This paper reviews the revenue impact of tax reforms implemented by the central government since 1991. Overall revenue has declined relative to GDP, owing to substantial tax cuts in recent years, but elasticity estimates point to small improvements in the revenue-generating capacity of the tax system. Partly based on a comparison with taxation in other Asian economies, the paper concludes by outlining elements of the remaining reform agenda to raise India’s tax revenue above the pre-reform level. In the short-term, the priority should be on base-broadening measures, while longer-term steps would include moving to a VAT, improving taxation of agriculture, and strengthening tax administration.


This paper reviews the revenue impact of tax reforms implemented by the central government since 1991. Overall revenue has declined relative to GDP, owing to substantial tax cuts in recent years, but elasticity estimates point to small improvements in the revenue-generating capacity of the tax system. Partly based on a comparison with taxation in other Asian economies, the paper concludes by outlining elements of the remaining reform agenda to raise India’s tax revenue above the pre-reform level. In the short-term, the priority should be on base-broadening measures, while longer-term steps would include moving to a VAT, improving taxation of agriculture, and strengthening tax administration.

I. Tax Revenue Performance in the Post-Reform Period1

1. Since 1991, tax reforms have been an integral part of economic reforms in India. Following a blueprint laid out by a Tax Reforms Committee, the system of central government taxation has been streamlined as rates have been reduced and rationalized, while some base-broadening measures were implemented and new tax categories have been introduced.2 The overall objective of the reforms has been to increase the revenue-generating capacity of the tax system, while at the same time facilitating economic growth by rendering the system more efficient, setting stronger incentives for saving and investment, and making taxation more equitable. States have also begun to reform their individual tax systems along similar lines, but the pace of implementation has been considerably slower and more uneven.3

2. In conjunction with other reform measures, changes in taxation have contributed to a rise in India’s long-term growth potential in recent years. The reform measures have also contributed to some improvements in tax elasticity. Nevertheless, overall tax revenue has fallen relative to GDP in the post-reform period as small gains in income tax revenue have been insufficient to offset losses on the indirect tax side. This, stronger revenue growth will require a decisive breakthrough in a number of areas where tax reforms have been lagging. In particular, there is considerable scope for removing tax exemptions for various key sectors of the economy; exploiting more fully the potential for presumptive taxation; and further streamlining the structure of indirect taxes. In addition, continued priority needs to be given to further strengthening tax administration, especially if progress should be made towards the introduction of a VAT.

3. This chapter focusses on reforms at the central government level where the thrust of recent reforms has taken place. Section A reviews the need for tax reform in India. Section B describes the measures that were introduced since 1991. Section C discusses the revenue performance in recent years, focussing on four major tax categories that constitute the bulk of central government tax revenue in India. Section D compares India’s tax system with other developing countries, and section E concludes by listing options for further reform.

A. Introduction: The Need for Tax Reform

4. The macroeconomic crisis of 1991/92 had its roots in the deteriorating fiscal situation of the late 1980s. By 1990/91, the fiscal deficit of the central government had risen to 8½ percent of GDP, and the overall public sector deficit reached well above 10 percent. The increase in the deficit originated mainly on the spending side and was not directly related to a weakening of tax revenue collections. In fact, gross tax revenue of the central government was at relatively high during the latter half of the 1980s (11 percent of GDP; Chart I.1). However, it had already become obvious that the tax system was in need of major reforms.


INDIA: Components of Central Government Tax Revenue

(In percent of GDP)

Citation: IMF Staff Country Reports 1998, 112; 10.5089/9781451818536.002.A001

5. India’s tax revenue has not only been low by international standards, but the tax system prior to 1991 also posed two fundamental macroeconomic problems for the Indian economy. First, even during the strong growth phase of the late 1980s, the tax system lacked the elasticity to generate sufficient revenue to keep up with public spending and thus prevent the fiscal deficit from soaring. Second, the distortionary nature of Indian taxes reduced overall economic efficiency and led to wide-spread tax evasion, driving a large part of the economy underground (Acharya et al. 1986).

6. These problems were caused primarily by the following:

  • High tax rates. Marginal income tax rates were at around 50 percent for both corporate and private income (albeit reduced from almost 100 percent in the early 1970s); excise taxes could reach as high as 125 percent, and import duties averaged above 80 percent (with the bulk of imports falling in a range of 50–150 percent).

  • A nontransparent tax structure. There were about 20–30 excise tax rates and tariff bands, and both tax categories had a number of auxiliary and specific duties. Different rates applied depending on the use (or user) of a particular product, and concessions and exemptions were widespread. Direct taxes were subject to surcharges that applied across different tax brackets. Moreover, with frequent changes in regulations, in practice it was all but impossible for economic agents to keep track of developments in the tax system, creating uncertainty and opening inroads for corruption (Tax Reforms Committee, 1991).

  • Insufficient coordination among tax authorities. On historical grounds, there has been a multitude of taxes levied by all government levels in India, including central and state governments as well as local authorities. Although the bulk of public revenues has been generated by only a few major tax categories (Table I.1), the lack of tax coordination with and across states has added to the economic burden of taxation.

Table I.1.

India: Combined Tax Revenue of Central and State Governments, 1990/91–1996/97

(in billions of rupees)

article image
Sources: Union Budget Documents; RBI Reports on Currency and Finance.

Central government revenues are actual data, revised estimates for state revenues only.

Revenue shared between central and state governments.

State taxes.

7. Efforts to address weaknesses in the tax system were already underway in the 1980s, primarily through the introduction of a modified value-added tax (Modvat) in 1986.4 However, it was the 1991 crisis that eventually gave the impetus for a broad-based and systematic review of the Indian tax system.

B. Tax Reforms Since 1991

The blueprint for reform

8. As part of its overall reform agenda, the government in August 1991 set up a Tax Reforms Committee (Chelliah Committee), charged with a comprehensive review of central government taxation. The objective was to propose a far-reaching reform agenda that, while initially being revenue-neutral, was to improve the elasticity of tax revenue and also increase the share of direct tax revenue as a proportion of both total revenue and GDP.

9. In taking up these objectives, the thrust of the Committee’s proposals was to simplify the tax system by rationalizing and lowering tax rates, while at the same time eliminating exemptions and widening the tax net (Tax Reforms Committee 1991, 1992, and 1993). The proposals centered around the following elements (Table I.2):

Table I.2.

India: Tax Reform At a Glance, 1991/92–1998/99

article image
article image
  • Transparency and burden sharing. A simple tax system with fewer and lower rates (both for direct and indirect taxes) would reduce economic distortions, become more acceptable to taxpayers, and diminish the discretionary power of lower tax officials. Moreover, the tax burden was to be distributed more equally through a widening of the tax net. This was to be achieved mainly through the elimination of exemptions, but the report also recommended presumptive taxation schemes for small traders and businesses. A broad range of services also was to be brought under the tax net.

  • Production incentives. The excise tax was to be gradually transformed into a VAT (with a minimum number of rates), beginning with a full expansion to the manufacturing and wholesale stage. Similarly, tariff rates were to be reduced and streamlined, while providing for an escalating tariff structure and maintaining some protection to domestic industry.

  • Saving incentives. The taxation of long-term capital gains and wealth held in productive assets was to be reduced to encourage saving and assist the development of capital markets.

  • Tax administration and enforcement. The report called for more efficient tax collection, including through improved taxpayer identification and more focussed efforts to combat tax evasion. Tax collectors were to be better paid and receive more training to boost morale, build human capital, and reduce incentives for corruption.

10. The Chelliah Committee reports have served as blueprint for reform in recent years, and many of its recommendations were implemented particularly in the 1992/93–1995/96 budgets. Tax and tariff cuts in the 1997/98 budget went even deeper than what had been proposed by the Committee, The following discussion summarizes the resulting changes in major central government tax categories and outlines the remaining agenda for reform.

Income tax reform

11. Changes in the system of direct taxes included the scaling back of the tax rate structure towards a maximum rate of 30 percent (35 percent for corporates) and reductions in the number of rates, including the elimination of surcharges (Chart I.2). Moreover, the lower limit for personal income taxation was raised in several steps to reach Rs 50,000 in 1998/99. A new long-term capital gains tax was introduced (at a 20 percent tax rate), the scope of the wealth tax was reduced to non-productive assets, the rate being lowered from 8 percent to 1 percent, and the gift tax was abolished in the 1998/99 budget.


INDIA: Recent Developments in Income Tax Rates 1/

Citation: IMF Staff Country Reports 1998, 112; 10.5089/9781451818536.002.A001

Source: Ministry of Finance, Economic Survey, various issues.1/ Including surcharges.

12. Significant steps were also taken to broaden the tax net, combat tax evasion and strengthen tax administration. The scope of tax deduction at source was expanded in several stages, and an “estimated income tax” was introduced for small businesses and traders that have generally been hard to tax.5 The 1996/97 budget also introduced a minimum alternative tax (MAT) for companies that would otherwise have avoided paying taxes altogether. Moreover, efforts have begun to fully computerize tax collection agencies, allocate unique tax identification numbers, and (in the 1997/98 budget) widen the scope of compulsory tax filing requirements based on certain wealth indicators. Finally, the 1998/99 budget introduced an incentive scheme to shorten tax litigation, aiming at reducing the large number of cases currently held up in tax arbitration or in the courts.6

13. While these measures have generally strengthened collection efforts, a wide range of tax exemptions still remains. For example, a vast array of tax incentives are still in place for promoting exports, foreign capital inflows, or small-scale industries (Shome 1997). In the 1997/98 and 1998/99 budgets, new tax holidays have been added for firms involved in infrastructure and housing projects, or the financing thereof, and export profits have again been excluded from the MAT (which is only calculated on book profits, and not on a presumptive rate of return on assets). The introduction of new depreciation schedules also had a perceptible negative impact on the tax base, and two tax amnesty schemes in 1991/92 and 1997/98—while partly successful in generating short-term revenue—have dented the credibility of the tax collection process.

Excise tax reform

14. Excise taxes on most manufactured goods have been converted to incorporate Modvat credits for taxes paid on inputs, including raw materials and capital goods.7 Wholesale dealers that fall within the chain of manufacturers have also become eligible for Modvat credits. The number of excise rates has been reduced to eight (with a maximum rate of 40 percent), including through a unification of basic and special excise duties; and the bulk of goods earlier subject to specific taxes have been converted to ad valorem rates.

15. Although these measures have brought about a simplification of the excise tax system, some major shortcomings have not yet been addressed. A number of important sectors remain fully exempt from excise taxation, including the small-scale industry sector, particularly in the areas of food processing and (hand-operated) textile manufacturing. In addition, a host of new exemptions have been granted in recent years, amounting to a revenue loss of almost Rs 20 billion, and the 1998/99 budget has further enlarged the exemption limit for small-scale sector enterprises from a turnover of Rs 300 million to Rs 500 million.8 Moreover, the remaining number of excise tax rates is still high, and progress still needs to be made to achieve the authorities’ objective of a streamlined system with a maximum of 2–3 rates.

16. The government has begun to levy a 5 percent tax on services, which had generally not been taxed before the 1994/95 budget. The service tax net has been somewhat expanded, and now includes most financial services, telephone, car rentals, and a few other selected services. A tax on road transportation was withdrawn in the latest budget following strong resistance from transport operators. The service tax is not part of the Modvat system, and a large number of services still remain untaxed.

Tariff reform9

17. Domestic tax reforms have been complemented by progressive tariff reductions. The maximum tariff rate (excluding auxiliary duties and surcharges) was lowered in several steps from 400 percent in 1991 to 40 percent in April 1997. Initially, particular emphasis was given to reducing tariffs on capital goods to give impetus to investment; subsequently, rates have also been brought down substantially on other products, maintaining an escalating tariff structure with duties sequentially increasing on raw materials, intermediate and finished goods. As a result of the tariff reductions, the import-weighted average tariff has declined from 87 percent in 1990/91 to below 25 percent in 1996/97, while the tariff collection rate has declined from 42 percent to around 30 percent over the same period (Chart I.3).10


INDIA: TARIFF REFORMS, 1990/91 – 98/99

Citation: IMF Staff Country Reports 1998, 112; 10.5089/9781451818536.002.A001

Source: World Bank staff estimates; and staff calculations.1/ Excluding auxiliary duties and surcharges.2/ Based on 1992/93 import weights.

18. Along with reductions in tariff rates, the structure of import taxation has also been somewhat simplified. Since 1991, the number of tariff bands has fallen from 40 to about 10, auxiliary customs duties were merged with basic customs duties, and surcharges were initially eliminated. However, the 1996/97 budget introduced a new surcharge of 2 percent to raise additional resources for financing increased investment in infrastructure, which was raised to 5 percent in September 1997 to offset a major increase in the government’s wage bill. Although the average tariff increased as a result, the tariff collection rate declined further to 27 percent. The 1998/99 budget imposed an additional 4 percent tariff on most imports, with the intention to offset a perceived disadvantage from local taxation for domestic producers, and thus create a level playing field between the taxation of domestic and imported goods.

19. With the partial reversal of earlier reform measures, the tariff structure has remained complex, particularly as tariff concessions and exemptions are still widespread—the result of attempts to encourage production and export of selected products and foster infrastructure development (Bhattacharyya and Palaha 1996). The complexity has been compounded by exemptions that apply only to portions of tariff lines, and by the reintroduction of surcharges with specific exemptions. As a consequence, the scope for inefficiencies and resource misallocation remains large, and officials are still vested with substantial discretionary powers (e.g., relating to the classification of imported goods) that encourage rent-seeking behavior.

Center-state relations

20. At present, 77½ percent of income taxes and 47½ percent of excise taxes collected by the center are transferred to the states. This system has resulted in the center’s preference for tax measures that are revenue neutral or revenue positive for the center, and is a major reason for the excessive reliance on customs duties. Under a proposed new revenue-sharing agreement, a proportion (tentatively 29 percent) of all taxes collected by the center will be transferred to the states. This proposal has been accepted in principle, but requires a constitutional amendment before being implemented.

C. The Revenue Impact of the Reforms

21. Apart from their contribution to higher economic efficiency and growth (which is not covered in this paper), the success of the tax reforms should be measured against the original objectives of the Chelliah Committee agenda.11 The following assesses the performance of tax revenue relative to these objectives, with a particular focus on developments in individual tax categories, including through the use of econometric time-series analysis.

Have the original reform objectives been achieved?

22. The main objectives of the tax reform agenda included: (i) to keep the reforms revenue-neutral; (ii) to bring about a shift from indirect to direct taxation; and, most importantly, (iii) to achieve an increase in total tax elasticity.

  • (i) Revenue neutrality. The overall impact of the reforms on revenue collection has so far been negative. Gross tax revenue of the central government fell from 11 percent of GDP in 1990/91 to 9½ percent of GDP in 1993/94, and has only partly recovered since, despite GDP growth being above 7 percent for three years in a row (Table I.3).

    Revenue performance in 1997/98 has been particularly disappointing, partly related to a cyclical downswing in imports (particularly oil imports), but also because expected supply-side effects in reaction to substantial tax cuts have not been as large as expected. After subtracting the receipts from the Voluntary Disclosure of Income Scheme (VDIS)—a one-off tax amnesty scheme that yielded 0.7 percent of GDP—traditional tax revenues amounted to only 9½ percent of GDP in 1997/98. Although there is likely to have been some substitution between regular tax payments and VDIS receipts, the overall revenue loss since 1990/91 amounts to 1–1½ percent of GDP by this account.

  • (ii) Composition of tax revenue. A striking development in the post-reform period has been the strong decline in indirect tax revenue, both in excise taxes which fell by 1¼ percent of GDP since 1990/91, and in customs duties which declined by 1 percent of GDP over the same period. By contrast, revenue from income taxes rose by ¾ percent of GDP since 1990/91. The share of direct taxes relative to indirect taxes has thus markedly increased in the post-reform period, as intended by the Chelliah Committee (Chart I.4); however, given the steep fall in indirect tax revenue, this has not been achieved in a way the Committee would have envisaged.

  • (iii) Tax elasticity. Although total tax elasticity (see Box I.1 for conceptual details) has improved in the post-reform period, returning to above unity in 1994/95, there has been a noticeable decline after 1995/96 (Chart I.5). With an average around 1.1 in recent years, tax elasticity would still fall short of boosting tax revenues sufficiently to create room for expenditure requirements. For example, at current growth rates, it would take about 6–7 years to increase the tax-to-GDP ratio by one percentage point (and thus return to the pre-reform revenue level) without additional tax measures.

Table I.3.

India: Central Government Tax Revenue, 1990/91–1997/98

(In percent of GDP)

article image
Source: Union budget documents

Indications are that actual gross collections will be ¼ percent of GDP below revised estimates.


INDIA: Composition of Central and State Government Tax Revenues, 1990/91–96/97

Citation: IMF Staff Country Reports 1998, 112; 10.5089/9781451818536.002.A001


INDIA: Gross Tax Revenue Elasticity 1/

Citation: IMF Staff Country Reports 1998, 112; 10.5089/9781451818536.002.A001

Source: Data provided by the Indian authorities; and staff estimates.1/ Excluding VDIS receipts.

Tax Elasticity

Tax elasticity measures the response of tax revenue to changes in the tax base that occur independent of changes in the tax system. In an elastic tax system (i.e., elasticity being greater than one), tax revenue would increase at a higher rate than national income—which would be particularly important for India, given its high needs for spending on physical and social infrastructure development.

The elasticity concept is closely related to the concept of tax buoyancy, which compares the change in overall tax revenue to the change in the tax base:


However, tax buoyancy would not adjust tax revenue for changes that were made to the tax system. For example, if it is estimated that tax increases would yield an additional revenue of M in the current year, the adjusted tax revenue would be equal to Ta = TM, and tax elasticity would be defined as:


The performance of major tax components

23. Tax reform measures have focussed to a large extent on four major categories: corporate and personal income tax, excise tax, and customs revenues. To better understand the impact of the reforms, the performance of these categories are discussed separately.

24. Income tax. The elasticity of corporate tax revenue has shown a clear upward trend in the post-reform period (Chart I.6). The elasticity measures improved strongly in the early years after 1991, but revenues have also responded well to tax relief measures in 1994/95 and 1997/98. Measured against corporate profits (derived from the national accounts), the corporate tax elasticity has shown a significant rebound since the mid-1980s—however, this is likely to be overstated as the national accounts appear to underestimate the increase in profits during that period. Nevertheless, independent of the tax base chosen, the elasticity is now well above one, compared to below unity values for most of the past decade.


INDIA: Developments in Corporate Income Tax Revenue

Citation: IMF Staff Country Reports 1998, 112; 10.5089/9781451818536.002.A001

Source: Data provided by the Indian authorities; and staff calculations.1/ Three-year moving average.

25. Similarly, personal income tax elasticity has improved sharply after 1992/93, rising from unity to a range of 1½–2, depending on the base chosen (Chart I.7). The cumulative revenue loss caused by cutting tax rates almost in half has amounted to some ½ percent of GDP, but revenues have nevertheless been growing strongly, including in 1997/98 when other taxes have performed relatively poorly. Elasticity has recently declined, but since VDIS receipts are not included in income tax revenue, this is probably somewhat exaggerated. The overall performance improvement in income taxes can to a large extent be linked to stronger administrative enforcement, such as through improved tax deduction at source (TDS) and presumptive filing requirements (Shome 1997). For example, the share of revenue collected through TDS increased from 37 percent in 1994/95 to 42 percent in 1996/97, and the introduction of MAT on corporate profits brought a number of the largest Indian companies under the tax net for the first time in several years.


INDIA: India: Developments in Personal Income Tax Revenue

Citation: IMF Staff Country Reports 1998, 112; 10.5089/9781451818536.002.A001

Source: Data provided by the Indian authorities; and staff calculations.1/ Three-year moving average.

26. Excise tax (Chart I.8). Excise tax revenues have declined relative to any of the tax bases used, and tax elasticity does not appear to have decisively broken a secular downward trend that began in the 1970s. Excise tax elasticity has generally been below unity, and revenues therefore had to be kept up through a steady sequence of tax measures. Since 1993/94, however, reform measures have been largely revenue-neutral, leading to an overall decline in revenue as tax elasticity has failed to improve.


INDIA: India: Developments in Excise Tax Revenue

Citation: IMF Staff Country Reports 1998, 112; 10.5089/9781451818536.002.A001

Source: Data provided by the Indian authorities; and staff calculations.1/ Three-year moving average.

27. The poor elasticity of excise revenues is related to the expansion of tax credit under the Modvat scheme that has led to sharp revenue losses in recent years, with Modvat claims increasing over-proportionally relative to gross receipts (Table I.4). Initially, this was mainly the result of not having adjusted excise rates sufficiently upwards to account for the loss of cascading revenues following the introduction of the scheme. However, there is also strong evidence for an increasing misuse of the credit system, facilitated by (i) procedural simplifications, particularly in relation to the extension of Modvat credit to capital goods and wholesale trade; (ii) poor verification facilities, resulting from lack of computerization; and (iii) insufficient determination on the part of lower tax enforcement levels to verify the validity of claims (Shome et al. 1997). Moreover, tax elasticity has been affected negatively by the increasing amount of tax exemptions for the small-scale sector which accounts for a major part of growth in the manufacturing sector. Along with other remaining exemptions, this has effectively offset a good part of recent efforts to widen the excise tax base.

Table I.4.

India: Excise Revenue and Modvat Credit

(in millions of rupees)

article image
Source: Data provided by the authorities; Shome (1997); and staff calculations.

28. Customs duties (Chart I.9). Customs revenues have been most strongly affected by the reforms, with tariff cuts leading to direct revenue losses of about 1 percent of GDP since 1990/91. As the share of imports at higher tariff rates has increased, customs revenue elasticity has improved somewhat after 1993/94—however, this has not yet led to a sustained increase beyond unity, and the overall revenue collection rate has declined strongly as a result.


INDIA: India: Developments in Customs Revenue

Citation: IMF Staff Country Reports 1998, 112; 10.5089/9781451818536.002.A001

Source: Data provided by the Indian authorities; and staff calculations.1/ Three-year moving average.

29. The relatively low elasticity in customs revenue is partly related to an ongoing shift in the composition of imports towards non dutiable goods or goods at zero-import tariff. For example, the share of non-dutiable goods increased from 17.1 percent of all imports in 1996/97 to 25.7 percent in 1997/98. Although this is partly explained by developments in international fertilizer prices and a shortfall in domestic cotton production that have led to strong increases in (non-dutiable) imports of these goods, it also appears that tariff concessions granted in past years have led to an excessive weakening of revenues.12

Econometric analysis of revenue determinants

30. What quantitative impact did tax reforms and changes in the economic framework have on tax performance and tax elasticity? In trying to answer this question, an econometric analysis was conducted to identify tax revenue determinants in each of the four major revenue components. The results indicate that the elasticity increase in the post-reform period has been statistically significant. However, it has not been possible to determine particular factors that have contributed to this increase, except that some tax policy variables were shown to have a significant short-run impact.

31. The approach followed the Engle-Granger (1987) two-step cointegration procedure (a more detailed discussion of data requirements and methodology is contained in the Annex):

  • In the first step, a regression of tax revenue on its economic tax base and other variables was used to estimate long-run revenue elasticities and possibly identify other long-run tax determinants.

  • The second step was to analyze changes in tax revenue in a regression with residuals from the first step and economic variables with short-term impact (including policy-related variables) as independent variables.

32. Long-run equation. The results for the first step are given in Table I.5, showing elasticities for each tax category over several periods. The relatively good performance of both income taxes is evident, while the elasticity of excise revenue is shown to have declined in the post-reform period. The elasticity of customs revenue has remained around unity which suggests that, in effect, the deep tariff cuts in recent years have not been offset by any base-widening measures.13

Table I.5.

Average Tax Revenue Elasticities by Period 1/

(Results of cointegration equation)

article image
Source: Staff calculations.

Periods refer to fiscal years, with 1969 equal to 1969/70.

Phillips-Perron test.

The estimated OLS-regression was; T = α0 + β0 B + ∑ (αi δi + βi δi B). T is adjusted tax revenue, B is the tax base (all in natural logs). Structural breaks were have been identified in various years t1, t2, etc.; δi is a dummy variable with δi = 1 for t >t1. The revenue elasticity for the first period is β0, β0 + β1 for the second period, etc.

33. The regressions have been disappointing in the sense that, except for a strong link between tax revenue and base for all revenue categories, no significant long-run relationships with other economic variables were found.14 However, model specification tests have consistently identified the post-reform period (beginning in 1991/92) as a period with significantly different tax elasticities, necessitating a split of the regression into pre- and post-reform periods through the introduction of dummy variables. The presence of structural breaks is by itself evidence that the fiscal framework has changed after 1991. However, owing to the lack of significant explanatory variables, it is unclear whether this change reflects budgetary measures or economic reforms more generally.

34. Short-run behavior. The second step of the Engle-Granger approach consists of an equation that specifies the short-term movement of cointegrated variables (Table I.6). Explanatory variables include the residual of the first-step regression and lagged first differences of tax revenue, as required by the error correction mechanism of the Engle-Granger approach. These variables are generally significant, showing a strong adjustment of corporate tax revenue to the underlying long-run relationship, and a weaker one for personal income tax and excise duties.15 The short-term equation for customs revenue yielded an insignificant error-correction coefficient and is therefore not presented.

Table I.6.

Results of Error-Correction Model

Regression (OLS): ΔT = γ+γ0 ECM−1 + Σ γiΔT−i + δx 1/

article image
Source: Staff calculations.

T is adjusted tax revenue, ECM is the cointegration residual. The vector x comprises other variables, e.g. the proxy for base-broadening measures (ΔB * / ΔB; ie. budgeted growth rate of tax base relative to growth rate of economic base), a measure for past tax revenue realization relative to budget ((T/T*)−1), and changes in imports (Im). All variables in logs. Numbers between parentheses indicate t-values.

Lagged once.

35. Two fiscal proxy variables were employed to capture the effects of budget policies (see Annex for the definition of the variables):

  • The first variable (ΔB* / ΔB) that relates budgeted base growth to actual base growth, has been found positive and significant for personal income tax and excise duties. This appears to confirm the positive impact of administrative tightening in income taxation, while also pointing to some short-term success in widening the excise tax net. As evident from the decline of elasticity in the long-run equation, however, these short-term effects were offset by factors that led to the large increase in claims under Modvat For corporate income tax, a negative coefficient indicates that expectations of business profits have systematically been over-optimistic.

  • The second variable (T-1/T-1*) has only been significant in the case of corporate income tax revenue, indicating that revenue shortfalls in this category have typically prompted efforts to recoup some of these losses in the following period (this is consistent with the strong error correction mechanism noted above).

36. Finally, most other variables that were included have not contributed to the explanatory power of the regressions. A notable exception was that an increase in imports would lead to an increase in excise tax revenue, presumably because imported goods tend to include relatively more high value added items than domestic goods.

D. Tax Revenue in an International Context

37. In addition to analyzing India’s revenue performance separately, what lessons can be drawn from comparing India with other developing economies? This section argues that, even after accounting for a number of adverse structural factors, the present tax collection ratio in India is low by international standards, and could be improved by adopting reform measures similar to those in other emerging economies.

Revenue comparison

38. India’s low tax revenue ratio is to some extent related to the structure of its economy. As outlined by Tanzi (1992, 1994), economies with a high share of agriculture (and, typically, low per-capita incomes) and a low share of imports relative to GDP tend to generate relatively less tax revenue:

  • The share of agriculture in India’s GDP is about 30 percent—a medium to high share among developing countries. Relative to other countries, India thus faces greater difficulties in generating revenue, given widespread difficulties in taxing the agricultural sector. Moreover, a larger share of rural population is also often associated with relatively lower pressures on spending on, e.g., urban infrastructure, housing, or the urban poor.

  • The share of imports in GDP has increased to 13 percent in recent years, but India still has a relatively closed economy. With imports constituting an easily accessible tax base, India is therefore at a relative disadvantage compared to other countries.

39. These two factors, together with the external debt ratio, explain statistically more than half of the variance among tax ratios in developing countries.16 Therefore, even among that group of countries, India’s tax revenues would not a priori be expected to be very high.

40. However, even after these factors have been taken into account, central government tax revenue in India is low relative to other countries. Based on a regression formula given by Tanzi (1994) that is applied to India’s structural characteristics, central government tax revenue would be projected above 13 percent of GDP, as opposed to the present level of below 10 percent of GDP.17 Indeed, the revenue increase over the medium-term would need to be even higher as India aims at improving average incomes. As shown in Table I.7, tax revenues in countries with a low to middle per-capita income level were almost double those of low-income countries. Thus, even if India’s tax revenue was raised by 3 percent of GDP, it would still be on the low side relative to countries with similar income levels.

Table I.7.

Level and Composition of Central Government Tax Revenue in Low and Middle Per-Capita Income Countries, 1994

article image
Source: World Bank, World Development Indicators; Union Budget documents.

Central government gross tax revenue (first column). Share in current central government revenue, including central taxes shared with states (last three columns).

Tax base and tax rates

41. A detailed comparison of India’s tax system with other countries is beyond the scope of this paper, and hence the following discussion is limited to a comparison of tax rates and tax coverage with a selected group of Asian countries—namely China, Korea, and the Asean-4 countries (Indonesia, Malaysia, the Philippines, and Thailand).

  • Corporate income tax. Despite the tax cut in 1997/98, India’s corporate tax rate still remains higher than that of these countries, with the exception of the Philippines and Thailand which also have a tax rate of 35 percent.18 By contrast, the Indian tax base is fairly narrow—comprising mainly trade, commerce, and manufacturing—compared to the other countries which cover most business activities carried out by commercial entities and private entrepreneurs, as well as independent professionals. India also has a fairly complex depreciation allowance and withholding system compared to these countries. However, the use of tax incentives favoring labor-intensive exports is fairly widespread, with the exception of Indonesia which perhaps has the most modern business tax, based on a broad base with only a few exemptions.

  • Personal income tax. The Indian rate structure is largely comparable to those in the Asean countries, while other Asian countries, for equity reasons, have chosen a higher degree of progressivity. However, the effective tax base in India is far less comprehensive than those of other countries, with the exception of China to which it is roughly comparable. For example, taxable income in Indonesia, Malaysia, and Korea covers employment income (including professional income) and most fringe benefits, with only relatively few and well-defined exemptions and deductions, India also has a relatively high minimum threshold for taxable income (Rs 50,000, or 3½ times per capita GDP) which compares to below ¼ times per capita GDP in Malaysia and the Philippines.

  • Excise tax. With the exception of Malaysia, all sample countries have a fairly well-designed and broad-based VAT in place. Their standard rates are generally low (7 percent for Thailand, and 10 percent for Indonesia, the Philippines, and Korea), with China at the high end with 17 percent. The excise systems in these countries cover a relatively small group of products and are relatively simple to administer. For example, Thailand and Malaysia only have 10–20 product groups that are excisable, and few exemptions are available.

  • Customs tariffs. Indian tariff levels remain significantly above those in most emerging markets. Following the introduction of tariff surcharges in recent budgets, the import-weighted tariff average in India has increased to around 30 percent from 25 percent in 1997/98, compared with levels around 10 percent or below in most of the Asean countries. Moreover, India’s tariff dispersion (representing the complexity of the tariff structure) also remains about twice as high as that of the other countries.

Reform trends in emerging economies

42, India’s tax structure shows a pattern broadly comparable to other developing countries (see Table I.7), but this structure has been relatively constant over the past years. By contrast, the tax structure in other developing countries has begun to change significantly during the 1990s: (i) the share of domestic taxes on goods and services has generally increased, partly as a consequence of a growing number of countries switching to a VAT; and (ii) in a number of countries in Asia and Latin America, the relative importance of customs duties has declined, reflecting progress in trade liberalization and other domestic reforms (IMF 1996):

  • The widespread increase in the use of the VAT has probably been the most striking development in the recent decade. The VAT has become fairly common in Latin American countries and has also progressed—at a slower pace—in Asia (Tanzi 1994). Most countries have been able to generate higher revenues following the introduction of the VAT, with Indonesia being regarded as a particularly successful case. In Indonesia, revenues from indirect taxes more than tripled relative to GDP in the 1980s after the country switched to a VAT in 1985 (World Bank 1991). At the same time, concerns that the VAT would lead to a regressive (and difficult to administer) tax structure were generally unfounded. Indeed, a range of countries managed to make their indirect tax structure more progressive through a well-designed structure of basic goods exemptions and supplementary excises on luxury goods; and the self-policing nature of the VAT has generally contributed to a strengthening of tax administration.

  • Trade taxes, and particularly import duties, still account for up to a third or more of total tax revenue in developing countries. This stability, however, hides important structural changes that have occurred in recent years. As in India, many countries have sharply reduced the level of import duties to encourage greater openness. At the same time, however, other countries have widened their tax base by imposing minimum tariffs on imports (e.g., Argentina, Chile, Morocco, and Turkey), or replacing quantitative import restrictions with import tariffs close to the maximum rate.

43. Other important reform areas include income taxation, which has also seen a trend towards base widening, particularly through reduction or elimination of tax incentives that have generally proven ineffective in achieving their stated objectives, while at the same time stimulating corruption and tax evasion (Tanzi 1994). There has also been a marked increase in the use of presumptive taxes, especially in taxing small businesses and agriculture, and minimum taxes on enterprises (Rajaraman 1995). These taxes are designed to yield revenue from entities that would otherwise be able to avoid tax payments through either evasion or loopholes in the tax law. Countries with presumptive tax schemes are located mainly in Latin America and the Middle East, but also include Italy, France, and the United States.

44. There have also been widespread attempts to improve tax administration, based on the realization that tax reforms need to be compatible with administrative capacity. Besides aiming at reducing the complexity of tax laws, such efforts have focussed on reducing the role of tax officials in tax assessment to minimize collusion between assessors and tax payers. For example, self assessment of tax payers has proved successful in the case of Indonesia, which has also streamlined its tax laws at the same time. Tax deduction at source has been highly effective, not only in India, but also in other countries that have implemented it on a larger scale. Finally, an important lesson from earlier tax reforms has been that insufficient computerization, low management skills, and a lack of morale among tax officials can have a significant impact on revenue collection, and need to be addressed in a comprehensive manner for improvements in the tax system to take hold (World Bank 1991).

E. The Remaining Agenda: Options for Reform

45. India’s disappointing revenue performance so far reflects the partial nature of the reforms. Although the Chelliah Committee had emphasized the need for a more streamlined tax system, the broad emphasis on tax measures has been to provide economic incentives and encourage better compliance through rate reductions, and less on the need to reduce exemptions and widen the tax base (see Table I.2). Although rate reductions have been desirable from an efficiency perspective—largely eliminating the rationale for exemptions—fiscal prudence would have necessitated stronger steps on the base-broadening part to ensure that revenues be maintained. Indeed, there are a large number of countries, primarily in Latin America, that have used tax reforms to increase revenue (Shome 1994).

46. The unfinished portion of the Chelliah Committee agenda contains many of the measures that are necessary to reverse the decline in the tax-to-GDP ratio and eventually increase revenues. The immediate focus of future reforms would thus need to be on base-broadening and accelerating the process of tax simplification and streamlining of rates. There also are a number of longer term issues that would put both central and state government revenues on a sustainable upward trend, including the introduction of a VAT, improved taxation of agriculture, and improvements in tax administration.

Base-broadening measures

  • Direct tax reform would aim at eliminating tax incentives for promoting exports, inflow of foreign exchange, and other exemptions (e.g., as contained in Section 80 of the Income Tax Act), as well as reducing the still vast scope for corporate income tax deductions. Moreover, tax concessions to small-scale sector enterprises could be largely abolished, and tax incentives to promote selected economic activities should be used more sparingly.

  • Use of presumptive schemes. Given the existing administrative difficulties to enforce regular tax laws for small businesses, better designed presumptive schemes would be a feasible way to enlarge the small corporate taxpayer base relatively quickly. Moreover, the existing Minimum Alternative Tax (MAT) could be strengthened by shifting the tax base from book profits to corporate assets, and by again including export income.

  • Tax payer identification. Efforts are underway to implement the presumptive tax filing scheme across the whole country, and allocate unique tax payer identification numbers which are to be quoted for major financial transactions. There is, however, a danger that these efforts might result in a large administrative burden without yielding much revenue.19 Therefore, the focus should be on measures to identify larger tax payers, such as through improving the compatibility of existing databases and upgrading computerization.

  • Reform of excise taxes would focus on eliminating exemptions for particular industries (in particular for the small-scale sector) while further reducing the large number of excise tax rates. The bulk of products would be taxed at a single rate, with special lower or higher rates for a limited number of mass consumption goods and luxury items. At the same time, traditional excise taxes (which would not allow for input tax credit) could be strengthened to address revenue needs. For example, envisaged changes in the tariffication and taxation of petroleum products should be reviewed to raise total revenue from the oil sector.

  • The service tax net could be expanded at a faster pace. The service sector has grown strongly in recent years and now contributes about 40 percent of GDP, but only a few selected services are taxed. Therefore, instead of gradually adding services to the tax net, the tax could be introduced on a universal basis, with a few exceptions specified in a negative list. Constitutional issues regarding responsibilities of the center and states in the taxation of services would also need to be clarified.

  • Tariff reforms should aim at minimizing tariff dispersion by reducing the number of tariff bands (including through the elimination of surcharges) and eliminating most tariff concessions, including those available under export-promotion and other incentive schemes. Replacing zero-rate duties by a minimum tariff would raise overall revenue and reduce incentives for rent-seeking. For efficiency reasons, the process of tariff reduction should continue with the objective of lowering rates to global levels over the next few years.

Longer-term issues

  • VAT. The introduction of a general VAT to consolidate consumption and production taxation at central and state levels—highly desirable for both revenue and efficiency reasons—would have to be preceded by a significant streamlining of the central excise tax and a sufficient number of states with own VAT systems in place. So far, however, the VAT experience in the states has been mixed (Box I.2), while progress towards a simplified central excise tax has been slow. Both center and states have so far not been able to avoid large revenue losses caused by the misuse of tax credits on inputs, necessitating stronger administrative efforts in this area.

  • Agriculture. As the implicit taxation of agriculture through import-protected industrialization and distorted terms of trade is increasingly being dismantled, the case for an explicit tax on agriculture is being strengthened. The states have so far considerably under-utilized their constitutional prerogative to tax agriculture, generating only minor revenues from land taxes and—in a few states—agricultural income taxes. Although the taxation of agriculture is both politically controversial and difficult to administer, there are examples of a successful implementation of land taxes, e.g., in Taiwan and Singapore. Khan and Khan (1998) and Rajaraman and Bhende (1998) discuss how a presumptive land-based taxation scheme could be implemented.

  • Tax administration. While there are immediate steps that could be taken to simplify tax administration (such as, e.g., widening the scope for tax deduction at source), measures would also need to address—over a longer term horizon—insufficiencies in the internal organization, functioning and management of tax collection agencies. In the public perception, the tax administration is not effective and delivers unsatisfactory services, with delayed refunds, harassment of tax payers, and protracted litigation being common experiences (Shome 1997). As this contributes to widespread unwillingness on the part of the public to cooperate with tax authorities, and thus feeds into tax evasion, improvements in these areas could go a long way in boosting tax morale and eventually improving tax compliance.

Experience with the VAT in Maharashtra and Andhra Pradesh

Maharashtra and Andhra Pradesh (AP) are among the first states to introduce elements of a VAT to supplement or replace state sales taxes. The experience of these two states has been mixed as revenue growth has remained weak, and problems with the administration of the tax have surfaced. However, many of the problems have been related to the partial nature of the VAT, and could be resolved if the tax base would encompass a wider range of products, and if tax harmonization with the center and other states would improve.

The design of the VAT

  • Maharashtra launched the VAT in 1995, with the aim of completely transforming its system of sales taxation within a period of five years. The initial moves consisted of (i) introducing the principle of VAT to traders; and (ii) extending the scope of credit for taxes paid by manufacturers on inputs. In 1996/97, the first full year with a VAT in place, growth of overall sales tax revenue (incl. VAT) slowed considerably, although this was partly caused by lower inflation and a decline in industrial output growth. The shortfall attributed to the VAT was about Rs 5.5 billion, or ½ percent of state GDP.

  • Andhra Pradesh opted to introduce an experimental VAT (EVAT) on trade margins in 1995/96, covering some 20 commodities and about 8,000 dealers (2.5 percent of all registered dealers). The revenue performance in 1996/97 was weaker than expected, with collections reaching only about Rs 100 million (¼ percent of total sales tax receipts), or about half of what was targeted.


The experience of these initial attempts, although broadly encouraging, point to a number of shortcomings that need to be addressed in order to turn the VAT into a success.

  • First, the gradual manner in which the VAT was introduced has helped to keep legislative and administrative requirements to a minimum, and overcome resistance from taxpayers who were about to face a higher tax incidence. However, the gradual extension of the tax net to an increasing number of dealers and a greater range of commodities, along with an array of exemptions that have been maintained, resulted in an increased complexity of the tax system that has partly undermined the success of the measures.

  • Second, revenue efforts were partly inhibited by intense inter-state tax competition. Although state finance ministers agreed to harmonize sales tax rates in December 1995, progress in this area has been slow. As a result, tax rates could not be adjusted upward in response to low revenue growth.

  • Third, especially in AP, the amount of transitional relief granted (e.g., through a tax credit on stock on hand at the introduction of the VAT) was significantly larger than planned, mainly because dealers were not required to maintain inventories and subsequent tax assessments were ineffective in uncovering wrongdoing.

  • Fourth, there were shortcomings in the preparing for the implementation of the VAT, such as a failure to introduce standard tax forms and adequately prepare dealers for the introduction of the VAT. In addition, cases of evasion or suspected evasion have not been dealt with promptly, hurting tax compliance in the initial phase.

  • Finally, administrative procedures have as of yet not been adjusted sufficiently to move to a full-fledged VAT in the near future. The most imminent measures required would be to beef up computerization (with the overarching aim to provide for a verification of invoice-based tax credits), and to properly educate and reorient staff across all levels, as opposed to only select groups as of now.

Sources: A. Bagchi (1997), VAT in Maharashtra, Report prepared for the World Bank; and World Bank (1997), India—Andhra Pradesh: Agenda for Economic Reforms, Report No. 15901-IN.


  • Acharya, S.N., et al. (1986), Aspects of the Black Economy in India, Report submitted to the Ministry of Finance, Government of India (Reprint); New Delhi: National Institute of Public Finance and Policy.

    • Search Google Scholar
    • Export Citation
  • Bhattacharyya, B., and S. Palaha (1996), Policy Impediments to Trade and FDI in India, New Delhi: Wheeler.

  • Chand, S.K. (1975), Some Procedures for Forecasting Tax Revenue in Developing Countries, International Monetary Fund, Fiscal Affairs Department, Departmental Memorandum 75/91.

    • Search Google Scholar
    • Export Citation
  • Chopra, A., et al. (1995), India: Economic Reform and Growth, International Monetary Fund, Occasional Paper 134.

  • Engle, R.F., and C.W.J. Granger (1987), “Co-Integration and Error Correction: Representation, Estimation, and Testing”, Econometrica, 55, 25176.

    • Search Google Scholar
    • Export Citation
  • Gerson, P. (1998), The Impact of Fiscal Policy Variables on Output Growth, International Monetary Fund, Working Paper 98/1.

  • Gregory, A.W., and B.E. Hansen (1996), “Tests for Cointegration in Models with Regime and Trend Shifts”, Oxford Bulletin of Economics and Statistics, 58, 55560.

    • Search Google Scholar
    • Export Citation
  • IMF (1996), “Fiscal Policy Issues in Developing Countries”, in: International Monetary Fund, World Economic Outlook, May 1996.

  • Khan, M.H., and M.S. Khan (1998), Taxing Agriculture in Pakistan, International Monetary Fund, Paper on Policy Analysis and Assessment 98/3.

    • Search Google Scholar
    • Export Citation
  • Rajaraman, I. (1995), “Presumptive Direct Taxation: Lessons from Experience in Developing Countries”, Economic and Political Weekly, 30, 110324.

    • Search Google Scholar
    • Export Citation
  • Rajaraman, I., and M.J. Bhende (1998), “A Land-Based Agricultural Presumptive Tax Designed for Levy by Panchayats”, Economic and Political Weekly, 33, 76578.

    • Search Google Scholar
    • Export Citation
  • Shome, P. (ed., 1997), India: Tax Policy for the Ninth Five Year Plan, Report of the Working Group on Tax Policy of the Steering Group on Financial Resources, New Delhi: National Institute of Public Finance and Policy.

    • Search Google Scholar
    • Export Citation
  • Shome, P., S. Mukhopadhyay, and H.N. Saleem (1997), “MODVAT Administration”, in: P. Shome (ed.), Value Added Tax in India: A Progress Report, National Institute of Public Finance and Policy, New Delhi: Centax.

    • Search Google Scholar
    • Export Citation
  • Tanzi, V. (1992), “Structural Factors and Tax Revenue in Developing Countries: A Decade of Evidence”, in: I. Goldin and L.A. Winters (eds.), Open Economies: Structural Adjustment and Agriculture, New York: C.U.P.

    • Search Google Scholar
    • Export Citation
  • Tanzi, V. (1994), “Taxation in Developing Countries”, in: L. Bernardi and J. Owens (eds.), Tax Systems in North Africa and European Countries, Deventer: Kluwer.

    • Search Google Scholar
    • Export Citation
  • Tax Reforms Committee (1991), Interim Report of the Tax Reforms Committee, New Delhi: Government of India, Ministry of Finance, Department of Revenue.

    • Search Google Scholar
    • Export Citation
  • Tax Reforms Committee (1992), Final Report: Part I, New Delhi: Government of India, Ministry of Finance, Department of Revenue.

  • Tax Reforms Committee (1993), Final Report: Part II, New Delhi: Government of India, Ministry of Finance, Department of Revenue.

  • World Bank (1991), Lessons of Tax Reform, Washington, D.C.

ANNEX: Data Adjustments and Econometric Methodology

Data adjustments

47. To calculate the revenue impact of tax reforms, it is necessary to separate changes in revenue that take place in response to growth in the tax base from changes that occur as the result of discretionary measures. Actual revenue data need to be converted into a series that shows what the revenues would have been had there be no changes in the tax system.

48. One method would be to apply current tax rates to the bases of earlier years, thus simulating a revenue series that corresponds to the current tax structure. However, this method imposes heavy data requirements, including detailed information on the distribution of the base by brackets or rate categories (Chand 1975). A more readily useable procedure requires only information on the revenue impact of tax measures, which is provided by the Indian budget documents. This method is described in the following.

Basic definitions

49. Revenue for a particular tax category is given by T0 = t0 B0, with t being the tax collection rate and B the tax base. The budget revenue estimate for year 1 is given by


50. The first term of the right hand side is tax revenue at unchanged rates. The second term is the expected revenue impact of tax measures, defined as M = Δt1* B1*, i.e., the change in the collection rate resulting from tax measures, times the new base, Indian budget documents contain estimates for M, but information on expected collection rates and the size of the actual tax base is generally not available.20

51. This information can be used to compute tax elasticities and construct a time series for econometric analysis. Define tax revenue elasticity as


where T1 − M is the part of tax revenue that is unaffected by tax changes. Unfortunately, this formula gives only an approximation for the true elasticity because M is just an estimate for the budgetary impact of tax measures. This is illustrated by comparing budgeted revenue T1* with actual tax revenue T1:


52. The effect of the discretionary change in the tax system is D = Δt1* B1, which cannot be calculated since there is generally no information on either tax base or collection rates.

53. If the base projection in the budget is close to the final outcome (ie., B1* is approximately equal to B1), M is likely to be a good approximation for D. In that case, the elasticity can be calculated reasonably well. Should base projections be overly optimistic or overly cautious, it would become difficult to calculate elasticities because M would not be a good estimate for the yield of tax measures. Therefore, the quality of elasticity calculations is closely linked to the quality of base projections and revenue forecasts in the budget.

Constructing adjusted time series

54. As noted above, an econometric approach to measuring elasticities requires time series that show tax revenue under an unchanged tax structure. Such series are generally calculated by a simple “proportional method” (see Chand 1975). The current year (year 0) is set as the reference year. Tax revenues of previous years are then adjusted according to:


55. While this remains a practical way of obtaining adjusted time series, the disadvantages of this method are clear. Forecasting errors contained in M accumulate over time, and the proportional adjustment is a rather crude way of taking changes in the tax system into account. These shortcomings should be kept in mind when analyzing the results.

Choosing a base

56. A further problem in calculating elasticities is the choice of a tax base. Data on the size of actual tax bases (e.g., total taxable income) are typically not available in India, at least not in historical form. However, the actual base is generally related to a macroeconomic aggregate (economic tax base), such as GDP or industrial production. As an obvious exception, actual and economic base coincide for customs duties, and data for customs imports are readily available. For the other tax categories, several time series were considered in choosing adequate bases. The final selection was based on goodness-of-fit measures from the econometric models, which led to choosing “non-agricultural GDP” (GDP from manufacturing and services) as economic base for both income taxes and excise duties.

The econometric approach

57. The standard regression approach to calculating tax elasticities is to estimate the following logarithmic equation:


where B is the chosen tax base. Since adjusted revenues are used on the left hand side, the estimated β-coefficient is equal to the average tax elasticity over the period.

Model specification

58. To take into account other possible revenue determinants, and to allow for the impact of tax measures on base growth (simultaneity), a wider model specification process was undertaken. Both co-integration and other model specifications (such as 2SLS) were tested, and a number of variables were included that could have either direct or indirect effects on tax revenue (e.g., growth, inflation, and financial variables). However, except for a strong link between tax revenue and tax base across all revenue categories, no significant long-run relationships with other economic variables were found. In addition, there was strong evidence of structural breaks in the estimated relationships.


59. In view of the data shortcomings and the structural breaks in the data, the preferred model specification was the Engle-Granger two-step cointegration method.21 This approach offered the advantage of obtaining separate elasticities for different periods in a first step, using dummy variables to take account of structural breaks. The estimated model was:


with δi a dummy variable denoting t > ti ti being the i-th structural break. The elasticities are calculated as β0, β0 + β1, β0 + β1 + β2, etc.

60. Subsequent tests on this regression did not reject cointegration (except probably for customs revenue), and thus the impact of tax policy variables and other regressors could be analyzed in a second step.22 In that step, residuals and differenced variables from the first step were included to account for the error-correction mechanism within the cointegration model. Besides, a number of variables were included that were found stationary and could not be included in the first step. Among those were two proxy variables to capture the effect of budget policies on revenue.

Proxy variables

61. The first of these proxy variables takes into account the important role of base projections contained in budget forecasts (see above). Exploiting equation (1) for the budget revenue forecast, and using t0 = T0/ B0, one obtains


62. In other words, although the actual tax base is unknown, budget data on revenue and expected impact of tax measures reveal the implicit assumption on tax base growth. In hindsight, this growth rate can be set in relation to the growth rate of the economic tax base, and thus the first proxy variable is ΔB* / ΔB, ie. budgeted base growth over actual base growth (in logs).

63. If the coefficient on this variable is significant and negative, budgetary assumptions are most likely excessively optimistic, and this variable would correct for the resulting shortfall in actual revenue. If the coefficient was positive, however, assumptions could either be overly pessimistic, or the coefficient would pick up short-term revenue gains from base-widening measures that would not be captured otherwise.

64. The second proxy variable relates actual tax revenue to budget estimates: (T-1/T-1*), which reinforces the estimation of the error-correction mechanism. A negative coefficient would indicate that current revenue would be boosted by a shortfall in the previous period (presumably through improved collection efforts and other adjustment measures).


Prepared by Martin Mühleisen.


The Tax Reforms Committee was chaired by Prof. Raja Chelliah. It submitted its interim report to the government in 1991, with final reports in 1992 and 1993.


States have the constitutional prerogative to levy taxes on a range of activities, including agriculture, retail sales, and certain services. For a discussion of reforms and revenue performance at the state level, see Chapter III of India—Selected Issues (IMF Staff Country Report No. 97/74, September 1997).


The Modvat (a synonym for the current central excise tax) was introduced to overcome the cascading nature of the old excise tax and to prepare for a general VAT. However, the Modvat is largely limited to the production stage, with retail sales being taxed by the states.


The estimated income tax and an earlier flat tax scheme have generated little revenue. These taxes initially had a presumptive element (taxable income for retail traders was fixed at 7 percent of turnover), but this element was dropped in 1993.


There are 500,000 income tax cases under litigation, and around 100,000 cases related to indirect taxes. The gross amount of taxes under dispute is Rs 540 billion (4 percent of GDP).


The 1998/99 budget restricted Modvat credits to 95 percent of duty paid on inputs, partly to offset revenue losses through fraudulent claims.


See Shome, 1997, for a comprehensive list of indirect tax exemptions.


A detailed discussion of tariff reforms and trade liberalization is contained in Chapter IV of India—Selected Issues (IMF Staff Country Report No. 97/74, September 1997).


Average tariff rates are based on 1992/93 import weights. The tariff collection rate (customs revenue divided by customs imports) includes special customs duties and a few other items but should move broadly in line with average tariffs. However, since 1991, a large number of items were moved off the restricted import list and have typically become subject to the maximum tariff. This process, together with a phasing-out of some end-use exemptions, has helped limit the decline in the tariff collection rate.


Economic reforms since 1991 have led to efficiency gains and a subsequent increase in growth rates (Chopra et al. 1995). There has been no study that attributed productivity growth to individual components of the reform program. Recent cross-country studies, however, point to a generally positive impact of fiscal reforms on growth (Gerson 1998).


For example, there are tariff incentives for the import of capital goods, provided the importer accepts the obligation to export goods of at least four times the import value of the capital good (the so-called EPCG scheme). Owing to a lack of controls, this incentive is frequently misused to import capital goods at low duty rates.


There are indications that the customs equation is not co-integrated (see Annex). However, the estimated elasticities have been broadly consistent with results from the 2SLS approach.


The search extended to variables that could have either direct or indirect effects on tax revenue (e.g., growth, inflation, and financial variables), and also included other model specifications (e.g., two-stage least squares).


The error-correction coefficient for corporate tax revenue is larger than one, which could have economic reasons (e.g., changes in the adjustment process due to structural breaks) but would warrant further analysis.


A high ratio of external debt to GDP creates pressure to generate revenues for debt servicing. India’s external debt has been well managed in the 1990s, and its external debt ratio was relatively low at 25 percent in 1997/98.


This result needs to be qualified in two respects. First, state and local taxes in India account for about 6 percent of GDP (see Table I.1) which is higher than in many other developing countries. A comparison of central government revenues therefore does not fully reflect India’s revenue position. Second, however, official GDP estimates in India are widely believed to under-report economic activity by a factor of 20 percent or more. According to such estimates, central government tax revenue could be as low as 8–8½ percent of GDP, and total public sector tax revenue would be around 12½ percent of GDP.


India’s tax rate on foreign companies is 48 percent, with other countries either taxing foreign companies at the same rate as domestic companies, or at a lower rate (Indonesia).


Under the current presumptive filing scheme, persons who fulfill one out of six wealth indicators (ownership of a house, car, telephone, credit card, or club membership; foreign travel) are required to file income tax declarations. This includes, e.g., many elderly people who have no taxable income.


In recent years, estimates for the revenue loss from tax cuts have frequently included offsetting revenue increases resulting from improved compliance and stronger collection efforts. Such effects would lead to higher tax elasticity and should not be included in M; however, the budget speech usually gives estimates for the direct loss caused by tax measures.


All variables included in the analysis have been found integrated of order one, using a Phillips-Perron test with critical values adjusted to allow for structural breaks.


There are no critical test variables for cointegration with imposed structural breaks in the literature. However, the critical value for the cointegration test should lie between the standard Phillips-Ouliaris values and those presented by Gregory and Hansen (1996) who tested for cointegration with unknown structural breaks. This would suggest a range between −3.07 and −4.68 at the 10 percent level of significance. Since the error correction term is significant in all short-term models except for the customs equation, a level around −3.30 appears likely.

India: Selected Issues
Author: International Monetary Fund