Photo by Anne Nygård on Unsplash

The belief that superannuation is generally lightly taxed, even subsidised, is widely held not only in official policymaking circles, but also amongst independent economists and social commentators. Especially emphasised are the tax advantages enjoyed by upper income earners. This virtual consensus helps explain the limits that have been imposed on voluntary contributions but also gives energy to those advocating increased taxes, and to those urging government to influence, even force, retirees to boost their own consumption.

In a recently published journal article, I project that over twenty-five years, upper-income earners will have paid over 50 cents for every dollar in their Superannuation Guarantee balances under current rules. An upper-income earner making 25 annual voluntary non-concessional superannuation contributions will have paid over 90 cents of tax for every dollar in the final balance.

Past conditions, more generous than current ones, generated results some find objectionable. My work is about current and prospective tax burdens.

Measuring superannuation tax burdens

The effective tax burden is the reduction in the taxpayer’s benefit from the activity being taxed. For the goods and services tax and the personal income tax, the salient benefit is the post-tax income; for superannuation, it is the post-tax balance, which is the primary motivation both for the system’s existence and for the decisions taken by savers.

Thus, the proposed definition of effective tax rate is (the effective burden) ÷ (the post-tax balance).

Do high income earners really enjoy large negative effective rates?

In influential estimates of effective tax rates commissioned by Treasury for the 2019 Callaghan Retirement Income Review, the Tax and Transfer Policy Institute (TTPI) generated the eye-catching finding (see Chart 5A-9 in the Callaghan report) that, despite paying taxes on (mandatory) Superannuation Guarantee contributions and fund earnings, high income earners enjoy large negative effective rates.

This occurs because TTPI excluded the personal tax paid on the income that funded the investment.

An implicit assumption behind the TTPI’s approach is that the decision to earn and the decision to save are independent. This is obviously wrong when the additional income is specifically to boost a superannuation account. More importantly, it ignores the fact that the excess burdens of taxes and the substitution effect both depend on effective rates that account for all the taxes that are necessary to finance eventual consumption. Since earning an income is a necessary step towards saving, all the taxes that bear on the marginal dollar saved need to be included in calculating effective tax rates.

The burden should therefore be measured against a zero-tax baseline. This is not to endow a zero-tax base with normative standing: ‘The marginal rate is 47 per cent’ is taken to mean that the rate is 47 per cent higher than zero, not that zero was chosen for its normative significance.

Why, if super is taxed heavily, has there been so much in voluntary contributions that limits have been legislated? A high income-elasticity of demand for consumption smoothing can manifest in high voluntary saving by medium- and high-income individuals. Despite the tax concessions, they would pay very high effective rates on superannuation; however, the rates are lower than for most close substitutes.

For example, if the upper-income earner mentioned above had invested in a term deposit, then instead of 90 cents tax for every dollar in the year-25 balance, the burden would be 120 cents.

Estimates of tax concessions

The Treasury’s annual ‘Tax Expenditure and Insights Statement’ provide estimates, for a large range of tax bases, of the quantum of ‘tax concessions’ as an aggregate across all taxpayers. As the normative counterfactual, Treasury uses its ideal, the ‘comprehensive income tax’, under which the person’s marginal rate of income tax is imposed on the income funding the (otherwise tax-free) contributions, and on fund earnings. However, Treasury does not provide an estimate of the rate of superannuation tax expenditures, relative to some apposite denominator.

For retirement savings, a sizeable majority of the Organisation for Economic Co-operation and Development (OECD) apply less punitive tax regimes (and the 2010 Henry tax review agreed). Of 42 OECD countries, only twelve, including Australia, levy taxes at two stages; twenty tax withdrawals only; five tax contributions only; four impose no taxes; and one taxes fund earnings only.

The OECD in 2018 estimated lifetime tax preference afforded to middle income earners in over forty countries’ private pension plans, including the Superannuation Guarantee system – the OECD’s index was the present value of the tax concessions, as a ratio of contributions made. The normative tax system was that applied to the country’s ‘traditional’ savings accounts. Australia, with a ‘tax advantage’ of about 25 per cent, ranked 24th out of 42 countries.

The OECD approach overstates how well-off Australian superannuants are compared to those elsewhere. In particular, in most other OECD countries, private savings comprise a small share of retirement plans, and most people benefit from generous public defined-benefit schemes where generally the only tax is on withdrawals. Moreover, in Australia, the age pension is highly means tested and one aim of the Superannuation Guarantee is to reduce eligibility for it even further, which will especially affect not only low-income earners but also many on medium incomes. The actuarial value of full entitlement is currently close to $1 million and even the part pension provides access to valuable benefits. A proper assessment of the burden being imposed on superannuants would take the foregone benefit into account, as is typically done in calculating effective rates on income.

To government and lobbyists, Treasury’s estimates of tax expenditures signal where extra revenue could be garnered; also, they are used for distributional analyses. Treasury compares the superannuation tax expenditures with the year’s actual expenditure on the age pension. It may be that such a comparison is useful in budget planning (and in stirring up envy); but otherwise, it has no obvious public policy relevance.

None of the calculations of the burden on contributors take account of the history and empirical research that shows that higher wages have been traded-off against an improvement in ‘the social wage’. Also, to the superannuation tax burden should be added the ‘tax-equivalents’ of the compulsion to contribute and of the restrictions on early access to the balance.

Another typically overlooked consideration is risk. The public commentary’s emphasis on the alleged generosity of tax concessions creates the impression that those who (as a result of a lifetime of forced saving under the Superannuation Guarantee scheme) will be entirely or largely ineligible for the pension, nonetheless enjoy the benefits of a splendid financial deal, well worth any sacrifices they have been forced to make.

In reality, the income stream they have been compelled to acquire is highly uncertain, and the risk could only be removed by acquiring an annuity—which is very costly.

Conclusion

I offer an economically meaningful but intuitive way of stating the effective burden of superannuation taxes, namely, as the tax paid per $100 of the post-tax balance.

When all the relevant taxes are included, for the Superannuation Guarantee balance, the tax burdens on lower income groups are heavier than those imposed by the income tax, but for upper income earners, appreciably lighter. However, when upper-income earners make voluntary contributions that qualify for the 15 per cent tax rate on earnings for twenty-five years or longer, they can expect to incur effective rates of 100 per cent or greater. These burdens are before any adjustments are made for losses of pension entitlements, wage rate suppression, compulsion and restricted access to the owner’s funds.

 

This blog draws on my recent article in Economic Papers, available at http://doi.org/10.1111/1759-3441.12434

 

Comments are closed.