Henry's plan is for taxes based on hard evidence

What happens when you ask perhaps the foremost tax expert in the country to conduct a "root and branch" review of the tax system? He recommends things no prominent businessman or retired judge would dream of proposing.

Few if any taxes are popular, but in all the submissions to the Henry review there was little agreement on which were the really bad ones - or on why they were so bad.

A lot of the arguments we have about particular taxes - whether they're achieving their objectives, whether they're doing more harm than good - are based mainly on the vested interests, ideology or some economic theory, freefall rather than objective evidence.

Has it ever occurred to you that the decisions governments make about what taxes to levy and the rates at which to levy them ought to be a lot more scientific? Based more on hard evidence than judgment?

Though the primary purpose of taxes is to raise revenue, most have stated policy objectives behind them. If nothing more specific, they're supposedly designed to minimise the tax system's distortion of the choices we make, to spread the burden of taxation fairly, to treat people in similar circumstances similarly, and to do all this in ways that leave people certain of their rights and responsibilities and don't generate high costs to comply.

Then we have the growing tendency to use taxes to correct the spillovers that occur when people make decisions but don't take into account their impact on others.

Sometimes the tax is intended more to change behaviour than raise revenue; sometimes the revenue raised compensates the losers from the spillover. Emissions trading schemes and carbon taxes are the latest examples of this, as is the push for congestion taxes.

But let's consider taxes on tobacco and alcohol. Ideally, tax on alcohol is set at the relevant "marginal social cost" - the level that transfers to drinkers the costs their actions impose on the community, as well as compensating the community for those costs.

How do we know the right level for the tax? We don't - not without collecting a lot more empirical evidence than we have now. More generally, we need to monitor the performance of the tax system better.

Henry says that "where possible, the performance of specific taxes and transfers should be measured objectively to identify whether they are meeting their policy objectives.

"An objective evidence base can reinforce public and government support for successful economic reforms and helps to determine when existing policy settings are no longer appropriate."

The paucity of information on specific taxes or cash transfers means data designed for other purposes are often used for analytical purposes, but this can be unsatisfactory.

Unbiased and systematically collected data on the tax system, based on widely accepted methodology and appropriate for tax policy purposes, are rare and often not publicly available.

Because such information is a public good - that is, the producers of it can't stop it being used by people who haven't paid for it - and even though society would benefit through improved tax policy based on it, the incentive for individuals or businesses to produce it is weak.

In any case, the capacity of non-government players to generate such information is limited because much of the data needed for the analysis is held by government.

All this says it's primarily governments' responsibility to generate - and make public - the needed information.

Henry adds that we would benefit from a system-wide study of taxpayers' compliance costs to monitor, on a continuous basis, the costs of complexity.

"Well-designed system-wide surveys are expensive," Henry says, "but they would provide valuable information on where simplification would yield the greatest returns".

Another area where we'd benefit from more information is on the extent of non-compliance with tax laws. The Tax Office doesn't derive estimates of non-compliance for key income and deduction items, nor publish these estimates.

Where people do examine the performance of taxes they tend to consider taxes separately. But this makes it difficult to get a sense of the system's combined performance and effects, and to determine whether it's making a coherent contribution to our national objectives.

So Henry recommends that federal and state governments systematically collect data on aspects of existing taxes and cash transfers - including compliance costs - according to consistent and transparent classifications and concepts, and make this information freely available for analysis and research.

He further recommends that every five years the feds publish a "tax and transfer analysis statement" on the overall performance and impact of the system, including estimates of efficiency costs and distributional impacts.

Academics, tax practitioners and the public should be encouraged to contribute to - and contest - the analysis in the statement. All data used and a full description of methodologies should be available to the public and subject to peer review. The government should also support one or more institutions to undertake independent policy research relevant to the tax and transfer system.

Only a man as possessed by the need for good tax design as Henry is would dare recommend such an egg-headed and expensive program just to make tax policy more "evidence-based", as the medicos like to say. But that doesn't mean he's wrong.