As usual, much of the pre-Budget discussions in the media has focused on the rates of major taxes, particularly the taxes on income. This is understandable as the rates of income tax constitute the most visible element of the tax structure and are believed to have a profound influence on taxpayer behaviour, especially compliance.

In general, the "deadweight" loss that taxes (other than a lumpsum tax) entail tends to increase with the rate while exorbitantly high rates provide an impetus for evasion.

Equity also suffers when compliance is uneven. That is why there has been a trend across the world in recent decades towards tax systems with broad base and moderate rates.

Tax reforms in India too have followed this precept with encouraging results.  Revenue from income taxes has surged impressively.

So, "go further along this road", says a vocal section of opinion makers. But how far should the finance minister go? One suggestion -- almost paralleling, if not inspired by, the "perennial campaign" of right-wing think tanks in the US -- is, let incomes be taxed at a flat rate doing away with the cobwebs that clutter the tax codes.

"Why not have a single rate of 15 per cent and make life simple for all?" say the proponents.   Also, why not offer an amnesty for those who have evaded in the past to come clean but are deterred by the fear of heavy penalties?

Despite the apparently persuasive logic underlying the proposals, the questions do not admit of a categorically positive answer. For that requires models that can predict the response of economic agents to alternative tax regimes with a measure of plausibility. Such models are difficult to come by.

Even so, in framing tax policies the finance minister often turns to expert opinion and findings of relevant research for guidance. The problem is that research results sometimes tend to gain currency even before they have undergone the test of peer review, especially when they accord with the ideological predilections of a powerful section of taxpayers.

The problem becomes more acute when expert opinions diverge and scholars belonging to different schools vent their views with indecent stridency, one school labeling the opponents' views as 'radical chic', jadu and worse, while claiming infallibility of their own judgements like self-evident truth, even when their methodology and numbers are open to question as seems to be happening this time.

Anyone familiar with the literature on economics of tax compliance would know that despite intensive research, why people pay taxes remains an enigma.

Plainly, while other things such as community values also matter, given the free rider instinct, a utility maximising individual, unless committed to moral codes of a high order, would choose not to pay a "non-benefit" tax such as the income tax so long as the cost of evasion is less than what compliance involves.

Cost of evasion depends on the scale of penalties prescribed in the law and the taxpayer's perception about the probability of getting caught and punished, while the cost of compliance consists of the tax payable if the correct income is reported and the associated expenses of preparing returns and maintaining supporting documents.

Hence, when evasion costs less than compliance, lowering the rate of tax should help to improve compliance.  This, it is claimed, is corroborated by the growth in income tax collections following the tax cuts that occurred during the 1990s and apparently constitutes the rationale for the plea for lowering the tax rates further. This argument is, however, flawed on several counts.

One, the surge in collections notwithstanding, whether or to what extent the lower rate regime has served to reduce the size of the black economy in India is difficult to say for certain. There has been no systematic study on the subject after the one done by Shankar Acharya in the late 1980s.

Back of the envelope calculations based on conservative assumptions suggest that the tax actually collected on personal incomes is probably still only around 50 to 55 per cent of the potential. There has been a significant improvement in the ratio over what prevailed before the reforms but the improvement had come about well before the maximum marginal rate was reduced to 30 per cent.

On the contrary there was a decline immediately after the rate cut of 1997.  The ratio has moved up again despite the surcharges that came later.

Granting, however, that compliance has improved following the reforms, what accounts for the improvement is not all that obvious. It may not be correct to attribute it entirely to rate reduction.

Studies such as the one by Surjit Bhalla that plead for a low flat rate based on an analysis of the compliance behaviour of different income groups carry no credibility when they equate compliance with filing of returns and want us to believe that the potential number of taxpayers with annual income of Rs 10 lakh (Rs 1 million) or more is no more than 174,000 for the country as a whole.

The fact of the matter is that along with the lower rates, the tax deduction at source net has expanded and now accounts for over two thirds of non-corporate income tax collections as against only 40 per cent earlier, and that may have played a major role in revenue growth.

Further, TDS from salaries has recorded the fastest increase. As far as one could gather, the number of returns filed has increased mainly in salary cases.

In the case of corporate income tax, the bulk of the revenue comes from public sector undertakings.  Thus, better compliance is probably the contribution of the expansion of the organised sector -- the "white economy" -- following liberalisation more than the lower rates.

Moreover, those who advocate rate lowering as the best way to induce compliance tend to forget that compliance can be improved also by measures that serve to make evasion more costly. This was brought home most tellingly by the Emergency.

Besides, the government has a resource constraint and cannot afford to take risks with revenue by lowering the tax rates in the hope that compliance will improve. The rates of income tax are now down to what would widely be regarded as reasonable.

Some progression in the rates is also desirable for vertical equity. So attention should be focused on better enforcement, for which there is ample room rather than going the easy way of rate cutting.

As for amnesties, there can be a case for reducing the cost of compliance for one who has already evaded the tax and is willing to disclose his unreported income but finds the cost too high because of penalties and accumulated interest.

For this purpose, the existing provisions in the tax laws that permit waiver of interest in deserving cases that seem to have fallen into disuse may be reactivated.

Although they serve to reduce the cost of compliance, tax amnesties cannot be conducive to better compliance as they bring down the perceived cost of evasion as well. Friendly taxpayer services, on the other hand, serve to reduce the cost of compliance but not of evasion and so deserve attention.

Given that the cost of evasion is widely perceived as low in our country -- the number of successfully prosecuted tax frauds is too small to have the required deterrent effect -- nothing should be done that sends out wrong signals that evasion is pass� as periodic amnesties tend to do or that our rates are too high to merit compliance, particularly at a time when the tax administration is gearing up for a more effective enforcement regime.