Many governments provide tax concessions for private pension contributions. These concessions are designed to increase households’ self-reliance in retirement and reduce government expenditure on public pensions. In Australia, tax concessions are mainly applied on contributions made by employers — mandatory 10% contributions under the Super Guarantee — plus voluntary salary-sacrificed contributions paid out of their employees’ salary on their behalf. Concessional contributions are taxed at a flat rate of 15 cents in the dollar instead of the marginal income tax rate. For high-income earners who face a marginal tax rate of 45 cents in the dollar, this means that for every dollar voluntarily contributed through salary sacrifice, they avoid up to 30 cents in tax.
However, concerns about the lost tax revenue from these concessions means that many governments impose individual caps on their use. In Australia, Treasury estimates suggest that the foregone tax revenue from these concessions was $20.5 billion in 2021-22, or around one quarter of the annual cost of the Age Pension. These concerns motivated reductions in age-specific caps between 2007-08 and 2016-17 in Australia, which saw caps fall from $100,000 to $35,000 for those aged 50 and older and from $50,000 to $30,000 for those under 50.
In a recent paper, we examined responses to the tightening of age-specific caps on concessional contributions since 2007-08 and the resulting impacts on tax revenue. Our analysis used data from the Australian Taxation Office’s (ATO’s) Australian Longitudinal Information File (ALife), which links tax and superannuation data over time. We compare the tax declarations of ‘high contributors’ (top 5%, who earn around $200,000 on average per year) aged 48 to 51 who experience a change in their cap from one year to the next to the declarations of individuals in adjacent birth cohorts who experience no cap change over the same years.
High contributors reduce total contributions and report lower income
Our estimates indicate that high contributors reduce their total contributions by $6,750 (19%) on average in response to a $25,000 reduction in the cap, which occurred in 2009-10 for individuals aged under 50 (from $50,000 to $25,000). This is driven by a decrease in concessional contributions, with minimal effect on non-concessional contributions. Restricting concessional contributions means that a larger fraction of high contributors’ employment income is paid directly into their bank accounts instead of into superannuation, which is taxed at the marginal tax rate instead of the 15 cents concessional rate. In response to a higher tax rate from a $25,000 cap decrease, high contributors are estimated to reduce their total taxable income by $4,375 (2.2%). The decline is from employment sources, lower reported wages and business income, with no change in unearned income or deductions. There is no change in income among lower contributors, who are not affected by the less generous tax concessions.
Lower employment income is found for both employees and the self-employed, with larger reductions for the latter. For both groups, the higher effective tax rate may have discouraged high contributors from working long hours, applying for higher-paid jobs or seeking promotions. Supporting this interpretation, we find no evidence that the effects on income are driven by individuals shifting income to their spouse (who may have more room to top up their superannuation without breaching the cap) or to periods when the cap is higher. Nonetheless, we cannot rule out other tax-minimising behaviours that may lead to changes in employment income, such as business owners drawing a wage to top-up their superannuation.
We find no change in tax revenue
We find that the tightening of concessional caps over time has had a net zero effect on tax revenue, with the increase in tax revenue from lower concessional contributions offset by a decrease in taxable income from lower employment income.
Our findings underline the importance of understanding how high-income earners respond to tax and superannuation settings. These taxpayers are extremely influential in determining overall income tax receipt — in 2018/19, one third of total income tax revenue came from the top 3.5% of taxpayers — and they often behave very differently to the average Australian. Compared to the average, they have high rates of financial literacy, are highly engaged in financial decisions and are well-resourced to respond in sophisticated ways to minimise their tax liability.
The release of the ATO’s ALife data provides new opportunities for research insight into the behaviour of these financially engaged individuals.
The study is published as: Chan, MK, Morris, T, Polidano, C, & Vu, H 2022, ‘Income and saving responses to tax incentives for private retirement savings’, Journal of Public Economics, vol. 206, 104598. Working paper available here.
Interesting paper well done! One very minor matter though super tax concessions are roughly 50% income concessions and 50% contributions. The $20b figure quoted is not 25% of age pension liability. Age pension liability is around $50b. I think you are including the costs of other costs such as age care which budget papers sometimes lump together.
Treasury is silly. Years ago, as Secretary to the IDC on Retirement Incomes, I argued for forward income averaging and that there should be no limits on deductions up to personal exertion income, no preservation and no taxation of income or investment earnings BUT –
1 all withdrawals must come out as fully assessable pensions or term certain annuities based on life expectancy at the time, and
2 all withdrawals should result in a dollar for dollar reduction in any social security or other benefits.
If we had done that, much of social security would be effectively replaced by voluntary self support as most people don’t want to live on the age pension. It would, however, allowed people to retrain or take time out raising children (which are more valuable as a social and economic investment than money in super.)