This Selected Issues paper discusses the issues related to reforms and growth in New Zealand. The paper analyzes the record on growth and productivity outcomes in a comparative perspective. The study provides a brief history of the industrial relations in New Zealand leading up the passage of the employment contracts act. The paper assesses the monetary policy framework, central bank decision-making processes, and also reviews the possible extensions to full funding of the country's future superannuation expenditures.


This Selected Issues paper discusses the issues related to reforms and growth in New Zealand. The paper analyzes the record on growth and productivity outcomes in a comparative perspective. The study provides a brief history of the industrial relations in New Zealand leading up the passage of the employment contracts act. The paper assesses the monetary policy framework, central bank decision-making processes, and also reviews the possible extensions to full funding of the country's future superannuation expenditures.

V. New Zealand Superannuation—Possible Extensions to Full Funding56

A. Introduction

151. New Zealand, like most OECD countries, faces an aging population over the coming century, which carries the implication of future strains on the public finances arising from both pension- and health-related costs. Particularly in the area of pensions, the policy challenge has long been recognized, but has been politically divisive. Indeed, the recent history of reforms of New Zealand Superannuation has been contentious and characterized by rather significant changes in policy, including some reversals (Box V.1).57

152. In its March 2000 Budget Policy Statement, the newly elected Labour-Alliance Government announced that it will raise the benefit ratio (for couples) from 60 percent to 65 percent of average national wages and that it will begin to allocate budgetary surpluses and build up assets in a Crown entity (or similar arrangement) beginning in 2001/02 with a view to “pre-funding” future superannuation expenses. Further details were released in June 2000 at the time of the announcement of the 2000/01 Budget. The approach to be adopted would aim to smooth the costs of the transition to a higher level (of GDP) of public pension payments. Such a policy would be a significant departure from the present system, which is funded on a purely pay-as-you-go (PAYG) basis.

153. Arguably, the decision to pre-fund superannuation expenses can be seen as the first step in a sequence of policy moves to reform the New Zealand’s pension system. That is, under the assumption that political momentum for further pension reform builds with the accumulation of funds in a public investment account of the type envisaged under the proposed pre-funding scheme, one can envisage additional policy actions aimed at further improving the long-term financing of the public pension system, including through a move to permanent (i.e., full) funding, perhaps combined with further parametric reforms aimed at enhancing incentives for private saving and further reduction in pension costs.

154. In this spirit, the present paper looks at ways to extend the pre-funding proposal with a view to “costing out” such possibilities.58 In particular, this paper looks at alternative ways of moving to a fully-funded superannuation system for considering: (i) various objectives as to what proportion of future pension expenditures should be fully funded; (ii) to what extent investments in equities could be used to bolster returns and achieve “politically feasible” (defined below) transition paths to a fully-funded system over the long run; and (iii) to what extent further parametric benefit reforms would contribute to such paths. This analysis is undertaken using long-term simulations for various combinations of objectives, instruments and rates of return.

New Zealand—A Recent Chronology of Key Public Pension Policy Developments

New Zealand Superannuation (NZS) is a universal, pay-as-you-go public pension system funded from general tax revenues—(here is no payroll tax. All pensioners receive the same pension depending on their marital status, and the pension is based on the level of nationwide average wages.

  • In 1977, the Government established New Zealand’s first “pure” universal public pension. Benefits began at age 60, and were set at 80 percent of the average wage for couples (48 percent for singles). Qualifications were liberal: only 10 years of residence were required, and there were no income or asset tests. Government pension costs quickly doubled to 7 percent of GDP.

  • To rein in costs, various Governments in the 1980s reduced the generosity of the system. In 1985, a surcharge was placed on other retirement income (effectively introducing means testing). In 1989, the 80 percent “replacement ratio” was suspended.

  • Further reforms followed in the early 1990s: it was agreed to raise the retirement age to 61 immediately, and to 65 by 2001; pensions were indexed to prices alone; and the surcharge was increased and its base broadened. Pension expenditure fell to 5 percent of GDP by mid-decade.

  • In 1993, an Accord on Retirement Income Policies was reached, which included: a transitional benefit to partially offset the effects of the increase in the retirement age; bounding a couple’s combined pension in a band of 65–72½ percent of average wages; and commissioning of a six-yearly Periodic Group Report (from 1997) to look broadly at the effects of demographic changes on pension spending. Also, the Office of the Retirement Commission was established to publicize the need to increase private retirement savings, and the public pension system was renamed NZS.

  • The National-New Zealand First coalition government posed a national referendum on superannuation in September 1997. A fully-funded, privately managed, defined contribution scheme with a government “top-up” was rejected by 92 percent of voters. Also, in April 1998, the surcharge on other retirement income was eliminated, returning New Zealand to a pure universal pension scheme.

  • In December 1998, the minority National Government set up a Superannuation 2000 Task Force, charged with developing a pension policy consistent with long-term fiscal sustainability (and to report its findings in 2000). The floor on the pension benefit rate was reduced to 60 percent.

  • In November 1999, a Labour-Alliance coalition government came into power and disbanded the Superannuation 2000 Task Force. In March 2000, the government announced the immediate restoration of the floor of the pension benefit rate to 65 percent, as well its plans to “pre-fund” future superannuation payments out of projected operating surpluses beginning in 2001/02. Cumulative contributions would be about 3 percent of GDP over the First three years.

  • The 2000 Budget spelled out further details of the pre-funding scheme. Annual contributions would be determined using a rolling, 40-year tax smoothing model, which implies the existence of an investment fund into the 22nd century. Also, the fund would be run on a commercial basis through a Crown entity, and be subject to income taxes. The precise investment objectives and instruments (and the extent to which assets could be invested abroad) have yet to be finalized.

155. The results of the simulations indicate that while fully-funding pension spending over and above current levels (in terms of GDP) going forward would be manageable, given the likely political constraints on the diversion of fiscal surpluses-to a public investment fund, moving to a system where all pension spending is fully-funded is likely to face some political problems. Specifically, a desire to fulfill the latter objective would require additional parametric benefit reforms beyond those already undertaken, most likely some combination of raising the retirement age and lowering the benefit ratio. Moreover, attaining this objective would require a long transition period in order to bring the annual financing costs down to politically feasible levels.

156. It should also be noted that for the purposes of this paper the sole objective of funding superannuation expenses is to finance the transition costs to a fully-funded pension system over the long term. All other possible future demands on the public finances—e.g., from future health liabilities which are assumed to be of a similar order of magnitude as pensions—are not considered here. Also, some authors (Carran, 2000; and Grant, 2000) have stipulated other possible objectives, including, but not limited to, improving savings performance and the current account, deepening local capital markets, and forming an investment pool for small business. While those goals may be defensible on both economic and public policy grounds, they are not considered here, except as they are affected by different policies for attaining a specific full-funding objective.59

157. The rest of the paper is structured as follows. Section B provides background and an international context to pension-related expenditure in New Zealand. Section C briefly outlines the new government’s approach to pre-funding superannuation expenses. Section D lays out the analytic framework used in this paper for assessing the issue of fully funding superannuation expenses in New Zealand, and specifies several interpretations of that goal as well as specific policy objectives to be evaluated. Section E describes the simulations and gives the results, which are then narrowed down to include only those that are likely to be politically feasible. Section F looks at the probable macroeconomic impact of fully funding superannuation expenses, including the effects on national savings, labor supply and the current account. Section G considers operational issues likely to be important in defining and implementing a full-funding strategy. Section H concludes.

B. Background

158. Projections of New Zealand’s pension expenditures point to increasing pressures on the public purse from roughly the 2010s onward. Data from Sarel (1998), which are presented in Table V.1, show that superannuation expenditures (net of recollected taxes, allowances for singles, and other costs) begin at an estimated 3.6 percent of GDP at present and remain roughly constant through the 2010s before rising to 5.0 percent of GDP in 2020, 7.3 percent of GDP in 2050 and 8.5 percent of GDP by 2100. That is, public pension expenditures on current policies in New Zealand are expected to rise by 5 percent of GDP this century. Underlying this increase is a jump in the old-age dependency ratio from 17/2 percent at present to 36 percent in 2050 and 42 percent in 2100.

Table V.1.

New Zealand: Selected Data on Projected Public Pension Expenditures

(As a percent of GDP)

article image
Source: Sarel (1998).

159. The pressures on New Zealand’s pension system going forward appear to be less than for the rest of the OECD, particularly in the first part of the 21st century. This reflects both the population structure as well as the reforms to the pension system undertaken by various governments over the past two decades. Table V.2 shows that for the OECD as a whole, baseline (gross) pension expenditure rises from 7 percent of GDP at present to over 12 percent of GDP by mid-century before dropping to 11½ percent of GDP in 2070. The non-G7 countries have a more favorable profile early in the century. New Zealand starts from a much lower base (4.8 percent of GDP) than the OECD average, and remains so through mid-century; however, this gap then closes.

Table V.2.

New Zealand: Gross Pension Expenditure of OECD Countries-Baseline Scenarios

(In percent of GDP)

article image
Source: Roseveare et al. (1996), Table 3.

160. Policy makers in New Zealand had already begun to address the pension issue prior to the current government’s pre-funding proposal. Most notably, an attempt at fundamental pension reform was made in 1997 by the coalition (National-New Zealand First) government, which presented a referendum on a new pension system to voters in September of that year. The initiative accorded a primary role to the private sector in superannuation policy; specifically, a Compulsory Retirement and Savings Scheme was proposed, which featured: (i) a pure defined-contribution, compulsory component with privately-funded and managed accounts; and (ii) a public “top-up” component to guarantee a minimum pension. The proposal was rejected by 92 percent of voters, implying that New Zealand would need to rely on its publicly-run pension system for the foreseeable future. In some ways, the proposal of the current government can be seen as an outgrowth of that referendum.

C. Pre-Funding New Zealand Superannuation—the Current Government’s Approach

161. The coalition Labour-Alliance Government that came to power in late 1999 announced in its March 2000 Budget Policy Statement (which preceeds the budget) that, in order to address the effects of an aging population on pension expenditures, it intends to run surpluses over the economic cycle to allow it to begin to pre-fund superannuation spending.60 Contributions would be made to a Crown entity or similar arrangement beginning in 2001/02. The working assumptions for the amounts of these contributions were specified as $NZ 0.6 billion in 2001/02, $NZ 1.2 billion in 2002/03, and $NZ 1.8 billion in 2003/04.

162. The 2000/01 Budget announced in June revealed further information on the proposed pre-funding strategy. Essentially, the objective is to smooth the transition of the projected 5 percentage point of GDP increase in net pension expenditures identified above. This smoothing will be achieved by using a rolling 40-year time horizon to determine the annual pre-funding requirement. In every year (beginning in 2004/05), a constant level of pension expenditure in terms of GDP will be chosen that fully funds superannuation outlays over the following 40 years. For the first 40-year period, the required pension expenditure would be around 6 percent of GDP (or 2 percent of GDP above current pension obligation), implying that a 2 percent of GDP surplus would be the pre-funding requirement. This amount would be placed into the Crown entity, where it would be invested on a commercial basis. The constraints—if any—on the portfolio manager in terms of instrument and investment location (domestic or foreign) have yet to be determined. Going forward, this process would be repeated every year until smoothing is no longer required. Specifically, when the pension expenditure to GDP profile flattens out, the required pension expenditure to GDP over the following 40 years would be (roughly) the same as the initial year pension expenditure to GDP ratio, implying that the investment fund would no longer be needed. That, in turn, would imply that at the end of the smoothing period, and in the absence of further policy actions, New Zealand would revert to a pure PAYG system with public pension expenditures on the order of 9 percent of GDP.

163. Before presenting the simulation model, it is worth emphasizing that the objective of the present paper is to anticipate possible future debate on moving to a fully funded system in New Zealand under the assumption that political momentum builds for such reform stemming from the implementation of the currently proposed pre-funding scheme.61

D. Permanently Funding Superannuation—Analytical Framework

164. The methodology used for the simulations in this paper will be that of Feldstein and Samwick (1996) (hereafter FS), who studied the transition path from a largely PAYG government-run pension system to a private, fully-funded plan for the case of the United States. In their model, mandatory individual retirement accounts are set up for each working individual and the assets of these accounts are invested solely in equities. FS note that these funds could in principle be collected and invested by the government, although they prefer a decentralized system. This sentiment has been echoed by a US expert panel on social insurance (Diamond (ed.) 1999) who argue that such a system would: (i) instill ownership in the retirement system on the part of citizens; (ii) offer individuals more choice regarding their investments, thereby better aligning their portfolios with individual preferences; (iii) reduce the influence of the government over both the investment of the funds and the workings of corporations whose stocks were held; and (iv) lessen the temptation to the government to spend the accumulated funds. However, unless additional assumptions are specified that would alter government behavior given the existence of such accounts, whether they are privately or publicly held and invested is immaterial. For the purposes of this paper, no such assumptions are made.

165. The thrust of the FS approach is to use the private investment accounts to exploit the difference between the historical returns to equity and the implicit return on a PAYG system to generate politically feasible transition paths to a fully-funded pension system. They assume that benefits are constrained to be the same as those that would be received from the existing US Social Security system. With benefits fixed, each individual’s pension is financed using: (i) the returns from their private investment account and (ii) the existing payroll tax. As time passes, pension benefits, given the assumed difference in the rate of return between equities and the PAYG return, are increasingly financed out of the returns from private investment accounts and the required contribution from the Social Security system (the residual) goes to zero. In the long run, all pensions would be fully funded from the individual investment accounts. Using US data, FS calculate that an additional 2 percent payroll tax invested in equities at historical rates of return would over the long run replace the existing 12.4 percent payroll tax used to finance Social Security. Importantly, they assumed a 9 percent real return to equities for the entire simulation period, 100 percent allocation of the portfolio to equities, and no expenses required to manage the funds.

166. The FS framework is equivalent to a government collecting an amount over and above that needed to fund current PAYG pension outlays and depositing it into a public entity, with this entity investing such funds on behalf of individuals to partially or fully finance future pension outlays. In the context of New Zealand, the FS approach would be appropriate for studying a case where the government used operating surpluses to contribute to Crown entity fund over time, had the proceeds invested, and utilized the returns (and perhaps some of the capital) to finance some portion of superannuation expenditures going forward. Thus, it would be compatible with an extension of the current pre—funding proposal to a fully-funded framework. The remainder of this section looks at possible objectives for fully funding superannuation expenditures over the long term, which sets the stage for the simulations.

167. The choice of objectives is critical for the exercise—there are two key issues. The first relates to the time-horizon of the increase in superannuation expenses to be funded. A temporary horizon would imply the need to finance a “bulge” in pension outlays. Under this objective, a fund would be built up in the period preceding the bulge, and run down during the peak expenditure years. Thereafter, pension funding would revert to the manner prevailing before the funding began. A permanent horizon would imply the need to finance a long-lasting increase in pension outlays. Under this objective, a fund would be built up over time so that returns from its assets plus any new contributions less any payouts would be sufficient to finance the relevant pension outlays indefinitely.

168. The second issue is: what portion of pension expenditures to fully fund? This paper considers two alternatives: (i) funding future net superannuation expenditures over and above those currently prevailing; and (ii) funding all future net superannuation expenditures. The second of these raises a further issue; namely, the transition period over which the move to full funding takes place. Three cases are considered: immediate transition (largely for illustrative purposes); transition over one generation (taken to be 25 years); and transition over the entire 100 year projection period. Therefore, all told, there are four different funding objectives.

169. In addition to the choice of objectives, another central question concerns what future parametric reforms to benefits, if any, would be undertaken during the simulation period. Under the baseline options listed in the previous paragraph, it is assumed that no future parametric reforms are undertaken other than the ongoing transition to a retirement age of 65, which is scheduled to be completed in 2001. For illustrative purposes, this paper will consider two more scenarios: in the first, the retirement age will be increased gradually from age 65 to age 68; in the second, the retirement age will be increased and the benefit ratio will be reduced gradually from 65 percent of the average national wage to 50 percent.

170. As there are a number of scenarios that are considered, for the convenience of the reader the table below lays out the notation for the labels as well as the scenario to which each label corresponds.

Notation Used for Scenarios

article image

E. Simulations of Fully Funding Pension Expenditures

171. This section will look at the resource requirements (i.e., fiscal surpluses) necessary for attaining the objectives of each of the options laid out above using simulations with reasonable parameter values for real returns and equity allocations in the portfolio. In contrast to FS, these scenarios use a variety of assumed rates of return and portfolio allocations to illustrate the extent to which the results are sensitive to the underlying parameters. Also, unlike, FS, the simulations include costs for managing the portfolios, taken to be ¼ of 1 percent, equivalent to major US-based international index funds (e.g., Vanguard).

172. The simulations will cover the period through 2100, and the contribution to the investment fund for 2001/02–2003/04 will be as in the 2000/01 Budget. The following parameter values will be adopted.

  • Initial real equity returns: 9 percent; 7 percent; and 5 percent. Real equity returns—historical and projected—should be seen in relation to real GDP growth. FS note that over the period 1960–94, the real marginal rate of return on capital in the US averaged 9.4 percent (Rippe, 1995), which is the basis for their assumed real rate of return of 9 percent. Over the same period, US real GDP growth averaged 3.4 percent, implying a “premium” of 5.4 percent over the real growth rate. The simulations in the present paper are generated by assuming that this premium over the real GDP growth rate remains constant over the projection period. Since US data were used for the assumed real equity returns, projected US GDP data are used to generate the future real equity returns in the simulations.62 This can be seen as a proxy for the “rest of the world” for the case of New Zealand. Therefore, while the real equity returns are set at 9, 7 and 5 percent at the beginning of the projections, they are in actuality premia over the US real GDP growth rate of 5.4, 3.4 and 1.4 percent respectively. These real returns then fall throughout the projections period to maintain the “premium” over the US real growth rate assumption.

  • Real debt returns: 5 percent. The assumption of 5 percent real debt return could be interpreted as an approximation for the real rate of return on either: (i) New Zealand government bonds (the Treasury’s Long-term Model assumes a 7 percent nominal rate of return, as against the 1½ percent mid-point of the RBNZ’s inflation target range); or (ii) “rest of the world” yields on corporate bonds of general investment grade quality, which would be assumed to trade at an average of 100 basis points over US Treasuries, which have a 3½–4 percent historical real return.

  • Portfolio allocations (equity/total): 90 percent; 70 percent; and 50 percent. FS assume a portfolio allocation to equity of 100 percent, which could be seen as compromising their results since (i) the liquidity needs of the portfolio would argue against holding all of the fund’s assets in relatively volatile instruments; and (ii) the simulations are extremely sensitive to variations in this parameter. The present exercise seeks to highlight this issue by looking at the above three portfolio allocations to equity. While somewhat arbitrary, this approach does give a flavor of the sensitivity of the funding requirement to non-trivial changes in this parameter.


173. In essence, the (recursive) simulation model operates as follows. First, for a given portfolio allocation and rate of return assumption, the year 2100 asset stock in the investment fund was chosen so that its yield was equal, as a percentage of GDP, to the level necessary to finance the requisite superannuation expenditures for that year. Then, the constant (in percent of GDP over the period 2004–2100) annual contribution to the investment fund was derived to generate, over the simulation period, the stock of assets consistent with the year 2100 payout subject to model’s equation of motion (i.e., the beginning of period asset stock plus interest thereon, plus new contributions, less payouts). The implication of this construction is that the stock of assets in 2100 would be commensurate with a permanent funding of pension expenditures with no further additional annual contributions from the Government or tax increases beyond the projection period. That is, the assumed rates of return would hold up, and that the pension expenditures as a percentage of GDP would stabilize.

174. In all, a total of 108 simulations were run, corresponding to nine return-portfolio allocation possibilities (using the parameters listed above) for twelve policy options (four each under the no reform, age only reform and both reform assumptions). Tables V.3 and V.4 show the results of these simulations in terms of the required fiscal surplus and the end-period investment fund stock for the “diagonal” of the matrix for these cases; i.e., the low-low case (50 percent allocation to equity and 5 percent real equity return), the middle-middle case (70 percent allocation to equity and 7 percent real equity return) and the high-high case (90 percent allocation to equity and 9 percent real equity return).

Table V.3.

New Zealand: Pre-Funding Superannuation, Required Fiscal Surpluses

(As a percent of GDP)

article image
Source: Staff estimates.

Assumes that no new reform efforts are undertaken.

Assumes increasing the retirement age from 65 to 68 years over 2015–27.

Assumes increasing the retirement age from 65 to 68 years over 2015–27, and reducing benefits by 23 percent (from 65 to 50 percent of the average wage) over 2015–50.

Table V.4.

New Zealand: Pre-Funding Superannuation, Investment Fund Balances in 2100

(As a percent of GDP)

article image
Source: Staff estimates.

Assumes that no new reform efforts are undertaken.

Assumes increasing the retirement age from 65 to 68 years over 2015–27.

Assumes increasing the retirement age from 65 to 68 years over 2015–27, and reducing benefits by 23 percent (from 65 to 50 percent of the average wage) over 2015–50.

Funding requirements with no parametric benefit reforms

175. Simulations for Option I(no) show that under the most optimistic parameter constellation (9 percent initial real return, 90 percent portfolio allocation to equities), the funding requirement—1.4 percent of GDP—lies in the range of the average planned contribution contained in the BPS over the coming three fiscal years. The underlying stock of assets would be just under 100 percent of GDP in 2100. In contrast, in the most pessimistic scenarios (a 5 percent real initial return on equities and a 50 percent allocation to equities), the annual funding requirement would be 3¾ percent of GDP and the underlying stock of assets would be nearly 300 percent of GDP in 2100.

176. Options II(no), which include three variants of fully funding of all future superannuation outlays, present a more problematic outlook in terms of their feasibility. Under the first variant, Option II(no)-A, which features an immediate transition to full financing, the annual funding requirement ranges from over 5½ percent of GDP in the most optimistic parameter constellation to 8½ percent in the scenario with no equity premium and a 50 percent allocation to equities. Clearly, generating fiscal surpluses of this magnitude over any time horizon would be very difficult to envisage. While Option II(no)-A simulations reflect the strong (and arguably implausible) assumption that all current net expenditure on superannuation is shifted to and financed immediately by the investment fund, Option II(no)–B assumes that this process takes a generation (25 years) to complete, which smooths the transition, and allows the fund to build up resources and benefit from the effect of compounding the equity premium. Nevertheless, the financing requirement for this variant remains onerous, as annual budgetary surpluses ranging from 4¼ percent of GDP to over 8 percent of GDP would be required. Allowing for a long-term transition path (100 years) improves the outlook dramatically, but the financing requirement is still very high, except perhaps under the most aggressive portfolio allocation and the highest returns. Simulations for Option II(no)-C show that a financing requirement ranging from 2¼ percent of GDP to over 6 percent of GDP would be needed to eliminate direct budgetary financing of superannuation expenditures by end-century in the absence of additional benefit reforms. Given that the end-period annual pension expenditure in terms of GDP is the same in all Options II(no)? the end-period stocks would be the same: these range from 167 percent of GDP to over 500 percent of GDP.

177. The tentative conclusion of this set of simulations is that the establishment of a hybrid system (Option I)—in which only the future projected increases in superannuation expenses would be funded—could be seen to be consistent with reasonable operating surpluses diverted annually over the long-term into a public investment fund. However, the transition to a system where all pension outlays are financed by an investment fund, even over a very long period, would be very large except in one optimistic case. Of note, the simulations in this section did not assume any future parametric benefit adjustments to the superannuation system. To the extent that additional reforms could be taken, they would alleviate some of the pressure on the system and, ceteris paribus, bring down the funding requirements to more acceptable levels.

Full funding requirements with an increase in the retirement age

178. The previous section looked at the financing requirements for New Zealand Superannuation expenses under the assumption that no additional parametric reforms were undertaken during the simulation period. Given the aging population profile and the identified (and publicly discussed) pressures on pension funding in the future, it is arguably very unlikely that the solution to pension financing would exclude some type of parametric reforms (e.g. Periodic Report (1997)). This section considers raising the retirement age from 65 to 68 years over the period 2015–27, which would lower pension expenditures by 0.5 percent of GDP in 2020, and about 5½ percent of GDP from 2030 and thereafter (see Table V.1).

179. For Option I(age), the required surplus in the most favorable scenario drops by over ½ percent of GDP to 0.8 percent of GDP, while for the least favorable case the drop is from 3¾ percent of GDP without reforms to under 2½ percent. The end-period asset stock falls to 69 percent of GDP in the optimistic case (from 96 percent of GDP) and 207 percent of GDP (from 290 percent of GDP) in the lease favorable case. For Options II(age)-A, B and C, the decline in the funding requirement is of the same order of magnitude as for Option I(age); i.e., just over ½ percent of GDP for the optimistic case and around 1¼ percent for the least favorable one. The end-period asset stock falls by a range of 30–90 percent of GDP. Of interest is Option II(age)—C, where the funding requirement falls to 1⅔ percent of GDP under the optimistic assumptions, not out of line with the working assumptions for prefunding allocations spelled out in the most recent Budget.

Funding requirements with both reforms

180. This section presents the effects of an additional reform to New Zealand Superannuation; namely, a gradual reduction in the benefit ratio from 65 percent to 50 percent of average national wages. While perhaps more contentious than the increase in the retirement age, the combined effects of both reforms are powerful. Reducing benefits by 23 percent over 2015–50 (i.e., by reducing the pension-average national wage ratio from 65 percent to 50 percent) would lower pension expenditure by 0.2 percent of GDP in 2020, 1.7 percent of GDP in 2050 and almost 2 percent of GDP by the end of the projection period. Together, the two reforms result in a reduction in pension expenditure of 0.7 percent of GDP in 2020, 1.7 percent of GDP in 2050 and 3.0 percent of GDP in 2100.

181. Under Option I(both), the contributions to the investment fund are lowered to a range of around 0.4 percent of GDP for the most optimistic parameters to P/a percent of GDP for the least favorable ones, down about 1 percent of GDP and 21A percent of GDP, respectively, over the non-reform Option I(no). The end-period stock of assets for this group ranges from 37 percent of GDP to 113 percent of GDP. In contrast to the results for Options I(no) and I(age), it would appear that even under the least favorable parameter assumptions the transition to a system where all future increases in superannuation expenses are fully funded would be politically feasible.

182. Regarding the cases where all future pension expenditures are financed by returns from the investment fund, under Options II(both)-A and B the simulations show that the annual contributions necessary to fully fund the entire amount of pension expenditures still fall in a range that is clearly not feasible (from 3¼ percent of GDP to 6 percent of GDP). For Option II(no)-C, which has a 100 year transition period to the funding of all pension outlays, the funding requirement ranges from 3¼ percent of GDP under the most favorable assumptions to 1¼ percent of GDP under the least favorable one. Option II(both)-C would therefore appear to be politically feasible only if something close to historical real returns holds up. The end-period asset stocks for these options range from just over 100 percent of GDP to 325 percent of GDP.

183. In sum, the addition of parametric benefit reforms—gradually raising the retirement age and gradually lowering pension benefits—decreases significantly the annual funding requirement in terms of GDP necessary to fully fund superannuation expenses, and takes the Option I scenarios to a range that appears feasible for all assumed parameter values. In contrast, parametric reforms do not have an impact on Option II scenarios large enough to take them to a feasible range, with the exception of Option II(age) and II(both)-C, with a long-term transition period, where benefit reforms reduce the required surplus to 1¼–1⅔ percent of GDP for the most favorable parameters.

F. Macroeconomic Effects of Permanently Funding NZ Superannuation

184. This section looks at the macroeconomic effects of fully funding New Zealand Superannuation expenses, focussing on savings, labor supply and current account. The intent is to obtain a broad direction of the effects of full funding on these variables. A more precise estimation of these effects would require a specified set of behavioral equations which allows for dynamic interaction between tax changes, labor supply, saving behavior and growth.63 Of note, the results hinge critically on what assumption is made regarding the counterfactual or “passive” scenario. A naive approach is adopted regarding this issue—that taxes are raised to finance pension expenditures by the amount necessary to continue the PAYG system under the current benefit structure.64


185. Two issues complicate the analysis of the response in national savings to the changes in superannuation funding detailed above. First, the effect of an increase in public saving on private saving is difficult to pin down. While international experience (Chand and Jaeger, 1996) suggests that there is some Ricardian offset, specifying a precise value for such an offset is problematic. In this particular exercise, the Ricardian offset could be smaller in the cases which include parametric reform, as lower benefit ratios could act as an incentive for higher private saving. By the same token, the offset could be larger, if the greater “certainty” associated with a dedicated pension fund reduces incentives for private saving. Second, under the assumed counterfactual there is no change in savings ratios from the Ricardian perspective since there is no shift in public or private savings—this assumption is adopted for convenience, and it is not clear that this would be the case even without prefunding of superannuation expenses. That is, other changes affecting public or private saving behavior are not considered.

186. Regarding the effect of increased public saving on private saving, an offset of 50 percent is used as an approximation, although given the increase in saving along the simulation path, the calculation of effects of other offsets is straightforward—this is done in Table V.5. Rather than discuss all of the options studied above, the exposition here is limited to those in the range of feasible surpluses (i.e., those with an annual funding requirement of less than 2 percent of GDP) in the most favorable parameter case. This limits the analysis to Options I(no), I(age), I(both), II(age)-C and II(both)-C.

Table V.5.

New Zealand--Effects of Pre-funding Superannuation on National Saving

(In percent of GDP)

article image
Source: Staff estimates.

187. For Option I(no), the average increase in public savings over the simulation period (contributions from surpluses, plus return on assets, less the payout for superannuation) is just under 1 percent of GDP, implying that national savings would increase by about ½ percent per year on average. For Option I(age) and I(both), net public savings would rise on average by 0.7 percent of GDP and 0.4 percent of GDP, respectively, over the simulation period, implying an increase in national savings of 0.4 percent of GDP and 0.2 percent of GDP. Finally, Options II(age)-C and II(both)-C imply increases in net public savings of just over 1.7 percent of GDP and 1.4 percent of GDP on average, respectively, implying an increase in national savings of 0.8 percent of GDP and 0.7 percent of GDP on average over the simulation period.

Labor Supply

188. The effects on labor supply from changes in the tax regime between the full funding options (including, where relevant, parametric reforms) and the counterfactual will only be sketched out here. A summary of the results appears in Table V.6. Assuming a tax ratio to GDP of one-third (the 2000/01 Budget figure is 32.1 percent), a three percentage point increase (decrease) in tax rates would be required to fund a one percent of GDP increase (decrease) in superannuation expenditure. Under current projections, pension expenditures as a percentage of GDP under both the no reform and the two reform scenarios will be roughly constant through the 2010s. Thereafter, under the no reform scenario pension costs will rise by some 3 percent of GDP by 2030, and by 5 percent of GDP by 2100. Under the “age-only” scenario pension costs will rise by 1½ percent of GDP by 2030, and 3½ percent by 2100, while under the “both” reform scenario, pension costs will rise by 1 percent of GDP by 2030 and by an additional 1 percent of GDP by 2100 (see Table V.1). As above, the focus will limited to those cases where the annual funding requirement is less than 2 percent of GDP under the most favorable parameters; i.e., Options I (all cases) and Options II(age)-C and II(both)-C.

Table V.6.

New Zealand: Estimated Labor Supply Response to Pre-funding Superannuation Expenses

article image
Source: Staff estimates.

Assumes a uniform current tax rate of 30 percent and no other policy changes going forward.

From Chaio and Walker [1992], p. 165.

189. Compared with Option I(no) (which has a full funding surplus requirement of 1.4 percent of GDP), taxes in the counterfactual would be some 3½ percentage points lower in 2010, 4½ percentage points higher in 2030 and 10½ percentage points higher by 2100. Relative to Option I(age) (which has a full funding surplus requirement of 0.8 percent of GDP), taxes in the counterfactual would be 2 percentage points lower in 2010, 2 percentage points higher in 2030 and 8 percentage points higher in 2100, while compared with Option I(both) (which has a full funding surplus requirement of about 0.4 percent of GDP), taxes under the counterfactual would be ¾ percentage points lower in 2010, 1¾ percentage points higher in 2030, and 4½ percentage points higher in 2100. For the cases where all pension expenditures going forward would be fully funded, in Option II(age)-C (which has a full funding surplus requirement of 1⅔ percent of GDP), taxes in the counterfactual would be 4½ percentage points lower in 2010, ½ percentage point lower in 2030, and 5½ percentage points higher in 2100, while for Option II (both)-C (which has a funding requirement of 1.3 percent of GDP), taxes in the counterfactual would be 3 percentage points lower in 2010, 1 percentage points lower in 2030 and 1¾ percentage points higher in 2100.

190. These implied tax changes can now be used to estimate the effect of fully funding superannuation expenses on labor supply. This exercise will be undertaken using the average labor supply elasticity figure from Chiao and Walker (1992) for New Zealand of 0.38 (with respect to gross wages in their study—the present exercise uses net wages). It will also be assumed that there is a uniform 30 percent tax rate at present and that the only tax changes going forward relate to superannuation. The outcome of this exercise appears in the final column of Table V.6. In all five case, the labor supply response in the first decades of the century is positive (in the range of ½ percent to 2½ percent) as taxes under the passive scenario taxes are raised by less than under the full funding plans owing to the slow rise in projected pension expenditures. By 2030, the labor supply has turned negative for the Option I cases reflecting the relatively low tax increase for full funding compared to the passive scenario as actual pension costs start to rise; in the cases where all pension expenditures are to be fully funded, the labor supply response remains positive owing to the higher taxes need to fully fund the entirety of future pension expenditure. By 2100 (and, implicitly, onward), the labor supply response is negative for all cases. Over the entire century, simple interpolation yields that labor supply would be some 3.3 percent higher than the counterfactual in Option I(no), 2.2 percent higher in Option I(age) and 1.3 percent higher in Option I(both). The labor supply response in the Option II(age)-C and Option II(both)-C is much more muted owing to the higher taxes relative to the counterfactual. The average labor supply response is ¾ percent for the former and zero for the latter.

191. Holding all else constant, the above suggests that, on average over the simulation period, labor supply, and hence output, would be lower under the counterfactual than under the full funding scenarios for the three Option I cases, and broadly unchanged relative to the two Option II-C cases. (Moreover, given that labor supply is higher under all full funding cases in the year 2100, this would carry through beyond the projection period.) Using a Cobb-Douglas aggregate production function with a labor coefficient two-thirds would result in an output increase of some 2.2 percent per year for Option I(no) relative to the counterfactual, and increases in 1½ percent and almost 1 percent per year for Option I(age) and Option I(both), respectively, relative to the counterfactual.

Effects on the current account

192. The composition of the current account could change significantly depending on whether the fund’s assets are invested abroad or not. In particular, the returns from the fund would have a significant effect on income receipts if invested abroad. Using the extreme assumption that all assets are invested abroad as an illustration, and taking all else as constant, Option I(no) would imply income inflows of 4.9 percent of GDP by the end of the projection period, while the corresponding figures for Options I(age) and I(both) are 3.5 percent of GDP and 1.9 percent of GDP. In the cases where all pension expenditures are (eventually) financed from the investment fund, for Option II(age)-C the income flows by end-period would be 7.1 percent of GDP, while for Option II(both)-C, the corresponding flows would be 5.5 percent of GDP.65

193. Also, the current account would improve, although by relatively small amounts given the changes in national savings estimated above under the assumption of an unchanged investment to GDP ratio. From the discussion above, and using the data from Table V.5 and assuming a 50 percent Ricardian off-set, the current account would improve by 0.2–0.7 percent of GDP for the politically feasible options. Given these modest effects on savings, the relatively large improvement in income receipts would be offset by a significant deterioration elsewhere in the current account relative to the counterfactual.

G. Operational Issues

194. The previous two sections have attempted to gauge the amount of fiscal resources necessary to move toward a fully-funded public pension system in New Zealand—should that objective receive the requisite political support at some time in the future—and the probable macroeconomic implications of such a transition. Underlying the change in policy toward any form of advance funding of New Zealand Superannuation, whether pre-funding or fully-funding, are a host of operational issues—institutional, instrument-based, and political—which this section seeks to discuss. While the proposed pre-funding scenario is used as the basis for this discussion, the points apply to a fully funded scheme of the type outline above as well.

195. The first issue concerns the creation of an institution to undertake advanced funding, its mandate and degree of independence. The March 2000 Budget Policy Statement notes that for the case of pre-funding this would be a Crown entity or similar institution. It would seem important that this entity have a legal mandate to undertake its role as well as the necessary independence to do so. Most importantly, the political environment in which it operates would have a major effect on market confidence regarding New Zealand, which hinges—at least in the foreseeable future; i.e., as long as there remain large current account deficits and net foreign liability positions—on fiscal policy performance.

196. One possibility would be to run the entity along the lines often stipulated for central banks, with separated goal and instrument independence. In the context of a Crown entity charged with undertaking the pre-funding of superannuation, the goals of pre-funding policy (say, returns relative to a benchmark adjusted for the duration of the liabilities) could be spelled out by the Treasury, and the independence to use the instruments necessary to attain those goals could be left to the managers of the entity. The entity could then report to the Treasury on a regular basis, and the fund’s management could be evaluated on its performance relative to the benchmarks laid out in its contract (say, as against market-wide performance). This could be characterized as an “activist” approach.

197. Another alternative is a less activist approach based on an indexing model of investing. Here, it would implicitly be assumed, over a long time horizon, that the market is efficient and that it would not be possible to outperform, systemically, the relevant benchmark. Operationally, a portfolio allocation could be chosen ex ante, and the entity would undertake a passive strategy; i.e., invest in way (as in index funds) so as to mirror the returns of a broadly-defined market aggregate without attempting to actively manage the portfolio in a way that would seek to outperform the market. This approach would have the advantage of minimizing turnover, and thus fund expenses. Indeed, the administrative costs assumed above conform broadly to those of international equity index funds. Also, an index approach could serve as an additional firewall to protect the fund from political influence, thereby strengthening the credibility of the fiscal regime.

198. An additional consideration would be the allocation of the portfolio between debt-equity and domestic-foreign instruments, both of which involve political as well as financial elements. The Government has signaled in the BPS that it envisaged investing some portion of the portfolio invested in equities. The long duration of the pension liability structure would suggest, from an optimal portfolio perspective, that it is appropriate to invest a relatively large proportion of the portfolio in equities, particularly early in the projection period before pension costs start to rise, and when payouts would be minimal A second issue involves the mix of domestic versus foreign instruments. While arguments have been advanced that prefunded balances could help stimulate and deepen local capital markets (Grant, 2000; and Carran, 2000), from a diversification point of view, a relatively small portion of the assets of the fund would be invested domestically. Given New Zealand’s weight in global capital markets of 0.2 percent (WEO, 2000), from a portfolio theory point of view a large proportion of assets invested domestically would not only negatively affect diversification through a “home country bias” (which would be much more serious for New Zealand than, say, for the US), but also compromise either the instrument independence or passive strategy laid out in the preceding paragraph. If there is a sizable foreign currency component in the portfolio, consideration would also need to be given to ways to hedge foreign currency exposure.

Political considerations

199. As noted above, perhaps the most important elements of any advance pension funding scheme would be the perception of political independence and of the will to carry on maintaining the required fiscal surpluses over a long time horizon. On the former, it would seem important to ensure that there were sufficient firewalls around the operation of the fund, particularly given that its size could approach 50 percent of GDP in the next generation and over 100 percent of GDP in the longer run. This would apply as well to the destination of investment of the fund’s resources. On the latter, while in the current environment of strong growth and rising surpluses would seem an opportune time to formulate a policy on the advance funding of superannuation expenses, the problem of time inconsistency may be severely tested if there were to be a major contraction in economic activity. This points to the need to consider a commitment technology to mitigate this problem.

H. Conclusion

200. This paper has looked at issues surrounding the possibility of fully funding part or all of New Zealand’s future superannuation expenditures as a possible subsequent step to the pre-funding proposal of the current government. In particular, it utilized the approach adopted by Feldstein and Samwick, whereby an assumed equity premium is exploited during the transition period (and beyond) to help finance the move to a fully-funded pension scheme. The following points are noteworthy:

  • While New Zealand, like most OECD countries, faces demographic pressure on its pubic pension scheme in the coming century, the size of these projected payouts in terms of GDP is somewhat smaller than average, particularly in the early part of this century. This in part reflects the pension reform efforts of the past 20 years.

  • While estimates of New Zealand’s future health liabilities are of a similar order of magnitude to those of pensions, they are not considered here.

  • A question arises as to what exactly should be fully funded if a political decision were made to go down that policy path: increases (in terms of GDP) in superannuation expenses going forward, or all superannuation expenses, phased in over an appropriate time horizon. The paper looks at both possibilities.

  • Using historical real returns on equities and bonds as a basis for simulations, fully funding the projected increases in superannuation costs looks plausible, even in the absence of reforms (on the order of 1.4 percent of GDP) if historical premia hold up.

  • With reforms involving gradually raising the retirement age and gradually reducing benefits (both of which are in the public domain), fully funding all superannuation expenses over an appropriately long time horizon would appear to be politically feasible.

  • The size of any public investment fund would be large by the end of the 100-year projection period used in this paper regardless of which option were chosen, ranging from about 40 percent of GDP under very favorable return assumptions (and including future benefit reforms) to over 100 percent of GDP under other politically feasible paths.

  • Assessing the macroeconomic implications of full funding is complicated by the counterfactual. Assuming that all superannuation costs were paid out as promised, maintaining the PAYG nature of the current system, full funding could increase the national saving ratio by around ½ percent of GDP per year. Depending on whether the assets were invested abroad, the composition of the current account could change dramatically.

  • The average annual labor supply response relative to the counterfactual case appears to be rather strongly positive for the cases in Option I—fully funding all future increases in pension expenditures. The interpretation is that the magnitude of the funding problem in terms of GDP may be larger than estimated. The average annual effect on output could be in the range of 1.0–2.2 percent of GDP per year for these cases.

  • A key issue is how to structure a public investment fund. Would it be actively or passively managed? How would firewalls be put up to prevent political pressures to spend resources of the fund? How would a credible, time-consistent policy be communicated to markets?

  • Consideration also needs to be given to what kind of independence, if any, the fund should have? Goal and/or instrument independence? Would the objective be solely to smooth the transition to a fully-funded system? Or would there be other goals?


  • Auerbach, Alan J., and Laurence J. Kotlikoff, 1987, Dynamic Fiscal Policy (Cambridge: Cambridge University Press).

  • Callen, Tim, and others, 1997, New Zealand—Selected Issues, IMF Staff Country Report No. 98/3 (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation
  • Carran, John, 1999, “Prefunding Pensions,” Unpublished mimeograph, New Zealand Treasury (September).

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

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

    • Search Google Scholar
    • Export Citation
  • Diamond, Peter A. (ed.), 1999, Issues in Privatizing Social Security: Report of an Expert Panel of the National Academy of Social Insurance, National Academy of Social Sciences (Cambridge: The MIT Press).

    • Search Google Scholar
    • Export Citation
  • Feldstein, Martin, and Andrew Samwick, 1997, The Transition Path in Privatizing Social Security.

  • Feldstein, Martin, 1997, “Transition to a Fully Funded Pension System: Five Economic Issues,” NBER Working Paper 6149 (Cambridge: National Bureau of Economic Research, Inc.).

    • Search Google Scholar
    • Export Citation
  • Grant, 2000, “Prefunding Government Liabilities: Fiscal and Macroeconomic Issues,” Unpublished mimeograph, New Zealand Treasury (June).

    • Search Google Scholar
    • Export Citation
  • International Monetary Fund, 2000, World Economic Outlook (Washington).

  • Mackenzie, G. A., and others, 1997, Pension Regimes and Saving, IMF Occasional Paper No. 153 (Washington: International Monetary Fund).

  • New Zealand Treasury, 2000, “Budget Policy Statement,” March.

  • New Zealand Treasury, 2000, Budget 2000/01, June.

  • Preston, David, 1999, Retirement Income in New Zealand: The Historical Context, written for the Office of the Retirement Commissioner (New Zealand).

    • Search Google Scholar
    • Export Citation
  • Rippe, Richard, 1995, “Further Gains in Corporate Profitability,” Economic Outlook Monthly, August (New York: Prudential Securities).

    • Search Google Scholar
    • Export Citation
  • Roseveare, Deborah, and others, 1996, “Ageing Populations, Pension Systems and Government Budgets: Simulations for 20 OECD Countries,” OECD Economics Department Working Paper No. 168 (Paris: Organization for Economic Co-operation and Development).

    • Search Google Scholar
    • Export Citation
  • Sarel, Michael, 1998, “Pension Reform in New Zealand,” IMF Staff Country Report No. 98/3, January.

  • U.S. Social Security Administration: 2000 OASDI Trustee Report. Available via the Internet:

  • U.S. Social Security Administration, 1995, “Social Security Programs Throughout the World,” Washington, DC.


Prepared by Paul Gruenwald (ext. 38430), who is available to answer questions.


For a useful chronology, see Preston (1999).


For simplicity, the simulations below use 2000 as the base year. This does not imply a suggested timetable for any of the changes discussed.


At a more fundamental level, some observers have questioned the rationale for prefunding as against using surpluses to reduce net debt, and, over time, lower taxes. This paper does not consider the effects of such alternative policy options.


The previous government’s policy was to achieve budgetary balance over the cycle until its debt objectives were achieved. After that, it would run surpluses on average over the cycle, with the size of these surpluses being influenced, inter alia, by the expected fiscal pressures resulting from population aging.


Analytically, the proposed pre-funded scheme could be converted to a fully funded scheme along the lines discussed below by specifying an appropriate asset stock target at the end of the smoothing period. Under the current proposal the end-period asset stock target is zero.


The real US GDP growth projection assumes 2 percent annual productivity growth and the “low cost” (i.e., high birthrate) labor force scenario from the US Social Security Administration (2000). The latter assumes that the labor force growth rate falls from over 1.0 percent this decade to 0.6 percent by 2075, and is taken to be constant thereafter.


Such an exercise is beyond the scope of this paper.


While the higher expenditure on pensions required to financing an increase in pension costs could also be financed by Crown borrowing, agents would perceive these as equivalent to future tax increases to the extent that Ricardian equivalence holds, and the results would be the same as those presented here.


The reader will note that these are simply the end-period annual funding requirements under the various funding objectives.