Union finance minister Pranab Mukherjee has finally admitted what many observers have held all along: the April 2010 deadline for a goods and services tax (GST) is unlikely to be met. GST fundamentalists might be dismayed but delay in not a bad thing.

GST wins over our existing tax system because it avoids cascading and hence improves efficiency. It beats sales tax because it secures revenue by collecting throughout the production chain: if a retailer escapes the sales tax net, the entire amount is lost but with GST not only is it more difficult to escape but also the amount lost is only the tax on the retailer's value-added. But the transition to GST is not easy.

In an IMF paper, 'What do (and don't) we know about the Value Added Tax? Micheal Keen, reviewing Richard M Bird and Pierre Pascal Gendron's The Vat in Developing and Transitional Countries, points to one of the main lessons from the experience of many countries that went in for the tax: the political difficulty of making sensible changes to a VAT once introduced.

Mistakes made at the time of introduction are hard to undo. (Incidentally, GST is nothing but a VAT on goods and services so Keen favours the term VAT; though given the common usage of GST rather than VAT in the Indian context, this piece uses 'GST' and VAT inter-changeably.)

That's not the only lesson, though. At a recent seminar hosted by National Institute of Public Finance and Policy (NIPFP), Keen drew on the VAT experience of different countries to provide some key takeaways. To start with, he finds the efficiency of a VAT-based tax system (what he calls the C-efficiency) varies from a low of 33 in Mexico to a high of 105 in New Zealand, with most countries in the range of 45-60.

While the low efficiency ratio in Mexico is attributed to the large number of items zero-rated, the fact that, with the exception of New Zealand, even countries like the UK have an efficiency ratio of just 49% is sobering.

Contrary to widespread belief, reduced rates are a bad way to help the poor; the rich gain more than the poor from any subsidy extended by way of lower GST rates. As the Empowered Group of State Finance Ministers struggles with never-ending demands to add to the list of items eligible for concessional rates, it might like to keep that caveat in mind.

Keen concedes it is almost impossible, politically, to remove zero rating of food and food items. (Zero-rating refers to a scenario where you get input tax credit). Nonetheless empirical evidence shows zero rating does not really help the poor. Income support measures are far better.

If, for whatever reasons, income support measures are not possible (in the Indian context leakages and poor targeting of beneficiaries are a perennial problem) spending measures - eg on health, education etc - could be a good alternative to zero rating. The success of the Ethiopian experience where fewer items were zero-rated but government supplemented this with more spending targeted at the poor seems to bear this out.

Multiplicity of rates is best avoided. In the Indian context though there is no final consensus it is likely that we will end up with three or more rates. It would be instructive for the Empowered Group to keep in mind that of the 50 new Vats introduced since 1995, 60% have one rate.

Environmental damage and sin taxes on things like fossil fuel, alcohol and cigarettes are best addressed, not by higher rates of GST but by additional excise tax that should be levied over and above the GST. Exemptions (where you don't get input tax credit) go against the very logic of VAT by breaking the chain of tax offsets. The only unavoidable exemption is for small businesses.

Treatment of such businesses, he points out, is very important; many VAT failures are due to inclusion of small businesses at low thresholds. A high thresh hold means loss of revenue. However a low one adds to administrative costs. Hobson's choice! But then whoever said GST was easy.