After struggling for almost a decade with it, the government finally managed to introduce the value-added tax. Some people cheered wildly, and some booed.
But first some broad facts. As of 2004, 136 countries had implemented a VAT and the size of VAT revenue in total tax revenue varies from 6.6 per cent in Japan to 40.5 per cent in Chile. And, as pointed out by Joseph Stiglitz, welfare gains from a switch to VAT vary inversely with the size of the informal sector.
Be that as it may, during the entire debate in India no one (most notably, not a single Indian economist) asked what conditions were required to make VAT succeed. People just droned on and on, in support or in opposition, to the tax.
Now comes a paper* by Joshua Aizenman and Yothin Jinjarak that tells us what to expect. It examines political and structural factors in 44 countries to indicate what determines collection efficiency. The data pertains to 1970-99.
We all know that a great deal depends on the penalties for evasion, not to mention corruption and the technical mechanisms for collection. The first two depend on political stability in the sense that if politicians worry that they may be out of power tomorrow, they don't enforce taxes. In India, many collect them for private use -- even when there is no such risk to office.
The debate in India raised the question -- unanswered, as I said -- about what one should be looking for in order to get probabilities of the success of collection efficiency.
Aizenman and Jinjarak give the answer. "A one standard deviation increase in the durability of political regime, and in the ease and fluidity of political participation, increases collection efficiency by 3.1 per cent and 3.6 per cent, respectively. A one standard deviation increase in urbanisation, trade openness, and the share of agriculture changes collection efficiency by 12.7, 3.9 and 4.8 per cent, respectively. In addition, a one standard deviation increase in per capitaGDP increases the tax efficiency by 8.1 per cent."
In short, efficiency gains are not automatic. They depend on other things. The authors have constructed a model, which discusses the agent's problem and the policy maker's problem.
The agent's problem is to figure out the risks and quantify the penalties associated with different levels of underpayment. This suggests that you can indeed underpay as long as you don't underpay too much.
In fact, the Indian tax system is based on this approach -- what I call the counter-part of the American plea-bargaining system in the criminal justice system.
The policy maker's problem is modeled as there being two parties that alternate randomly in office, depending what they do while in office. So there is an incentive not to annoy the agents. However, zero annoyance of agents also means zero public goods.
So how much risk the policy maker takes in this regard depends on his or her view of the composition of public goods and their stream over the future.
This seems plausible if one argues that had it not been for the various fiscal burdens imposed by the Common Minimum Programme, the UPA government too may have been unable to impose VAT.
It may have been less serious about it, anyway. There is nothing like the prospect of severe fiscal stress to concentrate the finance minister's mind. The idea that they may have no money to give away when the next election comes around would galvanise any politician.
To sum up, the main conclusions of the paper are as follows: