Photo by Hello I'm Nik GB on Unsplash

Professor Podger’s main criticism of my paper with Iskhakov is that the income and asset taper rates that we assume in our model are far too low.

He states ‘the paper refers to an income test taper of 27.7% whereas the test at present has a taper of 50%.’ Regarding the asset test, he states ‘the paper uses a simplified taper of 0.5% … whereas the current taper is 7.8%.’

He is absolutely correct about the taper rates we use. This is a point I tried to stress in my article, but I will restate the point here.

We are using effective taper rates

Our study utilized the HILDA data for the period from 2001 to 2016. During that period, the statutory taper rates applied to the Age Pension benefit were 50% on income and 3.9% on assets ($1.50 per fortnight, or $39 per year, reduction in the pension for every $1,000 of assets). However, the law provides a complex set of exemptions and deductions that enable people to avoid those high rates. The most significant of these is the exemption of the ‘principal home’ from the asset test.

As I explained in my article, it is not feasible to model the full complexity of the Age Pension rules in a life-cycle model like ours. A key reason is that the rules treat different assets – such as stocks versus the home versus other real estate – differently. No one has ever succeeded in estimating a life-cycle model with many types of assets that are taxed differently.

Thus, to make life-cycle modeling feasible, we had to develop a simplified version of the Age Pension rules. What we did was simply take the HILDA data from 2001-16 and run a regression of the actual Age Pension benefits that people 65 and over received on their income and their total assets – making no distinction among different types of assets. We actually ran a nonlinear regression that accounts for the fact that only the larger of the income or asset tests is applied.

The regression results indicated that in practice and on average, people’s Age Pension benefit is reduced by 27.7 cents for every dollar they earn, and 5 cents for every $1,000 in assets they hold. That is why I say the Age Pension applied “effective” taper rates of 27.7% on income and 0.5% on assets during the period of our study. These are the rates we assume in our model. They are much lower than the statutory taper rates.

Effective taper rates generate economic incentives

Effective taper rates are what generate economic incentives, not statutory taper rates that people can effectively avoid by utilizing exemptions and deductions (e.g., protecting assets in the form of the principal residence). Therefore, we don’t apologize for taking this approach.

If one wants to model a transfer program one has two sensible choices: 1) Model the statutory benefit rules in their full gory detail, or 2) assume simplified benefit rules that approximate the statutory rules and that capture the effective tax rates generated by the program.

Option 1 is ideal, so my article explained in the section entitled “Some drawbacks of the life-cycle framework” why option 1 was not feasible and why we had to adopt option 2.

Professor Podger seems to suggest a third option –  that we should have simply assumed the effective taper rates were as high as the statutory rates. However, such an approach would give a seriously distorted view of the incentives generated by the program.

Data up until 2016

Professor Podger criticizes us for not using the most up to date Age Pension rules, but that would be odd given that our data ends in 2016.

In 2017, the statutory taper rate for the pension on assets was raised from $1.50  to $3 per fortnight per $1,000 dollars of assets (which corresponds to an increase in the statutory taper rate from 3.9% to 7.8% annually). But the ‘asset free area’ was increased at the same time – meaning that exemptions were made more generous. Thus, whether the 2017 rule changes actually caused effective taper rates to increase is an empirical question. That question can be addressed when more recent HILDA data is released, so we can estimate effective rates for the post-2016 period.

Professor Podger says ‘Keane is similarly incorrect in stating that people with “well over $2 million in assets” (not including the home) can receive some pension.’ My article contains no such statement. When I said one could have over $2 million in assets and still receive the pension, I was referring to total assets including the home. This arises because the home is fully exempt from age pension means test. As I noted earlier, our model makes no distinction among different types of assets because that is not computationally feasible.

Income tax rules accounted

Professor Podger also argues that regarding the taper on income earned, once one accounts for the ‘interaction with the personal income tax’ the effective tax rate on pensioners is ‘generally between 65% and 70%.’ Contrary to his claim, we do account for income tax rules in our modelling.

This means that the overall tax rate on earnings in our model is indeed higher than just the effective taper rate once the income tax is also factored in. However, I note that we don’t model the actual income tax rules either, as that would be infeasible. We again use an approximation to the income tax rules, which we obtain by regressing the actual taxes people pay on their incomes based on the HILDA data. We use a nonlinear regression to capture the progressivity of the tax code.

One might reasonably ask why we insist on using a life-cycle model to analyze the Age Pension, when it requires us to make all these simplifying assumptions about the pension rules and the tax rules? Why not use a more traditional static simulation model, which could build in the full complexity of the rules?

The reason is that changes in Age Pension rules alter people’ incentives to work and save throughout their lives, not just their behaviour after they reach age 65. Our life-cycle model can account for those changing incentives, while a static simulation model would not. We think this is an important enough advantage that it is worth simplifying the rules so as to make a life-cycle model feasible.

We agree on the need to tighten the asset taper

Finally, Professor Podger reaches a conclusion that is completely consistent with our recommendations. He states

‘the one area where tightening the means test would clearly be sensible policy would be to include the home (main residence) in the assets test above some high threshold. However, this should be seen as a longer term initiative so as to avoid retrospectivity, allow people to plan for their retirement with reasonable certainty …’

which is perfectly consistent with our conclusion 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.’

To increase effective taper rates, generous exemptions and deductions should be removed

Perhaps I should have been clearer that the way to increase effective taper rates for the pension is precisely by removing exemptions and deductions, for example, by scaling back the main residence exemption. Just raising statutory taper rates obviously can’t do much to raise effective rates if one leaves generous exemptions untouched.

While we end up reaching a similar conclusion, it is worth noting there is a fundamental difference in how we get there. Our conclusion derives from model-based evidence, as our model says raising effective taper rates combined with reducing the income tax would be welfare enhancing. Professor Podger’s conclusion about what would “clearly be sensible policy” is based purely on abstract reasoning.

As I emphasized in my article, it is not a priori obvious if higher effective taper rates are a good or bad thing, as the theoretical impact of taper rates on labor supply is ambiguous. Thus, the impact of raising taper rates is an empirical question, and one needs a model to assess whether it is a good idea. Modelling always requires some simplifications of reality, which can always be easily criticized.


Further reading

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

Better Targeting of Australia’s Age Pension, by Michael Keane

Leave a comment

Your email address will not be published. Required fields are marked *