Design of a Pro-Poor Optimal Value Added Tax (VAT) System in a Federal Economy with Inter-Governmental Fiscal Transfer

Transcription

1 Poverty Monitoring, Measurement and Analysis (PMMA) Network Design of a Pro-Poor Optimal Value Added Tax (VAT) System in a Federal Economy wit Inter-Governmental Fiscal Transfer Aitava Raycauduri India A paper presented during te 4t PEP Researc Network General Meeting, June 13-17, 2005, Colombo, Sri Lanka.

2 POVERTY AND ECONOMIC POLICY (PEP) RESEARCH NETWORK Under Poverty Monitoring, Measurement & Analysis (PMMA) Network Submission of Revised Researc Proposal Proect Title Design of a Pro-Poor Optimal Value Added Tax (VAT) System in a Federal Economy wit Inter-Governmental Fiscal Transfer REVISED Proposal Date April 15, 2005 Autors - Team Leader Dr. Aitava Raycauduri, Department of Economics, Jadavpur University, Kolkata , INDIA Team Members 1) Sudip Kumar Sina, Assistant Commissioner, Commercial Taxes, Bureau of Investigation, Department of Finance (Revenue), Ministry of Finance, Government of West Bengal, 10, Madan Street, Kolkata , INDIA, 2) Ms. Poulami Roy, Junior Researc Fellow, Department of Economics, Jadavpur University, Kolkata , INDIA 1. Sort Abstract - In te sort run, economic policy cange in te economy affects te income levels of different income/social groups mainly troug prices of commodities tey consume or income transfers. However, in many developing countries, income transfer scemes for te poor are usually constrained by te lack of information tat government agencies ave on te distribution of well being. Tis makes indirect taxation a predominant tool for social welfare policy in tese countries. Te problem of ow to improve te design of indirect taxes and subsidies so as to meet bot poverty and efficiency criteria ten becomes an important element of poverty alleviation strategy. Te proposed study intends to develop a revenue maximization model to calculate socially optimal pro-poor Value Added Tax (VAT) rates in a federal economy, after endogenising inter-governmental fiscal transfer variables in an integrated teoretical framework. Te pro-poor dimension will be accommodated by setting alternative very low rates of taxes

3 for certain mass consumption commodities. It will, inter alia, contain a poverty and social impact analysis to study te impact of suc a socially optimal tax reform on different income or social groups, by comparing different existing tax rates wic do not address te poverty dimensions specifically (160 words) 2. Main Researc Queries and Core Researc Obectives Among different types of taxes, evidence suggests tat te case for direct tax reforms, suc as income tax reforms, is more on efficiency and equity considerations and is not in itself an instrument of a poverty reduction strategy as te poor are not liable for personal income tax (Gemmel and Morrissey, 2002). For all LDCs for te period , te study sowed tat te proportion of direct taxes to total revenue was 39.35% wile te proportion of indirect taxes in total revenue was 47.29%. For low income countries, proportion of direct taxes is furter lower at 30.26% of total revenue wile tat of indirect taxes was an impressive 57.69%. Income transfers and Social Security instruments are also not effectively used in developing and low income countries, tus limiting effectiveness of direct taxes as a welfare improving tool troug re-distribution. [Gemmel & Morrissey, 2002]. Evidence from te same study also suggests tat te principal taxes paid by te poor are sales taxes and tariffs on te goods tey consume. Te tax system could be made pro-poor if suc items are taxed at very low rates. Suc policy initiatives will not be feasible as indirect taxes are te main source of tax revenue for many, particularly sub-national governments, in developing countries. Given tis constraint, te basic question tat will be addressed is ow far welfare improving commodity taxes reforms, based on optimal commodity taxation, can play an important role in overall pro-poor tax reforms design. It is generally difficult to acieve significant redistribution troug indirect taxes, but efficiency in taxation can be significantly enanced by minimising dead weigt loss or te excess burden wit some provision for exemptions and lower rate on goods and services wic are predominantly consumed by te poor. Te study will try to assess te impact of a pro-poor indirect tax reform on different ouseold groups in te urban and te rural sectors. It is widely

4 eld tat sales tax-excise duty-import duty based indirect tax structure in developing countries is inefficient due to a large number of factors, cief among wic are: complex tax laws; a multiplicity and large dispersion of rates; overlapping, fragmented, partial and, in te case of excises, difficult-toadminister bases; te taxation of inputs; te erection of barriers to domestic trade; and administrative weakness [Burgess, Howes & Stern, 1993]. Introduction of Value Added Tax (VAT) as been recommended due to its basic caracteristics: enanced efficiency and buoyancy, removal of cascading effects of taxation, simple rate structure wit one or two tax rates and administrative advantages. For tese reasons, VAT as been te main domestic replacements tax for tariffs and a wide array of excises in many reforms, and were used in te maority of LDCs [Gemmel & Morrissey, 2002]. India as initiated maor reforms in indirect taxes wit te proposed introduction of Value Added Tax (VAT) wit effect from April 1,2005. However, te poverty alleviation is not viewed as an important goal of tis cangeover to a VAT regime, wic is someting tis proect wises to focus and analyse. Specific sense in wic VAT may be made Pro-Poor Te tax system could be made pro-poor if items mostly consumed by te poor are taxed at very low rates. Te design of te proposed optimal VAT, will require increased efficiency by decreasing excess burden wit some provisions for exemptions and lower rate on goods and services wic are predominantly consumed by te poor. Studies ave suggested tat altoug in general VAT is regressive if applied at a general rate to all commodities and services, VAT can be made progressive wit suitable low rates or exemptions of goods and services wic are mostly consumed by te poor. For example, Studies by Hossain [2003] indicate tat a revenue-neutral uniform VAT is regressive in its impact on different ouseolds- it benefits te ricer groups and adversely affects te poorer ouseolds. But te study found tat an alternative policy package consisting of a basic rate of VAT wit exemption and excise taxes for certain groups cosen on te basis of teir distributional caracteristics, is more progressive. To identify te distributional caracteristics of te commodities suitable for exemptions under VAT, it

5 proposed to look at te expenditure sare of tese commodities in te consumption basket of te urban and rural ouseold groups. Studies by Gemmel & Morissey (2002) also suggests tat commodity taxes can be made pro-poor by ensuring zero rates on goods tat are consumed predominantly by te poor rater tan by ric, and on activities tat are engaged in predominantly by te poor. Te study furter suggests tat since te poor predominate in te informal sector, one would not tax informal sector activities. Studies by Munoz & Co [2003] reports te results of a Poverty and Social Impact Analysis (PSIA) of te tax incidence of te new VAT in Etiopia and compares it to te incidence of te sales tax tat it replaced. It used te 1999/2000 Houseold Income, Consumption and Expenditure Survey to evaluate te tax incidence of te VAT. Te analysis sows tat te VAT is progressive wen total expenditure at te national level is analyzed. However, because it as fewer exemptions and only one rate, te VAT is less progressive tan te sales tax it replaced. In Etiopian case, most of te exempt goods and services are disproportionately consumed by te relatively well to-do, so te exemptions cannot be ustified on equity grounds. Te study suggests tat even toug exemptions complicate administration, erode te tax base and distort inputoutput decisions, exemption of some items to improve te distributional impact of te tax is a potentially reasonable trade off. In India, te exemption list of recently introduced VAT contains items mostly natural and unprocessed products, wic are in unorganized sector, and items, wic ave social implications. It also contains, unlike in te previous Sales Tax regime, foodgrains suc as paddy, rice, weat and pulses wic are disproportionately consumed by te poor. In India, te exemption list under VAT is mostly same across te states, wit a few variations due to different social and economic set-up in different States. Te motivation of our study will be, terefore, to set very low rates (starting from zero) for a few items mostly consumed by te poor, and to calculate te optimal VAT rates of remaining commodities, for a few States under federal fiscal structure wit different demand conditions. Since VAT is already implemented by te maority of states in India from April 1, 2005, it is

6 now timely to compare te actual tax rates and exemptions wit te optimal rates derived from te exercise done ere. Te present VAT structure in India covers only commodities and does not cover services. However, a compreensive VAT sould cover goods as well services. In India it as been decided tat services will be covered under VAT in a pased manner, as it requires constitutional amendments. If services are subected to VAT, exemption of certain services wic are predominantly consumed by te poor can significantly enance te progressivity of VAT. Oter Issues Addressed Te study also intends to address te question ow far te traditional welfare-maximising models of optimal commodity taxes are practical in te current economic scenario and te study proposes a revenue maximisation model to calculate optimal VAT rates. Altoug studies on Banglades (Hossain, 2003) suggests tat a revenue-neutral uniform VAT is regressive and a basic VAT wit exemptions and excise taxes for certain commodity groups cosen on te basis of teir distributional caracteristics is superior, no attempts appears to ave been made to calculate optimal VAT rates in a federal developing country. In a federal country, te options for VAT are : a federal VAT, a State VAT and a dual [oint federal-state] VAT. In many federal countries wit separate Constitutional provisions for revenue and expenditure for te federal and State governments, e.g. in India, a federal VAT is ruled out because it is politically impossible to impose suc taxes on all States, wic will destroy teir financial and political autonomy and make tem more dependent on te federal government. Te system of State VATs is also ruled out because te federal government will never agree to deny its sare of indirect taxes, wic is a significant part of federal revenue. In suc federal countries, te only form of a feasible system of VAT is some kind of dual VAT system operated by bot te federal and te State governments. In India te proposed dual VAT will convert te existing federal excise taxes and MODVATs into Central VAT and te States existing sales taxes into State VAT. Te current study will concentrate on a dual VAT as proposed in India. Answer to te basic question will depend crucially on te relationsip between marginal cost of commodity taxation and welfare weigts of taxpayers

7 (Bibi and Duclos, 2003), wic in turn depends on te planners' perception on measurement of inequality and poverty, usually referred to as te inequality aversion parameter. Tis is due to te fact tat te marginal social cost of commodity taxes, [denoted in te standard optimal indirect tax literature by { δw/δt i } /{ δr/δt i }), depends on te social welfare function(w) wic is again function of indirect utlities V. Measurement of { δw/δt i } furter incorporates β [={ δw/δv }] wic is te social marginal utility of income of ouseold. In te empirical work for calculation of optimal commodity taxes, te net social marginal utility of income is allowed to decline monotonically wit an increase in ouseold affluence at a rate determined by te inequality aversion parameter. Te proposed study will also address te problems of measurement of poverty and inequality based on te marginal cost of commodity taxation and welfare weigts of different ouseold groups and teir role in welfare improving marginal tax reforms. Te role of tax administration reform in design of suc a pro-poor tax policy reform needs to be studied in detail as a simple tax structure wit few and lower rates contributes to bot collection and efficiency, and is pro-poor (Gemmel & Morrisey, 2002). Reforms of indirect tax structure is part of te Tax Policy Reform (TPR) wic must be supplemented by effective Tax Administration Reform(TAR) as TPR and TAR are te two components of any tax reform agenda. It is widely eld tat tax administration as a crucial role in determining te real (effective) tax system, as opposed to te statutory tax system and te real tax system people and business face reflects not ust tax law but also ow tat law is actually implemented in practice. How a tax system is administered affects its yield, its incidence, and its efficiency (Bird, 2004]. As VAT will be extended to cover services as well, many of wic may be consumed by poor people, te question of accommodating services in te model also needs to be addressed. Finally, te proposed study intends to address te problems associated wit endogenisation of inter-governmental fiscal transfers variables to arrive at truly optimal Value Added Tax rates. Resource transfers from te federal government to te States in a federal economy, form a significant part of te

8 resources of te States as per Constitutional provisions. Tere are several economic rationales for inter-governmental transfers (1) tey address vertical fiscal imbalance by providing resources needed to balance te budget at te sub-national level; (2) tey address orizontal fiscal imbalances by providing resources to te regions wit ig negative net fiscal benefits (measured by te gap between fiscal capacity and fiscal need); and (3) tey address interurisdictional spill-over effects in certain public services suc as pollutioncontrol (water or air), inter-regional igways etc. were a local government tends to underinvest. Intergovernmental fiscal transfers, based on fiscal needs and fiscal capacities of subnational governments, use backwardness and/or poverty index as an important indicator of social welfare, to measure fiscal needs of a sub-national government (Jun Ma, 1997). Endogenisation of intergovernmental fiscal transfer variables is terefore crucial to any pro-poor tax policy reforms. Since calculation of optimal indirect tax rates will depend on te resource functions of te federal as well as te State governments, we need to include te intergovernmental resource transfer variables, suc as population, backwardness and/or poverty index in teir resource functions. To arrive at truly optimal tax rates we need to optimize tese variables and include teir optimal values in calculating optimal indirect tax rates. Tis automatically provides endogenously determined optimal intergovernmental fiscal transfer rates. Te core researc obective is terefore to develop a revenue maximization model to calculate socially optimal pro-poor Value Added Tax (VAT) rates, after endogenising inter-governmental fiscal transfer variables, and to assess its impact on different income or social groups. 3. Scientific Contribution As any tax, direct or indirect, affects individual utility and ence social welfare of an economy adversely by decreasing utility, te traditional view of te Economists is tat te commodity tax rates sould be decided in suc a way tat te decrement in utility is minimum wile fulfilling suc revenue requirements. Economists tus tried to calculate suc Optimal commodity taxes by maximizing social welfare subect to te revenue requirements of te

9 Government. Wile te basic model, proposed by Ramsey (1927), developed as a full fledged many person model by Diamond & Mirrlees (1971) and illustrated and empirical work done in India s context by Amad & Stern (1984) and R.Ray (1990), incorporates calculation of Optimal Commodity Taxes in a centralized economy, recent studies by R.Ray & M.N.Murty (1990) ave sown effect of federal (decentralized) set-up on calculation of Optimal Commodity Taxes by incorporating resource transfer (altoug partially) in te revenue constraints of te Centre & te States and calculating optimal excise tax and optimal sales taxes separately. Recent studies on Optimal Commodity Taxation ave also investigated effects of rationing, rural-urban differences in consumption pattern and demograpic variables on optimal commodity taxes. Studies by R.Ja (1990) & oters ave sown tat rationing of one or more commodities will ave serious implications for calculation of optimal commodity taxes. Studies by R.Ray (1990) ave worked out empirical implications of difference in rural-urban consumption patterns and ave suggested two sets of optimal commodity tax rates one for te rural sector and one for te urban sector. Studies by R.Ray (1993) ave also tried to work out effect of demograpic variables e.g. te number of cildren in a family and presence of a linear optimal income tax on te calculation of optimal commodity taxes. Te vast literature on optimal commodity taxation in general and calculation of optimal commodity tax rates for India in particular, owever, appear to ave failed to take into account te fact tat te aim of te Government is to maximize teir revenue subect to certain welfare considerations and not welfare maximization subect to teir revenue constraints, noting te fact tat te former is not te dual of te latter. Tis reality necessitates reframing of te traditional definition of optimality. Furter, various forms of resource transfers are to be included in te model and if possible, endogenised, to arrive at te actual revenue positions of te central and State Governments. Te alternative approac to indirect tax reform is to reframe te definition of optimality and calculate optimal commodity taxes as te tax rates wic maximize tax revenue subect to some welfare constraint. Studies by

10 Raycauduri & Sina [2004] proposed a model for calculation of optimal commodity tax rates, based on te revenue maximisation definition of optimality, in a decentralized federal set up like India were te Centre and te States levy commodity taxes independently. Te study sows tat uniform VAT rates across te country, in a federal economy, are not optimal altoug it is administratively convenient. However, poverty alleviating role of VAT was not te main concern of te aforesaid paper. As already mentioned, studies by Hossain [2003] also tried to investigate te poverty/income distributional implications of introducing a VAT. Likewise, study by Munoz & Co [2003] reports te results of a Poverty and Social Impact Analysis (PSIA) of te tax incidence of te new VAT in Etiopia and compares it to te incidence of te sales tax tat it replaced. Most of te exempt goods and services are disproportionately consumed by te relatively well to-do, so te exemptions cannot be ustified on equity grounds. Te progressivity of te VAT comes mainly from te ig ratio of in-kind transactions for poorer ouseolds, but tis sare is expected to decrease as te economy moves into a stable growt track and becomes more market-based. Terefore, te VAT in Etiopia is likely to become less progressive in te future, even at te national level. Moreover, te VAT is regressive or at most neutral in urban areas. Considering te growing urbanization in Etiopia and te fact tat most urban immigrants consist of poor people, tis could ave significant consequences for te poor. Te replacement of te sales tax wit te VAT as ad an adverse impact on te poorest 40 percent of te population. However, te impact is very small (about 1 percent of teir consumption). Tus, tis reform as not ad a maor adverse effect on te poor, especially in ligt of te iger expenditures on poverty-reducing activities tat can be financed out of tese revenues. Our estimates indicate tat if te additional revenues from te VAT were allocated for iger spending on primary education and ealt, te poorest 40 percent of te population would be net beneficiaries. Te study by Munoz & Co[2003] suffers from te same limitations as tat of Hossain[2003]. Te model assumes a single rate of VAT on all goods and services wit some exemptions. Te model considers a federal VAT as applicable to Etiopia and does not consider te case of a two-tier federal-state

11 VAT wit significant intergovernmental fiscal transfers. Te study concentrates on investigating progressivity of VAT versus sales taxes and does not calculate optimal VAT rates for different income groups and different commodity groups. In te ligt of above existing studies on te subect, te knowledge gap appears to be te absence of a robust teoretical model to calculate optimal, pro-poor (according to income differences), VAT rates in a federal developing country like India, wit two-tier dual VAT structure and intergovernmental fiscal transfers. It will also accommodate as far as feasible te desirability of a rural urban differential in taxes. Eac of te above aspects namely, a) optimality b) pro-poor caracter c) two-tier of te taxation and d) combination of VAT wit intergovernmental fiscal transfer, need to be studied in an integrated framework. Te model must be used to estimate suc optimal VAT rates and to compare tese rates wit existing commodity tax rates to investigate te distributional implications of introduction of suc a VAT, from te Poverty and Social Impact Analysis (PSIA) point of view. 4. Policy Relevance Many developing countries are introducing long-term fiscal reform policies aimed at fiscal prudence and enancing fiscal responsibility of different levels of te government. However, early experience suggests tat tere is an analysis gap tat is policies are being formulated and implemented witout adequate poverty and social impact analysis. Te proposed researc work is aimed at providing a socially desirable direction for poverty-alleviating tax reform. It also provides a framework for poverty and social impact analysis by comparing te pre and post reform effects of suc a tax reform policy on different income or social groups. Te study intends to compare te existing rates of sales tax and excise duties wit te proposed optimal federal and State VAT rates and will try to estimate te welfare effects of suc canges. Also, it will igligt te differences in tax rates wen income differences and urban-rural divide are accommodated. Te novelty of tis proposed researc work is tat of relating indirect tax reform to poverty alleviation efforts. Te maor aim of tax reform is seen to be aimed at increasing tax revenues by lessening te burden on te consumers.

12 Altoug tis migt be taken to be a positive step towards consumer and social welfare, tis as to be directly and explicitly related to te issue of simulating te possible impact of suc reform on te lower income groups. Tus, altoug VAT qualifies as a desirable tax on a number of counts, its impact on alleviating poverty troug targeted tax reduction on poor man s commodities is a rater under-researced area. Hence, it is felt tat a direct investigation on possible linkages between tax reform leading to VAT and reduction of poverty is urgently called for. Since India is moving towards a VAT regime soon, and also it is te ome of millions of poor people, te findings of te proect is expected to ave significant implications on policy of bot central and state governments in India. 5. Metodology to be used: Te proposed study intends to build up a teoretical model based on revenue maximisation subect to certain welfare considerations. Te proposed study will be based on earlier studies on te subect by te autors [Sina & Raycauduri(1998), Raycauduri & Sina(2004)]. Te proposed revenue maximisation model will ave two sets of constraints. Apart from te constraint of maintaining individual welfare weigts constant in respect of certain level of planners' perception of inequality, te model will also attempt to embed a propoor structure in te optimal tax rates by exempting/minimising te tax rates of certain commodities tat are predominantly consumed by te poor. Te federal governments revenue expression, after intergovernmental fiscal transfers to te States will look like R 0 = ci X i + τ w L ] Q [ ci X τ w L i i i Were, ~ [ + ~ X ~ i = gross value added in sector i (1,2,,n). p i w = price vector in State = gross wage rate of individual in State

13 p i = price vector in State w = gross wage rate of individual in State c i = rate of central VAT in commodity i M 1 = States' sare of Central VAT on te basis of relative population 1 M = States' sare of Central VAT on te basis of backwardness P P B 1 = population of State wit / P = relative population of State. = P, te = measure of backwardness of State. P National Population. 0 < B < 1, B = 1. Q = sare of E = 1 total sare of State of total Central VAT revenue States' of Central VAT T E 2 = sare of State of = total total sgare of states' of income tax income tax τw L τw L = relative sare of income tax of State. s I = rate of State VAT in State in commodity Te social welfare function takes te form 1 1 W = W ( v, v, v,...,v ) H Te State s revenue expression will be close to te expression

14 R = s i ~ X i + E1M 1 i ~ cˆ X i i + τ w L P P + E (1 M 1 * 1 ) i ~ cˆ X i i + τ w L B Were E 1, E 2, M 1, M 2 are te intergovernmental fiscal transfer variables Te teoretical model and empirical work in te proposed study will be based on India s federal fiscal structure as India is poised for maor reforms in indirect taxes wit proposed introduction of VAT wit effect from April 1,2005. Te study will elp to investigate te poverty/income distributional implications of introducing a VAT. In te empirical work, apart from calculation of optimal federal VAT, te motivation of te study is to calculate optimal State VAT rates for at least four States taking two from te lower income group and two from te ric industrial States. Suc a model will terefore be based on interactions of demand for commodities, income of te individuals, commodity tax (VAT) rates and welfare weigts of individuals. Demand functions will be estimated using Linear Expenditure System (LES). In addition, inter-governmental fiscal transfer variables will also be endogenied [refer to te earlier elaboration about endogenisation and tat wy it comes out of te same exercise] to arrive at truly optimal VAT rates. As as been stated in te preceding sections, te endogenisation in te proposed study will take te form of inclusion of intergovernmental fiscal transfer variables in te resource functions of te federal and te State governments. Endogenisation of suc variables will also include calculation of optimal values of suc variables and use of optimal values of suc variables in te calculation of optimal rates of federal and State VATs. One advantage of te proposed teoretical model is tat te endogenisation of intergovernmental resource transfer variables in te proposed study will use te same set of resource functions of te federal and State governments as well as te social welfare functions and will be carried out in te same simulation tecnique wic will calculate optimal values of suc variables and incorporate suc values in

15 calculation of optimal federal and State VAT rates, as as been done in earlier studies by Raycauduri & Sina [2004]. No separate elaborate modelling exercise will be required for te purpose of endogenisation. In contrast to calculation of optimal rates of sales taxes or excises, were data on consumption of commodities are used, for calculation of VAT we need to calculate value addition arising out of suc consumption of commodities. To calculate te value addition we need te input-output tables for te commodity groups. Te study proposes to use te input-output table of te federal government for te concerned States (wit suitable modifications) to calculate te value addition in commodity groups arising out of consumption of suc commodities by different groups. Tis is necessitated by te fact tat states do not ave separate input-output tables for te calculation of VAT. Te model will be based on a simulation exercise allowing interaction of te aforesaid variables, as demand, prices and taxes are affected by one anoter, till we get some convergence. Te proposed revenue maximisation model will first find out te optimal values of resource transfer variables by maximising te revenue functions of federal and State governments subect to te social welfare functions. For empirical work, te model will start wit a set of initial values of all suc variables including tax rates, prices of commodities, consumption, fiscal need indicators suc as backwardness/poverty index and population and fiscal capacity indicators suc as actual tax collection. Optimal values of resource transfer variables are calculated using a simulation tecnique described in Raycauduri & Sina[2004]. Optimal values of suc resource transfer variables along wit initial values of oter variables - suc as tax rates, prices of commodities and value added in consumption will be used to calculate te optimal federal and State VAT rates in te same simulation tecnique. Te simulation tecnique will be carried out till marginal cost of taxation, defined using our revenue maximisation model, of all commodity groups converges. Wit adustments in tax rates at eac step, te simulation tecnique will immediately invoke te in-built Linear Expenditure System (LES) to calculate te fres set of consumption and ence value added estimates for all commodity groups. Te new set of suc calculated variables will again be used in te same simulation tecnique to arrive at fres set of marginal cost of

16 taxation. Te exercise will continue till te marginal costs of taxation converge to give truly optimal VAT rates. To study te income-distributional effects of te proposed tax reform on different income groups, te simulation exercise will be carried on different income groups of te rural and te urban sector, classified on te basis of ouseold expenditure data. Tis ten can be compared wit te rates introduced very recently in te maority of Indian states. An earlier model by Raycauduri and Sina[2004] on calculation of optimal VAT in a federal country like India offers important observations on te possible optimal values of suc value added tax. Te results clearly sow tat uniform rates of VAT across te country will not be optimal, altoug te planners prefer uniform tax rates for administrative convenience, its less distortionary effects on te coice of production process and for minimizing te unealty tax wars among te States. However, te model quoted above concentrated on te efficiency aspects of marginal taxation mainly and did not address te poverty question directly. Te basic motivation of te proposed study is to address te poverty reduction issues of suc optimal taxation policy directly. Te model will be estimated using India s public finance statistics. Te model intends to estimate te Central Value Added Tax rates as well as te State Value Added Tax rates for a selected number of States. Te following section elaborates stepwise te metodology to be used in te proposed proect - Te existing literature on calculation of optimal commodity taxes (OCT) considered only calculation of optimal Sales Tax and Excise Taxes, wile our model proposes to calculate optimal Value Added Tax (VAT), wic is operational in more tan 130 countries. Te study will use input-output tables to calculate value addition in commodity groups arising out of consumption of suc commodities by different groups. Te existing literature on OCT uses a welfare maximizing model subect to a revenue constraint wic minimizes te excess burden of suc indirect taxation. Our model proposes to build a revenue maximization model subect to

17 certain welfare constraints based on individual welfare weigts based on planners perception of inequality. Te proposed model will also ave, unlike previous studies on OCT, an embedded pro-poor structure in te welfare constraint, by allowing very low rates of VAT on certain goods mostly consumed by te poor. Tis will be done across income groups and across regions like Rural and Urban in India. Unlike te existing literature on OCT in federal countries, te proposed model will try to endogenise inter-governmental resource transfer variables troug an integrated model and accommodate teir optimal values in te model to calculate rates of optimal VAT in a federal country like India. Te proposed revenue maximization subect to welfare constraints will be based on a simulation exercise allowing interaction of aforesaid variables including demand, prices, taxes (VAT rates). Te demand across different income groups and different regions will be calculated by Linear Expenditure Metod from Consumer Expenditure surveys periodically carried out by te National Sample Survey Organisation (NSSO) of India. Te simulation exercise will first calculate optimal values of inter-governmental resource transfer variables and ten maximize revenue expressions of te Centre and te States subect to welfare constraints and calculate optimal VAT rates. Tis will generate marginal cost of taxation of eac commodity group. VAT rates will ten be adusted to reduce te gaps of marginal costs of taxation of different groups. As tax rates cange, prices will cange and accordingly demand for suc commodities and consequent VAT revenue will cange. Te simulation exercise will again calculate fres set of optimal VAT rates and marginal costs of taxation for different commodity groups. Te simulation will continue till we get some convergence of marginal costs of taxes of different commodity groups. Te set of VAT rates for wic we get te required convergence will be te optimal VAT rates in our model. Te simulation will be carried out separately for rural and urban ouseolds to study te income-distributional effects of suc a optimal VAT structure, wit suitable low rates (starting from zero) of some commodities. Finally tis will be compared wit te existing aggregate VAT rates on different commodity groups.

18 6. Data requirements and sources Te data sets to be used in te empirical exercise of te model are data on consumer expenditure, tax rates, prices, income of individuals and intergovernmental fiscal transfers. Wile data on consumer demand for different commodities and commodity prices are available from consumer expenditure surveys publised regularly by te National Sample Survey (NSS) and te Central Bank (Reserve Bank of India) publications, data on tax rates, government revenue and intergovernmental fiscal transfer are normally available in official public finance statistics publised in Budget Documents of Government of India and State Governments. Te demand system to be used in te model, a LES version, will be estimated on a nine-commodity disaggregation of consumer expenditure, as followed in standard literature on optimal commodity taxation. Te demand system will be estimated using te popular econometric package SHAZAM. Te data set to be used is te time series on ouseold budget surveys from National Sample Survey Reports (Tables wit notes on Consumer Expenditure). In te absence of retail price data, wolesale price index data will be used, wic is available in Candok and Reserve Bank of India. Initial values of Central and State VATs are available in National Institute of Public Finance and Policy's report (1994). Te national input-output table, publised by te Central government, will be used to calculate value added in eac commodity group. 7. Sort list of key references Amad, E. and N.H.Stern (1984), Te Teory of Reform and Indian Indirect Taxes, Journal of Public Economics, 25, Baneree, Nirmala and Poulami Roy (2004), Gender in Fiscal Policies: Te Case of West Bengal, New Deli: UNIFEM and Sacetana Bibi, Saima and Jean-Yves Duclos (2003), Poverty-Decreasing Indirect Tax Reforms: Evidence from Tunisia,CIRPEE, May. Bird, Ricard. M.[2004], Administrative Dimensions of Tax Reform,Paper presented in a World Bank course on Practical Issues of Tax Policy & Administration in Developing Countries at NIPFP, New Deli, Marc.

19 Burgess, R., Stepen Howes and N. Stern (1993) Te Reform of Indirect Taxes in India, Discussion Paper EF No. 7, STICERD, London Scool of Economics, November. Burgess, R. and N. Stern (1993) A VAT in India: Problems and Options, Discussion Paper EF 4, STICERD, London Scool of Economics, July. Candok, H.L. and te Policy Group India Database, Te Economy, Vol. 1. Cellia Committee (1991) Tax Reform Committee Final Report, Ministry of Finance, Government of India, January Diamond, P.A. & J.A. Mirrlees (1971) Optimal Taxation and Public Production II: Tax Rules, American Economic Review, 61, Gemmel, N and O.Morrissey (2002), Te Poverty Impacts of Revenue System in Developing Countries, A Report prepared for te International Development, U.K. Hossain, Saabuddin M. (2003), Poverty and Social Impact Analysis: A Suggested Framework, IMF Working Paper, WP/03/195, October. Munoz, Sonia and Stanley Sang-Wook Co [2003], Social Impact of a Tax Reform: Te Case of Etiopia, IMF Working Paper, WP/03/232, November. Murty, M.N. and R. Ray (1990), Fiscal Federalism and Optimal Commodity Taxes: A Comparison of Centralised and Decentralised Decisions, Deli Scool of Economics Working Paper n. E/147/90, November. National Institute of Public Finance & Policy (NIPFP) (1994) Reform of Domestic Trade Taxes in India: Issues and Options, April. Puroit, Maes C. (1993a) Adoption of Value Added Tax in India: Problems and Prospects, Economic & Political Weekly, Marc 6. Ramsey, F.P. (1927) A Contribution to te Teory of Taxation, Economic Journal, 37, Ray, R. (1990), Optimal Commodity Taxes and Consumer Beaviour in India: Teory and Empirical Evidence, in Developing Policy and Economic Teory, eds by K.Basu and P. Nayak, Oxford University Press. Ray, R. (1990) Optimal Commodity Taxes and Consumer Beaviour in India: Teory and Empirical Evidence, Development Policy and Economic Teory, ed. By K. Basu and P. Nayak, Oxford University Press.

20 Ray, R. (1993) Optimal Demogrants and Taxes in a Federal Welfare State, Journal of Population Economics, 6, Raycauduri, Aitava and Sudip Kumar Sina (2004), Fiscal reforms Initiatives in India: An Indirect Tax Perspective in Purusottam Battacarya, Tridib Cakraborti and Sibasis Catteree (eds) Anatomy of Fear: Essays on India s Internal Security, New Deli: Lancer s Books. Sarma, Atul and Gupta, Manis [2002] - 'A Decade of Fiscal Reforms in India', Working Paper 02-04, International Studies Programme, Georgia State University, April Sina, Sudip [2002], Models of Optimal Commodity Taxation wit special reference to India, Unpublised P.D tesis, accepted for te award of doctorate degree by Jadavpur University. Sina, Sudip and Aitava Raycauduri (1998), Some Extensions of Optimal Commodity Taxation: Wit Special references to India, Journal of Economic Development, Vol. 23, No. 2, December. 8. Dissemination strategy Dissemination of te researc will be done troug several key mediums- (a) publication of ournal articles, bot academic and semi-academic (b) Seminar presentations (c) Implementation stages. Some of te related papers are already done by two of te autors and seminar presentations on te same were made in several seminars. Also, one of te autors of te proposal used tis in several documents prepared for te government. Te results of tis proect will be presented in two dissemination worksops wit local participants. One may ave a large scale worksop provided additional funding is available specifically for tis purpose. Te most important results coming out of te exercise will be sared wit te state tax autorities in te state of West Bengal, were te investigators are located. Te results may be publised as working papers from te Researc Centre of te Economics Department of Jadavpur University, were one of te investigators is associated. We intend to publicise te most important findings in couple of leading newspapers in Kolkata wit national editions. Finally, one expects couple of academic articles wic may be publised in related academic

21 ournals. Apart from tat, one expects additional dissemination troug PEP website. 9. List of Team Members Team Leader Aitava Raycauduri, Age 47 years, Sex- Male. Currently e is a Professor of Economics at Jadavpur University in Kolkata, India. Guided te doctoral dissertation on Optimal Commodity Taxation done by one team member in te proect, namely, Sudip Kumar Sina. Also, teacing public economics in Under graduate classes in Jadavpur University. Te Team Leader is well-conversant wit te tecniques used in te optimal commodity tax literature as well as te tax literature. His main task will be to lead te team and provide guidance for te overall proect implementation Team Members 1) Sudip Kumar Sina, Age 41 years, Sex- Male. He ad done is P.D on Optimal Commodity Taxation and as publised couple of papers. He works in te sales tax department of te State Government of West Bengal in India and knows te practical side of tax administration and implementation minutely. He will use is teoretical and practical knowledge in te actual formulation and simulation exercises. 2) Poulomi Roy, Age Below 30 years, Sex Female. Se is currently working for doctoral work on te role of Public capital in growt of a developing country like India. Knows te micro foundation of public economic teory very well. Se as also worked on a gender budgeting proect and te report is already publised. Will elp te team in data analysis, relating tem to some of te teories. 10. Description of te researc capacities tat team members and teir institutions are expected to build troug teir participation in tis proect: Te proposed researc ave team members from bot academic and government organizations. In case of te academic part, te researc will build capacities in te following ways- (a) Te youngest of te team member will utilize tis towards er doctoral level work (b) Jadavpur University, wit wic

22 te team leader is associated as introduced semester system aving public economics and development as special papers on wic empasis is laid. Tis proect will certainly elp im to introduce bot teoretical and empirical topics on public policy and poverty alleviation in relevant courses. Tis will elp te students learning some iterto neglected but very important policy issues (c) Te tird member of te team is associated wit tax department of a state government and is actively involved in te transition of te system to te VAT regime. Tus te work will certainly benefit in te actual implementation stages of te public policy related wit te VAT regime. Te following is an elaboration on te specific areas and tecniques tat te team members will learn in te process. Following te adoption by te international community of te poverty reduction strategy paper (PRSP) approac, tere as been a resurgence of interest in te poverty and social impact analysis (PSIA) of different policy reforms facing te countries. Wile many of te reforms in te area of taxation and expenditure policy, are aimed at reducing fiscal deficits, improving macroeconomic stability, and enancing economic efficiency, many of te measures affect different income and social groups differently and may urt or benefit te poor, vulnerable and low-income groups more. Given te poverty reduction goal of te PRSP, policymakers need to know te distributional impact to asses te desirability or feasibility of a policy cange, as well to address te possible adverse poverty/income distributional impact troug alternative policy instruments. From te proect, te team members will gain significant insigts on te issues relating to measurement of poverty indices, measurement of individual welfare weigts and role of planners perception of inequality, tax and expenditure structure of te Central and sub-national (State) Governments and basis of inter-governmental federal resource transfer. Team members will cover te maority of te existing literature and teories on te aforesaid topics. Te team members will also learn te tecniques of empirical model building for constrained optimization. Since te model will be based on a simulation exercise, te team members will be able to upgrade teir computer programming skills significantly.

23 Te proposed study will elp te tird member of te team, wo is a tax administrator and policy maker in te sub-national Government, to understand te PSIA approac to tax policy reforms, measure distributional impacts and suggest necessary canges in tax and expenditure policy reforms. Te team leader wises to incorporate te acquired knowledge in te course curriculum of Public economics and macroeconomics in te university were e teaces tese subects. Tis would be a long-term capacity building exercise for te younger generation. Also, tere is ample scope of supervising dissertation works in te doctoral and M.Pil levels utilizing te tecniques learnt in te process of tis proect work. Te youngest team member as started er doctoral work and se needs to upgrade er teoretical knowledge of poverty analysis related to tax reforms. Also, se is not very conversant wit te computer based simulation programmes wic se will learn in te process. 11. Any etical, social, gender or environmental issues or risks wic sould be noted- NONE 12. List of past, current or pending proects in related areas involving team members (name of funding institution, title of proect, list of team members involved) Aitava Raycauduri wrote te Infrastructure section for te State Development Report of West Bengal, India, funded by te Planning Commission, Government of India, He also wrote te paper on Land Reform in West Bengal for te World Bank funded proect on Reducing Poverty, Poulami Roy was a team member of te UNIFEM funded proect on Gender in Fiscal Policies, executed by te NGO Sacetana based in Kolkata, India, Se is also doing er doctoral work on Public Capital and Development, funded by State Government Scolarsip in Jadavpur University.