The present value of the fiscal cost of SEZs is thrice the infrastructure generated, and there is little to suggest this will result in a net addition to overall investment in the country.

With the report by ICRIER on the economic impact of SEZs having been submitted to the ministry of finance, the debate on the possible impact of this policy on the economy has resurfaced. In the interest of an informed debate, here is an alternative set of numbers.

Coverage: India's great rush for SEZs

Two important principles need to be mentioned upfront, before discussing numbers. First, the fiscal impact is contingent on whether the investment is incremental over and above the existing levels of investment or whether it is just a realignment. If the investment would not have occurred in the absence of the policy, there can be no revenue loss attributed to the policy. Revenue loss arises only if the investment is a re-alignment of normal investment levels in the country. If Rs 300,000 crore is the total investment expected in SEZs -- as reported in the "Factsheet on SEZ" put out on the SEZIndia website (http://sezindia.nic.in) -- it would be useful to compare this figure with the total investment in the country and the contribution of the corporate sector. In 2006-07, Gross Fixed Capital Formation (GFCF) was Rs 12,16,000 crore, of which if the corporate sector accounts for 40 per cent, its investment would be Rs 4,80,000 crore. If the investment in SEZs is to be realised over a long time period, there is no issue of concern, since on the one hand, small increments to investment are always feasible, and on the other, the policy is not likely to make a splash of an impact on the economy!

For the policy to be significantly effective, the investment should be upfront, say, spread over two years. The additional Rs 1,50,000 crore of investment implied would mean a step-up in GFCF by over 3 per cent of GDP in the first year itself. In a single year, this would be feasible, if the bulk of the investment is in the form of FDI. The profile of SEZs does not indicate a predominant FDI component. Domestic investment by the corporate sector expanding by over 25 percent in a year over and above normal growth seems unlikely, suggesting that the policy results in some fiscal costs. Following from the above, if the investment is not additional investment, employment generated too cannot be counted as a net gain either.

Second, since the losses are sought to be measured against some gains to the economy, the picture being painted depends on what are considered "gains". Clearly, one cannot measure the losses against the expansion in economic activity, since by definition, taxes can take away only a part of the total income generated, for any economic activity. No economic activity would be viable if taxes exceed the income generated. An alternative benchmark would be to assess the impact on the constraints facing the domestic producers. One of the major gains expected from SEZs is the augmentation of infrastructure. It would be interesting to assess the revenue foregone against this benchmark.

The data on the website of the ministry of commerce on proposal-wise information on area and proposed investment in SEZs indicates a wide variation in the investment per acre from Rs 50,000 per acre in Dahej to Rs 35 crore per acre in Quarkcity. The differences persist even if one segregates IT SEZs from non-IT ones. This suggests that the "infrastructure" implied in each of these proposals is widely different. To assess the impact therefore, it is useful to think of the projects broadly in two categories:

Category 1, where the developers of SEZs conceive of it as a normal production unit � like setting up a factory. IT-sector SEZs as well as some others envisioning large production units are of this type. In these cases, there is no difference between production and infrastructure investment. The entire investment would be used for production/exports. It should be mentioned that the infrastructure for this kind of SEZs is sector-specific at best and cannot be re-allocated to use in other sectors. As long as these sectors are profitable and viable, there will exist demand for the capacity created. Individual units would be willing to invest in these facilities since the gains from such investments accrue completely to them alone. In other words, these investments do not seem to ease the pressure on the existing infrastructure in the country.

Category II, where developers are planning to set up township-based zones involving basic infrastructure. In these cases, the developer invests in the infrastructure and lets others develop production facilities. A rupee of investment in infrastructure here, therefore, needs to attract further investment in production facilities and in township development, before the developer can earn any income from the zone. Using averages of Rs 30 lakh as the cost of development of infrastructure per acre, Rs 3 crore as the cost of development of township per acre, and with land and building amounting to 21 per cent of total investment in a production unit, the additional investment required can be derived. The required investment in plant and machinery, and township development, is over seven times the investment in infrastructure.

The accompanying table summarises the fiscal cost for each category, if total investment were dedicated to that category itself. The actual impact would lie somewhere in between, depending on the actual composition. These figures are derived by assuming 45 per cent equity participation in total investment and a 20 per cent return to equity capital. It is assumed that half of the raw material is sourced in India and the rest is sourced from abroad. Since all exports are entitled to zero-rating, there is no revenue foregone on account of either the machinery procured by the exporters or the inputs purchased.

Since the policy provides for incentives spread over 15 years, it makes sense to look at the total fiscal cost over the life of the incentive regime. Further, since such an estimate would overstate the impact, the present value of the impact with 10 per cent interest rate and 5 per cent inflation too is considered. These numbers assume that the investment is spread over two years and production and incomes are derived only from the third year onwards.

Since the first type of SEZs do not augment the basic infrastructure in the country, there appear no gains to measure the losses against. On the other hand, in the second type of SEZs, infrastructure is only 12.25 per cent of the total investment. The present value of the fiscal cost as reported in the table is three times the infrastructure generated. The bulk of the loss through indirect taxes is upfront. The revenue foregone in the first five years in indirect taxes alone exceeds the amount of investment in pure infrastructure. Whatever be the composition of SEZs, therefore, if the revenue foregone was actually collected and invested in infrastructure, the increment to infrastructure would have been greater. The question on why SEZs are a superior policy alternative remains unsettled.

The writer is Senior Fellow, National Institute of Public Finance and Policy. She can be reached at kavita@nipfp.org.in