SPECIAL ARTICLEEconomic & Political Weekly EPW june 7, 200847Reforming India’s Fiscal Transfer System: Resolving Vertical and Horizontal ImbalancesC Rangarajan, D K SrivastavaTwo central problems in a fiscal transfer system relate to resolving vertical and horizontal imbalances. This paper looks at the methodological background of fiscal transfers followed by recent finance commissions, particularly the Twelfth Finance Commission. It is noted that in India, there is long-term stability in the share of states after transfers in the combined centre and state revenues. This stability depends on linking the share of states in the transfers, particularly tax devolution, with the difference in the buoyancies of central and state taxes. In the context of horizontal imbalance, it is argued that some of the recent finance commissions have implicitly followed an axiomatic approach to tax devolution and brought in some normative elements in determining grants. In spite of large differences in fiscal capacities, a high degree of equalisation has been achieved. It is shown that for TFC recommended transfers, nearly 88 per cent of the needed equalisation was achieved while devoting 50 per cent of transfers to resolving vertical imbalance. A methodology is also developed to determine the weights of the vertical and equalising components of transfers through devolution.C Rangarajan (c.rangarajan@nic.in) is a member of the economic advisory council to the prime minister, New Delhi. D K Srivastava (srivastava@mse.ac.in) is at the Madras School of Economics, Chennai.With the recent Constitution of the Thirteenth Finance Commission, issues of resolving vertical and horizon-tal imbalances in the system of fiscal transfers in India have once again come to the fore. The Twelfth Finance Commis-sion (TFC) submitted its report1 at the end of 2004 against the backdrop of a severe fiscal stress affecting government finances, particularly states’ finances in India. The report contained, apart from the recommendations concerning the core tasks of the Finance Commission regarding tax devolution and grants, a detailed road map for the restructuring of India’s public finances including an incentive linked debt-relief scheme for the states. Government finances have shown significant improvements since then. The targets of achieving a fiscal deficit and revenue deficits under the restructuring programme seem well within reach for the central as well as state governments. Many states have en-acted fiscal responsibility legislations and others are following suit. In spite of these achievements, the fiscal transfers system in India requires further reforms concerning both its vertical and horizontal dimensions. These concerns primarily revolve around the following main questions:Stability in vertical transfers: Vertical transfers should be sta-bilised around an appropriate level. These should not be continu-ously changed in favour of one side or the other. The question assumes importance also because of the likely impact of the pro-posed goods and services tax (GST) on vertical imbalance in India. Composition of transfers: The composition of transfers should be changed towards grants compared to tax devolution and within grants, larger emphasis should be on grants on a statutory basis as recommended by the Finance Commission rather than grants at the discretion of the centre.Gap-filling approach to determining transfers: In the case of horizontal transfers, the long-term criticism of the Indian approach has been the so-called gap-filling approach in the assessment of needs and resources by the finance commission because of the implicit adverse incentives.Measurement of fiscal capacity: In applying the equalisation principle, measurement of fiscal capacity of states is a key requirement. The measurement of state level fiscal capacity in India is proxied by estimates of the gross state domestic product (GSDP) at factor cost. This provides an incomplete indicator of fiscal capacity though the Central Statistical Organisation (CSO)prepares comparable estimates of GSDP. We need a more compre-hensive indicator of fiscal capacity.Determination of relative weights of sharing criteria: The rev-enue sharing criteria used by the Finance Commission accounts

SPECIAL ARTICLEjune 7, 2008 EPW Economic & Political Weekly48by far the largest share of transfers. However, the relative weights assigned to different criteria remains by and large ad hoc. There is a need to develop a more objective framework for determining suitable weights for the alternative revenue sharing criteria. Bail outs and controls on borrowing: In a system where states have been borrowing heavily from the centre, there is a built-in expectation that the centre will provide a bail out from time to time. This leads to strong adverse incentives for the states to finance current expenditures through borrowing from the centre and other sources and expect that either a gap-filling grant or a debt-service write off will bail them out in future. Growing centralisation of expenditure on state subjects:This is an issue concerning the relative ambits of assignments of the two tiers of governments. There is a clearly noticeable trend of the central government getting involved in progressively spending more and more on subjects that are clearly under the concurrent or the state list in the Constitution, sometimes through the state governments and sometimes bypassing them. This paper attempts to address these questions. Section 1 summarises the evolv-ing fiscal scenario since the recommenda-tions of the TFC. Section 2 looks at the issue of resolving the vertical imbalance in the context of the proposedGST. Section 3 looks at the horizontal dimension of transfers, particularly the equalisation methodology and its adaptation in India in an axiomatic framework guiding tax revenue sharing. It also highlights the extent of equalisation achieved by the TFC recommendations by decomposing recommended transfers into vertical, equalising, and special needs com-ponents. Section 4 looks at the considera-tions relevant for determining suitable weights for the revenue sharing criteria used by the finance com-missions. Section 5 discusses the equalising health and education grants recommended by the TFC and the need for strengthening these. Section 6 looks at the steps relative to borrowing by the states in the wake of the TFCrecommendations. Section 7 provides the concluding observations. 1 Fiscal Developments since Recommendations of TFCTheTFC deliberated on fiscal transfer issues in the background of severe fiscal imbalances affecting both central and state finances. To reverse these trends, the TFC recommended a scheme that pro-vided for a major restructuring of government finances including borrowing and on-lending regimes for states. These changes were aimed at limiting the borrowing levels of both tiers of gov-ernments to sustainable levels, removing the adverse incentives in the on-lending mechanism and maximising growth by keeping the revenue account in balance and augmenting the saving rate. The evident progress towards lower revenue deficits since 2004-05 also led to the elimination of public sector dissaving. This was accompanied by moderate interest rates and an increase in investment, thereby leading to higher growth. During 2001-02 to 2006-07, the aggregate saving rate has gone up from 23.5 per cent to 32 per cent. Nearly half of this increase comes from the turnaround in public sector saving, which increased from (-) 2.0 per cent to 2.0 per cent, a turn about of more than 4 percentage points coming from the reduction in the government’s dissaving. This corresponds to a fall in the combined revenue deficit from about 7 per cent of GDP to a little over 2 per cent during the same period. Table 1 shows the profile of fiscal imbalance of the central and state governments, as indicated by revenue, fiscal and primary deficits from 1990-91 to 2006-07. Consistent with the restructur-ing plan suggested by the TFC, there are clear signs that both the central and state governments are likely to meet the fiscal responsibility targets of re-ducing the fiscal deficit to GDP ratio to 3 per cent. In 2006-07, the revenue deficit of the states relative to GDP fell to a near-zero level.The fiscal deficit of the states has also fallen to below 3 per cent of GDP. These changes indicate that the revenue deficit and fiscal deficit targets relative to GDP set out by the TFC for 2008-09 are well within reach. With the fall in fiscal deficit ratio, the debt-GDP has also started falling. For the centre, after reaching a peak at 63.8 per cent in 2004-05, it has fallen to an esti-mated level of 60.3 per cent in 2006-07. The state debt relative to GDP also fell in the corresponding period from 33.4 to 30.8 per cent. The combined debt-GDP ratio has fallen by nearly 6.6 percentage points from 82.4 to 75.8 per cent. Thus, although the fiscal deficit toGDP ratio has fallen, the fall in the debt-GDP ratio is slower and it will take some more years to reach the target of 56 per cent ofGDP as set out by the TFC. That is why, it is impor-tant to continue to adhere to the suggested reform path beyond 2008-09. As shown in Rangarajan and Srivastava (2005), the ad-justment period for reaching the debt level relative to GDP at which it should be stabilised could extend beyond the mid-1930s of the current century while maintaining the fiscal deficit of 6 per cent ofGDP throughout the period. It can be advanced by a few years by achieving a higher growth rate but the broad message is that fiscal responsibility targets will have to be adhered to for a long period.2 ResolvingVerticalImbalanceAn excess of federal revenues relative to its responsibility and a corresponding deficit in the state accounts where expenditures exceed own revenues is referred to as the vertical fiscal gap. The notion of a vertical gap conceptually contrasts with a benchmark situation in which responsibilities and resources perfectly match those for the two tiers of government. In federal systems, a vertical gap is often deliberately created for efficiency gains that result from the relative assignments and fiscal transfers that are used to Table 1: Fiscal and Revenue Deficits of Centre and States(% to GDP) Fiscal Deficit Revenue Deficit Primary Deficit CentreStatesCentreStatesCentreStates1990-91 7.763.273.230.924.021.761991-92 5.502.86 2.460.861.471.201992-93 5.312.762.450.681.201.011993-94 6.932.37 3.760.442.710.551994-95 5.632.70 3.030.601.330.811995-96 5.012.62 2.470.680.850.791996-97 4.822.69 2.361.160.520.841997-98 5.782.873.021.061.510.911998-99 6.444.22 3.802.482.01 2.181999-2000 5.35 4.67 3.45 2.75 0.74 2.362000-01 5.644.25 4.042.540.93 1.792001-02 6.184.21 4.392.591.471.472002-03 5.924.174.402.251.11 1.312003-04 4.474.46 3.562.22-0.031.502004-05 4.003.49 2.51 1.16-0.060.682005-06 4.113.202.590.480.390.702006-07 3.712.58 2.030.04 0.15 0.16GDP figures relate to the 1999-2000 base series. Figures prior to 1999-2000 have been adjusted by a conversion factor.State fiscal data are taken from RBI and follow RBI coverage and definitions.2006-07 data for the centre is RE and for the states, BE.Source (basic data): RBI and CSO.

SPECIAL ARTICLEEconomic & Political Weekly EPW june 7, 200849close the gap or convert it into a balance. The main justification for such transfers may be listed as follows: first, transfers may be purely passive responses to the asymmetric decentralisation of expenditure and revenue-raising authority (vertical transfers). Second, these may be used to equalise the fiscal capacity of the regions to avoid inefficient migration of persons and businesses among regions and to foster horizontal equity in the federation as a whole [Boadway et al 2002]. Third, these may also be used in conditional forms to neutralise fiscal externalities imposed by regional governments on other regions, as well as to achieve national standards in social programmes and to induce efficiency in the internal economic union of the federation [Dahlby 1996]. Finally, these may be used as instruments for insuring regions against shocks to their fiscal capacities [Lockwood 1999]. 2.1 SomeAnalyticalConsiderationsThe terms vertical gap, vertical imbalance and vertical fiscal imbalance are often used interchangeably in the literature. How-ever, in the more recent literature, following a normative approach, a distinction is made between vertical gap, optimum vertical gap and vertical fiscal imbalance. Boadway and Tremblay (2005) and Dahlby and Wilson (1994) define vertical fiscal imbalance in revenue-raising as a deviation from the optimum vertical gap where the optimum vertical gap is a situation in which the marginal cost of public funds is equalised across the levels of government. In many studies, the allocation of spending responsibilities is taken as predetermined and the issue is how revenue-raising and federal-regional transfers should be designed so as to achieve a second-best optimum in a decentralised setting, given that taxes are distortionary.In Boadway and Tremblay (2005), the notion of imbalance is related to the inability to achieve a second-best optimum in a decentralised federation and the distinction between the vertical fiscal gap and vertical fiscal imbalance (VFI) reflects that inability. Specifically, the vertical fiscal gap is taken to be the optimal level of transfers when the second best is achieved by a hypothetical central planner or equivalently a unitary national government that can take coordinated decisions for both levels of govern-ment. A VFI is then defined as any deviation whether positive or negative from the optimal vertical fiscal gap. These deviations will occur in a decentralised setting because of the fact that regional governments emit fiscal externalities on one another [Keen 1998] and are unable to coordinate their decisions. The existence of a VFI will be an optimal response of the federal government to this coordination failure between regional governments and will be efficiency-enhancing.Empirically, separatingVFI from the horizontal fiscal imbal-ance (HFI) is quite difficult although it is attempted here in Sec-tion 4 in the context of the TFC transfers. Bird and Tarasov (2002) observe that “it is important to understand that the two concepts of fiscal balance… – VFI andHFI – cannot be cleanly separated. One way to think of VFI, for example, is that it might be considered to be eliminated – that is vertical fiscal balance is achieved – when expenditures and revenues (excluding transfers) are bal-anced for the richest local government, measured in terms of its capacity to raise resources on its own. Even if this is achieved, fiscal gaps orVFI will of course still remain for all poorer local governments. Generally, it is common to discuss such gaps instead asHFI.” In Section 4, we follow the approach of taking the per capita transfer to the richest state as the benchmark for calcu-lating the vertical transfers. 2.2 Long-term Stability in Vertical ImbalanceAs shown in the TFC report, in India there has been long-term stability in the vertical distribution of resources after transfers. It is remarkable that while the relevant ratios and shares have been adjusted from time to time, there is a perceptible stability of the relative shares of the centre and the states in the combined reve-nue receipts and combined revenue expenditures. The main fea-tures of the resultant vertical distribution of resources may be highlighted since the period of the Seventh Finance Commission as follows: (a) prior to transfers, the centre collects, on an aver-age, about 63-64 per cent of the combined revenue receipts; after transfers, states get nearly 64 per cent of the combined revenue receipts; and (b) this enables the states to spend nearly 57 per cent of the combined expenditure, on an average, on the revenue account. The centre spends about 43 per cent of the combined revenue expenditure by retaining 36 per cent of revenues after transfers by borrowing relatively more. Table 2 summarises this picture for the period since the Seventh Finance Commission.The Eleventh Finance Com-mission (EFC) had suggested (for the first time) an indicative benchmark of 37.5 per cent cov-ering all transfers from the centre to the states with a view to achieving stability in the overall transfers from the cen-tre to the states. Given the his-torical trends and the current relatively high buoyancies of central taxes, particularly direct taxes and service tax, the TFC suggested a marginally higher benchmark of 38 per cent. The TFC also recommended an in-crease in the share of states in central taxes to 30.5 per cent of di-visible revenues. There has been an argument that this share should be fixed in nominal terms for a few decades or so. It can be argued that the objective of stability will not be served by fixing the share of the states in the central taxes in nominal terms as long as the central and state taxes are growing with different buoyancies. In particular, some upward adjustment is needed if central taxes are growing more than those of the states. At the present juncture, this was justified as the centre’s tax buoyancy is expected to be relatively higher due to their exclusive power to tax the base of growing services while for some time states will be undergoing adjustments on account of moving to the state level value added tax (VAT). As detailed in Annexure 1, it can be shown that between any two periods, the share of states in the total tax revenue of the centre and the states after transfers will be constant only if the share of states in the central taxes is increased by the margin by which the buoyancy of central tax revenues exceeds the buoyancy Table 2: Share of States in Combined Revenues (%)Average RevenueReceiptsRevenue (Award Period) ExpenditureFinance Before After Commissions Transfer Transfer Seventh 35.3 61.4 58.0Eighth 34.662.0 55.7Ninth 37.564.756.9Tenth 38.663.056.8Eleventh 39.0 63.9 57.1Source (basic data): Indian Public Finance Statistics.

SPECIAL ARTICLEjune 7, 2008 EPW Economic & Political Weekly50of combined tax revenues. This result can also be stated in terms of the buoyancies of central and state taxes. Representing the respective buoyancies of state, central and combined tax reve-nues as b, c and d and the share of states in central taxes as t and t’ between two periods, it can be shown that the share of states (or centre) in total tax revenues after transfers will be constant between the two periods ift’ – t = (c-d) g (1)where g is the GDP growth rate. Ifα is the share of states’ own revenues in total tax revenues, this condition can also be written as t’ – t = (c-b) α . g (2)This also implies that for stability, there should be no change in the share of states if the buoyancy of central taxes is equal to that of the states.2 Adjustments are also needed if the central govern-ment changes the size of the divisible pool by additional sur-charges and cesses that are not divisible. A scheme of assignment of resources, heavily in favour of the centre purely for efficiency reasons, is always prone to lead to a centralisation of expenditures in direct and indirect ways. There is a noticeable tendency in India for various expenditures in the concurrent list and often even if they belong to the state list, to be incurred by the central government. 2.3 Vertical Imbalance and GSTConsidering some important forthcoming tax reforms in India, it is also important to recognise that the vertical imbalance would be affected depending on the way the GST is implemented in India. In Australia, the implementation of theGST-led to a sub-stantial increase in the vertical imbalance because the states agreed to forego a number of taxes assigned to them in favour of a nationalGST. In India, the 2007-08 budget has mentioned the plan for implementing a nationalGST by April 1, 2010. The exact contours of the plan for the GST, which have not yet been spelt out by the government, would have a significant bearing on the verti-cal imbalance in the system. In this context, the following three, issues are of importance. 2.3.1 Nature of GST RegimeFirst, it is important to determine whether the proposal is for a centralGST, state GST, concurrent or dualGST. In the first two cases, the pre-transfer vertical imbalance would increase sub-stantially. The options may be as follows:Central GST: In this case, theGST is levied by the central gov-ernment and state VATs are all subsumed in this central levy. This would be like the Australian model. This option would deliver harmonisation by definition as only uniform rates will prevail. The whole country would be one common market and there would be no problems related to interstate trade. But this will increase the vertical imbalance tremendously. States will have to forego their power to levy a sales tax. A provision will have to be made for the distribution of the centrally collected VAT. Although a similar arrangement has been implemented in Australia, it will have a significant impact on the nature of fiscal federal relations. States will lose their autonomy to fix rates and collect their own revenues. It is doubtful that states will agree to such an arrangement. The scheme of redistribution would also be required to follow a principle different from the one normally used by the finance commissions so that states are adequately compensated for the revenues that they would have otherwise earned through the existing system of statevat or sales taxes. Concurrent or Dual GST: This seems the most practical route as it can be implemented while maintaining the current pre-and post-transfer profiles of vertical imbalance. It would require that states be enabled to tax services and the service tax rate should be the same as that for goods. Alongside, central government should be enabled to tax value added in the case of goods up to the retail stage. These changes would lead to a comprehensive and unified system of taxation of goods and services. The major problem in this case will be handling interstate transactions. In the literature, three main solutions have been suggested, viz, (a) system of compensating VAT (CVAT); (b) dual VAT, and (c) viable integratedVAT (VIVAT). The system ofCVAT is also known as the Versano proposal. McLure Jr (2000) suggested a modified ver-sion of the CVAT. InCVAT, uniform definitions and laws for the tax base in all jurisdictions are needed. States are allowed, however, to have their own tax rates with the proviso that all interstate transactions are zero-rated for state VAT. In addition to the cen-tralVAT, the central government levies a CVAT for all interstate transactions. The rate of CVAT is common across states. For inter-state imports, a system of deferred payment of state VAT and credit forCVAT is then put in place. The CVAT is an additional fed-eral level tax to ensure the tax revenues that might otherwise be lost to cross-border tax evasion. One alternative to CVAT in con-current tax regimes is the dualVAT as practised in Canada [Bird and Gendron 2000]. In dualVAT, central and state VAT rates are applied. States have autonomy to determine the state VAT rates. The central VAT is included in the tax base of the state VAT. States, therefore, have an incentive to collect the central component, if they are asked to collect it. The VIVAT system pertains to the ex-clusive state level VAT system.State Level VAT: This option takes one to the other extreme where the GST/VAT is levied exclusively by the state governments. This also changes the vertical balance equations drastically al-though in favour of the states. The centre will then largely lose power to undertake transfers for purposes of horizontal trans-fers. Even to provide centrally provided public goods, it may need to save some sumptuary excises for itself. Otherwise it may have to depend on reverse transfers. The problem of interstate harmo-nisation and interstate transactions will remain. For the case ofan exclusive state VAT regime, Keen and Smith suggested the system of VIVAT. In this case, for all intermediate purchases, that is, sales between dealers, a uniform tax rate regime is advocated for sales between dealers. This would be applied to transactions within a state as also across the states. There is no central VAT or GST.If the vertical imbalance in the system is not to be drastically altered, the concurrent or dualVAT regime seems to be most relevant in the current fiscal conditions of India.

SPECIAL ARTICLEEconomic & Political Weekly EPW june 7, 2008512.3.2 Determining Overall RateThe second issue is to determine a suitable GST rate. At present, goods are taxed at the core rate of cenvat at 16 per cent (changed to 14 per cent in the 2008-09 central budget) and state VAT of 12.5 per cent. This together would be very high, although it would be less than 28.5 per cent as the 16 per cent rate applies to value added only up to the manufacturing stage and the GST will have a larger base. The service tax rate is 12 per cent. The suggestion of the Kelkar Committee to aim at a 20 per cent combinedGST rate seems to be a suitable target as it compares well with some of the internationalGST rates. The highestGST rates are in Sweden and Denmark at 25 per cent. At the lower end, Switzerland, Japan, Thailand and Singapore have GST/VAT rates at 5 per cent or marginally above. 2.3.3 Determining GST Rate Components The third issue relates to decomposing the overall GST rate into its central and state components, making sure that the relative pre-transfer revenue levels are not disturbed. To achieve a 20 per cent composite rate, both tiers of governments have to jointly bring down the overall tax rate, which at present amounts to 16 per cent (14 per cent from 2008-09) and 12.5 per cent on the respective tax bases of the cenvat and state VAT as far as manufacturing and sales of goods are concerned. While the tax rate on goods can come down, that on services, which is at 12 per cent may have to be incrementally raised to bring it closer to the long- term desired norm. In the medium term, with a view to preserving our federal structure, a system of dual taxation consisting of a state GST (SGST) and a central GST (CGST) seems to be a viable option. The Kelkar Committee had suggested a division of the overall rate of 20 per cent into a 12:8 ratio in favor of the centre. This may need to be re-examined with current levels of revenues under cenvat and service taxes and the state vat and other related taxes that may be subsumed in theGST. 3 Resolving Horizontal Imbalance: Equalising TransfersIn theory as well as practice, a system of equalisation transfers is considered desirable as it is consistent with both equity and effi-ciency. The efficiency implications follow from two considera-tions: (a) locational inefficiencies that can result from inefficient migration induced by fiscal surpluses is neutralised by equalisa-tion transfers; and (b) the redistribution implied by equalisation transfers from richer to poorer states also gives a return to the richer states by avoiding congestion resulting from excessive migration in the context of services provided by these states that are in the nature of “congestible” goods.Courchene (1984, 1998) had argued that the efficiency case of equalisation depends on the existence of fiscally induced migration. If there is no fiscally induced migration, there is no efficiency case for equalisation. In a recent contribution, Dahlby and Wilson (1994) make a case for equalisation on efficiency grounds even in the absence of fiscally induced migration. They examine the role of equalisation grants as an instrument for maximising a social welfare function or minimising the “excess burden” of taxation. Optimal tax theory suggests that the social cost of raising revenues depends not only on the size of the tax base but also on the responsiveness of the tax base to tax rate changes. They argue that it is important to use “responsiveness” (or buoyancies in the formula for equalisation) rather than just the tax rate. The higher the demand and supply elasticities to tax rate changes, the larger is the marginal cost of public funds. On this basis they show that differences in fiscal capacities, even in the absence of fiscally induced migration, are sound grounds for arguing for equalisation. 3.1 Equalisation: Some International PracticesIn Canada, the “equalisation” payments have been mandated in the Constitution since 1982, which commits the federal govern-ment to the “principle of making equalisation payments to ensure that provincial governments have sufficient revenues to provide reasonably comparable levels of public services at reasonably comparable levels of taxation”. The equalisation transfer to a province in absolute amount is determined by applying the aver-age revenue effort to the difference between the standard base and actual base for that province with respect to the various revenue sources. This produces an estimate of revenue, which is higher than the actual revenue for provinces that have “below-average” capacity. This exercise is done for all revenue bases used by the provinces [see, for example, Rangarajan and Srivastava 2004a for a discussion]. In the Canadian system, there is no refer-ence to cost differentials and the states are free to use their equal-ised capacities in providing any mix of public goods and merit goods. The equalisation grants are supplemented by health and social sector transfers that are equally important in volume and are also of an equalising nature. The Australian system of equalisation transfers [see, Rangarajan and Srivastava 2004b] goes into the question of cost differentials relevant for comparison with some notion of equal efficiency in the provision of goods and services by the provincial authorities. The guiding principle of horizontal transfers system is fiscal equalisation, which is defined by the Commonwealth Grants Commission (CGC) as follows “state governments should receive funding from the pool of goods and services tax revenue and healthcare grants such that, if each made the same effort to raise revenue from its own sources and operated at the same level of efficiency, each would have the capacity to provide services at the same standard”. The Australian equalisation differs from theCanadian equalisation due to the reference to efficiency and standard of services. The Canadian system makes a reference only to equalisation in a fiscal capacity. In Australia, fiscal equalisation looks at both the revenue and expenditure sides. It may be noted that the typical methodology for determining equalisation transfers is not totally devoid of adverse incentives, as discussed in some recent literature [for example Garnaut 2002] on the subject. The ground conditions in India are different from those in Canada or Australia in two critical respects. First, the extent of difference in the resource bases is far larger than in Australia or Canada. For example, the ratio of maximum per capitaGSDP to minimum is 1.6 to 1 between Ontario (leaving Alberta as a special case) and Prince Edwards Islands; in Australia, the

SPECIAL ARTICLEjune 7, 2008 EPW Economic & Political Weekly52ratio of per capitaGSDP of New South Wales to Tasmania is 1.5 to 1. In India, this ratio between Maharashtra (leaving Goa as a special case) and Bihar is close to 6 to 1. The second difference is that the population that resides in the main “donor” states as compared to main recipient states is much larger in Canada and Australia. In India, it is the other way round. As a result, the amount of redistribution implicit in the equalising scheme is far larger when the recipients are more than donors, making it extremely difficult to achieve full equalisation. Thirdly, there are large interstate differences in cost conditions in India due to differences in density and composition of population, nature of terrain, etc. In India, the horizontal imbalance is resolved through a combi-nation of tax devolution and revenue-gap grants. In Canada, this is done by grants. In Australia, this is done by sharing the revenue under the goods and services tax (GST) topped up by the health-care grants. The Australian system has switched from grants to revenue sharing and back from time to time. Some economists consider grants as the right means of transfers. States themselves overwhelmingly prefer revenue-sharing. The transfer system in India has evolved in a manner that relies on both modes of trans-fers. Finding a suitable combination is the relevant problem. 3.2 India: Tax Revenue Sharing under an Axiomatic FrameworkAn explicit equalisation methodology was not developed or fol-lowed in India. Instead, an elaborate framework of tax revenue sharing was developed supplemented by revenue-gap grants. This methodology can also lead to an equalising system of trans-fers if some basic principles are followed. The evolution of criteria-based tax revenue sharing as recommended by the Finance Com-mission can be interpreted in an axiomatic framework. Fully equalising transfers are a special case under this axiomatic framework. The following five axioms may be proposed as desir-able axioms for criteria-based revenue sharing: (a) normalisation 1; (b) normalisation 2, (c) horizontal equity; (d) comprehensiveness, and (e) neutrality. The two normalisation axioms and horizontal equity can give rise to a system of fully equalising transfers. Axiom 1: Normalisation 1 – If two states have the same criterion values, their shares should be proportional to their populations.Axiom 2: Normalisation 2 – The sum of the shares of all states should add to 1.Axiom 3: Horizontal Equity – Between any pair of states, the state with lower per capita fiscal capacity should have higher per capita share and per capita shares should be equal for states with equal per capita fiscal capacity.Axiom 4: Comprehensiveness – In determining the share of any one state, information on all states should be used. A corol-lary of this is that under each criterion, every state should get a positive share. Axiom 5: Neutrality – The allocation criterion should be neu-tral with respect to the organisation of states. There should not be an incentive to bifurcate states with a view to benefiting from the allocation mechanism.These axioms are discussed in Annexure 2. It may be noted that the finance commissions have endeavoured to meet these criteria even though they were not explicitly stated. The same cannot be said, for example, of the dispensation criterion for determining grants under the Gadgil formula. Under their devia-tion formula, states with per capitaGSDP above the national per capitaGDP or average per capitaGSDP get a zero share, thereby not satisfying the comprehensiveness criterion. Similarly, Assam with a lower per capitaGSDP may some times get a per capita grant which is lower than Meghalaya’s, thereby violating the horizontal equity criterion. Since shares under the individual criteria under the Gadgil formula are not made public, it is not possible to subject these to a critical review. 3.3 Measuring Extent of EqualisationFor broad issues like those of resolving vertical and horizontal imbalances, the fiscal transfer scheme needs to be analysed in terms of the combined effect of all components of transfer. The total recommended transfers to states may be calculated based on the projections of the TFC regarding tax devolution and add-ing to these, the grants already specified in nominal terms. The grant profile tends to give larger grants in the initial years and to get a better picture of the interstate distribution of the recom-mended transfers, we have given the average annual transfers by dividing total transfers by five and considering the average annual transfers as centred in 2007-08. These are given in Appendix Table 1 in per capita terms using state-wise population of 2001 Census.The issue of determination of revenue-gap grants as “gap-fill-ing” has been raised by many authors from time to time. The con-cern arises from the implicit adverse incentive for a state to cre-ate a history of expenditure in the expectation of getting a grant later. For the period covered by the recent finance commissions, except for a very limited number of general category states and for some years, the revenue-gap grant is given mostly only for the special category states. The fact that the special category states get a large share of the revenue-gap grant follows from their large committed expenditures linked to the large plan assistance that they have obtained in the past.Let the per capita income (GSDP) of the states arranged in ascending order of per capitaGSDP be denoted by y1,y2 …,yn and the corresponding population be denoted by N1, N2, … Nn. If the vertical fiscal is measured with reference to the richest state, in per capita terms, it may be defined as [e-a.yn] (assuming that e>a.yn), where e is per capita expenditure norm, a is the average tax-effort, and yn is the per capita fiscal capacity of the highest income state. If e is exogenously or normatively determined, the total transfer to the highest income state is given by Nn. [e-a.yn].Since every state gets at least the amount [e-a.yn] in terms of per capita transfers, we may write total vertical transfer as ΣNi [e-a.yn] = [e-a.yn] ΣNi (3) All other states have a lower fiscal capacity and would get an amount, in per capita terms, higher than that obtained by the highest income state in a progressive scheme of transfers under the axiom of horizontal equity. The transfers to these states can be seen as consisting of the vertical component equal to the per capita transfer to the highest income state, an “equalising” com-ponent due to deficiency in fiscal capacity and a residual, which

Total benchmark transfers TFC recommended transfers

SPECIAL ARTICLEjune 7, 2008 EPW Economic & Political Weekly54equalisation grants, which use a relatively small amount of trans-fer but improve the equalising content of transfers significantly as they go only to those states that have a lower than average per capita health and education expenditure. The TFC has made a new beginning in this context. However, the gap covered was only 30 per cent in the case of health and 15 per cent in the case of education. These ratios will have to be increased to a larger extent to achieve full equalisation. Several qualifications may be noted in respect of the above comparisons. First, the equalisation benchmark is calculated with a revenue side approach and expenditure side considera-tions are not included. Secondly, a macro base reflecting fiscal capacity like theGSDP is used. Thirdly, the highest per capita GSDP among states, excluding Goa, is used as the benchmark. Fourthly, shortfalls from the equalisation benchmarks are equally weighted. Ideally shortfalls for lower income states would require a relatively higher weight. 4 Determining Weights of Tax Devolution CriteriaConsiderable time is spent by the finance commissions on deter-mining the relative weights that should be attached to the different tax devolution criteria, yet an explicit analytical framework has not been spelt out for this purpose. Given the preceding discussion, it is possible to develop a framework for determining suitable weights for alternative devolution criteria. Given the benchmark for equalisation (such as the highest or mean income), the total amount needed for equalisation gets deter-mined. Thus, the amount needed for equalisation (horizontal transfer) is given by H=a.ΣNi(yn-yi) (6)Here, a is the average tax effort. The vertical transfer (V) may be determined exogenously as it relates to the overall balance of responsibilities and resources. Both the vertical and horizontal transfers can be given through either route, i e, grants or tax devolution. In the case of both tax devolution and grants, three components can be distinguished: vertical component, horizontal component that is equalising, and horizontal component reflect-ing other considerations. Suppose these three components given through grants are referred to as gv (same for all states), ghi, and goi. The latter two have different per capita amounts for different states. The corresponding transfers through tax devolution are, say, dv, dh i, and doi.Thus, we have the following sets of decompositions:ti= tv+thi+toi (same as equation 4 given earlier)gi= gv+ghi+goi di= dv+dhi+doiwhere tv= gv+ dv, thi=ghi+dhi, and toi=goi+doi (7)Let a share W of per capita devolution (=W.d) be given for (total devolution D= d.ΣNi) equalising horizontal transfers (using criterion like the distance criterion), a share W1 of d for vertical transfer (under population or similar criterion). The remaining part of devolution constitutes a share W2, which is given for other considerations (like cost differentials), where the three weights add to 1. W2 may be taken as exogenously determined. Putting together the equalising horizontal transfers under grants and through tax devolution, the following condition should be satis-fied: W.D + ΣNighi = a.ΣNi(yn-yi). This can be solved to yield,W.D = a.ΣNi(yn-yi) -ΣNighi (8)Or, W.d = a.(yn- μ) – gh (9)where d is per capita devolution, yn is benchmark per capita income, μ is mean income (=ΣNiyi/(ΣNi), and gh is mean per capita equalising transfers given as grant. For the vertical transfer, we have V= (ΣNi) [gv+ dv ] or the per capita vertical transfer is, v=[gv+ dv ]. Per capita vertical transfers given through devolution is (1-W-W2)d= dv (10)Using equations (9) and (10), we have, (1-W-W2)/W= dv/ [a.(yn- μ) – gh]. This can be solved to yield,W= (1-W2)*[a.(yn- μ) – gh]/[ dv + a.(yn- μ) – gh] (11)Correspondingly d= [dv + a.(yn- μ) – gh] /(1-W2) (12)This indicates that given (1) the exogenously determined average per capita equalising grant (gh); (2) weight to be given to considerations other than vertical or equalising transfer in devolution (W2); (3) the benchmark (yn) and average per capita fiscal capacity (μ); (4) the average tax-GSDP ratio (a); and (5) the exogenously determined per capita vertical devolution (dv), the weight that needs to be given to horizontal equalising devolu-tion (under the distance or similar criteria) and the total amount of per capita devolution may simultaneously be determined in order to achieve full equalisation. In practice, in applying this to India, the tax devolution criteria in India may be considered in three parts: those meant for verti-cal transfers (population, and a large component of transfer under tax effort, fiscal discipline criteria), those meant for equal-ising horizontal transfers (distance) and those reflecting cost dis-abilities (like area). The effort and fiscal discipline criteria are efficiency-promoting modifica-tions of the population criteri-on and should be taken as part of the group of tax devolution criterion meant for vertical transfers and the deviations may be counted as part of the “other” considerations. Illustrating these considera-tions, with the TFC per capita amounts, centred in 2007-08, subject to some approxima-tions, Table 4 provides the rele-vant numbers. State-wise details are given in Appendix Table 3. The numbers are derived in a manner consistent with the decom-positions given in equation 7. It may be noted that the actual weight given to the distance formula in the TFC was 50 per cent. However, since even in the devolution formula an amount is given to the highest income state, there is a vertical component. The actual weight to the equalising horizontal component is estimated to be 45 per cent, obtained by dividing dh by d. The desired weight as derived for full Table 4: Illustrative Derivation of Weight for Equalisation(in Rs)gv gh go g114.2 38.5129.2 281.87dv dh do d632.5 541.91 37.181,211.58v h o t746.7 580.4 166.41493.4a.(yn- μ) 641.20 Weight to horizontal equalising devolution and per capita devolution Desired Actual (as per TFC)W 0.47 0.45d 1274.34 1211.58Variables defined as in text and amounts are in per capita terms.Source (basic data): TFC report (2004).

SPECIAL ARTICLEEconomic & Political Weekly EPW june 7, 200855equalisation weight is marginally more at 47 per cent. In addition, this is associated with higher per capita devolution amounting to Rs 1,275 instead of the actual amount of Rs 1,212. It may be also noted that increasing the amount of equalising horizontal grants would reduce the total amount needed for devolution almost by the same margin since the term gh occurs in the numerator. This would also affect the weight assigned to the equalising component of the devolution formula but the effect operates through both the numerator and denominator. In general, increasing the equalisa-tion component of grants makes it easier to achieve full equalisa-tion through devolution. That is why, the equalisation grants given in respect of education and health (which was a new type of grant given by the TFC but where equalisation was limited only to 30 per cent) needs to be strengthened. 5 Equalisation Grants for Health and EducationIn devolution formulae, it is difficult to use the mean income as the benchmark as states with per capita incomes higher than the mean income will get a zero share in tax devolution. The practice followed by the finance commissions in India has been to allocate a positive share to all states. It is easier to use benchmarks with reference to the mean of per capita incomes or service levels in determining grants for selected services. In this context, theTFC introduced equalisation grants for education and health with the aim of augmenting the equalisation content of fiscal transfers focusing on two high merit services. In devising a grant that is specific-purpose and aimed at the given sectors, it is important to make up for the deficiency in resources but not to underwrite the deficiency in priority accorded to the sector by the concerned state government. The TFC metho-dology entailed the following steps: (a) derivation of the average preference for allocation to health and education (say “a”); and (b) derivation of the gap of the state-specific expenditure on the concerned service (education/health) from the corresponding group average (general category/special category states) evaluated by applying the average preference to the state’s aggregate expen-diture. Thus, for any service, suppose that the group average per capita expenditure is z and state-specific expenditure for a state, zi. Here, z=Σzi Ni/Σ Ni (13)Subscript i varies over the states belonging to the relevant group. The per capita capacity of a state is given by ri and the average capacity is given by r=Σri Ni/Σ Ni (14)The average budgetary allocation for the given service is given by a = Σzi Ni/Σri Ni Thus, z=a.r and zi = a i.ri .Actual gap in expenditure between a state and the group- average can be seen as the sum of two components: one due to deficiency in fiscal capacity and the other due to giving the con-cerned sector less than average preference. It is only the first part, that is deficiency in expenditure due to lack of capacity, that is taken into account while the deficiency that results from giving less than average preference in budgetary allocation is ignored. Thus, the actual gap may be written as: z-zi = (z-ari) + (ari-zi) (15)or, z-zi = a(r-ri) + (a-ai )ri [where ai = zi /ri ] (16)Thus, the relevant gap is reflected in the first term, which is due to the deficiency in fiscal capacity, given the average allocation to the concerned sector. The second term is the difference due to allocating less than average share given the capacity of the state and this difference does not require to be made up under the equalisation principle. Thus, the total grant should be determined by Ni a(r-ri). In estimating the resources, r was proxied by resources devoted to expenditure excluding interest payments and pensions. In the TFC scheme, only 30 per cent of the equalising grants were provided. Clearly, there is a need for providing larger equalising transfers as this would ease the pressure on tax revenue sharing to accommodate a large share of equalising transfers. 6 Discontinuance of Further Debt ReliefRecommendations with regard to debt relief were formulated in the context of the overall programme for restructuring govern-ment finances in the country. Debt-relief was linked to the state governments meeting a set of conditionalities including the enactment of a fiscal responsibility legislation. According to available information, all state governments, except two, have already enacted their respective fiscal responsibility legislations.Two major recommendations regarding the state level borrow-ing from the centre were: one, to delink grants and loans in plan assistance and two, to discontinue the centre’s intermediation for state borrowing. Once the centre’s intermediation is discontin-ued, the moral hazard in expecting periodic bailouts would also be eliminated or at least significantly reduced. There is now no case for including a “debt-relief” clause in the terms of reference for future finance commissions. The Reserve Bank of India (RBI) has recommended that states should go more and more for the auction route as it leads to price discovery, promotes market dis-cipline, and improves secondary market liquidity. With some degree of flexibility in borrowing from the National Small Savings Fund (NSSF), the states should be able to adapt to the new market-oriented regime without much difficulty. With regard to theNSSF, the obligatory share of the states has been reduced to 80 per cent with effect from 2007-08. TheRBI has made arrangements so that state development loans (SDLs) are eligible for repo trans-actions under the liquidity adjustment facility and it has been decided to introduce the non-competitive bidding facility in respect of the primary auctions of SDLs. While the centre’s intermediation has been discontinued, loans and grants have not been delinked in Plan assistance except in the case of external assistance. This linking requires re-examination on the part of the Planning Com-mission particularly when the assistance is for social sector projects or for poverty alleviation, where the capacity to service the loan is not created as result of assistance. The TFC had also recommended the Constitution of a loan council, which may decide the overall annual borrowing limits for the state governments. While a full-fledged loan council has not been constituted, some steps have been taken towards achiev-ing the related objectives. In particular, the RBI has moved to set up a standing technical committee (STC) with representation from the central and state governments and the RBI. The STC will make annual projections of borrowing requirements of the state governments, build alternative scenarios and suggest alternative

SPECIAL ARTICLEjune 7, 2008 EPW Economic & Political Weekly56strategies and instruments for raising resources of the states. It will also advise a mechanism for annual allocation of market borrowings among the states. It will take note of actual and bud-geted borrowings of the state governments, develop a suitable data-base, assess fiscal risks from issuances of guarantees and advise state governments on various issues relating to their borrowings. It is useful to note that the reference to the debt relief clause has now been withdrawn from the terms of reference to the Thirteenth Finance Commission. This follows from the discon-tinuance of the centre’s intermediation in state borrowing and on-lending to states subject to limited exceptions like external assistance, which is also being passed on back-to-back terms except for the special category states.7 ConcludingObservationsThis paper has reviewed the fiscal transfer arrangements in India in the context of resolving the vertical and horizontal imbalances.Vertical Imbalance:In the literature, a distinction has been made between the concepts of vertical gap, optimum vertical gap and vertical fiscal imbalance. In the context of the Indian transfer system, in resolving the vertical fiscal imbalance, the following points have been made. First, in India, there has been a long-term stability in the share of the centre and states in the combined tax revenues of the system after tax devolution. It may be considered desirable to continue to maintain this stability as long as there are no basic changes in the division of responsibilities between the centre and states. It is further shown that maintaining such stability would requires an upward adjustment in the share of states in the divisible pool of taxes in periods where the expected buoyancy of central taxes is higher than that of the states.Second, the proposed move to a national GST will have significant implications for vertical imbalance. For maintaining the existing extent of vertical imbalance, a concurrent system of GST is recom-mended. The GST rates for the two tiers should be determined taking into account the present level of revenues of the two tiers from the concerned taxes. If, however, a central GST is adopted, vertical im-balance of resources prior to transfers will shift excessively in fa-vour of the centre and the resultant vertical gap will have to be re-solved by a corresponding increase in transfers. This will also have significant implications for the horizontal distribution of transfers. Third, for operational purposes, “vertical gap” is measured in this paper as the total transfer to all states. This reflects division of resources between the two tiers of government without look-ing into the inter se distribution of the share of states among the states. With a view to examining the inter se distribution, the transfers recommended by the Finance Commission are further decomposed into (a) the vertical component of transfers (as dis-tinct from the vertical gap) indicating per capita transfers to all states including the highest income state, (b) equalising transfers indicating the component of transfers only to the states with a fiscal capacity less than the defined benchmark capacity, and (c) a residual reflecting special needs and ad hoc components. Horizontal Imbalance: With respect of the horizontal dimension of transfers, the equalisation approach to transfers, which is followed by some of the important federal systems like Canada and Australia, is also suitable for India. While in Canada, atten-tion is focused on equalising revenue capacities only, in Australia, this is complemented by a comprehensive expenditure side equal-isation covering all services. In this context, the following points are made:Onesubject to certain assumptions, tax revenue shar-ing under an axiomatic framework will result in transfers that will be consistent with the concept of revenue side equalisation used in Canada. The difference is that in the Indian case, a macro base rather than a tax by tax approach is followed on the revenue side. While using a macro approach, there is a need to obtain a better measure of fiscal capacity asGSDP at factor cost is an in-complete indicator for this purpose. The CSO, which prepares comparable estimates of GSDP for the Finance Commission should be asked to prepare a more comprehensive indicator of fiscal ca-pacity taking GSDP at market prices and providing supplementary information on remittances and others influences that add to the spending capacity in different states.Two, a suitable methodology can be developed to objectively determine the relevant weights in the tax-devolution formulae, given the large share of tax devolution in total transfers, which it is not easy to scale down and substitute by grants. Three, using available information, it has been shown that contrary to the con-tention made by several economists, transfers in India are not necessarily “gap-filling” in nature in the recent past, at least for the general category states subject to the exception of one or two states for some years. Taking the TFC recommended transfers, it is shown that, under specific assumptions, systematic elements of transfers constitute a high proportion of transfers: 50 per cent of transfers are used as the vertical component of transfers and nearly 40 per cent is equalising in nature, consist-ent with the revenue side equalisation approach. The remaining is for assessed special needs and goes mainly to the special category states.Four, it is shown that for achieving full equalisation, subject to various assumptions, the weight to distance formula should have been marginally higher to ensure that an equalising transfer has a 47 per cent weight in total per capita devolution. In the TFC for-mula, the 50 per cent weight given to the distance formula pro-vided about 45 per cent equalising transfers. Further, this higher weight should be combined with a higher amount of per capita devolution. It is advisable to increase the amount of per capita equalising grants to reduce the necessity for using devolution to perform this task. Five, in the present exercise, equalisation has been viewed in terms of fiscal capacity equalisation only, which is the approach followed in Canada. Considerations of cost differ-entials and efficiency will modify the results. Finally, equalising grants may be extended particularly in services like health and education. These also provide cases where revenue side equalisa-tion should be supplemented by expenditure side equalisation, where cost and use disabilities should be fully neutralised, extending the methodology suggested by the TFC. Others:IfGST is levied and collected as a central levy, the princi-ples of distribution of transfers to states may need to keep divisi-ble revenues of GST as a separate category. States would need to

SPECIAL ARTICLEEconomic & Political Weekly EPW june 7, 200857be compensated on the basis of “return” requiring estimation of revenue foregone on account of sales and related taxes. As on-lending to states from the centre has been discontinued, there is no case for including a debt-write off clause in the terms of reference to the future finance commissions as has been done in the case of the Thirteenth Finance Commission. Notes 1 The subject of resource sharing in federal systems has important theoretical underpinnings and the TFC report also evoked a number of responses from economists, an example of which is EPW’s issue of July 30, 2005 that carried contributions of some well known experts on the subject. Notwithstanding several issues raised, in writing the overview of these contributions, Amaresh Bagchi, member of the Eleventh Finance Commission, observed in the summary of his overview, “The Twelfth Finance Commission has broken new ground in several key areas and made recommendations which, if fully recommended, will have a far reaching impact on the finances and functioning of government in the country at all levels.”2 If the power to levy the sales tax in respect of three commodities, namely, textiles, sugar and tobacco is reverted to states, the states’ share in the divisible pool of central taxes would be 29.5 per cent. Until this is done, it will be 30.5 per cent.3 Sometimes the area of a state is considered as a scaling factor but this is more appropriately taken as a determinant of per capita cost, which may be higher in states where large areas are sparsely populated. It may also be higher if population density is extremely high.ReferencesBird, R and A V Tarasov (2002): ‘Closing the Gap: Fis-cal Imbalances and Intergovernmental Transfers in Developed Federations’, Working Paper 02-02, Andrew Young School of Policy Studies, Georgia State University.Bird, R and P Gendron (2000): ‘CVAT, VIVAT, and Dual VAT: Vertical ‘Sharing’ and Interstate Trade’, International Tax and Public Finance, Vol 7, No 6, December, pp 753-61.Boadway, R, K Cuff and M Marchand (2002): ‘Equali-sation and the Decentralisation of Revenue-Rais-ing in a Federation’,Journal of Public Economic Theory,Vol5, pp 201-28.Boadway, R and Jean-Francois Tremblay (2005): ‘A Theory of Vertical Fiscal Imbalance’, IFIR Work-ing Paper Series. Courchene, T J (1984): Equalisation Payments: Past, Present, and Future, Ontario Economic Council, Toronto.–(1998): ‘Renegotiating Equalisation, National Polity, Federal State, International Economy’, CD Howe Institute, Toronto.Dahlby, B (1996): ‘Fiscal Externalities and the Design of Intergovernmental Grants’,International Tax and Public Finance,Vol 3, pp 397-411.Dahlby, B and L S Wilson (1994): ‘Fiscal Capacity, Tax Effort and Optimal Equalisation Grants’, The Canadian Journal of Economics, Vol 27, No 3, pp 657-72.Garnaut, R (2002): ‘Equity and Australian Develop-ment: Lessons from the First Century’, Australian Economic Review, Vol 35, No 3, pp 227-43.Government of India (2004):Report of the Twelfth Finance Commission, New Delhi. Keen, M J (1998): ‘Vertical Tax Externalities in the Theory of Fiscal Ferderalism’,International Monetary Fund Staff Papers,Vol 45, pp 454-85.Lockwood, B (1999): ‘Inter-Regional Insurance’, Journal of Public Economics, Vol72, pp 1–37.McLure, C E Jr (2000): ‘Implementing Subnational Value Added Taxes on Internal Trade: The Com-pensating VAT’, International Tax and Public Finance, Vol 7, No 6, pp 723-40. Rangarajan, C and D K Srivastava (2004): ‘Dynamics of Debt Accumulation: Primary Deficit, Growth and Interest Rate’,Economic & Political Weekly, No 38, November, pp 15-21. – (2004a): ‘Fiscal Federalism in Canada: Drawing Comparisons and Lessons’, Economic & Political Weekly,Vol 39, No 33.– (2004b): ‘Fiscal Transfers in Australia: Review and Relevance to India’, Economic & Political Weekly,Vol 39, No 33. – (2005): ‘Fiscal Deficits and Government Debt in India: Implications for Growth and Stabilisation’, Economic & Political Weekly,Vol 40, No 27.Annexure 1: Vertical Stability and Relative Tax Buoyancies We examine the conditions under which the share of the centre (or states) in the combined revenues of the centre and states remains con-stant over time. The following symbols are used. Central tax revenues prior to transfers: RCState tax revenues prior to transfers: RSTotal revenues: R = RC + RSThe buoyancies of central, states and combined tax revenues are given by c, b and d, respec-tively.Thus, c = [Δ RC / RC].[Y /ΔY] orΔRC = c.RC.ΔY/ Y = c.RC.g, where g = ΔY/Y is the growth rate between the relevant periods. Similarly ΔRS = b.RS.g andΔR=d.R.g.Let transfers (T) be a fraction t of central revenues in the initial period.T = t.RCShare of centre after transfers in total revenuesShare (centre)0 = (RC – t RC) / (R) = [RC (1-t)]/ [R]After a given period let the ratio of transfer be t’. The new share of centre in total revenues will be Share (centre)n = [(RC + Δ RC) – t’ (RC + ΔRC)]/ [R+ΔR] = [(RC + ΔRC) (1– t’)] / [R + ΔR] = [(RC + RC.c.g) (1– t’)] / [R + R.d.g] = [RC (1+ cg) (1– t’)] / [R (1+ d.g)]The relative shares of the centre and states between the two periods is constant of [RC (1- t)] / [R] = [RC {(1 + cg)}(1– t’)]/ [R(1+dg)]or 1-t = {(1+cg) (1-t’)} / (1 + dg)(1-t) (1+dg) = (1+cg) (1-t’)Ignoring 2nd order terms, t’ – t = (c-d) g.Thus, if the buoyancy of central taxes is higher than the combined tax revenues, the ratio of transfer to the states will need to go up between two periods in order to keep the rela-tive share of centre and states stable after trans-fers. The extent of increase will depend on the growth rate and the difference between central and combined tax buoyancies. This condition can also be written in terms of buoyancy of state tax revenues.We have, ΔR = ΔR C + ΔRS d R.g = cRCg + bRS.g d.R = cRC + bRS d = [c.Rc] / [R] + [b.RS ] / [R] = c [R– RS] / R + b [RS / R]Let the share of states’ own revenues in total revenues be α = RS / RWe have d=c(1-α) + b α the condition of stabilisation is t’ – t = [c – {c(1-α) + b α} ] g = [c– c(1-α) – b α ] g = [c – c + c α - b α ] g t’ – t = ( c-b)α gThus, an increase in the share of transfer is warranted provided the central tax buoyancy exceeds that of states taxes for keeping the share of the two tiers in total tax revenues stable.Annexure 2: Criteria-Based Tax Revenue Sharing: Axiomatic Basis In discussing this axiomatic framework, the following additional symbols will be used:Ni = Population of state i yi = per capita tax base of state isi= share of the ith states*i= corresponding per capita share Si= total transfer in absolute amount received by a state under a criterionS*i= per capita transfers received by a state in absolute amount by a stateLet the number of states be n. States are ar-ranged in ascending order of per capita tax base, i e, y1 ≤ y2 ≤ ……≤ ynThe per capita macro tax base (yi) is approxi-mated by the comparable per capita GSDP in determining Finance Commission transfers. We can write the state shares and per capita shares under the well known population and distance criterion as given below:Population criterion: share of ith state = qi (say)= Ni/∑Ni and per capita share as q*i= 1/∑Ni Distance criterion: share of ith state = ai (say) = Ni(yn-yi) /∑Ni.(yn-yi) and per capita

SPECIAL ARTICLEjune 7, 2008 EPW Economic & Political Weekly58share as a*i = (yn-yi) /∑Ni.(yn-yi). The state shares and per capita shares can be written in more general form as follows.Let the per capita share of a state in an allo-cation mechanism be a function of per capita income (tax base) and a set of other variables. The set of variables used in a criterion is re-ferred to in a general way by f(.). Let this be normalised by a function written asθ. Thus, the per capita share of the ith state iss*i = θ fi(.) (A1)Correspondingly, the share for the state as a whole is given bysi = θ fi(.)Ni (A2)The total and per capita transfers in absolute amounts can be written asSi=θ.fi(.)N.T and S*i=θ.fi(.)T (A3)where T is the total amount of transfers. Axiom 1: Normalisation 1If two states have the same criterion values, their shares should be proportional to their populations.Theentitlement of a state under any criteri-on should be determined in per capita terms. If two states have the same value of the allocation variable but different sizes of population, the share of the state with the larger population should be larger by the ratio by which its popu-lation is larger compared to the other state. The basic consideration here is that all transfers are aimed at citizens residing in the state, because all services are meant for the citizens. Popula-tion is the appropriate scaling factor in this con-text. This axiom means that, for two states i and j,si /sj = Ni/Nj ifθ fi(.) = θ fj(.) (A4)Axiom 2: Normalisation 2 The sum of the shares of all states should add to 1.Under any criterion, the share of all states should add to 1. This axiom ensures that the entire sum to be transferred to the states as a whole would be precisely exhausted among the states. If the total amount of transfer is T, we require that the sum of transfers received by each state should be equal to the total amount T, i e, the transfer received by each state is equal to (si.T)Therefore,s1T+s2T +……+snT = T (A5)∑siT = T or ∑si= 1Thus,∑θ .fi(.)Ni =1 orθ = 1/∑fi(i)Ni (A6)For a given set of Ni, and values of variables entering f(.),θ could be taken as given. This axiom also ensures that all allocation criteria satisfying it would be indifferent with respect to scalar changes in population. If population of all states increase by a factor k, we have the new set of shares, given as indicated below:fi(.)(kNi)/∑fi(.)kNi = fi(.)Ni/∑fi(.)Ni = si, so long as population is not a variable entering fi(.)Axiom 3: Horizontal EquityBetween any pair of states, the state with lower per capita fiscal capacity should have higher per capita share, and per capita shares should be equal for states with equal per capita fiscal capacity.The allocation mechanism should be consist-ent with horizontal equity. Horizontal equity requires that the allocation mechanism should treat equally two states if their criterion values are the same. It should treat them differently if their criterion values are different. This implies that if two states have the same per capita fiscal capacity, they would receive the same per capi-ta share, and if the two states have different per capita fiscal capacities, states with the lower fis-cal capacity would get a higher share. A criteri-on that satisfies this characteristic may be referred to as a progressive transfers mecha-nism, which ensures that a poorer state receives a higher per capita transfer according to the specified criterion. The poorer the state, the lower is its fiscal capacity to raise own resourc-es if these are assessed at a common tax effort. This condition requires that for a pair of states, bi and bi+1, where they have been arranged in an ascending order of per capita income if, bi<bi+1, s*i>s*i+1and if bi =bi+1, s*i = s*i+1. Considering the case where no two per capita incomes are equal, we require, for progressivity, θ fi(.) > θ fi+1(.)or fi(.)/fi+1(.) > 1 or fi(.) – fi+1(.)>0 (A7)Since fi(.) among other variables is also a function of per capita income, we may writes*i>s*i+1, if∂s*i/∂yi = θ∂f(.)/∂yi < 0, for a given value of θ. Axiom 4: Comprehensiveness The finance commissions have followed the practice that under no criterion, any state is given a zero share. This implies that informa-tion on all states is always considered together. The shares of any subsets of states are not to be determined independently ignoring relevant information pertaining to the remaining states. Suppose that some state (j) receives a zero share in the allocation mechanism. In this case, s*j = θ fj(.) = 0and sj = θ fj(.) Nj =0 or fj(.) Nj =0 sinceθ≠ 0The normalisation axiom indicates that 1/θ = fi(.) Ni +…..+fn(.) Nn (A8)If for some j, fj(.)Nj = 0, it will not enter in the allocation formula. As such no information on the jth state would enter into the allocation for-mula. Thus, to ensure that information on all states are used in the allocation exercise, we require, s* i > 0 for all iHowever if f(.) itself contains the relevant information of the jth state, in the criterion val-ues of all the states, the share of the jth state can still be set at zero.Axiom 5: NeutralityThe allocation criterion should be neutral with respect to the organisation of states. There should not be an incentive to bifurcate states with a view to benefiting from the allocation mechanism. As shown below, all non-linear cri-teria implicitly give an incentive either for split-ting a state into smaller states or for regrouping a state into larger states. All convex and pro-gressive allocation criterion provide an incen-tive for a poorer region in a state to break off and form a “new” state. If it does so, with its lower per capita income, it would ensure a higher per capita share in the transfer mecha-nism. Such fissiparous tendencies can however be neutralised by providing a mechanism of allocation of resources, which is neutral to the organisation of the states. However, even with such an intra-state mechanism, some of the devolution criteria may not be neutral to the organisation of the states under certain condi-tions. Consequently, they may encourage either disintegration of states into smaller units or their integration into larger units. It is a desira-ble property for a devolution criteria to be neu-tral to the organisation of states. The conditions required for this purpose may be stated as below.If, for any state i, if there are two regions 1 and 2, with per capita income y1i and y2i,and populationN1i and N22, such that Ni = N1i + N2iyiNi = y1i N1i +y2iN2iNeutrality of an allocation formula would require that the sum of transfer received by the two regions as separate states should be equal to the transfer received by the undivided state. We have s1i = θ f1 i (.)N1 is2 i = θ f2 i (.) N2 i

SPECIAL ARTICLEEconomic & Political Weekly EPW june 7, 200859The total number of states now being (n+1), neutrality thus requires; s1 i +s2 i = sior θ [f1 i (.)N1 i +f2 i (.)N2 i] = θ.fi(.)Ni (A9)An alternative way, in which this axiom could be stated is that under the allocation mechanism, no two states should either gain or lose if they joined up to form an integrated state. The neutrality axiom ensures that by itself, the devolution criteria do not give any incentives for states to bifurcate themselves or for two states to join together. The devolution criterion should be neutral to the existing organisation of states as a datum.Considering two important examples of specific criteria in use, namely, the population criterion and distance criterion, we can specify the values ofθ and fi(.) for each criterion as given below. For the population criterion:si = qi; f(.) =1; θ = 1/∑Ni; andFor the distance criterion si =ai; f(.) = yn-yi; θ =1/∑(yn-yi)NiUsing these axioms for analysing the devolu-tion criteria used by recent commissions it can be ascertained that all criteria used by the TFC, namely, population, distance, area, tax effort, and fiscal discipline meet the two normalisa-tion axioms. These also meet the comprehen-siveness axiom in the sense that information about all states is used to determine the share of any one state under all the criteria. The dis-tance criterion meets the horizontal equity axi-om. The population criterion and the pure form (where highest per capita income gets a zero share) of the distance criterion also meet the neutrality axiom under certain conditions. It may be noted that tax shares of different states under individual criterion are not pub-lished in some of the other transfer exercises such as those under the Gadgil formula. While in some components, the axioms may be satis-fied, in other cases, these may not be satisfied. It will be useful if the Planning Commission makes public the state-wise indices under different components of the criteria used under the Gadgil formula for a more informed debate on the subject. Using the notations given above, we can propose a set of axioms and their related properties for considering a desirable criteria-based system of tax revenuesharing.Appendix Table 1: Comparison of Equalisation Transfers AverageTaxRatio 0.0654 VerticalComponent: 746.67States Population Per Capita Equalisation Transfers Total Per Actual Per (2001) GSDP (Average Based on Data Capita Capita 1999-2000 to Centred in 2000-01 Benchmark Recommended 2001-02) Per Capita Per Capita Transfers Transfers (Crore) (Rs) Gap (Rs) Transfer (Rs) (Rs) (Rs) 1 2 3 4 5 6 7Bihar 8.300 6,539 20,455 1,337.8 2,084 1,821.5Uttar Pradesh 16.620 10,798 16,196 1059.2 1,806 1,605.0Orissa 3.68011,23415,7601030.71,7772,006.0Jharkhand 2.695 11,71715,277999.11,746 1,754.6Assam 2.66612,28814,706961.81,7081,824.2Madhya Pradesh 6.035 13,340 13,654 893.0 1,640 1,534.0Chhattisgarh 2.08313,71013,284868.81,6151,752.9Rajasthan 5.651 15,05911,935780.5 1,527 1,381.5Meghalaya 0.23216,03510,959716.71,4633,765.5Arunachal Pradesh 0.110 16,579 10,415 681.1 1,428 6,419.1Uttaranchal 0.849 16,9989,996 653.7 1,400 2,871.7Manipur 0.217 17,2649,730636.3 1,383 6,339.5West Bengal 8.018 17,377 9,617 629.0 1,376 1,268.2Jammu and Kashmir 1.014 18,132 8,862 579.6 1,326 4,217.5Andhra Pradesh 7.621 18,869 8,125 531.4 1,278 1,320.4Tripura 0.320 18,974 8,020 524.5 1,271 5,260.7Nagaland 0.19920,4696,525426.71,1737,489.5Karnataka 5.285 20,703 6,291 411.4 1,158 1,188.0Sikkim 0.054 20,9296,065 396.7 1,143 6,759.6Mizoram 0.089 21,2455,749 376.0 1,123 10,488.1Tamil Nadu 6.241 22,587 4,407 288.2 1,035 1,174.9Gujarat 5.067 22,7084,286 280.3 1,027 1,010.1Kerala 3.18422,8244,170272.710191,230.7Himachal Pradesh 0.608 24,762 2,232 146.0 893 4,754.1Haryana 2.11426,25673848.3795760.2Maharashtra 9.688 26,994 0 0.0 747 746.7Punjab 2.436 28,0300 0.0 747 1,057.3Goa 0.135 56,5990 0.0 747 2,557.6Vertical component is equal to the per capitarecommended transfer to Maharashtra. This is the minimum per capita transfer among all states. Equalisation transfers are calculated by applying the average all-state tax-rate to the difference between the three state average per capita GSDP(Maharashtra, Punjab and Goa) and the per capita GSDP of any given state.Source (basic data): Report of the Twelfth Finance Commission, 2004.Appendix Table 2: Decomposition of Recommended Per Capita Transfers(Rs)States Actual Vertical Equalisation Residual (for Shortfall in Recom-ComponentComponentSpecialNeedsEqualisation mendedandCost TransferDisabilities) 1 2 3 4 5 6Bihar 1,821.5 746.7 1,074.8 0.0 263.0Uttar Pradesh 1,605.0 746.7 858.3 0.0 200.9Orissa 2,006.0746.71,030.7228.70.0Jharkhand 1,754.6 746.7 999.18.8 0.0Assam 1,824.2 746.7 961.8 115.8 0.0Madhya Pradesh 1,534.0 746.7 787.3 0.0 105.6Chhattisgarh 1,752.9 746.7 868.8 137.5 0.0Rajasthan 1,381.5746.7634.90.0145.7Meghalaya 3,765.5746.7716.72,302.10.0Arunachal Pradesh 6,419.1 746.7 681.1 4,991.2 0.0Uttaranchal 2,871.7 746.7 653.7 1,471.2 0.0Manipur 6,339.5746.7636.34,956.50.0West Bengal 1,268.2 746.7 521.6 0.0 107.4Jammu and Kashmir 4,217.5 746.7 579.6 2,891.3 0.0Andhra Pradesh 1,320.4 746.7 531.4 42.4 0.0Tripura 5,260.7 746.7 524.5 3,989.50.0Nagaland 7,489.5746.7426.76,316.10.0Karnataka 1,188.0 746.7 411.4 29.9 0.0Sikkim 6,759.6746.7396.75,616.30.0Mizoram 10,488.1746.7 376.0 9,365.4 0.0Tamil Nadu 1,174.9 746.7 288.2 140.0 0.0Gujarat 1,010.1 746.7 263.4 0.0 16.9Kerala 1,230.7 746.7 272.7 211.3 0.0Himachal Pradesh 4,754.1 746.7 146.0 3,861.5 0.0Haryana 760.2 746.7 13.50.0 34.7Maharashtra 746.7 746.7 0.0 0.0 0.0Punjab 1,057.3746.70.0310.70.0Goa 2,557.6746.70.01,810.90.0Equalisation component is calculated by comparing actual recommended transfer net of vertical component with benchmark equalisation transfer (col 5 of Appendix Table 1). If the actual transfer net of vertical component is more than benchmark equalisation transfer, the benchmark equalisation transfer is entered in column 4. The excess becomes the residual given in column 5. If the actual transfer net of vertical transfers falls short of the benchmark equalisation transfer, the actual transfer is written in column 4 and the shortfall is given in column 6. Source (basic data): TFC report and estimates.

SPECIAL ARTICLEjune 7, 2008 EPW Economic & Political Weekly60 GrantsDevolution Vertical Equalising Residual Vertical Equalising Residual HorizontalHorizontal (Residual)1 2 3 4 5 6 7Bihar 114.278.00.0632.5996.80.0Uttar Pradesh 114.2 69.5 0.0 632.5 788.8 0.0Orissa 114.260.1112.3632.5970.6116.4Jharkhand 114.2102.18.8632.5897.00.0Assam 114.2 106.1 115.8 632.5 855.7 0.0Madhya Pradesh 114.2 56.2 0.0 632.5 731.1 0.0Chhattisgarh 114.250.726.0632.5818.1111.5Rajasthan 114.250.20.0632.5584.70.0Meghalaya 114.241.81,647.6632.5674.9654.5Arunachal Pradesh 114.2 39.7 3,048.7 632.5 641.4 1,942.5Uttaranchal 114.238.11,363.0632.5615.6108.2Manipur 114.2 37.14,139.6 632.5 599.2 816.9West Bengal 114.2 74.7 0.0 632.5 446.8 0.0Jammu and Kashmir 114.2 33.8 2,501.7 632.5 545.8 389.6Andhra Pradesh 114.2 31.0 -8.3 632.5 500.4 50.7Tripura 114.2 30.63,475.5632.5 493.9 514.1Nagaland 114.224.95,729.9632.5401.9586.2Karnataka 114.224.015.3632.5387.414.6Sikkim 114.2 23.11,475.7 632.5 373.5 4,140.6Mizoram 114.2 21.9 7,053.8 632.5 354.1 2311.6Tamil Nadu 114.2 16.8 1.5 632.5 271.4 138.5Gujarat 114.2 32.20.0 632.5 231.2 0.0Kerala 114.215.974.3632.5256.8137.0Appendix Table 3: Decomposition of Per Capita Grants and Devolution into Components(Rs Crore) GrantsDevolution Vertical Equalising Residual Vertical Equalising Residual HorizontalHorizontal (Residual)1 2 3 4 5 6 7Himachal Pradesh 114.2 8.5 3,578.3 632.5 137.5 283.2Haryana 114.222.60.0632.5-9.10.0Maharashtra 114.20.00.0632.50.00.0Punjab 114.20.0289.2632.50.021.4Goa 114.20.086.8632.50.01,724.1Average (population weighted) 114.2 46.9 120.8 632.5 533.4 45.7The per capita vertical transfer (column 3 of Appendix Table 2) is split between per capita vertical grant (equal to the lowest per capita grant among all states (for Maharashtra) and per capita vertical devolution (equal to the lowest per capita devolution, also for Maharashtra). All states get these amounts. The equalising components of grants and devolution are calculated in two steps. As a first step, per capita equalising grants for all states are calculated by applying the ratio of Bihar’s per capita horizontal grant (per capita total grant minus vertical component of grant) to benchmark equalising transfer (column 5 of Appendix Table 1). Per capita devolution is also decomposed into three components: vertical, equalising horizontal, and residual. The equalising component of devolution is taken by comparing two series. The first series is per capita total devolution minus vertical component of devolution. The other is the excess of the equalising total per capita transfer (column 4 of Appendix Table 2) over equalising grant referred to above. The equalising component of devolution consists of the lower number of the two series. This gives series 6 of Appendix Table 3. The residual of the devolution column is calculated as the total per capita devolution minus vertical component of devolution (column 5 of Appendix Table 3) and equalising component of devolution (column 6) of Appendix Table 3. Since the residual of grants and devolution should add to the overall residual as shown by column 5 of Appendix Table 2,the residual series for grants is derived as the difference between the overall residual and the residual of the per capita devolution series (column 5 of Appendix Table 2 minus column 7 of Appendix Table 3.This gives the residual grant series (column 4 of Appendix Table A3). Finally adjusted equalising grant series (column 2 of Appendix Table 3) is derived as per capita total grant minus per capita vertical grant minus per capita residual grant. These procedures satisfy the conditions of equation 7 in the text. There are two negative numbers in columns 4 and 7 of Appendix Table 3. These numbers may be adjusted by following a suitable rule in calculating the residuals, but since the magnitudes involved are small, further adjustment has not been done. Source (basic data): TFC Report, 2004.SAMEEKSHA TRUST BOOKSInclusive GrowthK N Raj on Economic DevelopmentEssays from The Economic Weekly and Economic & Political WeeklyEdited by ASHOKA MODYThe essays in the book reflect Professor K N Raj’s abiding interest in economic growth as a fundamental mechanism for lifting the poor and disadvantaged out of poverty. He has also been concerned that the political bargaining process may end up undermining growth and not provide support to those who were excluded from access to economic opportunities. These essays, many of them classics and all published in Economic Weekly and Economic & Political Weekly, are drawn together in this volume both for their commentary on the last half century of economic development and for their contemporary relevance for understanding the political economy of development in India and elsewhere.Pp viii + 338 ISBN 81-250-3045-X 2006 Rs 350Available fromOrient Longman LtdMumbai Chennai New Delhi Kolkata Bangalore Bhubaneshwar Ernakulam Guwahati Jaipur LucknowPatna Chandigarh Hyderabad Contact: info@orientlongman.com