Implementing Subnational Value Added Taxes on Internal Trade: The Compensating VAT (CVAT)

Transcription

1 Implementing Subnational Value Added Taxes on Internal Trade: The Compensating VAT (CVAT) Charles E. McLure, Jr. Hoover Institution Stanford University Please send proofs to: Charles E. McLure, Jr. Hoover Institution Stanford, CA Phone:

2 Implementing Subnational Value Added Taxes on Internal Trade: The Compensating VAT (CVAT) Charles E. McLure, Jr. Hoover Institution Stanford University Abstract This paper describes an ingenious and elegant scheme for implementing a destination-based value added tax (VAT) on cross-border trade within a nation or group of nations. Sales to local purchasers (registered traders, households, and unregistered traders) would be subject to the local VAT, but sales to purchasers in other states would be zero-rated for state VAT and subject instead to a compensating value added tax (CVAT). Credit would be allowed for tax on purchases by registered traders: for the local VAT on intrastate purchases and for the CVAT on interstate purchases.

3 Implementing Subnational Value Added Taxes on Internal Trade: The Compensating VAT (CVAT) Charles E. McLure, Jr. * Hoover Institution Stanford University 1. Introduction Bird and Gendron (1998) offer a useful overview and analysis of two similar problems, to which they assert there is a common solution: a) the imposition of value added taxes (VATs) by both national and subnational governments in a single country and b) implementation of a VAT on cross-border trade (e.g., between nations) where only one level of government imposes the tax. The solution, they say, is a dual VAT levied either by two levels of government or by one level of government and an organization representing such governments. 1 A dual VAT is currently used by the federal government of Canada and the provincial government of Quebec, and Bird and Gendron contend that a variant of it could be employed in the European Union (EU). This paper extends the Bird-Gendron analysis by describing, with some suggested modifications, an ingenious and elegant scheme for dealing with cross-border trade that is internal to a nation or to a group of nations that wish to form a single market without internal fiscal borders, such as the EU. The technique was first proposed by Varsano (1995) as a way to convert the origin-based state VATs of Brazil to the destination basis, but is largely unknown in the English-speaking world. 2 My intent is to stimulate discussion and analysis of the CVAT and its competitors, such as the VIVAT discussed Section I do not pretend to have answered all the tough questions or to have designed an air-tight system. Essentially, Varsano would impose what I call a compensating VAT (hereafter CVAT) on sales to registered traders located in other states, rather than simply zero-rating them, as in the systems currently used in Quebec and (ostensibly as a transitional measure) by the member states of the European Union. 4 The CVAT minimizes the risk that households and unregistered traders will attempt to masquerade as registered traders located in other states, in order to make zero-rated purchases. 5 While perhaps not needed in Canada, where there is high-quality tax administration and an adequate level of communication and cooperation between federal and provincial tax authorities, or in the EU, the CVAT would provide

4 valuable protection of revenues, especially in less-developed countries (LDCs) and countries in transition from socialism (CITs), where neither tax administration nor cooperation between tax authorities is as highly developed, and does so in a way that I believe is superior to the superficially similar VIVAT method proposed by Keen and Smith (1996). 2. The Varsano Proposal Varsano proposes a dual federal/state VAT system what he calls a shared VAT with the following characteristics: 6 Uniform definition of the base of the federal and state VATs, Uniform administration of the federal and state VATs, 7 The same state VAT rate in all states, Zero-rating of interstate exports to registered traders (state VAT only), Deferred payment of tax on interstate imports by registered traders (state VAT only), Application of the CVAT to interstate sales to registered traders, Credit for CVAT on interstate imports by registered traders, Administration of the CVAT by the federal government (with state administration of their own VATs), Application of the VAT of the jurisdiction where the vendor is located to interstate sales to households and unregistered traders, and The possibility of intermediate solutions that combine the results of origin and destinationbased taxation. I will not discuss most of these features. Zero-rating/deferred payment and the need for both a uniform federal and state tax base and uniform administration are absolutely essential, but are so obvious as not to deserve detailed discussion. Some other features are more controversial, but are not central to the issue at hand.

5 Varsano assumes that the CVAT is not a separate tax; instead, it is merged into the federal VAT applied to interstate trade. (The federal VAT on interstate trade is higher than the federal VAT on intrastate trade by the amount of the uniform state VAT on intrastate trade; there is no state VAT on interstate trade.) This feature is important to minimize the necessity to pay refunds on interstate trade. It is, however, unrealistic in the case of the EU, where there is no federal tax. The following discussion treats the CVAT and the federal VAT as separate levies, but considers the possibility of combining them to reduce the need to pay refunds. The choice of a uniform state tax rate would simplify administration and minimize distortions of consumer choices (discussed below), but would sacrifice one of the primary objectives of assigning revenues from the VAT to state governments, the power of the state of destination of interstate trade to set the tax rate. 8 The following analysis is based on the assumption of state autonomy over tax rates; if the states choose uniform rates, that is a special case of this more general formulation. The possibility of a hybrid solution combining origin and destination-based taxation might be attractive, in theory, as a means of recognizing the claim of the state of origin to tax interstate trade. 9 After all, not all public services are provided to individuals where they live; some are provided to businesses and to individuals where they work. 10 Moreover, in the Brazilian context, where the VAT on interstate trade currently follows the origin principle, such a hybrid approach might be necessary for political reasons, at least as a transition device. Even so, I do not discuss the hybrid, as it raises issues that go well beyond the intended scope of this paper, including the politics of changes in tax policy that would cause substantial shifts of tax revenue among jurisdictions and means of ameliorating such shifts. Finally, imposing the CVAT only on sales to registered traders seems to serve no useful purpose; this limitation would complicate compliance and administration, application of the VAT

6 of the state of origin to interstate sales to households and unregistered traders would yield revenue to the wrong state, 11 and it would probably be less likely than the CVAT to prevent over- and undertaxation of sales to such purchasers. I assume that the CVAT applies to all interstate sales. 3. The Modified Varsano Proposal In summary, then, the proposal being analyzed here has the following essential features, as applied to interstate sales: Uniform definition of the state and federal tax base, Uniform administration, including all relevant laws, 12 VAT rates set by the individual states, Zero-rating of all interstate exports (state VAT only), Application of the CVAT to all interstate exports, Deferred payment of state VAT on interstate imports by registered traders (state VAT only), and Credit for CVAT on interstate imports by registered traders. It may also be appropriate to treat all sales of digitized content over the Internet as interstate exports subject to CVAT, due to the difficulty of determining the location of purchasers. Federal administration of the state and federal VATs and of the CVAT, treated initially as a separate tax, is assumed for expositional convenience. 13 Under this system sales local merchants make to local households and unregistered traders bear tax at the local tax rate, but sales to households and unregistered traders in other states are subject to the CVAT. Credit is allowed for tax on purchases by registered traders: for the local VAT on intrastate purchases and for the CVAT on interstate purchases. Thus, as is common of credit-based VATs, tax on sales to registered traders is irrelevant, as far as the aggregate VAT on sales to households is concerned. Table 1 summarizes these relationships. Vendors would need to deal with only three rates (plus zero, for exports from the country), only two of which would be relevant for any one domestic sale, as in any system involving taxation by two levels of government: the ordinary federal rate on all sales, the local rate on intrastate sales, and the compensating

7 VAT rate on sales to other states. 14 There would be no need for merchants to deal with the tax of any other state or to engage in interstate clearing of tax credits (discussed further in section 5.1). Of crucial importance for ease of compliance and administration, it would be necessary to distinguish only between intrastate and interstate sales; there would be no need to differentiate between sales to households and unregistered traders and sales to registered traders. (For reasons to be explained below, it might be thought desirable to ask merchants to distinguish interstate sales to registered traders from interstate sales to households and unregistered traders. This feature, which is not central to the proposal, would have no impact on tax liabilities, and thus would provide no incentive for cheating.) Various countries may answer the following questions differently: Should the federal and state governments each administer its own VAT? Could a consortium of states administer the CVAT? If there are separate state and federal administrations, which administers the CVAT? If both federal and state VATs and the CVAT are administered by the same government, which should administer them. 15 For convenience, following Varsano, I initially assume federal administration of both state and federal VATs and of the CVAT. 16 See also section 4.3 on the problem of excess credits. An illustration. Table 2 illustrates the working of the CVAT. It assumes a three-stage productiondistribution process, with value added of 100 in each stage and total value of sales to households of 300. To isolate the mechanics of the CVAT, it is assumed that all sales at the first two stages are to registered traders and that all sales at the third stage are to households. The first column summarizes these transactions. Gross liability for VAT (federal, state, and CVAT) at each stage is calculated by application of the relevant tax rates to sales. Net liability is calculated by deducting credits for VAT (federal, state, and CVAT) paid on purchased inputs; these equal taxes paid on sales at the previous stage. It is assumed that state A imposes a VAT of 4 percent and state B imposes a VAT of 8 percent. The federal government imposes two taxes: the ordinary federal VAT of 20 percent and a CVAT of 6 percent on interstate sales. The rate of the CVAT is assumed for convenience to lie midway between the

8 two state rates. Alternatively, it could equal the highest state rate, the lowest state rate, or any rate between, without affecting the basic conclusions presented here. The choice of this rate is discussed further below. The second set of columns shows the calculation of liabilities for state and federal VAT if all three stages occur in state B; it serves as a useful benchmark. The state collects VAT of 24 (8 percent of total sales to households of 300) and the federal government collects VAT of 60 (20 percent of 300). The third set of columns illustrates the operation of the CVAT, assuming that the first two stages occur in state A and only the third occurs in state B. 17 Thus, Stage 1 occurs in state A, Stage 2 involves an interstate sale by producers in state A to registered traders in state B, and Stage 3 involves sales to households by registered traders in state B. State A collects VAT of 4 on the production of 100 in Stage 1. Since State A zero-rates the interstate sale that occurs at the second stage, it refunds to the producer at Stage 2 the VAT of 4 collected at the first stage. State B does not tax these interstate imports; rather, it defers payment of tax on the imports and collects VAT of 24 on the entire sale of 300 to households, replicating the result in the second column, as is appropriate under a destination-based VAT. This is exactly the result that occurs under the system of zero-rating and deferred payment currently employed in the European Union. 18 The federal government collects its ordinary VAT at each stage and allows credits for it at each subsequent stage, so that, in total, the federal VAT on sales to households is 60, as in the second set of columns. In addition, the federal government collects the 6 percent CVAT of 12 on the interstate sale of 200 occurring at the second stage, but allows a credit for it at stage 3, thereby eliminating any net liability and any net revenue for the federal government when the two stages are consolidated. 19 In the example of table 2 all sales at the first two stages are to registered traders. It is a simple matter to modify the analysis to allow for interstate sales to households and unregistered traders. The 6 percent CVAT on such sales is a final tax, whether the sale is to a purchaser in state A or to one in state B. Table 3 summarizes the working of the CVAT system in the more general case in which registered traders in both states make interstate sales both to registered traders and to households and unregistered traders, as well as intrastate sales. It indicates the VATs applied to intrastate and interstate sales, credits

9 available to registered traders, and net tax burdens on purchases by households and unregistered traders. Intrastate sales are subject to the state s VAT rate, while interstate sales are subject to the CVAT; the federal VAT applies to both. Credit for tax on purchases of registered traders eliminates all VAT (and CVAT) collected on stages prior to sales to households and unregistered traders. Thus intrastate sales to households and registered traders pay the VAT of the state, while interstate sales to such purchasers pay the CVAT. In short, for sales to households and unregistered traders passing through registered traders in the state of destination, the final result for all three governments (A, B, and federal) is the same as if a) all production occurred in state B or b) zero-rating/deferred payment had been used without the CVAT. 20 That is, use of Varsano s CVAT implements a destination-based state VAT for such sales, but does so in a simple and elegant way that protects revenues from the risk of zero-rating interstate (or intrastate) sales to households and unregistered traders, which are subject to the CVAT. 4. Other Issues 4.1 Choosing the rate of the CVAT Several considerations would seem to be relevant to the choice of the rate of the CVAT. Setting the rate at the level of the lowest state VAT rate would minimize the need to pay refunds because credits for CVAT on business purchases imported from other jurisdictions exceed VAT due on sales, an issue that is discussed below. If this rate were applied to sales to households and non-registered traders, as assumed, there would be some discrimination against local merchants in states with VAT rates above the lowest rate. Moreover, this option would leave some latitude for evasion through diversion of products to household use and to non-registered traders. Finally, there would be an incentive for households to use mail-order purchases (subject to CVAT) as a surrogate for cross-border shopping in high-tax jurisdictions. 21 These problems would be serious primarily if the lowest state rate were well below the typical rate, as it is in the Canadian province of Alberta. 22 Setting the CVAT rate equal to the highest state rate would prevent favoritism to mail-order sales and tax-induced diversion of products to low-taxed uses and it would increase the incentives for traders in the importing state to register, in order to be eligible for credit for the CVAT, but it would increase the

10 incidence of excess credits for CVAT. Moreover, it would burden interstate commerce directed at states with low tax rates, if applied to sales to households and non-registered traders. Again, this problem is not likely to be severe, unless tax rates differ substantially across states. On balance a CVAT rate that is near the (weighted) average of state rates seems appropriate. 4.2 Taxation of interstate sales to households and unregistered traders The treatment of interstate sales made directly to households and unregistered traders deserves further discussion, as it poses a dilemma. It would be possible in theory, but probably not desirable, especially in LDCs and CITs, for vendors to apply the tax rate of either the state of origin or the state of destination to such sales of tangible products and remit revenues to that state. 23 The first option gives the wrong answer (origin-based taxation for such sales, which is problematic if such sales are concentrated in commercial centers), could result in tax exporting, and could distort economic relationships, by giving a competitive advantage to states with low tax rates. 24 The second option would be needlessly complicated, as it requires merchants to differentiate between shipments to various states and to collect and remit taxes due each state where it makes sales. 25 It seems preferable, especially if state VAT rates do not differ much, simply to levy the CVAT on these sales at the same rate as on sales to registered traders, as assumed above. This approach is probably inevitable in the case of interstate sales of digitized content downloaded from the Internet; indeed, it may be necessary to subject all sales of digital content to the CVAT, since their destination often cannot be known. 26 Under the CVAT system, vendors would need to distinguish between only three types of sales: intrastate sales (taxed at the federal and state VAT rates); sales to customers in other states, whether registered traders, households, or unregistered traders (taxed at the ordinary federal rate and the CVAT rate, but zero-rated for purposes of the state VAT); and sales for exports from the nation (zero-rated). This seems much simpler than applying the CVAT only to interstate sales to registered traders and levying the VAT of the state of origin on all other domestic sales, as in the original Varsano proposal. Under that alternative it would be necessary to differentiate between interstate sales to registered traders and other interstate sales. Under the modified proposal all interstate sales would be treated the same way. 27

11 As noted above, this solution is not perfect. But it is probably acceptable, since interstate sales of tangible products to households and unregistered traders are not likely to be important in LDCs and CITs, except where large cities straddle state boundaries, making cross-border shopping an attractive alternative; 28 mail-order sales are probably relatively unimportant. Moreover, it seems likely that the threat of cross-border shopping will prevent substantial divergence of tax rates between adjacent states, where this problem would otherwise be the greatest. If state VAT rates were to diverge substantially, there would be an incentive to make mail-order purchases from low-tax jurisdictions, if such sales were taxed at the rate of the state of origin of sales. Imposing the CVAT on these sales would reduce this incentive. The CVAT on interstate sales of tangible products (and perhaps all sales of digitized content) to households and unregistered traders would be a final tax. Since net CVAT revenues from these sales would initially flow to the federal government under the proposed scheme, it would be necessary to divide them among the states. A natural division of revenues would be in proportion to estimated intrastate sales to households and unregistered traders that are subject to VAT in each state. 29 This would combine the simplicity of a single rate on all interstate sales with a fair division of revenues. In order to calculate net CVAT revenues accurately, it would be necessary for vendors to differentiate interstate sales to registered traders from other interstate sales, even though both would be taxed at the same rate. This is probably an unnecessary complication. It would probably be satisfactory simply to periodically divide CVAT revenues that had not been claimed as input credits among the states. This would allow vendors to ignore the identity of those making interstate purchases. 4.3 The problems of excess credits/refunds One potentially troublesome feature of the scheme proposed here is the likelihood that many firms would have excess credits for, and thus deserve refunds of, the CVAT, at the same time they owed state or federal VAT on sales. 30 If CVAT were a separate system, firms whose interstate imports exceeded their interstate exports would require refunds. Because of the risk of fraudulent claims for refunds, in many countries this would, as Varsano notes in correspondence with the author, sink the little boat. This is one reason Varsano combines the CVAT and the ordinary federal VAT in one levy on interstate trade. Combining the CVAT and federal VAT into one rate would sacrifice knowledge of the net revenues from

12 CVAT to be divided among the states. This problem could be avoided by asking vendors to segregate the two taxes, but remit them together. Even this approach might not eliminate refunds, since net CVAT on interstate transactions could exceed net federal VAT on intrastate transactions. This problem would probably be vastly more important in Brazil, where for historical reasons the state VAT rate far exceeds the federal rate, than in countries where the federal rate is far above the state rate. 31 It seems that it would be desirable to offset credits and liabilities for federal tax (including the CVAT) against credits and liabilities for state tax, with clearing of net liabilities between the various governments involved. Thus, for example, in the example of Table 2, the firm at stage 3 might be allowed (and required) to net the credit of 12 for CVAT on interstate imports against its net liability of 20 for the federal VAT. There would be a transfer between federal and state governments equal to net credits. In more complicated cases the same procedure could be followed, with the taxpayer either writing only one check or receiving only a single refund (but not both), with the various governments netting credits and liabilities. (The netting of taxes and credits would occur more-or-less automatically if the federal government administered state taxes, as in the example of Table 2.) It must be emphasized that this netting of tax liabilities and refunds would not be the type of clearinghouse described below; it would be much simpler. 4.4 Problem industries As noted earlier, the proposal outlined here works best for trade in tangible goods. The distinction between intrastate and interstate sales underlying the proposal is more difficult to make in the case of services and intangible products, such as those delivered on-line in digital form in electronic commerce. But this may be a strength of the proposal, rather than a weakness. As noted earlier, electronic commerce in digitized content inevitably poses serious problems for a destination-based system, since the location of the buyer may be unknown and unknowable. But this suggests that revenue from sales of digitized content to households and unregistered traders should be shared among the states, which is precisely what the CVAT does. There is no need to throw the baby out with the bath, as the Canadian HST does, by sharing all revenues from the provincial VAT, not just those from services and intangibles.

13 Of course, electronic commerce in digitized products is the extreme case; there are other cases (e.g., telephone service) in which it may be difficult to allocate the tax base related to interstate transactions among states, but it is possible to distinguish between interstate and intrastate transactions. In such cases CVAT performs satisfactorily, because it utilizes the distinction that can be made (between interstate and intrastate transactions) and does not require the one that is difficult (between states). The CVAT proposal should be used where it works; it should not be discarded just because it does not work in all circumstances, especially since a) the cases where it does not work may be relatively unimportant in many LDCs and CITs and b) it will be difficult to find a solution that works better. 5. Comparison with Other Proposals. Those who do not examine the CVAT proposal carefully and do not understand how it operates may confuse it with other highly visible ways to implement a destination-based VAT. This section notes that it differs significantly from two such proposals and argues that it is superior to both. 5.1 Comparison with clearing-house arrangements In a clearinghouse system, some variant of which would be required for implementation of the VIVAT, to be discussed below, interstate exports would be subject to the VAT of the state of origin. 32 Registered traders importing from other states would be allowed credit for the origin-state s VAT in calculating liability for tax in the destination state. Destination states granting these credits would be reimbursed by states of origin that had collected tax for which credit is granted. Thus it would be necessary for vendors to identify credits by state collecting the input tax so destination states can be reimbursed for the credits they grant for taxes paid to states of origin. This system is so complicated that it has rightly been roundly rejected or endorsed with substantial modification. 33 By comparison, the Varsano system is simple and elegant. Interstate exports are zero-rated for state tax purposes, but subject to the CVAT, whether sold to registered traders or to households or unregistered traders. The CVAT is levied at a single rate, and it is not necessary for vendors to identify the state of destination of sales. 34 State tax on interstate imports by registered traders is deferred. Registered traders who make interstate imports are allowed credit for the CVAT, as for any VAT. For households and unregistered traders the CVAT is a final tax. Only net revenues from the CVAT are divided in proportion to

14 taxable consumption. It would be desirable to allow taxpayers to offset tax credits and liabilities involving the federal government and the various states, to minimize the need to pay refunds. But there is no tax credit clearinghouse and thus no need to trace tax credits. 5.2 Comparison with the Keen-Smith approach Keen and Smith (1996) propose an alternative method of protecting the VAT from evasion of tax on trade between members of the EU. They propose to distinguish between sales to registered traders and sales to households and unregistered traders and levy a VAT (which they call a VIVAT, for viable integrated VAT ) on the former at a rate that would be uniform throughout the EU. (Sales to households and registered vendors would be subject to the local VAT rate.) Registered traders would be eligible for input credits for the VIVAT. Interstate sales to households and registered traders would pay the tax of the state of origin, except under certain circumstances (e.g., when sales by remote sellers or to tax-exempt buyers exceed certain limits and when new means of transport most notably cars are sold). Table 4 summarizes how the two proposals would treat interstate and intrastate sales to registered traders and to households and unregistered traders. Whereas the CVAT distinguishes between intrastate and interstate transactions, the VIVAT proposal distinguishes between sales to registered traders and to other purchasers. Thus the CVAT proposal would apply the local VAT to all intrastate sales (cases 1 an 2 in Table 4) and the CVAT to all interstate sales (cases 3 and 4), while (subject to the exceptions noted above) the VIVAT proposal would apply the local VAT to all sales to households and unregistered traders (cases 2 and 4) and the VIVAT to all sales to registered traders (cases 1 and 3). The design of the VIVAT is motivated by the desire for what Keen and Smith call compliance symmetry application of the same tax rate to all sales to registered traders, whether or not they cross internal borders within the EU. I do not find the Keen-Smith proposal attractive, for several reasons. 35 First, it would require vendors to treat each sale differently, depending on whether it is to a registered trader or to someone else, even if the sale is to a local customer. Keen and Smith (p. 406) acknowledge, The one significant additional administrative cost... is that it would be necessary for traders and revenue authorities to

15 distinguish between sales to final consumers and sales to VAT-registered traders... This is not a trivial burden. (Emphasis added) Indeed, this burden is not trivial. The distinction is not one traders would ordinarily make in the course of business for non-tax reasons. Certainly they would not make it for all local sales, as would be required under the VIVAT. As Bird and Gendron (1998, pp. 432) note, the Keen-Smith scheme in effect rectifies one asymmetry, between trade within a member state and trade between member states, by creating a new one, between sales to registered taxpayers and sales to final consumers. By comparison, vendors would more-or-less automatically make or could make with relative ease the distinction between intrastate and interstate sales required to implement the CVAT proposal, at least in the case of tangible products. (See also the discussion of services and intangible products in Section 4.4.) In this regard it is important to reiterate a point made earlier: that there would be only one CVAT rate, which would be applied to all interstate transactions. The need to determine whether out-of-state purchasers claiming credit for the CVAT or VIVAT are actually registered traders would be handled by the federal tax administration (or by the states acting in cooperation), which would be in a far better position to make such a determination than the state of origin, which must make the determination if a system involving zero-rating of interstate exports is not protected by the CVAT or VIVAT. Second, Keen and Smith acknowledge (p. 406) that there is weakening of the chain of credits on intrastate transactions with registered traders, if the VIVAT rate lies below the in-state rate which is inevitable in some states if the rate on sales to business is set to equal the lowest rate on sales to households, as Keen and Smith suggest. There is also some weakening of the chain of credits under the CVAT, where the CVAT rate lies below the state VAT rate. But this occurs only for interstate trade with registered traders in some states; by comparison, it happens for all sales to registered traders in states with VAT rates below the VIVAT. Finally, under the Keen-Smith proposal destination states allow credits for taxes collected by origin states. This implies the need for a mechanism for clearing tax credits between states, an undesirable feature discussed above. 36 Again, this is not a minor problem. 5.3 Could there be a CVAT/VIVAT hybrid?

16 One reader of this paper has suggested limiting the CVAT to interstate sales to registered traders (block in 1 Table 4), as in the original Varsano proposal; other sales would be subject to the local VAT. The transition system in place in Europe can be seen as a special case of such a system, with a CVAT of zero. This approach, which we might call a hybrid CVAT/VIVAT, has both advantages and disadvantages. It would limit application of the CVAT to the case where it is most needed, interstate sales to those claiming to be registered traders. Since the CVAT would yield no net revenue, there would be no need for a formula to distribute CVAT revenues among states. The hybrid CVAT/VIVAT would violate the principle of compliance symmetry, by treating sales to registered traders differently, depending on whether or not they cross state borders. Moreover, for interstate sales to households and unregistered traders the hybrid would substitute differences between VAT rates in various states for differences between the CVAT and state tax rates. Differences would, on average, be greater under the hybrid than under the basic CVAT proposal with a CVAT equal to the average of state rates. Finally, revenues from taxing such sales would go to the wrong states. On balance, this approach is not attractive. 6. Summary Appraisal It appears that the use of a CVAT on interstate sales would cut the Gordian knot that has long stymied assignment of the VAT to subnational governments throughout the world. 37 States could retain sovereignty over VAT rates, while employing destination-based taxation. 38 The CVAT would allow the exporting state to zero-rate interstate sales and the importing state to defer payment of tax on imports from other states, without breaking the chain of credits that is the administrative hallmark of the VAT. The CVAT is vastly superior to both the incredibly complicated proposal for a tax credit clearinghouse and the revenue-sharing definitive solution proposed by the Commission of the EU, which involves total loss of state autonomy over VAT rates. I believe it is also superior to the Keen-Smith proposal, which requires vendors to distinguish between sales to registered traders and other sales.

17 The potential importance of the CVAT proposal is indicated by the following quotation from a recent paper by Richard Bird (1999): This simple system seems to make subnational VATs feasible and potentially attractive... Indeed, over time, as with respect to earlier changes in tax administration such as the introduction of income tax withholding and indeed the VAT form of sales taxation itself, this new idea in fiscal technology may prove to be one of the key innovations in tax thought of the century.

18 ENDNOTES

19 REFERENCES Bird, Richard M. (1999) Rethinking Subnational Taxes: A New Look at Tax Assignment. Working Paper No. WP/99/165 of the International Monetary Fund. December. Bird, Richard M., and Pierre Pascal Gendron. (1998). Dual VATs and Cross-Border Trade: Two Problems, One Solution? International Tax and Public Finance 5, pp Commission of the European Communities. (1985). Completing the Internal Market. Brussels. Commission of the European Communities. (1996). A Common System of VAT: A Programme for the Single Market. Brussels. Fenochietto, Ricardo. (no date). El IVA Compartido: Una Herramienta Útil para el Reemplazo del Impuesto sobre los Ingresos Brutos y la Descentralización de Tributos. xeroxed. Gonzalez Cano, Hugo. (1996). La Reforma Tributaria de Brasil y Posible Aplicación del Nuevo IVA Federal y Estadual para el Reemplazo del Impuesto a los Ingresos Brutos. Boletin de la DGI 513 (September), Keen, Michael. (1999). CVAT, VIVAT, and All That: New Forms of Value-Added Tax for Federal Systems. xeroxed; forthcoming in Canadian Tax Journal. Keen, Michael, and Stephen Smith. (1996). The Future of the Value Added Tax in the European Union, Economic Policy 23 (October), and McLure, Charles E., Jr. (1987). The Value-Added Tax: Key to Deficit Reduction? Washington: The American Enterprise Institute. McLure, Charles E., Jr. (1998a). Electronic Commerce and the Tax Assignment Problem: Preserving State Sovereignty in a Digital World. State Tax Notes 14 (April 13), McLure, Charles E., Jr. (1998b). Achieving a Level Playing Field for Electronic Commerce. State Tax Notes 14 (June 1),

20 McLure, Charles E., Jr. (1998c). The Tax Assignment Problem: Conceptual and Administrative Considerations in Achieving Subnational Fiscal Autonomy. a paper presented at the Intergovernmental Fiscal Relations and Local Financial Management Course, OECD Multilateral Tax Centre, March 16-27, 1998, Vienna, Austria. McLure, Charles E., Jr. (1999a). Electronic Commerce and the State Retail Sales Tax: A Challenge to American Federalism. International Tax and Public Finance 6 (May), McLure, Charles E., Jr. (1999b). Implementing a State VAT: Breaking the Logjam in Tax Assignment. forthcoming in the proceedings of the annual meeting of the National Tax Association, Atlanta, October 25. McLure, Charles E., Jr. (forthcoming, a). The Taxation of Electronic Commerce: Background and Proposal. forthcoming in Nicholas Imparato, editor, Public Policy and the Internet: Privacy, Taxes and Contracts. Stanford, Calif.: Hoover Institution Press. McLure, Charles E., Jr. (forthcoming, b). Rethinking State and Local Reliance on the Retail Sales Tax: Should We Fix the State Sales Tax or Discard It? forthcoming in Brigham Young University Law Review. Poddar, Satya. (1990). "Value-Added Tax at the State Level." in Malcolm Gillis, Carl S. Shoup, and Gerardo P. Sicat, editors. Value Added Taxation in Developing Countries. Washington: The World Bank, Ring, Raymond J., Jr. (1999). "Consumers' Share and Producers' Share in the General Sales Tax." National Tax Journal. 52 (March), Varsano, Ricardo. (1995). A Tributação do Comércio Interestadual: ICMS versus ICMS Partilhado. Texto para Discussão No. 382, Instituto de Pesquisa Econômica Aplicada, Brasilia. Varsano, Ricardo. (1999) "Subnational Taxation and the Treatment of Interstate Trade in Brazil: Problems and a Proposed Solution." Presented to the World Bank s Annual Bank

21 Conference on Development in Latin America and the Caribbean (ABCD-LAC) Decentralization and Accountability of the Public Sector, Valdivia, Chile, July 1999.

22 Table 1 Summary of Tax Treatment of Key Transactions: State VAT and CVAT Intrastate Transactions Interstate Transactions Tax applied Local VAT CVAT Credit allowed registered traders Local VAT CVAT Net tax paid on sales to households and unregistered traders Local VAT CVAT

23 Table 2 Illustration of Compensating VAT on Interstate Sales to Business: Tax rates: Ordinary Federal VAT Rate = 20%; State VAT Rates: 4% in A and 8% in B Compensating VAT Rate = 6%; Assumptions: Column 2, All Production in State B Column 3: Stages 1 and 2 in State A; Stage 3 in State B Transactions (Purchases, Sales, and Value Added) Calculation of Tax (VAT on sales; credit for input VAT: net VAT liability) Intrastate Sales Interstate Sale (from A to B) at Stage 2 State Federal State Federal Tax Tax: Tax Tax: Compen- Ordinary Total (A) (B) sating Federal Federal Stage 1: a. Purchases/Credits {This stage occurs in state A} 0 n.a. n.a. 0 0 b. Sales/Tax n.a. n.a c. Value added/ Net tax (c= b-a) n.a. n.a Stage 2: d. Purchases/Credits n.a. n.a (d=b) {Interstate sale from A to B}

24 e. Sales/Tax n.a f. Value added/ Net tax (f=e-d) n.a Stage 3: g. Purchases/Credits (g=e) {This stage occurs in state B} n.a h. Sales/Tax n.a. 24 n.a i. Value added/ Net tax n.a (i=h-g) j. Total tax n.a (j=c+f+i) n.a. : not applicable The first column shows calculation of value added; remaining columns show calculation of tax. Gross tax liability at each stage is calculated by application of the relevant tax rates to sales. Tax credits equal taxes paid on sales at the previous stage, where applicable. Algebraic notation indicates the calculation of value added and net tax liabilities at each stage.

25 Table 3 Illustration of VAT and CVAT Applied to Various Transactions Intrastate sales to households Interstate sales to Sales of imports by Direct interstate sales to and unregistered traders registered traders registered traders households and unregister (a) (b) traders (c) (d) ational VAT 20 percent 20 percent 20 percent 20 percent sales ational VAT 20 percent purchases redit for national (20 percent) T on purchases tate VAT 4 percent in A zero-rated 4 percent in A zero-rated sales 8 percent in B 8 percent in B VAT on sales 6 percent 6 percent VAT 6 percent purchases

26 redit for CVAT (6 percent) otal VAT 24 percent 26 percent 24 percent in A 26 percent 28 percent 26 percent 28 percent in B 26 percent ational 20 percent 20 percent 20 percent 20 percent tate A 4 percent 4 percent B 8 percent 8 percent VAT 6 percent 6 percent

27 Table 4 Comparison of CVAT and VIVAT Proposals Sales to registered traders Sales to households and unregistered traders CVAT proposal Intrastate sales 1 2 Subject to local VAT Interstate sales 3 4 Subject to CVAT VIVAT proposal Subject to VIVAT Subject to local VAT

28 * The author acknowledges his debt to those who first made him aware of the Varsano proposal or have discussed it with him, including especially David Rosenblatt, Ricardo Fenochietto, Richard Bird, Ricardo Varsano, and Fernando Rezende. Richard Bird, Amaresh Bagchi, Sijbren Cnossen, Pierre-Pascal Gendron, Michael Keen, Satya Poddar, Ricardo Varsano, David Wildasin, and an anonymous referee provided useful comments on earlier drafts of this paper (or on other papers from which this one derives). The Argentina Country Department of the World Bank supported the initial work on this proposal. As always, the views expressed here, and any errors, are the author s. 1 This is not meant to be a comprehensive description of the Bird-Gendron analysis. It is intended only to set the stage for what follows. 2 Varsano (1999) was presented after the present paper was essentially completed. The Varsano proposal has thus far been rejected in Brazil. Gonzales Cano (1996) and Fenochietto (no date) have proposed use of the Varsano technique in Argentina. 3 My success in doing so is indicated by Keen (1999), which reached me after the present paper was essentially completed. 4 I use the term state in what follows, since it is equally descriptive of the top tier of subnational government in important federal countries such as Brazil, India, and Mexico, where these government are called states, and of the member states of the EU. In other contexts one could refer to provinces (Argentina, Canada or South Africa), to subjects of federation (Russia), or to oblasts

29 (other parts of the former Soviet Union). I use the terms national and federal interchangeably, despite the inaccuracy of doing so in some contexts. 5 One referee asks, Why not use a retail sales tax? The distinction between registered and unregistered traders can be solved (as it is in the United States) by the distinction between wholesalers and retailers. This question and the description of the situation in the United States shows a woeful ignorance of the defects of the retail sales tax (RST), as it is actually implemented in the United States, as well as inherent problems of the RST. This paper is not the place to describe these defects, some of which I describe in McLure (1987), (1998a), (1999a), and (forthcoming, b). I make only a few major points. First, the RST is structurally inferior to the VAT as a means of safeguarding revenues and eliminating tax on purchases by business. Under the VAT, revenue is collected as products move through the production-distribution process, not all at one stage, as under the RST. Whereas a purchaser interested in diverting products to personal use need only lie to a supplier (and be ready to withstand audit) to obtain exemption from RST, it is necessary to lie directly to the tax authorities to obtain credit fraudulently for VAT paid on products diverted to personal use. Second, reflecting the superiority of the VAT, most economically important LDCs and CITs collect VATs at the national level. They could not afford the waste of resources necessary for independent implementation of a state RST. Third, American states have not made a serious effort to exempt all purchases by business. (They do not make a legal distinction between wholesalers and retailers, as the referee suggests.) Ring (1999) estimates that sales to business account for about 40 percent of revenues from RST. The uniform resale exemption certificate proposed by the Multistate Tax Commission is by no means uniform, because state laws are not uniform. Finally, the on-going debate over the tax treatment of

30 mail-order sales (and now electronic commerce) does not instill much confidence that the RST is the best way to handle interstate sales in a destination-based sales tax. This list of defects of the RST might lead readers to believe that I would favor replacing the state RSTs with a VAT or increased income taxation. McLure (forthcoming, b) discusses the relative merits of a conceptually pure RST, the RST actually employed in the United States, the VAT, and the individual income tax as sources of revenues for state and local governments. I conclude that the VAT is not a suitable source of revenue for local governments, because of crossborder shopping and the difficulty posed by cross-border trade. While the income tax might be preferable to the RST for the finance of state and local governments in a system being designed de novo, the wrenching adjustments that would be involved make it inappropriate to substitute it for the existing RSTs. 6 In Portuguese Varsano calls the tax he proposes the ICMS partilhado. (ICMS is the Brazilian state VAT.) Gonzales Cano (1996) and Fenochietto (no date) employ similar terminology in Spanish: IVA (impuesto sobre valor agregado) compartido. I prefer the term compensating VAT to emphasize that the CVAT compensates for the state VAT not collected at the state border. In my proposal there is only limited sharing of revenues between state and national governments. 7 As noted below in the discussion of the modified Varsano proposal, administration could be by only the federal government, by only state governments (acting singly or in concert), or by both state and federal governments. In each case the assumption is that the administrations apply the same legal structure and administrative standards. As a practical matter, federal administration is more likely than state administration to produce uniformity, given interstate differences in budget, training, and competence of tax authorities.

31 8 Thus I believe the Varsano approach, modified to allow the possibility of divergent state VAT rates, is superior to the definitive system recently proposed by the Commission of the European Communities (1996) and to the Harmonized Sales Tax (HST) recently introduced in Canada and three of the maritime provinces. (Both are described in Bird and Gendron, 1998.) While administratively simpler than a state VAT, the Commission s proposal (sharing of revenues from an EU-wide VAT on the basis of aggregate consumption figures) and the HST (provincial sharing of revenues from the federal VAT) allow no state or provincial autonomy over tax rates. 9 The technique Varsano proposes to use in implementing the hybrid is relatively simple. Rather than zero-rating interstate sales, the state of origin would subject them to a uniform (and presumably relatively small) VAT on interstate sales; the CVAT that would otherwise apply would be reduced by this amount, so that the aggregate burden on sales to households and unregistered traders would not be affected. Registered traders would be allowed credit against the VAT of the state of destination for the VAT on interstate sales, as well as for the CVAT. The state of origin would receive revenue from the VAT on interstate sales and revenues of the state of destination would be smaller (compared to revenues from a pure destination-based tax) by that amount. 10 I argue in McLure (1998c) that destination-based taxation is probably generally a better surrogate for benefit-related taxes than an origin-based tax. In theory, it might be appropriate to allow individual states to choose their own rate of origin-based taxation, in order to fine-tune the split of tax burdens between households and producers, based on the nature of publicly supplied services. In fact, this seems to be a counsel of perfection, given the risk of tax exporting, the fact that originbased taxation is likely to distort the location of economic activity if differences in tax rates do not reflect differences in public services provided to businesses, and the incentive for manipulation of

32 transfer prices if origin-based taxes are not levied at uniform rates. All things considered, it seems best to constrain the states to impose any origin-based VAT on interstate sales at a uniform rate. 11 In correspondence with the author, dated November 22, 1998, Varsano notes that he has subsequently proposed the possibility of transferring some of the revenue from interstate sales to households and unregistered traders to the destination state 12 One referee has noted the need for uniform rules for the treatment of purchasers making exempt sales, as well as taxable sales. Credits should be allowed only for the VAT on purchases related to exempt sales. I assume complete uniformity of rules in the state, federal, and compensating VATs. 13 One referee has suggested that, One of the key conditions for a CVAT is the existence of a central VAT. While it is doubtless true that the CVAT would generally work best where the central government imposes a VAT, this does not mean that the existence of a central VAT is either necessary or sufficient for the CVAT to function. It is not necessary because the CVAT could be implemented by a consortium of states. It is not sufficient because the cooperation between state and federal authorities may not be adequate. 14 Alternatively, if the federal VAT and the CVAT were merged, a given sale would be subject to either the federal and state rates on intrastate sales or the higher federal rate on interstate sales. 15 The states of Brazil, having administered the state VAT (which is quantitatively much more important than the federal VAT) since its inception in the late 1960s, are likely to resist surrender of administration to the federal government. Quebec administers both the federal and provincial VATs, because of the unique ethnic situation of that province, which manifests itself, inter alia, in