-7 countries and Sweden). Their conclusion is that the fiscal costs of undertaking a shift to a privatized fully funded scheme may be very high and it may be preferable to “fix” the existing public schemes, by adjusting levels of benefits and contributions. However, this result appears to depend on the sample of countries and the horizon period that is being considered. First, the cost of required fiscal adjustment largely reflects the cost of servicing the “recognition” bonds to current members in the scheme which, in turn, depends on how generous the current scheme is. For example, in Italy the current scheme is very generous (high pensions and low retirement age), implying a huge cost of transition to a fully funded system.14 Second, the horizon period considered by Chand and Jaeger is only up to 2050. This period, although it appears long by fiscal planning standards, is quite short in terms of demographic dynamics, which are the main force driving the fiscal costs of pensions. Limiting the horizon period to 2050 understates the magnitude of the problem with the pay-as-you-go scheme and also the potential long-term gains from switching to a fully funded system.15

International experience

26. Defined-contribution pension schemes have been tested in several countries. Some of these countries introduced the scheme as their only old-age pension, avoiding a public scheme altogether, while others started with a defined-benefits public scheme, but then switched to a defined-contribution private scheme. Naturally, the experience of these countries is the focus of much interest, especially among countries that are now contemplating reforming their pension system. First, the international experience can help them decide whether or not to reform their pension system. Second, assuming they decide to go ahead with reform, they can use the previous international experience in their reform design, in order to avoid problems that earlier reformers did not anticipate.

27. International experience appears to confirm some of the merits of privatizing public pension schemes, but also points out some important problems. The remainder of this section summarizes the main lessons that can be learnt from the experience of three countries: Australia, Chile, and Singapore (more detail is provided in Appendix II.3).

28. The Chilean experience is, in general, very positive. In particular, the rates of return in the privatized scheme are very high, providing support for the Feldstein thesis. Equally important, it provides a strong degree of stability and continuity, which the old system lacked. Although the short-term fiscal costs were high, the discipline that the pension reform helped to impose on other fiscal expenditure probably contributed to fiscal consolidation in the long term. The effects on private saving, work participation, and growth rates were probably beneficial, although it is difficult to separate the effects of pension reform from the other measures taken in the context of structural reform and stabilization policies that preceded the Chilean economic miracle. Two important lessons that appear to emerge from the Chilean experience are: (i) the strict rules of the pension funds market, designed to protect individual contributors, can easily slip into overregulation that inhibits competition, increases costs, and limits personal choices; and (ii) the design of the publicly provided minimum pension needs to be done carefully, taking into account possible disincentives to work and saving.

29. The Singaporean experience is not really with a “privatized” pension system, given that in the past (and to a large degree also in the present) the pension funds were centralized and invested in risk-free government bonds. The Singaporean system, nevertheless, includes privately owned defined-contribution based individual accounts. The main lesson from the Singaporean experience is that it is possible to use a pension system to significantly increase saving rates. However, this requires very high contribution rates and strict withdrawal rules; once these are relaxed, individuals offset much of the pension saving by adjusting other voluntary saving.

30. The Australian experience is quite recent and, at this stage, it is difficult to draw definite conclusions. However, the general impression so far has been of some disappointment regarding the effects of the scheme on private saving (which was one of the key reasons for adopting the scheme). The main lessons appear to be two: (i) to raise national saving, it is important to abstain from large tax concessions which are intended to increase private saving (because it is difficult to accurately estimate private responses which are based on unknown offset coefficients); and (ii) despite the reform, future policy decisions that change the rules of the game are almost inevitable (and, in this sense, the degree of stability in a privatized scheme is limited).

D. The Proposed Scheme: Could It Have Solved the Problems?


31. The government’s proposed RSS is described and analyzed in this section. Although rejected in a national referendum in September 1997, the RSS warrants analysis because it was designed as a possible solution to New Zealand’s demographic pressures and can be regarded as a departure point for other future policy reforms in New Zealand, as well as for future pension reform in other industrial countries.

32. The proposed RSS was designed to achieve the following objectives: enhance income in retirement; ensure intergenerational equity; enhance national saving; introduce stability into retirement incomes policy; and increase labor participation and employment.

33. The most important characteristics of the RSS were the following:16

  • It was based on a fully funded defined-contribution compulsory private scheme.

  • Funds were managed by the private sector.

  • Savings had to be locked away until retirement age (unless they exceed 110 percent of “target savings”) and had to be taken out as an annuity.

  • The government provided a top-up to the private savings (cover any shortfalls from the prespecified “target savings” level).

  • There was a long period of transition (40 years) during which the existing NZS was supposed to be phased out, and the public benefit was supposed to be a weighted average between the NZS benefits and the top-up benefits (with weights that change over time).

34. The top-up part was intended to ensure a minimum income during retirement, and its magnitude for each individual depended on the balance accumulated in the personal fund at the time of retirement.17 Therefore, it is analytically easier to examine the scheme by dividing it into two distinct stages:

  • A private saving scheme (compulsory, universal, defined-contribution, fully funded).

  • A public top-up scheme.

The private saving scheme

35. The main assumptions underlying the simulations of the private saving scheme are fully consistent with the RSS proposal (the first five assumptions, in particular, are key points of the official proposal), but also include several additional assumptions which are required for simulating the long-term dynamics of the scheme:

  • Full participation in the scheme of all income earners, with the first $NZ 5,000 of wages being exempted from contributions, and small contributions (of less than $NZ 2) being excused.

  • A gradual increase in the contribution rates, starting from 3 percent of gross wages in July 1998 and increasing up to 8 percent of gross income after April 2003.

  • A parallel gradual reduction in tax rates, up to 5 percent of gross income.

  • A gradual transition from the existing NZS to the new RSS system, with an overlapping period of 40 years.

  • The retirement age is 65, and retirees fully invest their savings at retirement in annuities.

  • All males live for 15 years after retirement, and all females for 19 years.

  • The (gross) real rate of return on private savings is equal to the real rate of interest in the economy, determined by the marginal product of capital and the depreciation rate.18

  • The market for annuities is perfectly competitive.

  • The real interest rate earned (both through the savings accounts during the preretirement period and through the annuity accounts during the retirement period) is taxed at an average rate of 33 percent.19 There is no additional taxation of the stream of withdrawals during retirement, and there are no tax concessions for contributions to the scheme.

  • Participation rates by age and gender, in both foil-time and part-time employment, remain constant at their 1995 levels. Also, hours of work remain constant for foil-time and part-time workers, by gender.

  • Relative hourly wages by age and gender, and by foil- and part-time participation, remain at their 1995 levels.

36. The economic assumptions (used to predict growth of wages and of GDP) broadly follow the steady state of the neoclassical growth model with exogenous labor-augmenting technological progress. The rate of technological progress is assumed to be initially high and to decrease over time, gradually bringing the New Zealand economy to a level of productivity that asymptotically approaches the level of productivity in the United States.20 The level of productivity (per worker, adjusted for demographic effects) in the United States is assumed to initially be 43 percent higher than in New Zealand, and its annual growth rate is assumed to be 0.742 percent (and to stay at that level in the future). The rate of convergence of the productivity level in New Zealand to that in the United States is assumed to be 2.5 percent per year, implying that the growth rate of productivity in New Zealand (adjusted for demographic dynamics) is currently around 1.5 percent (and is gradually decreasing).21 The rate of growth of productivity can be divided into two components, an exogenous rate of technological progress and an endogenous rate of accumulation of capital per effective worker.22 The growth rate of wages for a specific type of worker (say, a 50-year old male in full-time employment) is determined by the exogenous productivity rate. The average wage in the economy is determined, in addition to productivity, by the changing demographic structure of the labor force.23 The share of total wages in GDP (including implicit wages of self-employed) remains constant (around 52 percent). The capital/output ratio and the real rate of return are assumed to behave according to the steady-state values of the neoclassical growth model with logarithmic preferences.24

37. The results of the simulations are presented in Table II.3. The main results are:

  • Contributions to the scheme (as a percent of GDP) will increase gradually, following the phasing-in of the scheme. By 2004, they stabilize at 3.8 percent of GDP.

  • Withdrawals from the scheme will be very small during the scheme’s first 20 years. Afterward, they will gradually increase, reaching 3.2 percent of GDP by 2050 and 3.7 percent of GDP by 2100.

  • Total assets in the scheme—including both savings assets (which workers accumulate in their private accounts until they reach the retirement age) and annuity assets (which generate the stream of monthly payments during the retirement)—will increase quickly, reaching 74 percent of GDP by 2020, and 112 percent of GDP by 2050. After 2050, they will remain roughly constant (as percent of GDP).

  • Net increase in assets—the key variable with respect to the potential of the new scheme to increase private saving—will increase to 4.3 percent of GDP by 2010, and then decline to less than 1.5 percent of GDP after 2050. The net increase in assets includes net accumulation (contributions less withdrawals) and reinvested after-tax profits. While in the short term the first component dominates, in the long term (when the scheme matures and the demographic dynamics approach a steady state) the net increase in assets is generated mainly by reinvested profits.

  • The average retirement pensions generated by the scheme are likely to be very modest during the first 20-30 years, because the gradual phasing-in of the scheme and the short contribution period for retirees graduating from the scheme. In the long run, the average pension for men that can be generated by the scheme is comparable to the current (unreformed) superannuation pension. It is important to remember, however, that these are averages per person, not per worker. The people that retire after a lifetime of work (and contributions to the scheme) are likely to accumulate significantly larger amounts than these averages suggest.

  • There is a large difference between men and women: the average pension accumulated by men will be about 2.5 times its average for women. This difference reflects mainly higher labor participation rates and higher earnings history for men, but it also reflects a shorter retirement time (because men have a lower life expectancy, they can buy a larger annuity for the same amount).

Table II.1.

New Zealand: Demographic Projections

article image
Sources: Statistics New Zealand; and World Bank.

Assuming medium fertility, medium mortality, and annual net migration of 5,000.

Assuming medium fertility, medium mortality, and annual net migration of 10,000.

Table II.2.

New Zealand: Projected Expenditure on the Public Pension Scheme

(As percent of GDP)

article image
Source: Author’s simulations.

Includes a transition from 62+ to 65+ during 1995-2001.

Includes recollected taxes, allowances for singles, and other costs. (Singles receive about 10 percent more if sharing accommodations, and about 15 percent more if living alone.)

Excludes recollected taxes, allowances for singles, and other costs.

Assuming constant net migration rates of 10,000 per year after 2000, instead of diminishing rates.

Figures are approximate, based on Figure 6.1 in Periodic Report Group (1997).

Table II.3.

New Zealand: The Proposed Private Scheme

(As percent of GDP)

article image
Source: Author’s simulations.
The public top-up scheme

38. The proposed public top-up scheme addressed the obvious inadequacy of the private saving scheme to cover the retirement needs of many people with low income or who do not have a long earning history when they retire. It effectively complemented the private compulsory defined-contribution scheme with a means-tested minimum pension. The top-up scheme required the government to supplement the savings accumulated through private contributions and reinvested profits up to a certain predefined level. The predefined level was intended to enable the retiree to purchase a “standard” annuity (given the market conditions at the time of retirement). The standard annuity was defined as being equal to 33 percent of average net wages during the first year of retirement, and indexed to inflation thereafter. It also needed to include a guarantee to make payments during the first ten years after retirement (from age 65 to 75), even if the retiree dies before age 75.25

Fiscal effects of the RSS

39. The net fiscal expenditure on the RSS scheme (described at the bottom of Table II.4) is projected to stay around 4 percent of GDP during the next century.26 This masks two opposing trends:

  • The costs of pensions to current retirees, which dominate in the short term, are declining over time.

  • The costs of top-up to new retirees, which are initially small, increase over time.

40. These projections compare favorably with the projected net expenditure on the existing NZS (also described at the bottom of Table II.4) which are expected to increase from 4 percent of GDP in 2000 to 9 percent of GDP by the second half of the 21st century.

Table II.4.

New Zealand: Macroeconomic Effects of the Proposed Scheme

(As percent of GDP)

article image
Source: Author’s simulations.

Adjusting for taxes collected on interest earnings.

Assuming each additional dollar in private pension assets increases private savings by 50 cents.

Assuming each additional dollar of decrease in public savings increases private savings by 50 cents.

Zero offset of both public savings and private pensions.

Full offset of both public savings and private pensions.

Zero offset of public savings, full offset of private pensions.

Full offset of public savings, zero offset of private pensions.

Using projections of Polackova (1997, p. 15); includes health, education, unemployment, domestic purposes, other social welfare, and defense; estimate for 2100 based on figures, assuming growth rate of health expenditures similar to growth rate of pension expenditures.

41. However, the net fiscal effect goes beyond the difference in projected net expenditure. The reason is that the contributions to the RSS were planned to be partly offset by reductions in taxes. For example, in 2003, when the contribution rates reach the top level of 8 percent of gross income (less the $NZ 5,000 exemption), the planned reduction in taxes reaches 5 percent of gross income (with no exemptions). At the same time, public expenditure on the existing NZS would continue almost unchanged. As a result, the government’s total stock of debt would increase faster (or decrease slower) than in the absence of the RSS system. In addition to this effect, there is also an opposite effect: the difference between the expenditure on the RSS scheme and the (larger) expenditure in the absence of the RSS scheme can be invested (or used to repay debt), generating additional savings over time, especially in the long term.

Main scenario

42. A full calculation of the net fiscal effects of RSS over NZS reveals the following results:27

  • The total net fiscal effect of RSS (compared with NZS) is negative until 2050.

  • The negative effect reaches 3 percent of GDP in the early 2000s, and peaks around 2010, at 3.1 percent of GDP.

  • After 2050, the net fiscal effect is positive and is increasing (reaching almost 2 percent of GDP by 2100).

43. These results are caused by the interaction of several factors:

  • The cuts in taxes on income come into effect during the first few years of the new scheme, but there are virtually no reductions in expenditure until close to 2010.

  • The taxes on interest earnings from the private schemes are relatively small in the first few years, increasing only gradually before reaching around 0.8 percent of GDP in 2030.

  • Until 2030, the government runs primary deficits (or smaller surpluses), generating an increase (or a smaller decrease) in the stock of public debt. The interest payments on the additional public debt, which reach 1.6 percent of GDP by 2030, more than offset (until 2050) the primary surplus generated by the new scheme.

An alternative scenario

44. An alternative scenario of analyzing the net fiscal effect of the RSS would be to assume that the tax cuts that formally were part of the RSS proposal would be given in any case, with or without the introduction of the RSS. The rationale for this scenario is that the New Zealand budget is at present in surplus, and it would be politically difficult to abstain from cutting taxes in the short term, even if the fiscal balance is projected to worsen in the longer term.28 Results of simulations using this alternative scenario are described in Table II.4. Under this alternative scenario, the net fiscal effect of the RSS would be positive and very large, increasing gradually to 7 percent of GDP by 2050 and to 14 percent of GDP by 2100.

Effects on national saving

45. The net effects on national saving (also described in Table II.4) are likely to be positive, but modest. The range of possible projections depends on the effects of the new private pension scheme and of changes in public saving on private saving. International experience suggests that it is reasonable to expect some effects, but it is difficult to pinpoint specific offset coefficients.

46. Our central scenario regarding effects on national saving assumes a coefficient of 50 percent offset on private saving of both compulsory pensions and changes in public saving. In this case, the net increase in national saving are projected to be about 0.5-0.7 percent of GDP during 2000-50, and then to increase to 1.7 percent of GDP by 2100.

47. The assumption of 50 percent offset coefficient of compulsory pension saving is used as an approximation for the aggregate effect (alternatives are discussed below). The individual offset coefficients, however, are likely to differ according to income levels. While high-income individuals are likely to significantly reduce contributions to other saving instruments, low-income individuals, who do not save much in the first place, would find it difficult to offset the compulsory saving.

48. Given the high degree of uncertainty regarding the private saving offset coefficients, Table II.4 also presents four alternative scenarios. These scenarios cover the most extreme values of offset coefficients. They suggest that the main uncertainty concerns the period until 2030: while it is possible that national saving will increase by as much as 4 percent of GDP, a decrease of 3 percent of GDP also cannot be ruled out during this period. In the long term, however, the range of possible results is much narrower. By 2050, the possible net effect on national saving ranges from zero to +1.5 percent of GDP, and by 2100 from zero to +3.3 percent of GDP.29

Disincentives for work

49. The disincentives for work implied by the top-up scheme are significant. Because the top-up scheme simply complements existing savings, the effective marginal tax rate on income for people who would need top-up is 8 percentage points higher than in the absence of the top-up scheme. The reason is that, on the margin, every dollar earned in gross wages will result in 8 cents’ contribution to the private saving scheme; during retirement, however, the government will reduce its top-up contribution by exactly 8 cents (converted to their net present value). Therefore, these 8 cents would be regarded by the individual wage-earner as pure income tax.30 This calculation, however, does not apply to the high-income people whose retirement savings will exceed the target level and will not be given top-up assistance.31 For them, the contributions made for each additional dollar in gross income will increase one-for-one their retirement pension.32

50. In general, it is difficult to obtain accurate estimates of labor elasticities, which are necessary in order to project the impact of tax rates on labor supply, and estimates vary across different studies. In one major study of labor markets in New Zealand, Chiao and Walker (1992, p. 165) estimate that the average labor supply elasticity in New Zealand (with respect to gross wages) is 0.38, with females having a much higher elasticity than males (0.64 compared with 0.22).33 Chiao and Walker’s estimates can be used for a rough calculation. Assuming for simplicity a uniform tax rate of 30 percent for men and 20 percent for women, a reduction of 8 percentage points in income tax rates would increase net wages by 11.4 percent for men and by 10 percent for women, increasing labor supply by 2.5 percent for men and by 6.4 percent for women. Assuming constant average working hours, the net effect of an 8 percentage point reduction in income taxes would be to increase labor force participation (for ages 15-64) from 77.1 percent to 79 percent for men, and from 59.9 percent to 63.7 percent for women.34

51. It is interesting to compare the disincentives for work in the top-up scheme with those in the current superannuation scheme in the short term. The envisioned tax cuts (up to 5 percent of gross wages on the margin) do not cover the full amount of contributions to the scheme (up to 8 percent of gross wages on the margin). As a result, the effective marginal tax rate will increase by 3 percentage points for many wage earners. Only for the top earners will the effective marginal tax rate decrease, by 5 percentage points. The net effect on aggregate labor supply depends on the relative elasticities of labor supply across different sectors of the labor force. For example, if low-wage earners have low elasticity and high-wage earners have high elasticity, then the new scheme may increase aggregate labor supply. However, if the elasticity is uniform across income groups, then the new scheme is not likely to generate a significant increase in labor supply, and it might even decrease it. In the simulations discussed in this section, it is assumed that the introduction of the RSS would have no net effects on the labor supply.

Other potential problems with the RSS

52. Other potential problems with the RSS include the usual array of problems inherent in every transition to a defined-contribution privately managed pension system (as discussed in Appendix II.1): increased administrative costs, increased compliance costs, increased uncertainty, a transition which might be quite costly for the current generation, and potentially large fiscal costs as a result of fraud, incompetent management, or excessive exposure to risk by the privately run investment funds.

53. Although it is impossible to completely eliminate these problems, a clever design of the scheme and the regulatory framework can go a long way toward reducing their magnitude. In these issues, learning from the experiences of other countries can be beneficial.35

54. The RSS seemed to be well designed, with one exception: the combination of the generous top-up scheme (which will provide a public top-up to a significant number of retirees) and the almost unlimited freedom of individuals to choose their investment funds and investment strategy (subject only to some very general diversification criteria) seemed to raise two problems:

  • Individuals that are not likely to reach the savings target (probably the majority) will not particularly care about their investments in the scheme, because they will end up in any case with the same amount of pension. As a result, there will be no market mechanism to ensure that the investment funds manage the savings of these individuals in a responsible and efficient way, or that they charge reasonable management fees.

  • Furthermore, such individuals may choose (in a perfectly rational approach) to expose themselves to the maximum possible risk. The reason is that, in taking extreme chances, they can only gain, since their minimum savings level at the time of retirement is guaranteed through the top-up scheme.

Possible improvements to the RSS

55. Identifying and discussing possible improvements to the RSS design may benefit other countries which are likely to propose similar schemes in the near future. The main problems with the RSS were caused by an ineffective design of the top-up scheme. As explained in previous sections, the proposed top-up scheme suffered from several major problems. These problems can be constructively addressed, without fundamentally changing the basic motivation for the top-up scheme—providing a safety net for old-age people with insufficient earnings history. The key elements of addressing these problems are:

  • Reducing the minimum top-up ceiling.36

  • Establishing a significant link between contributions to the scheme and net investment earnings, on the one hand, and levels of pension benefits, on the other hand.

56. One simple way to achieve these elements is to replace the horizontal top-up ceiling with a diagonal ceiling.37 This change would guarantee a reasonable minimum pension level to the very-low-income earners, and still provide a strong incentive for low- and middle-income earners to maximize their savings in the RSS, through greater contributions (meaning more labor participation) and improved net returns on investments (meaning a more careful choice of investment funds, according to their investment skills and management fees, and a more responsible investment strategy). This second channel also provides a natural mechanism to increase demand for efficient and responsible management of these funds, and increase their overall efficiency and financial sophistication. In addition, such a change in the top-up scheme would improve the fiscal position (and this improvement could be partially used to address possible adverse effects on income distribution). Finally, by reducing the overall level of publicly provided pensions, this change probably would also increase voluntary private saving (as individuals will adjust their life-time profile of consumption) and, therefore, also improve the external position.

E. Further Reforms


57. The existing public pension scheme (NZS) suffers from three main problems: disincentives for work, disincentives for private saving, and, most importantly, fiscal sustainability. These problems were not affected by the rejection of the proposed RSS. However, the public debate and the referendum on pension reform achieved two important results: (i) they focused the attention of the public and of the policymakers on this problem and on the need to address it in a timely manner; and (ii) they dramatically narrowed the possible channels of addressing the fiscal sustainability problem.

58. A previous section of this paper discussed the main two alternatives of addressing the problem of fiscal sustainability of public pension schemes in countries facing adverse demographic dynamics: establishing a private defined-contribution scheme and reforming the existing pay-as-you-go public scheme. It is now clear that the first alternative has been strongly rejected in New Zealand, and is very unlikely to be proposed again, at least for a very long time. Therefore, the remaining alternative, reforming the existing NZS, has now become the main priority.

59. The fiscal sustainability problem can be successfully addressed by reforming some parameters of NZS, without changing the basic principles of the system. Furthermore, reforming the scheme by reducing the benefit levels of the public pension can also go at least some way toward addressing the other two main problems (work and saving incentives).

Addressing the fiscal sustainability problem

60. A simple way to describe the fiscal sustainability problem is to say that in the long run current tax rates will not be enough to finance current levels of benefits. One obvious solution to this problem is to increase taxes in the future as much as needed to cover fiscal costs in each period. However, this solution has three critical disadvantages. First, it generates significant intergenerational transfers.38 Second, it causes economic inefficiencies as a result of nonconstant tax rates.39 Third, and most importantly, it ignores the fact that tax rates will need to be raised in any case in order to finance the growth in other government programs (such as health care), and there is a practical limit to the rates of taxation that can be effectively levied on the income-earners.

61. Another possible option is means-testing (for example, reinstituting the tax surcharge on additional income). This option, however, would obviously have strong negative effects on incentives to work and save. In addition, it is perceived as unfair, because it introduces a very visible inequality in benefits that are viewed as an universal entitlement.

62. Therefore, other possible ways of addressing the fiscal sustainability problem need to be examined. The two main alternative solutions are either to adjust the level of benefits in the future in order to reduce fiscal liabilities, or to prefund these liabilities.40 A third possibility is, of course, to implement some combination of the two approaches.41 The fiscal effects of possible reforms are described in Table II.2.42

An increase in the retirement age

63. A significant increase in the retirement age has some attractive features. First, it significantly reduces fiscal costs of pensions, while still providing a safety net for elderly people and fully insuring against unexpected longevity. Second, it creates an important role for private saving (requiring individual provision of savings for the first few years of retirement, assuming people will still prefer to retire at age 65). Third, it encourages greater labor participation (both by young people, in order to accumulate the necessary savings, and in particular by people older than 65). Fourth, there is a clear rationale for such an increase because of the expected increase in life expectancy.

64. An increase in the retirement age from 65 to 68, phased in gradually from 2015 to 2027, is examined. Such an increase in the retirement age would reduce net fiscal expenditure on “basic” superannuation by 1.4 percent of GDP per year after 2027.

A temporary change in the method of pension indexation

65. Changing the indexation system can significantly reduce fiscal costs and provide incentives for private saving, while providing a minimum guaranteed basket of consumption during retirement. Another (less radical) possibility is to index the pensions to wages across cohorts, but to prices during the retirement period of each cohort (as was suggested in the RSS proposal).

66. A temporary change in the method of pensions indexation is examined. The change is assumed to involve indexation to the average of prices and wages (instead of full indexation to wages) during the period 2015-50. After 2050, the pensions will again be fully indexed to wages. The effect would be to reduce pension levels after 2050 by around 23 percent, compared with what they would have been if no change in indexation had occurred. Such a temporary change in the indexation method would reduce net fiscal expenditure on “basic” superannuation by 1.7 percent of GDP per year by 2050 (and by 1.9 percent of GDP per year by 2100).

Implementing both reforms

67. The total effect of implementing both reforms (raising the retirement age and changing the method of indexation) can be significant. It would reduce net fiscal expenditure on “basic” superannuation by 2.7 percent of GDP per year by 2050 (and by 3.0 percent of GDP per year by 2100).


68. Adjustments in pension benefits can go some way toward reducing the mounting fiscal pressure resulting from demographic dynamics. However, they can only make a modest contribution in this regard. This is particularly clear when one takes a more general view on future fiscal liabilities, including expenditure on other major programs, and especially on health care.43

69. Another possibility to address the problem caused by the deteriorating demographic structure is to set up (or contract) a fund which will accumulate assets during the “good” years and use the returns from these assets during the “bad” years. In the short term, this approach requires the government to renounce or significantly scale back any planned cuts in taxes and increase in expenditure. In the medium term, it requires the government to manage a large investment fund.

70. Table II.2 presents a possible scenario of the prefunding approach, assuming that it comes on top of the other two measures discussed above (raising the retirement age and changing the indexation method).44 The scenario assumes that the objective of the government is to limit pension-related fiscal outflows (net expenditure on “basic” superannuation and net transfers to the fund) to 4.8 percent of GDP.45 The government gradually increases its fiscal outflows from 3.6 percent of GDP in 2000 to 4.8 percent of GDP in 2006, an increase of 0.2 percentage point of GDP a year. The funds that are not needed for current expenditure on superannuation are deposited in an investment fund, which will presumably invest mainly abroad. After 2010, the net transfers to the fund start to decrease (as a percent of GDP), and after 2060 they become negative (the government starts to withdraw money from the fund). Because of reinvested profits, the fund continues to grow, until it reaches about 40 percent of GDP in 2080.46

71. One important potential problem with the prefunding approach is the low likelihood that the rate of return achieved by a government-run investment fund would match the rate achieved by private investors. Based on experience from other countries (e.g., Singapore’s Government Investment Corporation), it is possible that such a government-run fund would undertake a conservative strategy and will invest mainly in assets with minimal risk. Although investment in a more diversified portfolio is certainly possible, this might not happen, because of lack of competition (in achieving higher rates of returns) and a high degree of risk aversion (incurring losses might cause political difficulties).

F. Tentative Lessons From New Zealand’s Reform Experience

72. The experience of New Zealand’s attempt to reform its pension system will probably receive a fair amount of attention in coming years. In particular, future studies will need to explain why the proposed scheme was rejected by such an overwhelming majority of voters. At this very preliminary stage, an explanation appears to include the following tentative lessons:

  • There was no clear alternative to the proposed scheme. The voters were not presented with the real alternative to the scheme, namely the significant adjustments in the level of benefits that will be needed in the existing public scheme. By default, they compared the proposed scheme with the current scheme, and they saw a large reduction in benefits for most retirees.

  • There was no strong political support for the scheme. Most Members of Parliament rejected the scheme and actively campaigned against its approval. The public is unlikely to endorse a painful and risky reform if the political leadership is not united in supporting it.

  • There was a significant confusion regarding the “sustainability” of the current scheme. The proposed scheme was advanced mainly on the ground that the existing scheme is not fiscally sustainable. However, during the public debate, it became clear that the existing scheme is sustainable, if “sustainability” is defined broadly enough to include possible adjustments to the scheme’s parameters.

  • The proposed scheme was labeled and perceived as “compulsory.” The official characterization of the proposed scheme as “compulsory” strongly reduced its popular support. The extensive public debate that preceded the referendum somehow failed to demonstrate that it was no more compulsory than the existing public scheme and, at least, it allowed some consumer choice regarding possible investment alternatives.

73. A pension scheme is expected to provide income insurance for old-age people, to be fair, to avoid excessive distortions and disincentives, to be fiscally sustainable, and to be compatible with national saving targets. Designing such a scheme is not an easy task. In addition, because most industrial countries already have a pension scheme in place, replacing it with a new scheme necessarily requires a very careful design of the transition procedure.

74. Paradoxically, the main weakness of the existing pay-as-you-go pension schemes and the main argument for replacing them with new and better schemes, namely the fiscal sustainability problem, makes it extremely difficult for the current generation to embrace a change. The reason is that if future liabilities are explicitly recognized, the adverse demographic dynamics make terminating a pay-as-you-go scheme an extremely expensive affair, and the current working generation must carry a significant part of the cost; and if these liabilities are not recognized, then the current working generation must lose a significant part of the benefits which it already “paid for” through large tax contributions in the past. In any case, the current working generation needs to make a significant sacrifice when replacing a pay-as-you-go system.

75. For these reasons, governments are in general slow to seriously address the issue of pension reform, and they are extremely reluctant to propose a clear-cut reform plan. The government of New Zealand proved to be a remarkable exception in this respect. The scheme that it proposed, despite its significant drawbacks (in particular, an imperfect design of the top-up part of the scheme), was a reasonable one and, at least, it successfully addressed the problem of fiscal sustainability.


  • Carling, Robert, and Geoffrey Oestreicher, 1995, “Singapore’s Central Provident Fund,” IMF Paper on Policy Analysis and Assessment 95/11 (December) (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation
  • Chand, Sheetal K., and Albert Jaeger, 1996, Aging Populations and Public Pension Schemes IMF Occasional Paper No. 147 (December) (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation
  • Chiao, Yen-Shong, and Ian Walker, 1992, “Labour Market Behavior of Prime Age Individuals,” Paper Five in Prebble and Rebstock (eds.), Incentives and Labor Supply: Modeling Taxes and Benefits (Wellington: The Institute of Policy Studies).

    • Search Google Scholar
    • Export Citation
  • Cook, Diana, and John Savage, 1995, “The Fiscal Impacts of an Aging Population: Report to the Office of the Retirement Commissioner,” NZIER Working Paper No. 95/33 (October) (Wellington: New Zealand Institute of Economic Research).

    • Search Google Scholar
    • Export Citation
  • Costello P., and J. Newman, 1997, Savings: Choice and Incentive, Statement by the Treasurer, Treasury, Canberra (May).

  • Diamond, Peter, and Salvador Valdes-Prieto, 1994, “Social Security Reforms,” Chapter 6 in Bosworth, Dornbusch, and Laban (eds), The Chilean Economy: Policy Lessons and Challenges (Washington: The Brookings Institution).

    • Search Google Scholar
    • Export Citation
  • Economist, The, 1997, “Pension Reform: A Model Shows Its Age,” (September 13), p. 78.

  • Edwards, Sebastian, 1996, “The Chilean Pension Reform: A Pioneering Program,” NBER Working Paper No. 5811 (November) (Cambridge, Massachusetts: National Bureau of Economic Research).

    • Search Google Scholar
    • Export Citation
  • Feldstein, Martin, 1995, “Would Privatizing Social Security Raise Economic Welfare?,” NBER Working Paper No. 5281 (September) (Cambridge, Massachusetts: National Bureau of Economic Research).

    • Search Google Scholar
    • Export Citation
  • Feldstein, Martin, and Andrew Samwick, 1997, “The Economics of Prefunding Social Security and Medicare Benefits,” NBER Working Paper No. 6055 (June) (Cambridge, Massachusetts: National Bureau of Economic Research).

    • Search Google Scholar
    • Export Citation
  • Gallagher, P., 1997, Assessing the National Saving Effects of the Government’s Superannuation Policies” Retirement Income Modeling Task Force paper, Treasury, Canberra (June).

    • Search Google Scholar
    • Export Citation
  • Gwartney, James, and Richard Stroup, 1983, “Labor Supply and Tax Rates: A Correction of the Record,” American Economic Review, Vol. 73, No. 3, pp. 446-51 (June).

    • Search Google Scholar
    • Export Citation
  • Holzmann, Robert, 1997, “Pension Reform, Financial Market Development, and Economic Growth: Preliminary Evidence from Chile,” IMF Staff Papers, Vol. 44, No. 2, pp. 149-78 (June) (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation
  • Housman, Jerry A., 1981, “Labor Supply,” in Aaron and Pechman (eds.), How Taxes Affect Economic Behavior (Washington: Brookings Institution).

    • Search Google Scholar
    • Export Citation
  • Husain, Aasim M., 1995, “Determinants of Private Saving in Singapore,” Chapter VII in Bercuson (ed.), Singapore: A Case Study in Rapid Development, IMF Occasional Paper No. 153 (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation
  • Kotlikoff, Laurence J., 1995, Privatization of Social Security: How it Works and Why it Matters, NBER Working Paper 5330 (Cambridge: National Bureau of Economic Research).

    • Search Google Scholar
    • Export Citation
  • Mackenzie, G.A., Philip Gerson, and Alfredo Cuevas, 1997, Pension Regimes and Saving, IMF Occasional Paper No. 153 (August) (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation
  • New Zealand Government, 1997, “You and Your Retirement Savings: The Proposed Compulsory Retirement Savings Scheme.”

  • Roseveare, Deborah, Willi Leibfritz, Douglas Fore, and Eckhard Wurzel, 1996, Ageing Populations, Pension Systems and Government Budget: Simulations for 20 OECD Countries, OECD—Economics Department Working Papers No. 168.

    • Search Google Scholar
    • Export Citation
  • Periodic Report Group, 1997, “1997 Retirement Income Report: A Review of the Current Framework,” Interim Report (July).

  • Polackova, Hana, 1997, “Population Aging and Financing of Government Liabilities in New Zealand,” World Bank Policy Research Working Paper No. 1703 (February).

    • Search Google Scholar
    • Export Citation
  • Sarel, Michael, 1994, “On the Dynamics of Economic Growth,” IMF Working Paper No. 94/138 (Washington: International Monetary Fund).

  • Sarel, Michael, 1995, “Demographic Dynamics and the Empirics of Economic Growth,” Staff Papers, Vol. 42, pp. 398-410 (June) (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation
  • Sarel, Michael, 1996, “Growth and Productivity in New Zealand,” New Zealand—Selected Issues and Statistical Appendix, IMF Staff Country Report No. 96/144 (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation
  • World Bank, 1994, Averting the Old Age Crisis: Policies to Protect the Old and Promote Growth (New York: Oxford University Press).


Prepared by Michael Sarel.


This approach reflects the Fiscal Responsibility Act, which requires a forward-looking perspective in the formulation of fiscal policy.


See Chand and Jaeger (1996) for a more in-depth discussion of these issues.


The demographic projections are based on the latest World Bank’s projections, and are presented in Table II.1. They were done in June 1997 and were kindly supplied by Eduard Bos.


The main features of NZS are described in Appendix II.2.


Detailed simulation results for the future of NZS are presented in Table II.2. More generally, the projections presented in this paper are derived from a long-term model of the economy (explained in a latter section) and from the World Bank’s demographic projections. They are also based on a specific set of assumptions regarding individual behavior and responses. Therefore, these projections do not necessarily coincide with the official projections of the authorities. Furthermore, the projections in this paper cover the period up to 2100, while most official projections in New Zealand go only as far as 2050.


These projections are in broad agreement with the World Bank study by Polackova (1997), the New Zealand Institute of Economic Research (NZIER) study by Cook and Savage (1995), and the Todd-St. John projections published by the Periodic Report Group (1997). A comparison with these projections is provided in Table II.2.


These estimates are slightly higher than a previous study by NZIER’s Cook and Savage (1995). They projected an increase in health expenditure from 5.7 percent of GDP to 10.3 percent of GDP during 2000-50.


For example, the Periodic Report Group (1997, p. 194) reports that the old-age dependency ratio in 2050 has been projected to be 33 percent in New Zealand, compared with 38 percent in Australia, 36 percent in Canada, 56 percent in Japan, 37 percent in Sweden, 38 percent in the United Kingdom, and 34 percent in the United States. However, these projections are not very recent and, at least for New Zealand, they had been substantially revised. The World Bank now projects a dependency ratio of 36 percent in 2050, while according to Statistics New Zealand (mid-projections), the ratio will reach 42 percent. In any case, the expected increase in pensions expenditure in most OECD countries is very significant (for a cross-country comparison, see, e.g., the OECD paper by Roseveare and others (1996)). Even in the case of the United States, where the old-age dependency ratio will increase less than in most other industrial countries, Feldstein and Samwick (1997, p. 2) report that the total cost of social security and Medicare programs is now projected to increase from its current level of 8.4 percent of GDP to 18 percent of GDP by 2050. They also note that many experts consider these official projections too optimistic.


Assuming as a reasonable approximation that pensions are proportional to average wages, that participation rates (for the population at working age, on average) are constant, and that average wages do not depend on the size of the workforce, the percentage change in the tax rate required to finance pensions in each period are proportional to the change in the old-age dependency ratio (ratio of retirees to working-age population). This ratio, as described in Table II.1, is projected to more than double during 2010-40.


Three case studies of the experience of Australia, Chile, and Singapore, are presented in Appendix II.3.


These very large gains, however, depend on their projected values of the rate of return and the rate of growth. Their specific projections assume, perhaps unrealistically, that the rate of growth will decline by 2 percentage points (compared with past values), while the rate of return will remain unchanged.


These figures, however, are simulated assuming an initially large degree of distortion in the pay-as-you-go system (no tax-benefit linkage) and an efficient transition to a privatized system. Changing these assumptions strongly reduces the potential gains that can be achieved by privatizing the pension system. In the extreme case, if there is full tax-benefit linkage in the existing system, privatizing it makes future generations worse off.


This point raises a more general question—if it is relevant to compare the fiscal costs of a transition to a privatized system (that usually implies full recognition of current levels of benefits) with fiscal costs of a public system after parametric adjustments (that implies a significant reduction in levels of benefits).


This point is also recognized by Chand and Jaeger (1996, p. 30).


For a full description of the proposed scheme and the transition procedure, see the official proposal by the New Zealand Government (1997).


It is important to notice that, as a result of the public top-up scheme, the proposed RSS is not a typical fully funded defined-contribution private scheme.


The marginal product of capital is itself determined by the growth rate of productivity (as described below), and it decreases over time.


This, of course, implies a lower taxation rate of nominal interest earnings (e.g., in the case of an inflation rate of 2 percent and a real interest rate of 3 percent, 33 percent tax rate on real interest implies about 20 percent tax rate on nominal interest). This assumption approximates the RSS proposal of taxing interest income at existing personal marginal tax rates.


This assumption captures the idea that a technological catch-up is partly responsible for the faster current growth rates in New Zealand. This “technological convergence” assumption avoids “leapfrogging” in productivity levels across countries.


These assumptions are based on results estimated during previous work by Sarel (1995) and Sarel (1996). These studies converted output per person to output per person adjusted for demographic dynamics. Potential productivity growth in the United States is derived from the average growth rate of output per person adjusted for demographic dynamics during the periods 1985-95 and 1990-95, and is assumed to remain constant in the fixture. The same variable was calculated for New Zealand. The rate of convergence of New Zealand productivity to U.S. productivity is estimated using the two numbers and the gap between the levels of productivity in the two countries in 1995.


The first component is the dominant one, and it accounts for around 90 percent of the increase in productivity. The second component reflects the dependence of the output/capital ratio in the steady state of the neoclassical growth model on the rate of technological progress (this issue is discussed in Sarel (1994)).


This effect comes from differences in levels of productivity of workers at different ages. This demographic effect is currently strong in New Zealand, resulting in a positive effect on productivity of around 0.4 percent per year. It is projected to decline in the future, becoming insignificant after 2020.


The rate of return is equal to the population growth rate, plus the rate of time preference (assumed to be 1 percent), plus the rate of labor-augmenting technological progress. The capital/output ratio is equal to the capital share (one-third) divided by the sum of the depreciation rate (5 percent) and the rate of return.


The approximate cost of such an annuity for men would be around four times the average net annual wages. Cost for women would be about 20 percent higher than for men.


This expenditure includes allowances for singles and the costs on the existing superannuation scheme during the phasing-out period.


These projections and the underlying calculations are summarized in the top section of Table II.4.


Many people in New Zealand view independent tax cuts as a viable option. For example, the proposed scheme was criticized on the grounds that it is unfair to force people to save for their retirement when the government can instead cut their taxes and let them choose their own saving priorities.


The intuition is probably the following. In the short term, fiscal effect is negative, and compulsory saving positive. If the offset coefficients to these factors are very different (say, zero and one, or one and zero), then the total effect on saving can also be significantly different. In the long term, both fiscal effect and compulsory saving are positive, and differences in offset coefficients matter less.


This analysis abstracts from potentially important noneconomic factors, such as the possibility that people will try to avoid getting a top-up, because of the stigma attached to it.


The RSS proposal is very clear regarding the fact that individuals are not required to continue to contribute to the scheme if their savings reach a certain predefined level (110 percent of “target savings”). However, the proposal is not very clear regarding the applicability of the tax cuts for individuals that reach the savings limit and stop contributing to the scheme. Here, it is assumed that tax cuts are given only to contributors, and people will keep contributing even after they reach the savings limit, in order to continue to benefit from the tax cuts.


As a result, the profile of the marginal effective tax rate on wages will not be monotonous: it will first increase until the wage level reaches a certain point which would suffice to cover the minimum required retirement pension, and drop by 8 percentage points at that point (and, possibly, increase again at higher levels of wages, if the tax system is progressive).


These elasticities are for gross wages. They would be true also for net wages, if tax rates would be uniform. They are generally in line with international estimates. For example, Housman (1981) estimates for the United States an elasticity of around zero for men, around 1 for married women, and around 0.5 for unmarried women.


An important issue is whether it is valid to use coefficients estimated for individual behavior in order to project aggregate elasticities. For example, Gwartney and Stroup (1983) examine this issue and conclude that “it is invalid to generalize from the individual work-leisure analysis to the economy as a whole when analyzing the impact of changes in tax rates on the supply of labor.” However, Gwartney and Stroup’s analysis appears to suggest that the aggregate response to tax cuts may be actually stronger than the individual response. The intuition for this result is that the individual response to higher wages depends on two conflicting effects: the substitution effect (the opportunity cost of leisure is higher) and the income effect (more demand for leisure, assuming leisure is a normal good); in the aggregate, however, the income effect is questionable (because the economy as a whole, as opposed to an individual, in general cannot consume both more leisure and more other stuff simply because the government changes its tax policy). Using this argument, it is possible that the positive macroeconomic effects of higher net wages on labor supply could be much stronger than those calculated above.


The experience of Chile, in particular, is extremely valuable in this respect. See previous section, as well as Appendix II.3, for a discussion of the Chilean experience and its lessons.


Although the design of the top-up’s benefit and indexation effectively reduces pension benefits by 15 percent from existing NZS levels, alternative reforms that are being proposed are more significant. For example, the Periodic Report Group (1997) proposes to reduce replacement rates for couples from 65 percent to 50 percent of average wages, an effective reduction of about 23 percent from existing NZS levels. The general point, of course, is that benefits of the top-up scheme should be compared with fiscally sustainable reforms of NZS, not with NZS as it exists today.


For example, the ceiling could increase from 80 percent to 100 percent of the proposed top-up ceiling, as a positive function of income received during retirement through the private saving scheme.


For example, because the dependency ratio is projected to double between 2005 and 2040, the people that will work in 2040 would pay about twice the amount of taxes (for retirement purposes) compared with people that will work in 2005. The two groups, however, will receive a similar amount of pension during their retirement.


As argued by public finance text books, the distortion created by taxes increases with the tax rate. Therefore, a constant tax rate will create less distortion than a nonconstant tax rate, even if they both generate the same amount of revenue in present value terms.


A possible adjustment in the level of benefits was explored by the Periodic Report Group (1997). This Group, chaired by Jeff Todd, was set up as part of the Superannuation Accord (see Appendix II.2). Its first (interim) report was issued in July 1997. The specific possibilities of reform described below (an increase in the retirement age from 65 to 68 and a reduction in the levels of pensions from 65 percent to 50 percent of average wages) closely follow the proposals of the Group. The Group, however, did not propose implementing both reforms simultaneously, and also did not propose prefunding pension expenditure.


A potential additional possibility is to increase immigration rates, in order to reduce the rate of increase of the old-age dependency ratio. However, given the sensitivity analysis with respect to immigration rates described in Table II.2, it appears that such an increase needs to be very large in order to have a significant fiscal impact. Such a large increase is probably unlikely to overcome social and political constraints.


The simulations in Table II.2 report net expenditure on “basic” superannuation, excluding allowances for singles and recollected taxes (total gross expenditure are about 36 percent higher). Total savings as a result of possible reforms will also be proportionately higher than the numbers reported here. An additional fiscal improvement, as explained in a previous section, can be achieved by interest earnings from investing the resulting fiscal savings (or using them to reduce public debt).


Some broader fiscal projections are presented in Table II.4.


This scenario is only illustrative, and other scenarios are certainly possible. In general, the government has to decide on two parameters: what is the desired ceiling on total fiscal outflows, and how fast they get there from current levels. Because total assets in the fund should probably be kept at a positive but modest level in the long run, there is some trade-off between the two parameters.


This ceiling has to be compared with the projected expenditure on superannuation in the absence of funding (described in Table II.2). It increases above 7 percent of GDP by 2040, and reaches 8.5 percent of GDP by 2100. In case the government decides to smooth the total on old-age pensions (including allowances for singles), such a fund would be somewhat larger.


Of course, an alternative way to prefund would be to first pay the existing public debt (currently about 20 percent of GDP). In this case, the positive accumulation in the public fund which will be required in the future would be much lower.