In brief

  • As the tax reform discussion focuses more sharply on the choice between the goods and services tax (GST) and our personal tax system as potential funding sources, it is worth considering their relative efficiency costs.
  • Most economists would argue that the GST is a more efficient tax than personal income tax, but acknowledge that, in the absence of compensation for lower income people, increases in the former can be viewed as unfair.
  • Most reviews of the Australian tax system advocate greater reliance on the GST, together with targeted compensation because of this efficiency benefit.  Modelling presented in the Government’s Re:think, Better tax system, better Australia Tax Discussion Paper does not tell a different story. The intuitive view of most economists remains valid - the GST is a more efficient source of revenue than the personal income tax. However, when examining modelling, the caveats spelt out by modellers should always be taken into account.

A tale of two taxes

At the PwC Connect conference in Canberra in July 2015, a key panellist referred to Treasury research on the relative efficiency of personal income tax and the GST. This research, he claimed, showed that the efficiency losses associated with each tax were broadly the same (at around 20 cents for every dollar raised).

This statement was a response to the Premiers of our two largest states, Mike Baird and Daniel Andrews, who have tabled competing proposals for funding future health budgets. Baird has proposed an increase in the GST rate to 15 per cent. Andrews wants to raise this money through the personal income tax system, specifically by an increase in the existing Medicare levy by 2 per cent.

Assuming the same amount of money can be raised from either tax base, what are the relative efficiency costs of each approach? Most economists would argue that consumption taxes are less economically damaging than progressive income taxes, acknowledging however that the latter are viewed as fairer. Has the Treasury research quoted by our esteemed panellist changed this?

Taxation and efficiency: what do we mean by this?

Before looking at this research, a brief primer on taxation and efficiency might be useful. Taxation lowers efficiency in the economy by distorting the behaviour of employees, consumers, investors and firms in ways which make them worse off (we are ignoring cases of market failure, which correctly calibrated taxes can correct, boosting efficiency). While taxes raise revenue for needed government services and address income inequality, this comes at an economic cost. This cost, for minor tax changes, can be represented by the concept of ‘marginal excess burden’, that is, the efficiency cost associated with raising an additional dollar of a particular tax.

As a general rule, the more a given tax alters individual behaviour, the greater its efficiency cost (or the higher its excess burden, per dollar raised). Taxes which are difficult to minimise or avoid are the most efficient and have the lowest excess burden. Examples include land taxes and municipal rates. Those which trigger the largest behavioural responses, in contrast, will be less efficient and have a higher excess burden. Corporate income tax, given the mobility of global capital, falls into this category.

Where do the GST and our progressive personal income tax system rank? While both lie partway along the spectrum, most economists would argue that the latter is less efficient, particularly for earners (at both ends of the income spectrum) who face high marginal rates and have discretion over how many hours they will work. Bracket creep, moreover, will increase these economic costs over time. A broadbased consumption tax, in contrast, might be expected to distort behaviour by less. This is because, for many, the scope for discretion over consumption may be lower. This applies particularly to ‘basic necessities’ in household budgets, if not to other types of consumption like restaurant meals.

Australia’s heavy reliance on income tax is inefficient

Reviews of Australia’s tax system have been critical of Australia’s relatively heavy (by OECD standards) reliance on income tax. For this reason, they have favoured a change in the tax mix away from income and in favour of consumption. The OECD Economic Survey of Australia (released in December 2014), for example, recommended that Australia “shift away from income and transaction taxes” and “make greater use of efficient tax bases such as the GST and land tax”. The concluding statement of a recent IMF mission to Australia in June 2015 called for a shift “towards more efficient and simple taxes”, stressing the need (amongst other things) to prevent bracket creep and broaden the base and possibly raising the rate of the GST.

Both reviews, incidentally, acknowledge that those on lower incomes would need to be compensated for any GST increase. As the OECD points out, targeted compensation for this group is a better way to address fairness concerns than current arrangements, whereby all households benefit from GST exemptions. These can be seen as an unmeans tested benefit.

So does recent Treasury research suggest the efficiency costs of personal income and consumption taxes are broadly similar? The Government’s Tax Discussion Paper includes Chart 2.9 which shows that a ‘flat rate income tax’, at least under one scenario, is no more inefficient that the GST. This is drawn from modelling released in a Treasury Working Paper in April 2015.1

But the short answer is no. The Treasury modelling is estimating the efficiency cost of a flat rate of income tax earned by a representative household. Treasury makes clear that the diversity of households, and the effect our progressive rate scale has on their work decisions, are not taken into account in this modelling. This, Treasury acknowledges, “will likely underestimate the marginal excess burden of personal income tax”, in essence, the excess burden of our present personal income tax likely exceeds GST. The Henry Review, using a different model, estimates the efficiency cost of the GST as less than half that of labour income tax. Other models will show different results.

A caution on modelling

This prompts a concluding observation. Modelling should always be interpreted carefully. The results these techniques derive will be affected by the type of model used (partial equilibrium, static general equilibrium, dynamic general equilibrium), the assumptions built into them and the quality of the data used. Modelling can provide invaluable insights, to be sure, but the fine print should always be examined. And current Treasury modelling is not implying that the economic burdens of our personal income tax system and the GST are the same.