Better Targeting of Australia’s Age Pension

The Australian retirement income system has been praised as one of the best in the world. In the Melbourne Mercer Global Pension Index, it was typically ranked third from 2012-16, behind only Denmark and The Netherlands. But there is still room for improvement. In particular, in a recent paper with Fedor Iskhakov of ANU, we argue that Australia’s old age pension could be better targeted towards those most in need, while also enhancing work incentives.

The current retirement system in Australia combines private accounts with a publicly funded safety-net pension meant to help seniors with relatively low income.

The private account system, known as ‘superannuation,’ is a defined contribution pension scheme. Employers are required to make contributions equal to a percentage of workers’ earnings (currently 9.5%) into privately held ‘super’ accounts. Workers can then draw from the accumulated contributions and earnings in these tax advantaged accounts to help finance their retirement.

The safety-net, known as the ‘Age Pension,’ is a traditional means-tested public transfer program, financed by current government revenues. The Age Pension provides financial support for senior citizens with relatively low income and assets, including for low-wage workers (or those not in the work force), who do not have adequate superannuation to fund their retirement.

Australia’s Age Pension is not well targeted

A key advantage of the superannuation system, which stands in sharp contrast to the US and many other OECD countries, is that the private accounts take a large part of the cost of financing retirement off the federal government budget. Australian government spending on the Age Pension is only 2.9% of GDP, less than half the OECD average spend on public pensions. Consequently, unlike most other OECD countries, Australia has not experienced a rapidly growing budgetary burden arising from the need to fund government old age pensions. In fact, the cost of the Age Pension as a share of Australia’s GDP has remained fairly stable since 1995.

Despite its advantages, the Australian pension system is not perfect. The Age Pension is poorly targeted. Roughly 80% of eligible Australians aged 65 and over receive at least some Age Pension benefit – not what one would expect of a safety-net program! The reason this occurs is that the rates at which age pension benefits are reduced as seniors have more income and/or assets are quite low, largely due to generous income and asset exemptions. For example, a couple can earn about $75,000 per annum and/or have assets well over $2 million, and still be eligible for some benefits. Partly due to this loose means-testing regime, the Age Pension is the largest social welfare program in the federal budget, costing around $50 billion annually.

In a report released in February 2014, the National Commission of Audit (NCOA), an independent body established by the Australian Government to review a wide range of government programs, recommended that Age Pension ‘taper rates’ – the rates at which benefits are reduced as income and/or assets increase – should be substantially increased to improve program targeting. Specifically, the NCOA called for ‘tighter targeting of eligibility’ by ‘increasing the income test withdrawal (taper) rate from 50 per cent to 75 per cent,’ with the change applied prospectively to new recipients of the Age Pension from 2027-28 onwards. The report also called for higher effective taper rates on assets, to be achieved by reducing asset exemptions

Would higher taper rates increase or reduce labour supply?

Obviously, higher taper rates would improve program targeting, by reducing Age Pension benefits paid to seniors with relatively high income or assets. On the one hand, a higher taper rate may reduce labour supply because program participants face higher effective marginal tax rates if they work. Economists call this the ‘substitution’ effect.  If this effect is strong enough, then increasing taper rates would probably be a bad idea.

But economists have understood since the 1980s that transfer program taper rates have ambiguous effects on labour supply. In particular, higher taper rates may increase labour supply, simply because they tend to reduce the fraction of the population who rely on the program, causing more people to work. This may be called an ‘eligibility’ effect.

Furthermore, if one adopts a life-cycle perspective, it becomes clear that the very existence of the Age Pension may reduce labour supply over the working life, because the Age Pension reduces the need to save for retirement. Economists call this the ‘income’ effect. Higher taper rates would reduce this income effect by making the pension less generous, thus inducing people to work more.

Clearly, whether an increase in Age Pension taper rates is really a good idea depends on the labour supply effects of such a tax increase. And this is an empirical question, as it depends on the balance of the three effects described above.

A life-cycle model

In order to study the design of the Age Pension, we built a model of life-cycle labour supply, saving, and consumption that incorporates the main features of the Australian retirement income system. We used the model to simulate the impact of changes in Age Pension rules on both program targeting and labour supply. In this way, we hoped to find ways to improve the design of the program.

Using a dynamic life-cycle model to evaluate a means-tested transfer program like the age pension is quite novel. Such programs are typically evaluated using static simulation models, and the literature using dynamic models is very small indeed. Our dynamic model has the key advantage that it can capture not just how changes in age pension rules affect behaviour at ages 65 and over, but at younger ages as well. For example, our model can capture how tighter age pension means tests affect not only labour supply of the 65+ population, but also the labour supply and rates of saving for retirement by younger workers. This is important, as a policy that increases labour supply of the over 65 population may be very misguided if it reduces labour supply of younger workers.

Our life-cycle model incorporates several important features of the economic environment that have not been previously captured in one model, including human capital accumulation, liquidity constraints, and the “bunching” of work hours at discrete levels. The fact that people are not typically free to choose any level of work hours, but instead face a constrained choice among a discrete set of hours levels offered by employers, has not previously been incorporated in life-cycle labour supply models.

We calibrate our model to fit data on male household heads from the “Household Income and Labour Dynamics in Australia” survey, known as HILDA. The model provides a good fit to life-cycle profiles of consumption, labour supply (i.e., the proportion of men working at each discrete hour’s level at each age), wage levels, wage dispersion, wealth profiles, superannuation balances, and annual transition rates between employment states. The good fit of the model to a wide range of behaviours and outcomes gives us some confidence in using it to predict the potential impact of changes in Age Pension rules.

Some drawbacks of the life-cycle framework

The downside of dynamic life-cycle models is they are very hard to solve, calibrate and simulate. This means we must introduce some important simplifications of reality to make the modelling feasible. First, we only model the behaviour of male household heads. Unfortunately, if we tried to also include women and model the formation of households, it would be an even more complex enterprise that would severely strain current computational limits.

Second, again due to computational limits, we can’t model the full complexity of the benefit rules. During the period of our study, Age Pension benefits were reduced by statutory taper rates of 50 cents per dollar on income in excess of a tax free limit (for example, about $7,000 per annum for a couple), and 1.5 cents per dollar on assets (excluding the family home) above tax free limits (e.g., roughly $286,000 for a home-owning couple), with only the stricter of the two tapers applied. It is not feasible for us to model the complex system of income and asset exemptions in detail, especially as they differ depending on household structure and the types of assets held.

Instead, we fit a relatively simple, but still fairly accurate, approximation to the age pension rules using the HILDA data, and use that approximation within our model in place of the statutory rules. Our approximate age pension benefit rule has a taper rate of 27.7 cents per dollar of earnings, and a taper rate on assets of only 1/2 cent per year per dollar of assets over $117,000 (with only the larger of the two reductions applied). This implies one can have assets of almost $2.6 million, and/or earnings of about $45,000 per year, before age pension benefits are reduced to zero. Notice how the effective taper rates we fit from the data are much lower than the statutory taper rates – the difference arising primarily from the generous system of exemptions and deductions.

Most workers would be better off if…

Our model implies that the Age Pension system, as currently designed, significantly reduces male labour supply because of the ‘income effect’. The program reduces the need to save for retirement, which in turn reduces peoples’ incentive to work.

To quantify the labour supply effect, we predict that if the Age Pension were entirely eliminated, and the savings used to finance an across the board cut in income tax rates, aggregate labour supply would increase by 5.8%. Our model predicts most workers would be better off under this scenario. A important condition for that result to hold is that workers must be aware of the policy change from the time they first enter the labour market, so they can plan their retirement savings accordingly.

Protecting low income workers

Not surprisingly, however, workers at low education and skill levels would be worse off, and these are precisely the people the program is meant to protect. To address this, we used our model to simulate a doubling of the income and asset taper rates for the Age Pension. This means increasing the effective taper rate on earnings from 27.7 cents on the dollar to 55.4 cents per dollar, and increasing the effective taper rate on assets from ½ cent on the dollar to one cent per dollar.

Our model predicts this program change would increase labour supply of age 65+ college graduates, reducing their dependence on the Age Pension. But the reverse pattern holds for high school dropouts – i.e., they work less at ages 65+ and rely more on the Age Pension. Thus, the program becomes better targeted towards the low-income population, which is its original intention.

Our simulations suggest that a doubling of taper rates would raise enough additional revenue to fund a 5.9% reduction of income tax rates. Furthermore, all workers – including low skill workers – would be better off if they faced this environment of higher taper rates combined with lower tax rates – provided, as before, that early awareness allowed workers to plan their retirement savings accordingly. Thus, we predict that a doubling of effective taper rates combined with a 5.9% reduction in tax rates would leave everyone better off. This is what economists call a ‘Pareto improving’ policy change.

In the meantime, the Age Pension still needs to be generous

Our results should be viewed in historical context. The Superannuation Guarantee was created in 1992 and contributions were not ramped up to 8% of earnings until 2000. Hence, current retirees were not able to contribute to super for their whole working life. Thus, the current Age Pension program needs to be rather generous because we are still in a transition period where super balances have not reached their full level.

However, starting in roughly 2030-2035, we can expect that new retirees who have worked full time for their working life will have made a full 40 to 45 years of super contributions. Thus, our policy advice is that Age Pension effective taper rates should be gradually increased over the next 15 years until the program is better targeted as a safety net for the low income population. This is very similar to the NCOA recommendation noted earlier. Crucially, the taper rate increases must be combined with income tax rate cuts for the policy to be Pareto improving.

A final caveat to our conclusions is that our model only applies to male household heads, so one should carefully consider whether the proposed policy change would also benefit women. Given that the consensus of the labour supply literature is that women’s labour supply is even more responsive to tax rates than men’s, we conjecture that women would also benefit from the combination of taper rate increases and income tax cuts that we propose. But one would obviously want to analyse that conjecture more carefully before changing the rules.

 

Further reading

Modelling the Age Pension Taper: Reply to Andrew Podger, by Michael Keane

No Case for Tightening the Age Pension Means Test: A Response to Michael Keane’s Analysis, by Andrew Podger