Models only give part answer to real tax reform

Economic models are not likely to give Prime Minister Malcolm Turnbull any magic answers on tax reform. Lukas Coch/AAP

The current debate on tax reform in Australia is strong on rhetoric but largely devoid of hard evidence. Also missing are clear proposals from the government that can be assessed on their respective merits.

The purported benefits of reform centre on worthy outcomes such as a larger economy, reduced government deficits, more stable revenue sources, less complexity and a fairer system.

But at this point we only have vague suggestions, covering everything from GST to income tax. Reform could be anything from highly regressive to highly progressive and the economic benefits could also vary wildly depending upon just how such a package is put together.

Do personal income tax changes benefit those groups who are more likely to respond to work incentives such as young mothers, or do they largely benefit high income earning males who would work full-time regardless?

In an ideal world a tax system should be transparent, hard to avoid (few distortions) and equitable. Tax reform inevitably leads to winners and losers and is often influenced heavily by powerful lobby groups with incentives to benefit their members but not necessarily the country. Political parties rarely miss the opportunity for a full blown scare campaign - the Liberals with the carbon tax and Labor with the GST are recent examples.

What the models tell us

The most recent modelling of taxation impacts in Australia is that provided in the government’s tax discussion paper. The modelling calculates the “marginal excess burden” of the major forms of taxation available. The marginal excess burden is an estimate of the loss in economic activity from increasing each form of taxation. As an example the modelling suggests that for every dollar raised in personal income taxation 20 cents of economic activity is lost - mostly through lower employment participation and productivity.

The modelling suggests stamp duties have the largest burden at around 70 cents in lost activity whereas land taxation and municipal rates have the lowest burden with a 10 cent gain in economic activity. Company taxation is estimated to have a relatively high burden at around 50 cents in the dollar wasted whereas personal income taxation and the GST are both around 20 cents in the dollar.

In theory, this all suggests that Australia, with a relatively high dependence on income taxation could make gains in economic activity by lowering company taxation but increasing more efficient forms of taxation such as land taxation. The modelling strongly suggests removing stamp duty and shifting towards land taxation (as is happening the ACT).

The modelling is based on “long run” models of the Australian economy using what are called Computable General Equilibrium (CGE) models. These models are based on input-output (I/O) tables that describe the structure and linkages within the economy. In addition to the I/O tables the CGE models also have estimates of behaviour change through elasticity parameters. These parameters are either estimated based on actual data or assumed based on past studies or theory - there is rarely consensus on these parameters and they heavily influence the outcomes of the modelling work.

The elasticity tells us the responsiveness of demand to changes in prices. If a certain product is very responsive to price change then a tax applied to that product would reduce demand significantly for that product and welfare for the consumer would be lower - hence a large excess burden. Where elasticity is low then the excess burden would also be low as economic activity changes little.

Models are only as good as the assumptions

All of this is interesting in theory, however, in spite of the enormous complexity of such models the results are very sensitive to the underlying assumptions and based on assumptions that can over-simplify.

As an example, an increase in personal income taxation is based on a “representative agent” paying a flat rate of taxation and no account is made for social security payments. With a GST there would also need to be significant compensation payments. The modelling does not take into account the impact these payments may have on participation decisions. The modelling also does not take into account the reality that the responsiveness of lower income taxation rates will likely differ depending on income level and how those compensatory tax reductions are applied.

All this does not imply that the CGE modelling is not a useful tool for thinking about taxation reform, but the results should not be taken as the final story.

The modelling also says little about the fairness of reform. For that, microsimulation models such as Treasury’s CAPITA model are very useful as they are based on survey data with actual households and people. A similar model (STINMOD) was used extensively to demonstrate the distributional impacts of the GST package in 2000 and the carbon price in 2012. In both instances the distributional modelling did show that such tax packages could be implemented with “fairness”.

No revolution

In thinking about tax reform it is important to keep in mind that the gains from modest tax reform are not likely to be a revolution in Australia. The models themselves only estimate relatively small gains from tax reform.

Over the past 25 years Australia’s living standards have increased by around 60% whereas the sorts of gains estimated from tax reform are expected to be little more than 1 or 2%. It remains important that in securing such modest gains we don’t ignore fairness. Models such as CAPITA provide much information on the fairness of reform. Reform can live or die on the outcomes of such modelling - think about the 2014-15 Federal Budget where such models showed the unfair nature of the budget.

The most talked about policy change at the moment, a tax switch mix from personal income taxation to a GST was only estimated to gain the economy 0.5%. Another popular choice among the business lobby is a lower company tax rate. In a model world this would be expected to provide a stronger growth dividend and higher wages. In theory this is terrific but selling this theory to a sceptical public and getting a “rising tide lifts all boats” result in wages will be a challenge.

Tax reform may well lead to some gains in economic activity and can be developed in such a way that is fair to low and middle income families. There are significant challenges in modelling the impacts of tax reform. To some extent it is the process of modelling that is often more useful and insightful than the actual estimated impacts that the models derive.

There may well be some modest economic gains to be had from tax reform but in the absence of any concrete proposals it is very difficult to judge the merits of tax reform.

Love them or hate them, the GST package in 2000 and the carbon price package in 2012 were both provided with great policy detail and detailed modelling of both the anticipated distributional and economic impacts. Both modelling exercises were imperfect but ultimately that evidence was not far from the lived experience. The current tax reform debate could well do with some concrete proposals and modelling work to better understand the costs and benefits.