تجزیه و تحلیل بین زمانی تعادل عمومی از آزمون سن بازنشستگی استرالیا

The Australian age pension is somewhat unusual among developed countries in that it is means tested against both the claimant’s income and assets. While means testing of age pensions facilitates the aims of directing public pensions to those senior individuals most in need and of containing pension expenditures by governments, it also has the effect of changing the incentives of individuals to work and save. This paper examines the implications of the Australian means tested age pension for incentives of individuals to save and work, for government financial commitments and for the welfare of individuals. To this end, we develop an overlapping generations model of the Australian economy that incorporates the essential features of the Australian pension, superannuation and taxation policy settings and use it to explore the implications of several hypothetical policy changes that relax the means test of the age pension. Our results confirm that the existing means-tested, age pension represents a disincentive for some older Australians to work.

The Australian age pension provides a safety net payment to older people unable to support themselves financially in their retirement. The Australian pension is somewhat unusual among developed countries in that it is means tested against both the claimant’s income and assets. Although 18 OECD countries have targeted pension programs, many of these countries also pay some flat-rate basic pension that is not means tested (OECD, 2005). Furthermore, in most countries targeted public pensions are only income tested and not subject to both income and asset means tests as in Australia. While the targeting or means testing of age pensions facilitates the aims of directing public pensions to those senior individuals most in need and of containing pension expenditures by governments, it also has the effect of changing the incentives of individuals to work and save. By reducing the pension payments as private income from labour earnings or interest income from past asset accumulations increase, the effective marginal tax rates are increased for those in receipt of pensions. This has the potential to alter retirement and labour supply decisions by eligible pensioners and, moreover, to alter their labour supply and savings decisions throughout the life cycle. The purpose of the present paper is to examine the implications of the Australian means tested age pension for incentives of individuals to save and work, for government financial commitments and for the welfare of individuals. One specific aim is to determine whether the age pension means test provides disincentives for senior Australians to engage in some part-time work and incentives to bring forward retirement. Other aims are to determine the effects upon saving over the life cycle and upon individual welfare, and to draw out the macroeconomic and fiscal implications. To this end, we develop an overlapping generations model of the Australian economy that incorporates the essential features of the Australian pension, superannuation and taxation policy settings and use it to explore the implications of several hypothetical policy changes that relax the means test of the age pension. In contrast with Australian micro-simulation models with limited or no behavioural policy effects (Atkinson et al., 1996 and Rothman, 1998), our inter-temporal general equilibrium model with overlapping generations of heterogeneous households allows us to deal with life-cycle saving, consumption and hours of work decisions of households and, moreover, allows us to numerically determine the inter-generational, welfare and macroeconomic implications of the age pension policy and potential changes to this policy. We explore the implications of several hypothetical policy changes that relax the means test of the Australian age pension. The first of these is the complete removal of the means test, which amounts to installing a universal pension for all seniors of at least 65 years of age. As variations on this theme, we also give brief consideration to the removal of the income test alone and then removal of the assets test alone. Second, we simulate the implications of several hypothetical policies that partially relax the income means test, since it is the income test that is binding for most households. We consider three partial relaxations to the income test: (i) the halving of the taper rate applied to the income test, (ii) halving the labour earnings to be included as assessable income for the test and (iii) halving the investment earnings to be included as assessable income for the test. The first of these policies relaxes the income means test in a manner neutral to the source of income, while the other two bias the relaxation to labour and investment earnings respectively. Some of the policy changes to the age pension that we examine in this paper have been the subject of debate in the broader literature. First, our examination of the consequences of a complete removal of the means test is motivated by the fact that many countries do not have such a targeted pension (e.g., New Zealand) and by the considerable literature on whether publicly provided pension payments should be paid to all citizens from a certain age (i.e., universal age pension) or targeted to those in need (i.e., means tested pension); see, for example, Mitchell et al. (1994).1 Arguments for the universal age pension are that it would simplify the public pension pillar, reduce administrative costs and avoid very high effective marginal tax rates, thus removing disincentives to work and save for the elderly. On the other hand, government expenditures on targeted pensions are substantially lower than on universal pension payments. Second, others have suggested only partial adjustments to the Australian age pension means test to encourage pensioners to provide some work and to avoid substantial costs that would result from the shift to the universal pension. Dunsford and Rice (2004) propose the removal of labour earnings from the income test, while the Australian government enacted a halving of labour earnings (with a limit) to be counted as assessable income as part of its 2009/10 budget, thus motivating policy change (ii) above. Empirical literature generally supports the view that labour supply responds positively to relaxation of earnings testing of social security benefits. The literature that examines aggregate labour supply responses by individuals eligible for pension benefits to changes in the earnings test includes studies by Baker and Benjamin (1999) for Canada, Disney and Smith (2002) for the UK, and Burtless and Moffitt (1985) and Friedberg (2000) for the US. Using a dynamic programming, structural model of households, Rust and Phelan (1997) conclude that the discretionary effects of the US progressive tax schedule and the transfer features of the social security system combine to encourage retirement earlier than otherwise would be the case. French, 2005 and Benitez-Silva and Heiland, 2007, also using structural models of retirement, show positive effects of the earnings test removal in the US on labour supply of those at and near the threshold of the earnings test. Similarly, Määttönen and Poutvaara (2007) simulate lower aggregate labour supply and welfare when they incorporate an earnings test of social security benefits into their general equilibrium overlapping generations model that is calibrated to the US. In addition to work disincentives, the age pension means tests have the potential to distort saving decisions over the life-cycle of individuals. This is because the age pension declines as assets (other than the family home) increase beyond an assets threshold under the assets test, and because income from assets (such as interest, dividends and rent) is included as assessable income under the income test. Simulation results by Sefton et al. (2008) demonstrate that the reduction in the taper rate of the means tested benefits from 100% to 40% implemented in the UK in 2003 would increase labour supply and saving of low-income households but have the opposite effects on high-income households, resulting in higher overall employment and lower aggregate saving. Similarly, Neumark and Power (1998) find empirical evidence that high Supplementary Security Income benefits for the elderly in the US, which are means tested, reduce pre-retirement saving of those who are likely participants in the program. Our simulation results demonstrate that the hypothetical pension policy changes considered have a significant behavioural impact on middle-income households that are income tested in the pre-reform case with the existing means test. These households work longer hours at older ages and postpone their full retirement as the penalty of high effective marginal tax rates is eliminated or reduced. Low-income households on the full age pension in the benchmark steady state are affected only through general equilibrium impacts on the wage rate and the consumption tax rate. High-income households, for whom the assets test binds, are directly affected only by the means test removal as they then receive the universal age pension, which has a negative income effect on their labour supply. Thus, the analysis indicates that the existing means test represents a labour supply disincentive for some older Australians. The macroeconomic simulation results show that the means test abolition would raise the government costs on age pension substantially relative to the costs arising from the investment income and taper rate changes to the income tests. Only the policy of a 50% exemption for labour earnings results in lower pension expenditures and a lower consumption tax rate. All policies yield an aggregate efficiency gain, allowing for potential Pareto improvements in welfare. The rest of this paper is organized as follows. In order to put our modeling and simulation results into context, the next section provides a brief overview of the Australian retirement system including a discussion of age pension, superannuation and income taxation policies. Following that, Section 3 describes the overlapping generations model that embodies the essential aspects of the retirement system and provides a dynamic general equilibrium framework for the policy simulations. The calibration of the model to the Australian economy and a discussion of the benchmark model solution are dealt with in Section 4. Section 5 reports and discusses the results from the simulation of a policy change that completely removes the means test for the age pension. Section 6 then deals with several policies that relax the income means test for the age pension, since it is this test that binds for most households. The discussions of each policy reform include consideration of household life cycle and welfare effects as well as macroeconomic and fiscal implications. The final section offers some concluding remarks.

This paper applies an overlapping generations model to simulate and evaluate several hypothetical policies that change the means testing of the Australian age pension. The policies included the complete abolition of the means test, and several relaxations of the income test component of the means test – a halving of the taper rate for all assessable income, halving the taper rate for only labour earnings and halving it for only investment income. There are several main conclusions that arise from the analysis. First, the results demonstrate that the existing means-tested age pension provides a strong disincentive for older middle-income households to work. These disincentives arise primarily because of the reduction in the age pension that a pensioner receives as a result of receiving income from extra labour effort or investments via the taper rate and are reflected in a high effective marginal rate of taxation. The numerical simulations of all policy changes to the pension means test, except the reduction in the taper rate for investment income, show that middle-income households work longer hours at older ages and delay their retirement from the workforce as the penalty of higher effective marginal tax rates for working beyond the age pension age is completely eliminated (in case of the means test abolition policy) or significantly reduced (in case of the policies of the 50% reductions in the income taper rate and in the labour earnings taper rate). Low-income households, for whom the means test is not binding (so they get the full pension) and productivity at older ages is low, are not affected by the pension policy changes directly, but only through the general equilibrium impact on the consumption tax rate, wage rate and bequests. Similarly, high-income households are assets tested at the early age pension ages, do not get the pension and respond little to changes to the income test. Overall, the result is that the positive labour supply of middle-income households yields a higher aggregate labour supply for the economy resulting in greater capital stock, output and consumption levels. The complete means test removal, however, generates a pure income effect on high-income households, reducing their labour supply. Similar effects on the labour supply of the elderly were found by Friedberg (2000), who finds a significantly positive impact of the earnings test removal in the US on hours worked and earnings of those individuals at or near the earnings test threshold, but a negative effect on labour supply of more affluent elderly. The second main conclusion is that the ranking of the different means test policies contemplated in the paper depends very much upon the objective of the policy maker. (a) If the objective is to increase the aggregate labour supply and to thereby generate a higher steady state capital stock and output level then the most effective of our policies is to completely remove the means test and so make the pension universal. This policy yields a simulated 1.134% increase in the size of the economy measured by these macroeconomic variables. The means test removal eliminates the distortion arising from the reduction in pension payments to senior Australians as a result of higher labour earnings or investment income. The next ranked policy is that of halving the taper rate for labour earnings, which directly targets labour supply, followed by the blunter policy of reducing the taper rate for all income sources. Reducing the taper rate on investment income is quite an ineffective policy to increase labour supply. (b) On the other hand, if the objective is to minimize government expenditures on the age pension the ranking changes in an interesting way. The labour income taper rate reduction policy now moves to top place, since it encourages seniors to work longer and reduce reliance upon age pension income. The other income test relaxation policies move up a rank and, perhaps not surprisingly, the means test abolition policy moves directly to bottom rank, since all seniors get the age pension. (c) Finally, if the objective is to maximize aggregate efficiency (greatest potential welfare gain for future generations in our framework) then the means test abolition policy regains the top rank, since it has removed the means test distortions, while the previously top ranked labour income policy moves to the bottom. The combined taper rate reduction policy is second placed, since it is reducing a distortion but doing it evenly with respect to source of income. In short, these observations demonstrate that the different policies have different effects on different variables of interest and, hence, that their assessment depend on the objective. A desirable extension of the model would be to implement non-stationary demographics with an increasing share of older households in the total population to account for population ageing. Because the growth rate of population is expected to decline and the proportion of those aged 65 years and over is projected to increase significantly in the next 50 years (Productivity Commission, 2005), the fiscal pressures on government age pension expenditures are likely to be much stronger than found in the present paper, which assumes stationary demographics and a constant population growth rate. Another desirable extension of the model would be to include wage rate uncertainty, which introduces a reason for precautionary saving on the part of households. It also provides the means testing of the age pension with a social insurance role against income risk.