Argentina - Provincial Tax and Revenue - Sharing Reform

CONFIDENTIAL Report No. 19395-AR ARGENTINA: Provincial Tax and Revenue-Sharing Reform Country Distribution Draft June 30, 1999 This document has a restricted distribution and may be used by recipients only in the performance of their official duties. Its contents may not otherwise be disclosed without World Bank authorization. FILE COPY EXCHANGE RATE Currency Unit Peso (As of June 1999) USSI=ARGS1 Fiscal year January I - December 31 Glossary of Acronyms ATN National Treasury (Discretionary) Transfers CVAT Compensating Value Added Tax FEDE National Fund for Electricity Development in the Interior FONAVI National Housing Fund GDP Gross Domestic Product INDEC Argentine National Statistics Institute MCBA Municipality of the City of Buenos Aires NBI Unsatisfied Basic Needs Index VAT Value Added Tax Vice President Shahid Javed Burki, LCRVP CMU Director Myrna Alexander LCC7C SMU Director. Guillermo Perry, LCSPR Lead Economist: Paul Levy, LCC7A Lead Specialist Eliana Cardoso, LCSPR Task Manager- David Rosenblatt, LCSPR - ii - TABLE OF CONTENTS 1. FINANCING PROVINCIAL GOVERNMENTS IN ARGENTINA...................... A . PURPOSE OF THE REPORT .................................................................................... B. BRIEF HISTORY AND OVERVIEW ......................................................................... C. KEY PROBLEMS TO BE ADDRESSED BY THE REFORMS.........................................3 D . FE)ERALISM PRINCIPLES.....................................................................................5 2. PROVINCIAL REVENUE-SHARING...................................................................7 A . B RIEF H ISTORY ...................................................................................................7 B. TAX SOURCES FOR REVENUE-SHARING: STREAMLINING THE SYSTEM ............ C. THE PRIMARY DISTRIBUTION AND VERTICAL BALANCE ...................................14 D. SECONDARY DISTRIBUTION: ISSUES AND OPTONS..........................................16 E. TRANSITIONAL ISSUES .......................................................................................27 3. PROVINCIAL TAX REFORM .............................................................................28 A. CURRENT STRUCTURE OF TAXES ......................................................................28 B. TAX ASSIGNMENT PRINCIPLES ..........................................................................29 C. ALTERNATIVES FOR A GENERAL PROVINCIAL CONSUMPTION TAX ...................34 D. DECENTRALIZATION OF ADDITIONAL TAX POWERS..........................................41 4. CONCLUDING COMMENTS .............................................................................42 5. R E FER EN C E S ......................................................................................................44 ANNEX 1: Equalization Transfers.............................................................. 48 ANN EX 2: Consultants' Reports..................................................................50 Preface This report is based on an ongoing dialogue with Argentine officials on the reform of provincial taxes and provincial revenue-sharing. Over the period April 1998 to June 1999, several missions were led by David Rosenblatt (LCSPR) to engage in technical discussions on these topics. During these missions, Charles McLure, Jr. (consultant) led the discussions on tax reform and provincial tax assignment, and Thomas Courchene (consultant) led the discussions on the reform of the revenue-sharing system. In a separate mission, Robert McLarty (consultant) drew on his experience in the administration of revenue- sharing in Canada to advise the Argentine technical teams. Peer reviewers were Maria Freire (WBIGF) and Anwar Shah (OEDCR) The process has been a truly cooperative one, with knowledge flowing in both directions. The main organizer of the work on the Argentine side was Carola Pessino, Secretary of Fiscal Equity, Chief of Cabinet's office; however, technical staff were actively involved from the offices of Rogelio Frigerio, Secretary of Economic and Regional Planning, Ministry of Economy and Rodolfo Vacchiano, Secretary of Financial Assistance to the Provinces, Ministry of Interior. Despite the highly cooperative nature of the work, the views presented in this report are the author's and should not be attributed to the World Bank or to the Argentine officials involved in this dialogue. General features of the reforms discussed here were included as policy conditionality of the Argentina Special Structural Adjustment Loan approved in November, 1998. - 111 - Executive Summary Purpose of the Report. 1. This report is a synthesis of an ongoing dialogue with the Argentine authorities on provincial tax and revenue-sharing reform. A variety of complex issues are to be addressed by the reforms, and the struggle to reach a final proposal necessarily will involve not only the Argentine federal government authorities, but also provincial authorities and civil society. The purpose here is not to make a specific recommendation, but rather to report on the progress to date, analyze some of the advantages and disadvantages of various alternatives, and highlight issues for further discussion. Background. 2. The contemporary Argentine Republic is divided into 23 autonomous provinces and the autonomous Municipality of the City of Buenos Aires (with legal status similar to the provincial governments). The provinces are the principle providers of health, primary and secondary education, and public safety functions in the Argentine public sector. They are also responsible for most transportation and housing investment and share responsibility with the federal government for regulation of privately provided infrastructure and utilities. Provinces execute approximately 38 percent of total public sector expenditures in Argentina. 3. Given the country's large geographic area and its diverse socioeconomic conditions, the high degree of decentralization of expenditure responsibilities seems natural. There have been some proposals to increase the degree of decentralization of expenditure responsibilities in Argentina - for example, turning the federal universities into provincial universities and creating a greater role for municipalities. For the purposes of this report, however, the current distribution of expenditure responsibilities will be taken as a constant fact. The focus here is on the financing side of the decentralization issue. 4. Revenue-sharing ("Coparticipation"). The current system of "coparticipation", as it is known in Argentina, is a complex mix of one large revenue-sharing pool (generating $10.7 billion or about one-third of provincial current revenues in 1998) and a series of tax-sharing programs. Many of the latter, although distributed automatically, are earmarked legally for financing specific sectoral expenditures, like housing projects or home mortgages (FONAVI -- $900 million) or providing electricity service to remote, rural areas (FEDEI). There are a number of discretionary transfers, as well: the special national treasury transfers (ATNs -- $538 million) as well as some specific social sector assistance programs and investment programs. Total budgetary transfers to the provinces were $18.3 billion in 1998, just under 6 percent of GDP, and financing about 56 percent of total provincial government expenditures. Aside from the three or four largest provinces, the rest of the provinces depend on transfers for 70 to 90 percent of their revenues. - iv - 5. The distribution of the general revenue-sharing fund is according to percentages that were fixed in place by a 1988 law, and these percentages were determined by historical patterns and political debate. Some of the smaller sectoral programs follow need-based criteria, while others do not. The final result is a large diversity in transfers per capita -- a nine-fold difference in total transfers per capita for the highest and lowest provinces. This dispersion is difficult to justify in terms of needs and/or cost differentials. The 1994 Constitutional Reform called on Congress to enact a new "coparticipation" law by the end of 1996. Congress later extended this deadline by two years, and now it has been delayed again. 6. A series of Bank reports on provincial finances in Argentina dating back to 1989 analyzed problems in the system of provincial finances. The most recent analysis was a chapter in The Fiscal Dimension of the Convertibility Plan report. The chapter included simulations of the outcome of greater decentralization of tax powers (or derivation-based transfers) in an attempt to improve the degree of correspondence between provincial revenues and expenditures. It also included preliminary simulations of the distributional impact of implementing Canadian style equalization transfers in Argentina. 7. A number of consensus views have emerged from the series of Bank sector work, Argentine academic papers, and most significantly, in recent government proposals for reform. Three problems that need to be addressed by any reform of intergovernmental fiscal relations are: (1) simplify the complicated multiple tax- sharing/revenue-sharing system, consolidating most programs into one fund, to improve transparency; (2) introduce a rational formula -- almost any rational formula -- for distributing these resources across provinces; and (3) increase the degree of correspondence between provincial tax payer and the provincial services he/she receives. Chapter two will examine the prospects for simplifying the transfer system and the prospects for developing a rational distribution formula. 8. Provincial Tax Reform and "Tax Decentralization." The third key problem to be addressed is the lack of correspondence between provincial tax payer and provincial recipient of public services. The only two ways to accomplish this (assuming a fixed assignment of expenditure responsibilities) is to either increase the share of provincial revenues raised by provincial own taxes or make the distribution of revenue-sharing across provinces based largely on the origin of where federal taxes are collected - or some combination of these two approaches. The advantage of decentralizing additional tax powers to the provinces is that it would give the provinces the ability to raise revenues at the margin (or reduce revenues) by changing the provincial tax rate. Demands by the provincial population for increased levels of government expenditures could be met, and financed through own tax revenues. Similarly, in the absence of discretionary federal transfers at the margin, political leaders would need to face the local taxpayer with the bill any time they wish to increase expenditures at the margin (or present a future bill, in the form of increased indebtedness). 9. In addition, the main own-source revenue of provincial governments is the gross receipts (ingresos brutos) tax. This is a traditional "turnover" sales tax applied to the sale of goods at all stages of production - without crediting producers for the tax paid on inputs, as would occur under a VAT. The result is that the incidence of this tax accumulates over the stages of production, providing an incentive for companies to integrate vertically, and leading to differential final tax rates on consumer goods, depending upon the degree of vertical integration involved in the production of those goods. The government and the provinces have already recognized the need to change this tax to a more efficient form of sales taxation. Chapter 3 provides a brief history of these efforts and presents a proposal for developing a provincial VAT, which in parallel to the federal VAT, could provide a buoyant and efficient source of tax revenues. This tax instrument could first be used to substitute for the gross receipts tax, but it may be powerful enough to be used for the additional decentralization of tax powers referred to in the preceding paragraph. Main Findings and Recommendations. 10. Revenue-sharing. We support the government's proposal to simplify the revenue-sharing system by consolidating all major federal taxes in the revenue-sharing pool. On the more detailed question of the inclusion of all payroll contributions, or just employer payroll contributions, it depends upon the appropriate degree of burden sharing and risk-sharing across levels of government for the short run fiscal losses from the transition period of social security reform. With regards to the primary distribution (the division of revenue-sharing across levels of government), the government should consider the option of setting the size of total provincial transfers equal to a moving average of recent years, as a means of achieving greater macroeconomic stability. 11. With regards to the secondary distribution, the government is proposing a formula based on four components. Component (1) is derivation ("devolutivo " component) of federal tax revenues. The idea is that part of the revenues collected by the federal government should return to the jurisdiction where they are collected. The government is considering using measurements of provincial GDP as a proxy, since precise data on the source of VAT collections are not available. Component (2) is equalization of potential access to the same level of revenues per capita. This would be based on a Canadian-style equalization formula. Component (3) is compensation for special needs. This component would provide additional resources to provinces in proportion to the percentage of the population living with "Unsatisfied Basic Needs" and in proportion to low population density. Component (4) is comprised of transitional adjustments to ease the burden on provinces that would receive a much lower level of transfers under the new system, compared to historical levels. These adjustments would be eliminated over time. The current government approach is to promise the provinces that these transitional transfers will assure that no province receives a nominal amount less than it has received in the most recent year. 12. Chapter 2 then provides a simulation of how the equalization component could be calculated and a simulation of a "stylized Canadian" formula based just on population and equalization. The simulation serves the purpose of displaying some of the technical .issues involved in developing the new distribution formula. We then make the following comments on the prospects and principles for the reform. - vi - 13. Any secondary distribution formula that does not directly return revenues to the jurisdiction where they are collected will be re-distributive. Even equal per capita would be highly re-distributive given that the collection of federal revenues per capita is highly dispersed Argentine society will have to decide on the degree of re-distribution that should occur: how much of the health, education, police and road services of the neighboring province is a local tax payer willing to subsidize? Once that decision is reached, there remains the equally difficult task of determining what criteria should be used to determine the pattern of relative re-distribution across provinces. It seems, though, that these decisions should be based on guiding principles and not on the basis of comparing percentages from different formulas and simply choosing the one that looks "about right" - or worse still, the one that would be closest to the current situation. Four sensible criteria seem to appear and re-appear throughout Argentine history, the academic literature, and throughout international experience: (a) Equal per capita transfers to provide financing for a basic package of services (in particular with regards to the social sectors). Perhaps this component could be refined to take into consideration student-aged population, along the lines of the recommendations in the Argentina: Provincial Finances Study; (b) Equalization of revenue-raising capacity through some sort of compensation to jurisdictions with lower per capita ability to generate own tax revenues. The calculation could be made through a variety of measurement techniques; (c) Special needs/special costs compensation to provide additional resources for jurisdictions that face unusual demands due to the particular characteristics of their population, or that face higher than average cost of service provision (due to inability to capture economies of scale, or economies of scope, or simply a high local cost of living). (d) Derivation -- simply to return federally collected revenues to the source of collection, under the understanding that those revenues are provincial revenues, collected by the central government's tax agency for efficiency reasons or practical administrative reasons. This approach is the only one that implies no re-distribution. 14. Moving from just (a) to adding on a (b) component almost certainly increases the degree of re-distribution, unless there is no correlation between federal and provincial per capita tax bases. Moving from (b) to (c) could actually move back towards less re- distribution, depending upon how the component is calculated. For example, if it is based on costs, rich urbanized jurisdictions might be forced to pay higher salaries to public workers. If this type of factor weighs heavily in (c), then including (c) in the formula would lessen the degree of re-distribution. One word of caution: the more elements one adds to the formula, the more difficult it becomes to follow the internal logic of the final outcomes of the formula. In any case, additional components should have clear objectives. For example, the special needs component (iii) in the - vii - government's proposal: is the purpose to adjust for the higher costs of public services in low population density areas, or is it to improve the interpersonal distribution of income (with the inclusion of "NBI")? It it is the latter, then perhaps there are better instruments (federal social programs) than intergovernmental transfers for achieving this goal. If there are specific public investment issues, like a lack of access to drinking water or sewerage services in specific regions, then the issue may be resolved through regulation (or contract design) of private water companies, or via specific purpose matching grants (or loans). 15. Derivation and Correspondence. An additional element in this discussion is the high degree of provincial dependence upon transfers. This implies a low degree of correspondence, which breaks down the link between the individual who finances expenditures and the individual who receives public services. Transfers represent well over 50 percent of revenues in most Argentine provinces, so the degree of re-distribution inherent in the transfer system will have an enormous impact on the distribution of total revenues across provinces - much more so than in a country with a higher degree of provincial financial independence. This fact provides a strong argument for the derivation component of the government proposal outlined above. In the absence of further decentralization of tax authority, re-distributive transfers perhaps should be tempered by simply returning a particular share of revenues collected to the source of collection - as a kind of proxy for a system in which provinces raise a greater share of revenues from taxing their own local populations. On the other hand, if greater decentralization of tax powers occurs, then a derivation component to revenue-sharing might not be necessary. 16. To find a reasonable formula that is acceptable to all will require a difficult consensus building process with provincial authorities. According to the Constitution the new coparticipation law would have to be a law-agreement (convenio-ley) ratified by both houses of the national legislature and approved by all 23 provinces (Article 75, item 2). There must be at least a dozen secondary distribution rules that would generally satisfy the needs for horizontal balance in the Argentine federation. Unfortunately, there are infinitely many bad secondary distribution rules. (Perhaps the best example would be a component that allocates resources according to which jurisdiction spends the most, or hires the most workers.) The value of the comprehensive analytical work conducted by the federal government is to present some of the advantages and disadvantages of the sensible criteria. Hopefully this will assist them in reaching an agreement with the provincial authorities. 17. Provincial Tax Reform. Chapter 3 provides a brief overview of the structure of provincial taxes in Argentina. Two provincial tax instruments are highly distortionary: the gross receipts tax (mentioned above) and the stamp tax (a charge on property sales and a variety of financial transactions). A number of alternatives are evaluated for substituting for these distortionary taxes and potentially decentralizing greater tax power to the provinces. Based on the work of Charles McLure, two particularly promising alternatives are considered: decentralization of federal excise taxes via a provincial surcharge on that tax, and the implementation of a provincial VAT (perhaps via - viii - surcharge) in conjunction with the existing VAT. The latter is a more controversial step, so below we present the general outline of how it would work. 18. Active academic debate, originating primarily in Latin America, now regards a provincial VAT or VAT surcharge as a promising alternative for broad-based intermediate-level government taxation. The general features of the provincial VAT that have emerged from the discussion are: (1) destination based; (2) zero-rating of inter- provincial sales by the province of origin with deferred payment in the province of destination (ala Europe); (3) provinces set their own rate; (4) a new federal "compensating VAT" on inter-provincial trade to reduce the incentives for evading the provincial VAT; and (5) continuation of the current regular federal VAT on all transactions (w/ reimbursement of foreign exports). This idea draws upon proposals that have come out of both Brazil and Argentina for an IVA compartido or IVA partilhado, as well as the dual VAT currently functioning in the province of Quebec, Canada. 19. On sales within a province, the provincial VAT would be charged along with the traditional federal VAT. On sales across jurisdictions, no provincial VAT would be charged in the province of origin (zero-rating), but when a final sale (or next stage of production sale) occurs in the province of destination, then the destination province's VAT rate would apply to that sale (deferred payment). To eliminate the incentive for households (final consumers) in other provinces to pose as registered traders, a federal "compensating VAT" (CVAT) rate would be applied to all inter-provincial sales - whether they be to households or registered traders. The "compensating VAT" not only would discourage evasion, but it would establish a chain of credits when goods move across boundaries between registered traders, thus maintaining one of the most desirable features of the VAT. 20. Evaluating the Alternatives. Any final evaluation of the prospects for the provincial VAT should be taken in the context of all the possible alternatives. There are five main alternatives: (1) Continue with the gross receipts tax (and stamp tax); (2) Implement the provincial VAT/ CVAT idea as a substitute for the gross receipts tax; (3) Implement a retail sales tax as a substitute for the gross receipts tax; (4) Eliminate the gross receipts tax and make the provinces even more dependent on federal revenue- sharing (with federal VAT or some other federal tax increasing to make up for the lost revenues, overall); (5) Abandon general sales taxation at the provincial level and implement a provincial surcharge on the income tax (or payroll taxes), or some other tax. The first option implies the economic distortions from this tax would continue to hamper the productivity of Argentine businesses. At some point, an empirical investigation, estimating the economic costs of this tax in Argentina should be conducted. With regards to the retail sales tax, a consensus view is emerging that for practical administrative reasons, it would be difficult to implement in Argentina. Eliminating the gross receipts tax and making the provinces more dependent upon revenue-sharing would cause a further deterioration in the low degree of correspondence and fiscal autonomy of the provinces. Option (5) does not appear promising, since the VAT is the only tax in Argentina (other than payroll taxes) that generates a significant share of GDP in revenues. If there are no clear reasons why the provinciallfederal dual VAT would not - ix - work in Argentina, the government should move ahead with plans for implementing it as the most viable option. 21. Tax Decentralization. Two potentially strong tax instruments for decentralizing greater tax authority to the provinces are discussed in Chapter 3. One is a surcharge on federal excise taxes: tobacco, fuels and alcohol. The practical issue is one of how to allocate the revenues. Once again, given that these revenues would finance public services primarily in the location where the tax payer resides, it seems that the best solution would be to allocate the revenues according to where the goods are consumed. At the location of production, an identifier as to the location where goods are shipped would then identify where the revenues from that sale should go. The second strong tax instrument would be the provincial VAT described above. If successful as a substitute for the gross receipts tax, it could be used as a way to increase provincial fiscal autonomy with the federal government vacating tax room to the provinces, in exchange for a smaller share of revenue-sharing. It should be noted that a third instrument that in principle could be successful is a surcharge on the federal personal income tax. The main reason for limited enthusiasm for this instrument is that it continues to be underused at the national level: via a combination of exemptions and poor compliance. Despite this short-run problem, a provincial surcharge on the federal personal income tax should remain an interesting option for the future reform agenda. 22. Sequencing. The sequencing of reforms is another critical element. Experimenting with the provincial VAT as a substitute for the gross receipts and stamp taxes may be the key to later decentralization of tax powers. In addition, decentralization of tax powers may only be viable if a new system of revenue-sharing is in place - to assist in a reasonable way those provinces that will not be able to collect taxes at a level comparable to national norms. There is a strong argument, then, for proceeding with the replacement of the gross receipts tax and the creation of new criteria for revenue-sharing, to be followed thereafter with greater decentralization of tax power. 23. With decentralization of greater revenue-raising capacity -- via excise taxes, the IVA Compartido and eventually, the personal income tax - the problem of fiscal correspondence would be reduced. Simultaneously, the accountability of provincial governments to their local tax payers would increase. At this stage, unconditional block transfers (via revenue-sharing) to the provinces may become irrelevant - with the exception of the fiscal equalization transfers discussed in Chapter Two. These would continue to be important for provinces to be assured that tax decentralization does not leave behind the low per capita tax base provinces. At this latter stage, with fiscally strong provinces, the central government may still choose to enact specific, conditional transfers for supporting additional provincial actions in those specific areas where there is a clear national interest. 1. FINANCING PROVINCIAL GOVERNMENTS IN ARGENTINA A. PURPOSE OF THE REPORT 1. This report is a synthesis of an ongoing dialogue with the Argentine authorities on provincial tax and revenue-sharing reform. A variety of complex issues are to be addressed by the reforms, and the struggle to reach a final proposal necessarily will involve not only the Argentine federal government authorities, but also provincial authorities and civil society. The purpose here is not to make a specific recommendation, but rather to report on the progress to date, analyze some of the advantages and disadvantages of various alternatives, and highlight issues for further discussion. B. BRIEF HISTORY AND OVERVIEW 2. The Argentine nation was born out of the union of various colonial regions with differing economic and social characteristics. The first two decades of independence (1810 to 1830) were characterized by struggles between unitarian and federalist groups. Under the strong hand of Rosas (1830 to 1851), a federalist nation emerged. This process was consolidated with the Constitution of 1853, establishing a constitutional federal republic. The provinces were granted autonomy in the administration of their territories, a limited set of federal government functions was established, the provinces were granted residual powers over any matters not specified to be federal functions, and the internal economic union was promulgated with the elimination of internal customs controls.2 Despite some modifications (including the recent reform of 1994), the essential federalist structure of the 1853 Constitution remains in force today. 3. The contemporary Argentine Republic is divided into 23 autonomous provinces and the autonomous Municipality of the City of Buenos Aires (with legal status similar to the provincial governments). The provinces are the principle providers of health, primary and secondary education, and public safety functions in the Argentine public sector. They are also responsible for most transportation and housing investment and share responsibility with the federal government for regulation of privately provided I Alemann (1992) and Skidmore and Smith (1984). 2 The establishment of constitutional federalism was not accomplished immediately, as the Province of Buenos Aires refused to endorse the new Constitution until the 1860s. -2- infrastructure and utilities.3 Provinces execute approximately 38 percent of total public sector expenditures in Argentina.4 4. Each province has autonomy to organize its territory into local municipal jurisdictions, according to its provincial constitution and supplementary provincial law. There are approximately 2,150 municipalities and towns (comunas) in Argentina.s The legal autonomy, expenditure responsibilities and financing arrangements of local governments vary across provinces. Municipal governments are responsible for about 7 percent of total public sector expenditures in Argentina.6 In most cases, their activities are restricted to traditional urban functions of local street maintenance, street lighting, municipal parks, and solid waste disposal. 5. The provinces have diverse geographic and demographic characteristics. The Province of Buenos Aires is by far the largest, with nearly 14 million people. Then there are three jurisdictions with about 3 million people each (C6rdoba, Santa Fi and MCBA) Mendoza has 1.5 million inhabitants, followed by 7 provinces with populations in the 0.7 to 1.3 million range (Tucuman, Salta, Entre Rios, Corrientes, Chaco, Misiones and Santiago del Estero). Another 12 jurisdictions have populations that range from 160,000 to 550,000. Finally, there is Tierra del Fuego, at the tip of the continent with about 100,000 inhabitants. The level of economic development varies dramatically, as well. The high population provinces are at the top of the scale, while other provinces are blessed with particularly strong natural resource bases (prime land in the humid pampas, or the oil riches of the south). There is a concentration of lesser developed provinces in the northwest and northeast. In addition, even in the more developed regions, migration has led to pockets of poverty (for example, in the province of Buenos Aires, there are densely populated slums outside the federal capital). For example, see World Bank Report, Argentina - Transport Privatization and Regulation: the next wave ofchallenges, or Argentina-Reforming Provincial Utilities: Issues, Challenges and Best Practice. See Argentina- Provincial Finances Study, World Bank Report No. 15487-AR, July 12, 1996. 5 Argentine Ministry of Economy, Informe Economico Regional; July, 1998. 6Argentna: Provincial Finances Study, page 5. -3- TABLE 1: Basic Characteristics of the Argentine Provinces Povinc' ~ 1PopuPop'n i grji ateo.~~1eay na Buenos Aires 13.879.575 87 17.1 307,571 11.8 to 15.1 97.4 20.9 Catamarca 306.430 60 28.2 102,602 12.5 95.0 26.4 Cordoba 3.027.113 97 15.1 165,321 10.3 to 12.8 96.5 19.4 Corrientes 896.418 53 31.4 88,199 12.0 89.7 26.1 Chaco 929.878 49 39.5 99,633 11.3 87.6 34.4 Chubut 428.405 120 21.9 224,686 12.8 94.9 18.0 Entre Rios 1.096.121 75 20.6 78,781 10.0/14.3 94.6 19.5 Formosa 480.979 44 39.1 72,066 6.9 90.6 31.4 Jujuy 584.267 53 35.5 53,219 15.6 92.1 24.4 La Pampa 296.748 109 13.5 143,440 11.3 95.5 12.4 La Rioja 266.846 73 27.0 89,680 7.6 95.6 25.7 MCBA 3.040.292 299 8.1 200 8.6 99.2 14.7 Mendoza 1.568.461 94 17.6 148,827 5.7 94.9 17.5 Misiones 950.262 51 33.6 29,801 4.8 90.8 24.1 Neuquen 520.423 150 21.4 94,078 12.2 93.9 15.7 Rio Negro 594.794 97 23.2 203,013 11.3 93.7 16.2 Salta 1.022.846 56 37.1 155,488 12.0 92.3 25.5 San Juan 569.527 64 19.8 89,651 7.1 95.3 21.6 San Luis 346.591 108 21.5 76,748 6.0 95.3 206 Santa Cruz 196.438 185 14.7 243,943 3.9 97.5 12.8 Santa Fe 3.038.867 93 17.6 133,007 13.5 96.0 15.7 Santiago del Estero 715.891 42 38.2 136,651 4.8 90.4 17.1 Tierra del Fuego 104.823 283 22.4 21,571 8.7 98.8 9.7 Tucuman 1.262.936 58 27.7 22,524 14.9 94.5 28.8 Total 36.124.931 102 19.9 2,780,400 12.4 96.0 20.9 Source: INDEC and Ministry of Economy (GDP per capita, preliminary unpublished data-to be revised.) Notes: Population is INDEC's projection based on 1991 census and forecasted growth rates. % of population in NBI is the percent of the population living in dwellings with unsatisfied basic needs (lacking in some basic access to service/minimum quality of construction). Unemployment rate is only for urban areas (where more than one urban area is surveyed, then both rates or range is reported). Literacy is % of population over 15 years of age that is literate. Infant mortality is per 1,000 live births. C. KEY PROBLEMS TO BE ADDRESSED BY THE REFORMS 6. Given the country's large geographic area and its diverse socioeconomic conditions, the high degree of decentralization of expenditure responsibilities seems natural. There have been some proposals to increase the degree of decentralization of expenditure responsibilities in Argentina - for example, turning the federal universities into provincial universities and creating a greater role for municipalities.7 For the See FIEL, Hacia Una Nueva Organizacion del Federalismo Fiscal en la Argentina, 1993. -4- purposes of this report, however, the current distribution of expenditure responsibilities will be taken as an historical fact.8 The focus here is on the financing side of the decentralization issue. 7. Like intermediate level governments in most federations, the Argentine provinces finance their expenditures through a mix of own tax revenues, non-tax revenue and transfers from the federal government. When these levels of revenues are not sufficient, they borrow, either through formal contracts with Banks, voluntary bond issues, running arrears in payments to suppliers and employees, or involuntary bond issues to consolidate informal debts. 8. "Coparticipation". The current system of "coparticipation", as it is known in Argentina, is a complex mix of one large revenue-sharing pool (generating $10.7 billion or about one-third of provincial current revenues in 1998) and a series of tax-sharing programs. Many of the latter, although distributed automatically, are earmarked legally for financing specific sectoral expenditures, like housing projects or home mortgages (FONAVI -- $900 million) or providing electricity service to remote, rural areas (FEDEI). There are a number of discretionary transfers, as well: the special national treasury transfers (ATNs -- $538 million) as well as some specific social sector assistance programs and investment programs. Total budgetary transfers to the provinces were $18.3 billion in 1998,9 just under 6 percent of GDP, and financing about 56 percent of total provincial government expenditures. Aside from the three or four largest provinces, the rest of the provinces depend on transfers for 70 to 90 percent of their revenues. 9. The distribution of the general revenue-sharing fund is according to percentages that were fixed in place by a 1988 law, and these percentages were determined by historical patterns and political debate (see chapter 2). Some of the smaller sectoral programs follow fairly reasonable need-based criteria, while others do not. The final result is a large diversity in transfers per capita -- a nine-fold difference in total transfers per capita for the highest and lowest provinces. This dispersion is difficult to justify in terms of needs and/or cost differentials. The 1994 Constitutional Reform called on Congress to enact a new "coparticipation" law by the end of 1996. Congress later extended this deadline by two years, and now it has been delayed again. 10. A series of Bank reports on provincial finances in Argentina dating back to 1989 analyzed problems in the system of provincial finances. The most recent analysis was a chapter in The Fiscal Dimension of the Convertibility Plan report. The chapter included simulations of the outcome of greater decentralization of tax powers (or derivation-based transfers) in an attempt to improve the degree of correspondence between provincial revenues and expenditures. It also included preliminary simulations of the distributional impact of implementing Canadian style equalization transfers in Argentina. 8 These broader issues were examined in World Bank, Argentina: Provincial Finances Study, 1996. 9 Using cash data, as reported in the "national public sector accounts", Boletin Fiscal, 4h Quarter, 1998, Table 2. -5- 11. A number of consensus views have emerged from the series of Bank sector work, Argentine academic papers, and most significantly, in recent government proposals for reform. Three problems that need to be addressed by any reform of intergovernmental fiscal relations are: (1) simplify the complicated multiple tax-sharing/revenue-sharing system, consolidating most programs into one fund, to improve transparency; (2) introduce a rational formula -- almost any rational formula -- for distributing these resources across provinces; and (3) increase the degree of correspondence between provincial tax payer and the provincial services he/she receives. Chapter two will examine the prospects for simplifying the transfer system and the prospects for developing a rational distribution formula. 12. Provincial Tax Reform and "Tax Decentralization." The third key problem to be addressed is the lack of correspondence between provincial tax payer and provincial recipient of public services. The only two ways to accomplish this (assuming a fixed assignment of expenditure responsibilities) is to either increase the share of provincial revenues raised by provincial own taxes or make the distribution of revenue-sharing across provinces based largely on the origin of where federal taxes are collected - or some combination of these two approaches. The advantage of decentralizing additional tax powers to the provinces is that it would give the provinces the ability to raise revenues at the margin (or reduce revenues) by changing the provincial tax rate. Demands by the provincial population for increased levels of government expenditures could be met, and financed through own tax revenues. Similarly, in the absence of discretionary federal transfers at the margin, political leaders would need to face the local taxpayer with the bill any time they wish to increase expenditures at the margin (or present a future bill, in the form of increased indebtedness). 13. In addition, the main own-source revenue of provincial governments is the gross receipts (ingresos brutos) tax. This is a traditional "turnover" sales tax applied to the sale of goods at all stages of production - without crediting producers for the tax paid on inputs, as would occur under a VAT. The result is that the incidence of this tax accumulates over the stages of production, providing an incentive for companies to integrate vertically, and leading to differential final tax rates on consumer goods, depending upon the degree of vertical integration involved in the production of those goods. As will be discussed in more detail below, the government and the provinces have already recognized the need to change this tax to a more efficient form of sales taxation. Chapter 3 provides a brief history of these efforts and presents a proposal for developing a provincial VAT, which in parallel to the federal VAT, could provide a buoyant and efficient source of tax revenues. This tax instrument could first be used to substitute for the gross receipts tax, but it may be powerful enough to be used for the additional decentralization of tax powers referred to in the preceding paragraph. D. FEDERALISM PRINCIPLES 14. In order to advance in the reform agenda, federal and provincial authorities will need to agree on the basic principles that should guide the restructuring of provincial finances in the context of a country that has undergone a period of rapid economic reform and a radical revision of the role of the state. The country has embarked on a more -6- market-based development path, following the gradual accumulation of both human and physical capital.'0 The public sector now has a much more focused role in the former rather than in the latter, as large government owned enterprises have been privatized and grandiose public investment schemes abandoned. The provinces are the key players in the new Argentine public sector, given that they are the primary providers of public education and health services. The provinces need a reliable source of finance for these programs, but they need to finance these programs in a way that does not create excessive burdens on a private sector that is now competing in relatively open markets." 15. Argentina is also a country with persistent regional inequalities in income and in the general level of economic development. The federal government retains responsibility for a variety of social safety net programs, and those programs that are well-targeted to the poor naturally will follow a regional distribution of spending that is in line with the regional distribution of poverty. In addition, provinces with poorer than average populations generally are disadvantaged from the point of view of public finances: on the revenue side, because of their below average tax bases, and on the expenditure side, since local populations may be more dependent on publicly provided services compared to other provinces. Federal to provincial transfers that attempt to correct for these public finance problems likely will be a part of Argentine federalism in the future. How to determine the appropriate level and distribution of these transfers is not a trivial task. 16. In concrete terms, the Argentine people and their elected government officials need to ask basic questions about who should pay for whose services. For example, should an accountant in C6rdoba pay for part of a policeman's salary in Chaco? If not, would it be more acceptable if the Cordobes' contribution were to finance a primary school teacher's salary, or a young Chaquehlo child's medical treatment? These are the real issues that come into play with any system of intergovernmental transfers. 10 See Argentina: Second Generation Reforms for a broader discussion of these issues. Courchene (1998) provides a captivating discussion on the evolution of Canadian federalism and its economic progress. This will be discussed in the Argentine context in more detail later. -7- 2. PROVINCIAL REVENUE-SHARING A. BRIEF HISTORY 17. During the period immediately following the first national Constitution of 1853, up until 1934, the provinces were largely independent in the collection of their own revenues.12 The provinces had full autonomy to levy any type of tax, except external trade, which was reserved for the federal government. Up until the debt crisis of the 1890s, the situation could be generalized as: (1) a federal government financing itself from taxes on international trade; and (2) provincial governments financing themselves' primarily from taxes on real estate and inheritances. Over the period 1890 to 1935, continued problems with international trade motivated the central government to assume a greater share of tax room from the provinces, in terms of taxes on domestic economic activity. In parallel, discretionary transfers to the provinces grew to nearly 12 percent of provincial expenditures by the end of this period. Federal and provincial excise taxes existed side-by-side, with the federal government raising an increasing share of these revenues. 18. The first system of centralized tax collection and revenue-sharing was implemented over the period 1935 to 1947. Excise taxes were unified at the federal level, and federal profits and sales taxes were created. The revenues from the sales and profits tax was to be shared between the federal government and the 14 provinces (plus the federal capital), with 82.5 percent remaining with the federal government and 17.5 percent to the provinces. The distribution across provinces was based on four factors: (1) 30 percent according to population; (2) 30 percent in proportion to each province's expenditures; (3) 30 percent in proportion to each province's own-source revenues; and (4) 10 percent in proportion to the amount of federal sales tax collected in each jurisdiction. During this period, total transfers to the provinces financed less than 30 percent of provincial expenditures. Meanwhile, excise taxes had their own tax-sharing arrangements, mainly following historical revenue levels in each province. 19. The reform of 1947 shifted the primary distribution in favor of the provinces, with their share increasing to 21 percent and the federal share declining to 79 percent. For the distribution of resources across provinces, the reform subtracted a small percentage off each of the above four indicators in order to create a fifth indicator: the inverse of population, with a 9.58 percent weight. During the early 1950s, the share of excise taxes going to the provinces also increased, and the criteria for distribution across provinces became largely based on population (78 percent) and provincial production of the particular taxable good (20 percent). 20. The reform of 1959 further consolidated the revenue-sharing system to include most federal taxes. It also increased the provincial share of the total, and it established new criteria for the distribution across provinces. Now the distribution would be based 12 See Annex 2 of Argentina: Provincial Finances Study, and Annex I of Hacia Una Nueva Organizaci6n del Federalismo Fiscal en la Argentina, FIEL, 1993. -8- on the following four factors: (1) 25 percent according to provincial own-source revenues; (2) 25 percent according to provincial expenditures; (3) 25 percent according to population; and (4) 25 percent in equal amounts for every province. 21. In 1967 the federal government managed to increase its share of total revenue- sharing as part of a fiscal adjustment effort. The fiscal adjustment program began to unravel in the early seventies, and the federal government increased dramatically the amount of discretionary transfers to the provinces - reaching a level equivalent to 60 percent of revenue-sharing.13 22. Reminiscent of the situation today, the reform of 1973 was inspired by the desire to increase provincial financial independence and simplify the system of transfers. The Law 20.221 created a system whereby all national taxes would enter the revenue-sharing pool, with exception of external trade, some excise taxes and 15 percent of automobile taxes.14 Shared revenues would be divided 48.5 percent to the Central Government (responsible in those days for MCBA and the territory of Tierra del Fuego), 48.5 percent to the provinces, and 3 percent to a Regional Development Fund (established for financing public works). A rational formula for the distribution across provinces was introduced: 65 percent based on population, 25 percent based on a measure of the provinces' "development gap", and 10 percent on a measure of low population density. A province's development gap was measured by taking an average of three indicators: housing quality, education level of the population and per capita automobile ownership. In addition to these dramatic reforms in the distribution across provinces, the reform led to an increased provincial share of the revenue-sharing pool. 23. In the early to mid-1980s, a combination of the debt crisis, political crisis, re- structuring of the tax system15 and eventually, hyperinflation resulted in the gradual collapse of the entire revenue-sharing apparatus. Over the period 1985 to 1987, no coparticipation law was in effect, and federal transfers became purely discretionary. Adding to the lack of transparency, the macro-economy was entering a period of extremely high inflation, bordering on hyperinflation, making it even more difficult to trace real resource flows. 24. In 1988, Law 23.548 and a series of more specific laws re-created a general system of revenue-sharing and sectoral tax-sharing programs. This new set of laws ended the brief period of purely discretionary transfers; however, it also locked in place the existing pattern of transfers across provinces - after a series of negotiations.'6 The general coparticipation pool was comprised of the revenues from federal VAT, income and non-fuel excise taxes. Other specific laws established funds for financing FIEL (1993), page 142. 14Argentina Provincial Finances Study, page 44. 15 It is interesting to note that the restructuring of the tax system in 1980 was based on the model of reducing high payroll taxes (even higher in that era, compared to today) and generalizing the VAT. 16 See Saiegh and Tommasi (1998) for an historical description, as well as an interesting analysis of the politics involved. -9- investments in particular sectors, like housing, roads and rural electricity development. Modifications in the system in the early and mid-i 990s established "pre- coparticipations"; that is, partial earmarking of particular tax revenues to address a particular problem, before those tax revenues entered the coparticipation pool.'7 One example is the 10 percent deviation of the income tax (with a maximum of $650 million per year) to the Province of Buenos Aires for financing social programs in the poor suburbs surrounding the federal capital. Another example, is the 20 percent deviation of the income tax for financing the national social security system (facing high transition costs, post-reform). 25. The final result of this process is a system that is difficult to follow, as shown in the diagram below. In general terms, the left-hand side of the diagram displays that the main federal taxes, VAT and Income tax feed the general coparticipation pool after several pre-coparticipations. On the right-hand side of the diagram, one can trace the sector-specific programs, financed by tax-sharing of fuels taxes, and these transfers are tied to public works in sectors like housing, roads, electricity development and general infrastructure. At the bottom of the diagram, one sees that payroll taxes are earmarked for social security, and social security receives some "pre-coparticipation" earmarking of the income, fuels and VAT taxes. Total budgetary transfers to the provinces were $18.3 billion in 1998,18 just under 6 percent of GDP, and financing about 56 percent of total provincial government expenditures. Aside from the three or four largest provinces, the rest of the provinces depend on transfers for 70 to 90 percent of their revenues. About 70 percent of these transfers are unconditional general coparticipation. The rest are earmarked for the special programs mentioned above. It should be noted that some of the confusion in the system may be attributed to efforts to adapt a highly inflexible system of revenue-sharing to an Argentine public sector that undertook tremendous structural reforms over the last decade: privatization, the final stages of decentralization of secondary education and health, social security reform, re-construction of tax administrations at the federal and provincial level, and re-structuring of the public administration at both federal and provincial levels. As documented in Provincial Finances Study, Annex 1 and Provincial Finances Update, 1997 (internal Bank report). Also, see Schwartz and Liuksila (1997), "Argentina" in Fiscal Federalism: in Theory and in Practice. 18 Using cash data, as reported in the "national public sector accounts", Boletin Fiscal, 4h Quarter, 1998, Table 2. -10- Figure 1: General Revenue-Sharing System TM P_WM_CWWW_ GUARANTY ,mottewhm CLAUSE YEAR 1997 ASSIGNMENT OF FEDERAL TAX REVENUES o ' by FORMATION OF THE REVENUE-SHARING POOLS le L INCOMETAX u0 PROVINCE 60% Exm o S.Aw a dee pen.ss Enec SEnbsidies 12PROVINCES F. '7- SHED TAXES K ontNy a nine-fol d differee 1993 t 41dMs &% Educaon provIncE organized in decnigavlo rvnil au-de eaia' h (HA qesdualnerl sas) GROSS ~~~35 -2 10-7, 1.91 1 XENALS7 ITRADE TAXES VAs STATITICA TAXL I,i% - TAES L Deseltl krsno nd EXCISETA ON seriesN haveese bentasomdit nie eaiet h ainal aveae Iis INRestNE Source Argentne Mmisty ofInterror 26. Total transfers per capita and total revenues per capita vary dramatically. In terms of transfers per capita, there is a nine-fold difference between the highest and lowest provinces (even excluding the City of Buenos Aires). In addition, total transfers per capita bear little resemblance to provincial GDP per capita. The chart below has the provinces organized in descending level of provincial value-added per capita. The series have been transformed into indices relative to the national average. It is interesting to note that the current system of transfers, in some cases, appears to be re-distributive in the "wrong" direction: favoring rich provinces on the left-hand-side of the figure. Then there are some surprising spikes in transfers per capita in the middle of the scale, and only one poor province on the right-hand side that receives more than the average level of transfers per capita. A simple regression actually yields a positive relationship between transfers per capita and value-added per capita, although it is not statistically significant at traditional confidence levels. Finally, note how total revenues per capita largely follow total transfers per capita, with the notable exception of the City of Buenos Aires (the richest and most fiscally independent jurisdiction). Preliminary, unpublished data. - 11 - Figure 2: Provincial Value-Added per Capita, Transfers per Capita and Total Revenues per Capita 300 . 250 200 100 50 * Total Transfers per capita O Total Revenues per capita * Value-added per capita B. TAx SOURCES FOR REVENUE-SHARING: STREAMLINING THE SYSTEM 27. The first step for simplifying the system, allowing for greater clarity of resource allocation, would be to consolidate the various tax-sharing arrangements, "pre- coparticipations," and coparticipation into one revenue-sharing pool. This would make it easier to follow the sources of provincial revenue, and facilitate the provincial budget- setting process in coordination with federal authorities. Under the current system, gradual shifts occur in the distribution of transfers across provinces, for example, as oil prices fluctuate and consequently fuels tax revenues, since some provinces receive a relatively greater share of their transfers via the sectoral programs financed from fuels tax revenues. With one pool, it would be easier for the provinces and federal government to work together on estimates of the projected size of the pool for the upcoming year. It would also clarify the division of resources between federal and provincial levels, and put an end to the perverse incentives for federal officials to amplify or improve administration of particular taxes that are shared with the provinces at a lower rate than other taxes.20 28. One idea that has been gaining much support from a number of federal officials is to set up a single fund where all major federal taxes feed the revenue-sharing pool: VAT, 20 This argument also was made in the Provincial Finances Study and the Fiscal Dimension of the Convertibility Plan. - 12 - personal income, corporate income, payroll taxes and fuels taxes and other excise taxes.2' The idea would be to convert the "battleground" map presented in the previous section into the simplified map below. One noteworthy issue is the inclusion of payroll taxes. Since the social security reform of 1994, employee payroll contributions accrue to either a private pension fund, or the public pension fund, depending upon the choice of each worker. The employer contributions all go to the public social security system to partially finance current pensioners and to finance future pensioners, all of whom will receive a basic minimum pension from the government. As an increasing share of workers choose the private pillar, the system is moving towards fully-funded private pensions with the addition of a minimum public income support program for retired people. During the transition to this system, there is a significant deficit to the public social security fund (about 2 percent of GDP)22, since there are millions of current retirees from the old pay-as-you-go system. This deficit must be financed from general federal revenues, so from this point of view, it makes sense to include payroll taxes in the revenue-sharing program and consider the federal burden with the social security system in the primary distribution; that is, the division of revenue-sharing between federal and provincial levels of government (more on this in the next section). Even in terms of the future, when the "transition deficit" disappears, the federal pension program will be largely a minimum income support program for the elderly, and it is not clear that such a program should be financed from earmarked payroll tax revenues. Lowering payroll taxes, as a way to reduce distortions in labor markets and improve competitiveness, is firmly on the reform agenda in Argentina, with initial reductions implemented this year. Figure 3: Streamlined Revenue-Sharing System VAT, Income, Assets and Payroll Taxes Fuels and other Excise Taxes Federal Govermment(inldu&i9 Social Security), 21 The only exceptions would be import duties (which are relatively small 0.8% of GDP and declining over time) and a few other minor taxes ( about 0.5% of GDP). 22 See Chapter 2, Section A, of Argentina- the Fiscal Dimension of the Convertibility Plan.. - 13 - 29. Streamlining and macro-economic policy. General principles of fiscal federalism (Musgrave, 1959 and Oates, 1972) state that macroeconomic stabilization should remain a federal government function. In Argentina, due to the currency board arrangement, macroeconomic policy is, for the most part, fiscal policy.23 In any federation with some degree of decentralization, by definition, the federal government does not have full control over fiscal policy. Control is constrained by the fact that revenues from federally collected taxes are shared with sub-national governments, as these "federal" revenues are automatically spent, via transfers, without allowing for any discretionary federal fiscal controls. On the other hand, the only general alternative to revenue-sharing is greater decentralization of tax power to sub-national governments, which actually would imply even less control of fiscal policy by federal authorities (although it has other desirable features to be discussed in the next chapter).24 In effect, weakening of federal control over fiscal policy is an inevitable part of the decentralization process. In order to keep fiscal policy from spinning out of control, the focus of efforts probably should be in two areas: (1) eliminating the scope for sub-national influence over money creation by privatizing sub-national government-owned banks; and (2) restricting implicit or explicit federal guarantees on sub-national borrowing, to the greatest extent possible, in order to establish clear inter-temporal budget constraints.25 Efforts also could be made to establish fiscal stabilization funds at the provincial level. In addition, simply establishing clear rules for intergovernmental transfers can help maintain stability.26 30. One feature of the proposal that would improve federal control over fiscal policy is the elimination of automatic financing for sectoral programs, to be discussed in more detail below in the "primary distribution" section. Sectoral investment programs (the housing fund, electricity fund, etc.) would become federal budget line-items that could be adjusted according to federal fiscal policy objectives. 31. Finally, another way to approach the macroeconomic stability issue is in the design of the primary distribution. The traditional approach in Argentina (and most other countries) is to fix percentage shares of federal tax collection that go to each level of government. Another approach would be to set the total amount as some moving average of previous years.27 In this way, the sub-national level of government will not immediately receive (and perhaps spend) cyclical increases in expenditures. Of course, on the downside of the business cycle, this scheme would imply a greater fiscal burden 23 There is some scope for active monetary policy on the part of the Central Bank, to the extent that it can change reserve requirements and to the extent that the Argentine system is not a pure currency board: base money must be at least 70 percent covered by liquid international reserves (rather than 100 percent), with Argentine dollar bonds making up the difference. 24 A third option would be purely discretionary transfers, which would be undesirable in terms of predictability, transparency and objectivity in the distribution of revenues. 25 See Chapter 4 of the Argentina: Provincial Finances Study. Administrative controls may also be effective (see Ter-Minassian and Craig, 1997). Also see Sanguinetti, Zentner and Rotstein (1997). 26 See Sanguinetti (1994). 27 See Annex 1, page 33, of the Argentina- Provincial Finances Study. -14- on the federal government; however, this may be a reasonable price to pay to assure greater macroeconomic stability. C. THE PRIMARY DISTRIBUTION AND VERTICAL BALANCE 32. In principle, the distribution of revenues between the federal and provincial levels of government (the "primary distribution") should be set to correct for "vertical imbalance" - the mismatch between the expenditure responsibilities and revenue-raising powers of each level of government. As will be discussed in the tax reform chapter, almost all taxes can be well-assigned to the federal level; however, for the sub-national level(s), only a small subset seem to function well. As a result, vertical imbalances are common in decentralized federations. 33. The problem is how to measure the true degree of vertical imbalance. One might begin by determining whether there are chronic fiscal deficits in one level of government, but not the other. In this case, the current transfer program might have the wrong primary distribution, and hence, is not satisfactorily solving the vertical imbalance problem. In recent years in Argentina, the federal government has sustained fiscal deficits in the 1.5 to 2 percent of GDP range; however, much (or all) of these deficits can be attributed to the transition costs of social security reform. Meanwhile, the provinces have on aggregate suffered from deficits in the 0.4 to 1.1 percent of GDP range. In any case, this information is only of limited usefulness, since the deficits could be attributed to a variety of factors that have nothing to do with the primary distribution: poor tax administration, inefficient production of the expenditure functions allocated to that level of government, or simply tax rates that are too low. In order to determine precisely the appropriate primary distribution, one would need to calculate the reasonable cost for producing reasonable levels of the expenditure functions assigned to each level of government. In addition, one would need to calculate the yield from the revenue sources assigned to each level of government, given reasonable (perhaps "second-best") tax rates and reasonable levels of tax compliance. 34. One practical approach as part of the reform process would be to simply base the primary distribution on the historical division of resources across the levels of government. Under the reform proposal, the streamlining mentioned above implies that a variety of calculations must be made, since new taxes would be entering the revenue- sharing pool and all taxes would enter 100 percent into the pool (no "pre- coparticipation"). A first and very preliminary attempt to do these calculations is shown below, based on data for 1998 from the Ministry of Economy. One controversial issue should be noted: the final decentralization of secondary education and health that occurred in 1993. At that time, both federal and provincial revenues were growing extremely rapidly, thanks to the newly stabilized economy. As part of a broader negotiation of fiscal policy and provincial reforms (the "Fiscal Pact"), it was agreed that the provinces would assume these responsibilities without receiving a greater share in the primary distribution. This was in the context of rapidly increasing revenues for everyone, as well as fiscal losses suffered by the federal government from economic reforms - 15 - including the lowering of tariffs (with these revenues belonging exclusively to the federal government). 35. One mechanism that could be used to address a perceived inadequacy in the primary distribution would be the creation of a Fiscal Federal Forum.28 This organization, with federal and provincial representation, could conduct technical evaluations of the primary distribution, as well as play a role in the calculation of the secondary distribution (see below). Finally, future decentralization of tax powers gradually could close the vertical imbalance, making transfers and the primary distribution less important over time. Table 2: Historical Primary Distribution, An Approximation29 Primary Distribution, 1998 (Arg$, billions) Federal Collections Transfers to Provinces Revenues from Shared Taxes 35.8 - Income 9.5 12.6 General Coparticipation - VAT 20.3 0.2 Prov'1 Pension System - Fuels 3.7 ,1.5 Special Infrastructure Funds - Other excise 1.5 1.7 "Pre-Coparticipations" - Personal Wealth 0.8 0.7 Other Automatic 0.2 MCBA (automatic) Social Security 12.0 Total Shared and Soc. Sec. 47.8 16.7 Total Automatic Transfers 34.9% Provincial Share (incl. MCBA) Memo: Trade taxes 2.8 (But Needs to Calculated More Precisely) Others not shared 1.7 Sources Boletin Fiscal, 4th Quarter 1998, Table 14 for transfers, Gross Tax Revenues from Informe Economico Note there are another $1.6 billion in special discretionary or sectoral transfers that would bring the total transfers up to approximately $18 3 billion (total from Boletin Fiscal, 4h Quarter 1998, Table 2) **Social Security Taxes above include BOTH employee and employer contributions Another alternative would be to include only the employer contributions. 36. Note that if the special infrastructure funds become discretionary rather than automatic transfers, then this item would be removed from the right-hand side above - lowering total automatic transfers to about $15.2 billion. Probably, as part of the negotiations process, the federal government would commit to maintaining the value of those funds in the short-term. Once again, these numbers would need to be calculated with great precision. The table above is a first approximation. Government technical staff are working with their own calculations and continue to refine them. 28 The creation of a federal fiscal organism is called for in Article 75 of the 1994 Argentine Constitution. 29 The arrows give some additional information on which taxes feed which transfer type - without copying all the "spaghetti" of figure 1 (page 9). -16- 37. Social Security. Recent government proposals allocate only the employer payroll contributions to the revenue-sharing pool. In their proposals, employee contributions for those workers still affiliated with the public pension pillar, as opposed to the private pillar, would remain 100 percent with the federal government. Assume that the primary distribution for the provinces is set as a percentage share of the pool (as opposed to a moving average of the amount transferred in recent years). Then, as employer payroll reductions are reduced in the future to improve labor market conditions, the provinces will share in those fiscal losses. This would make payroll tax reform more palatable from the federal government's point of view, but it might also create a natural lobby of 23 Governors against this particular reform in the future. If employee contributions are included (as they are in table 2), then the provinces will share in the fiscal losses associated with future migration of individuals from the public system to the private system.30 Negotiations with the provinces necessarily will bring to light these issues, and some form of agreement would need to be reached. D. SECONDARY DISTRIBUTION: ISSUES AND OPTIONS 38. Principles. Courchene (1998)3 discusses two general rationales for intergovernmental transfers: the "constitutional rationale" and the "citizenship rationale." The constitutional rationale is the idea that each level of government must have access to sufficient revenues for it to carry out its constitutional responsibilities-that is, transfers exist to resolve a vertical imbalance between constitutional expenditure responsibilities and revenue-raising powers. The citizenship rationale is the argument that citizens ought to have rights to basic levels of certain services. The failure of a provincial government to meet its citizen's demands co.uld be due to a fundamental vertical imbalance between levels of government, or due to a horizontal imbalance. The latter implies that a particular province may have a smaller economic base relative to its population size and thus not be able to deliver the minimum standards of service, unless it receives additional resources from the center. 39. Courchene then discusses the principle of fiscal equity by paraphrasing the Canadian Constitution in this regard: all provinces should have access to a level of overall revenues such that they can provide reasonably comparable levels of public services at reasonably comparable tax rates. Some form of equalization transfer, discussed in more detail below, can be used to meet this objective, eliminating horizontal imbalances. In many federations, vertical imbalances are not fully resolved, intentionally, so that the center can exercise influence over provincial government expenditure decisions via conditional transfers. It may be desirable that the conditions are fairly broad (for example, simply tying certain funds to a particular sector), since the federal interest in the particular condition stems from the fact that there may be inter- provincial externalities associated with the particular good or service. Often for the 30 Already in 1996, 61 percent of employed persons and 46 percent of the self-employed had chosen the fully funded private option. See his background paper (Annex 2), and also his C.D. Howe Institute Commentary. -17- social sectors, though, this idea is similar to the citizenship rationale or fiscal equity principle described above. 40. The discussion in the preceding paragraph combines practical issues, relating to the limitations on sub-national own-revenue generation, with equity concerns - in terms of providing a comparable level of services across jurisdictions in a decentralized setting. Boadway and Hobson (1993) have argued for intergovernmental transfers on efficiency grounds, based on the notion of "net fiscal benefits."32 The theory starts by considering a resident of any particular jurisdiction, who receives government benefits with a utility value that may surpass, exactly equal, or be less than the taxes paid by that individual for receiving those services. If the benefits do not equal the taxes paid, then a positive or negative "net fiscal benefit" accrues to that individual. In the absence of intergovernmental transfers, negative net fiscal benefits, other things being equal, could cause an individual to migrate to another jurisdiction. In other words, there could be fiscally induced migration that surpasses the natural level of migration existing in a world without sub-national governments. Equalization transfers (see box 1) would eliminate inefficient levels of migration by eliminating non-zero net fiscal benefits (on average) and allowing governments to provide the same level of government services per capita at similar per capita levels of taxation. 33 41. Clearly, intergovernmental transfers in Argentina attend to the constitutional rationale described above. As the provinces ceded tax powers to the center, revenue- sharing became the main source of financing the provinces' expenditure responsibilities. At various points in history, the formula for the secondary distribution appeared to address "citizenship" purposes as well. For example, the reform of 1973 put in place a formula which assured minimum basic levels of services, with transfers largely determined by population, and partially determined by special needs or costs (development gap and population density). 42. The Argentine Constitution establishes a set of principles for the system of revenue-sharing. The following paragraph from Article 75 establishes the "constitutional" and "citizenship" rationales for the design of coparticipation: "The distribution between the Nation, the provinces and the city ofBuenos Aires and across them, will be made in direct relation to the competencies, services and functions of each one of them contemplating objective criteria of distribution; it will be equitable, solidarity enhancing, and it will give priority to the accomplishment ofan equivalent level of development, quality of life and equality of opportunity in all the national territory. " 32 Dahlby and Wilson (1994) construct an alternative efficiency argument for equalization transfers even in the absence of migration. Instead, they approach the subject from the marginal social cost of raising revenues - equalization allows provinces to refrain from over-exploiting a particular tax base from a national margmal cost minimization point of view. 33 One critique of the equalization approach is that higher per capita tax bases are often capitalized in higher per capita wages, rents, etc. In other words, high per capita revenue jurisdictions may also face high per capita costs. - 18- 43. Recent Government Proposals. An inter-ministerial team (Chief of Cabinet's Office, Ministry of Economy and Ministry of Interior) has been discussing a variety of proposals for reforming the coparticipation system. A consensus proposal has emerged that would use the following four criteria for determining the secondary distribution; that is, the distribution of federal transfers across provinces: (i) Derivation ("devolutivo " component) of federal tax revenues. The idea is that part of the revenues collected by the federal government should return to the jurisdiction where they are collected. The government is considering using measurements of provincial GDP as a proxy, since precise data on the source of VAT collections are not available. (ii) Equalization of potential access to the same level of revenues per capita. This would be based on a traditional Canadian equalization formula (see Box 1, below). (iii) Compensation for special needs.34 This component would provide additional resources to provinces in proportion to the percentage of the population living with "Unsatisfied Basic Needs" and in proportion to low population density. (iv) Transitional adjustments to ease the burden on provinces that would receive a much lower level of transfers under the new system, compared to historical levels. These adjustments would be eliminated over time. 44. With regards to (iv) above, the government has decided that the most effective strategy for securing support for the reform is to work under the assumption that no province will receive nominal total transfers that are less than the level received in the most recent year. Starting from that basis, the question remains then, how to phase out the transitional transfers. One possibility is to set the transitional transfers to a fixed nominal amount, so that with future revenue growth, they would shrink as a share of total transfers and gradually disappear. A second (perhaps more controversial) option would be to set the size of the total transfers to a fixed nominal amount. Clearly, this would lead to a much faster transition. These transitional issues are discussed in more detail in the next section of the chapter. 45. The government is in the process of analyzing a variety of simulations, attaching different weights to the components listed above. It is a tricky process, in that the equalization component is costed out first, based on measurements of the value of provincial tax bases. Total equalization transfers are a function of the degree of inequality in the values of per capita tax bases across provinces. In order to calculate a value that brings each province's total transfer up to the level of the most recent year, one must iteratively try different transfer values of the combined share of components (i) and 34 Referred to as equiparaci6n territorial in the government proposals for reform. -19- (iii) to reach a secondary distribution that satisfies this constraint (really 24 constraints). There is a problem of 24 equations with 25 unknowns (each individual province's. transitory transfer (24), plus the combined share of components (i) and (iii)), so one must use a spreadsheet's iterative solution function to find a value of the combined share of (i) and (iii) that fulfills the constraints with feasible values for the transitional transfers. 46. To date, the government's numerous simulations have allowed them to begin evaluating the options. There are a number of key decisions that would need to be made, in conjunction with the provincial authorities. For example, the relative weight between components (i) and (iii) would need to be based on the core beliefs of the Argentine people with respect to what degree of re-distribution should be part of the co- participation system, or to what extent the formula should be addressing special expenditure needs. There are important and difficult agreements that need to be forged in order to finalize this and other parameter values. Another example would be the decision of how many taxes to include in the equalization formula. In principle, all provincial taxes should be included; however, perhaps not all the parties involved feel comfortable with the reliability of the estimates of the value of the tax bases. 47. A Stylized Canadian Approach. Another approach with one less component to the formula would be to model the system on the general approach taken in Canada. With many changes over the years, the Canadian system of intergovernmental transfers can be characterized by two simple components: (1) equal per capita transfers to all provinces to assure a minimum level of service provision; and (2) equalization transfers, to compensate Provinces that have poorer per capita tax bases than average (see Box 1 on the next page). In addition, a transitional component could be added, as in the government proposal mentioned above. In brief, this approach would be fairly similar to the government proposal, except that population would be used instead of GDP for (i) above, and there is no special needs component ((iii) above). The issue is really how far one wants to use the formula to address particular expenditure needs. 48. The most difficult factor in calculating this formula is to arrive at acceptable measures of the per capita tax base across provinces, for all existing provincial taxes -as well as new provincial taxes suggested in the next chapter. For property-type taxes (like automobiles and real estate) the quantities can be derived from existing registries. Arriving at reasonable and consistent estimates of values across the country is a bit more difficult, but a variety of sources of information can be used - along with some simplifying assumptions. Argentine government officials have used regional price data to arrive at what appear to be usable calculations of the value of taxable provincial automobiles and real estate. For the most important provincial tax, the gross receipts tax, one can use regionalized national accounts data - although the most recent year is 1993. Data exist for both provincial value-added and provincial gross production. Since the tax is levied on gross production, I used the latter as the measurement of the provincial bases. For a description of the Canadian system, see Courchene (1998A, 1998B and 1983), Bird (1980) and Boadway and Hobson (1993). It should be noted that in the Canadian case, total transfers comprise a relatively small share of total provincial revenues. There are a number of national social policy conditions attached to the per capita transfers as well. -20 - An important source of revenues for a few provinces is oil and other natural resource "royalties" (actually a gross production tax). The rate is set by national law, so I used the actual royalties collected as a substitute for the national average rate times the estimated base. Another important tax that is gradually being eliminated is the stamp tax on property transfers and financial transactions. One could use one particular sector of the regionalized national accounts; however, this tax has been left out of the calculations below. BOX I: Canadian equalization formula. One common justification for intergovernmental-transfers is to avoid inefficient, fiscally induced migration. The idea is to avoid differences in "net fiscal benefits" for individual citizens from living in large per capita tax base jurisdictions, where net fiscal benefits to an individual are defined as the difference between the benefits the individual receives from government expenditures and the taxes the individual pays. This theory has been part of the inspiration the implementation of federal-provincial transfers in Canada which are based upon a simple equalization formula6: (Ei) = tas * (Be/Pc - Bip); (1) where,province i receives a per capita entitlement,.E)Pi, to compensate it for the amount (per capita) it could collect in taxes, using an-average national tax rate, if its per capita tax base, BlPi, were as large as the national per capita tax base, Bc/Pc. The average tax rate can be calculated as just the total tax revenues, TR, divided by the total consolidated tax base, Bc. Clearly these values differ depending upon the particular tax in question, whether it be the income tax, or a retail sales, or VAT tax. Equalization transfers can be calculated for each tax instrument and added together to determine the total equalization transfers. In Canada, the number of taxes that are equalized has increased over time, but it currently stands at over 35. In Canada, the negative values for rich tax base provinces are zeroed out; however, in principle (and in Germany, in practice), one could finance the positive transfers with explicit negative transfers from the rich provinces. It should be recalled that if it is federal government revenues that finance the equalization program, then it will be population of the rich tax base provinces that finances the program (unless there is no correlation between the values of federal tax bases and provincial tax bases - which is unlikely. In the contemporary Canadian system, the national aggregate per capita tax bases is no longer used for the standard. Instead, they have adopted a five province standard (leaving out particularly rich or poor provinces). This change was made in response to windfall oil royalties for Alberta a couple of decades ago, dramatically increasing the positive transfers from the formula-above. 49. The results, displayed in the figures below, are very similar to simulations of the equalization component calculated by government technical staff.37 It is interesting to note that the total resources required to equalize these four tax bases (that is, bringing the poorer provinces up to the average) is a relatively modest $1.5 billion - less than 0.5 percent of GDP, or about 10 percent of total transfers. The figures below assume that the $16.7 billion mentioned in the primary distribution discussed above is distributed 36 For a discussion of these formulas, see Courchene and Copplestone, "Alternative Equalization Programs: Two-tiered systems," in Fiscal Dimensions of Canadian Federalism, Richard Bird, editor. Also, see Boadway and Hobson, "Intergovernmental Fiscal Relations in Canada," Canadian Tax Foundation, Canadian Tax Paper No. 96, and Courchene (1983). One difference is that the government is already using a value-added tax base for sales taxation, assuming that the provincial VAT surcharge (chapter 3) is implemented as a replacement for both the gross receipts tax and the stamp tax. In addition, they have not included royalties - so far. -21- according to two criteria: (1) $1.5 billion in equalization; and (2) $15.2 billion in proportion to population (equal per capita).38 It should be noted that this is a highly re- distributive approach to transfers. Federal tax collections that finance the transfers are certainly not equal per capita,3 and in addition, disparities in revenue-raising capacity are being addressed. The latter is the appealing re-distributve logic of Canadian-style equalization - even if one ignores the efficiency argument discussed in Box 1. Figures 4 A and B Transfer Formula of Equal per Capita and Equalization A. Total Transfers per Capita 600 500 400 300 200 100 0 * Equal per Capita a Equalization B. Total Revenues per Capita 2000 1800 1800 9 __ 1400 1200 800 60 0 400 200 0 IM Equal per Capita U Equalization 0 Own Rev's (ncl. Roy's) The sectoral programs would have to be deducted if these become discretionary line items in the federal budget, rather than automatic transfers. See the calculations conducted by Juan Antonio Zapata for the Bank report, The Fiscal Dimension of the Convertibility Plan. -22 - 50. One major problem with this approach is that it is incompatible with the "transition rule" currently proposed in Argentina; that is, no province will receive less nominal transfers than it currently receives. The above distribution of transfers is so different from the current distribution that approximately $3.9 billion would be needed in additional transfers to pay the required "transitional" transfers (although, if lucky, part of this could come from federal revenue growth). Recall from paragraph 43 the discussion of the 24 constraints imposed by this transition rule. Iteratively trying different values can produce a result that would fit into the overall total amount of transfers. From the preliminary calculations, the only way to bring the whole system within the $16.7 billion threshold would be to shrink the equal per capita component to less than $3.5 billion and use the rest to finance equalization and huge compensatory transfers. In other words, either another component would need to be added to the formula (e.g., item (iii) in the government proposal) in order to make the distribution closer to the current distribution, or a different transition rule would be required. A decision would need to be taken as to whether one wants to mimic the current distribution, and on what basis, or whether a politically viable alternative transition rule can be arranged. 51. Additional refinements are required for the measurement of tax bases and fine- tuning the calculations above. Numerous simulations by government technical teams, however, lead to very similar results and to the same general conclusions discussed in the preceding paragraph. It is not clear, however, that abandoning the "no-one receives less" clause would be considered politically viable by the government. 52. "Capitation" Grants: Another Approach. A similar approach was proposed in the Argentina: Provincial Finances Study. The argument starts from the fact that education and health expenditures represent approximately half of all provincial expenditures, and these social expenditures are precisely the areas where there may be some inter- jurisdictional externalities, justifying federal transfers. In effect, transferring revenues on the basis of equal per capita may actually represent a needs-based approach to transfers. In order to provide a basic package of social services to each provincial resident, each province receives a per-person amount from federal revenues. To further refine this approach, the authors of the report split the per capita transfer between education and health components. The education component would be in proportion to school-age population rather than total population, while the health component would be in proportion to total population. In addition, an estimate was made of national average educational costs to determine a reasonable size for the per-student grant. Finally, simulations were run of one scenario ("Scenario A") that would add on an extra special needs component (based on the inverse of GDP, population with NBI and population). In addition, "Scenario B" involved a simple calculation of transfers required to provide equal total revenues per capita for all provinces (data on tax bases for the formula calculated above were not readily available at that time). In brief, the alternatives analyzed in the Argentina: Provincial Finances Study were very much in line with the philosophy behind the government proposal and the stylized Canadian approach described above. 53. Other Argentine Proposals. Piffano (1998) develops an interesting alternative approach, using some basic regressions. He begins by estimating the normal - 23 - expenditures per capita for the provinces given a variety of socioeconomic and geographic characteristics (GDP per capita, population, population density, average temperature, degree of urbanization and a political variable - Congressional deputies per capita). It is essentially an indirect way to arrive at an approximated cost function for a general level of expenditures. He then uses some regressions on physical and national accounts data to estimate expected revenues per capita for each province based on these characteristics (e.g., number of homes, regionalized GDP, number of automobiles). This part is in a similar spirit to the equalization transfers above. Finally, federal transfers are distributed so as to close the gap between expected expenditures per capita and expected revenues per capita for each province. It is an interesting approach that shares some similarities with the generalized "Representative Expenditure System" of Shah (1994A), combined with similarities to the "Representative Tax System" embodied in the Canadian equalization program described above. 54. Piffano also compares this approach with the Provincial Finances Study and with another Argentine proposal. In particular, he describes a technique used by a group from the University of La Plata, led by Prof. Alberto Porto. Piffano reports that the approach was to take the simple average of five alternative distributions: (1) 65 percent population, 25 percent inverse of tax capacity, and 10 percent inverse of population density; (2) 65 percent population, 25 percent development gap, and 10 percent inverse of population density;40 (3) 60 percent population, 30 percent NBI ("Unsatisfied Basic Needs" Index), and 10 percent inverse of population density; (4) 60 percent population, 30 percent NBI, and 10 percent fiscal effort; and (5) 50 percent population, 30 percent NBI, 10 percent fiscal effort, and 10 percent inverse of population density. By averaging, perhaps one can arrive at a kind of statistical consensus approach. 55. The FIEL (1993) research institute proposed a general re-structuring of expenditure responsibilities and tax assignment in Argentina. With regards to federal to provincial transfers, they proposed: per student grants for education; per capita grants for health insurance; transfers for extreme poverty, based on NBI; and general purpose grants, distributed in inverse proportion to tax capacity. 56. Summary: The Challenge Ahead. Argentine society faces the difficult task of reaching a consensus on how to convert an admittedly unfair revenue-sharing system into something that reflects society's values and preserves the cohesion of the federation. A new fiscal covenant41 needs to be reached between the federal government and the provinces, agreeing on the overarching social policy objectives of the country and the way that the system of fiscal federalism can serve those objectives. 57. Any secondary distribution formul.a that does not directly return revenues to the jurisdiction where they are collected will be re-distributive. Even equal per capita would be highly re-distributive given that the collection of federal revenues per capita is highly 40 Note the similarity with the 1973 reform. See Courchene (1998). - 24 - dispersed.42 Argentine society will have to decide on the degree of re-distribution that should occur: how much of the health, education, police and road services of the neighboring province is a local tax payer willing to subsidize? Once that decision is reached, there remains the equally difficult task of determining what criteria should be used to determine the pattern of relative re-distribution across provinces. It seems, though, that these decisions should be based on guiding principles and not on the basis of comparing percentages from different formulas and simply choosing the one that looks "about right" - or worse still, the one that would be closest to the current situation. Four sensible criteria seem to appear and re-appear throughout Argentine history, the academic literature, and throughout international experience: (a) Equal per capita transfers to provide financing for a basic package of services (in particular with regards to the social sectors). Perhaps this component could be refined to take into consideration student-aged population, along the lines of the Provincial Finances Study; (b) Equalization of revenue-raising capacity through some sort of compensation to jurisdictions with lower per capita ability to generate own tax revenues. The calculation could be made through a variety of measurement techniques; (c) Special needs/special costs compensation to provide additional resources for jurisdictions that face unusual demands due to the particular characteristics of their population, or that face higher than average cost of service provision (due to inability to capture economies of scale, or economies of scope, or simply a high local cost of living). (d) Derivation -- simply to return federally collected revenues to the source of collection, under the understanding that those revenues are provincial revenues, collected by the central government's tax agency for efficiency reasons or practical administrative reasons. This approach is the only one that implies no re-distribution. 58. Moving from just (a) to adding on a (b) component almost certainly increases the degree of re-distribution, unless there is no correlation between federal and provincial per capita tax bases. Moving from (b) to (c) could actually move back towards less re- distribution, depending upon how the component is calculated. For example, if it is based on costs, rich urbanized jurisdictions might be forced to pay higher salaries to public workers. If this type of factor weighs heavily in (c), then including (c) in the formula would lessen the degree of re-distribution. One word of caution: the more elements one adds to the formula, the more difficuli it becomes to follow the internal logic of the final outcomes of the formula. In any case, additional components should have clear objectives. For example, the special needs component (iii) in the government's proposal: is the purpose to adjust for the higher costs of public services in 42 See the calculations made by Juan Antonio Zapata for Chapter 3 of the The Fiscal Dimension of the Convertibility Plan. - 25 - low population density areas, or is it to improve the interpersonal distribution of income (with the inclusion of "NBI")? If it is the latter, then perhaps there are better instruments (federal social programs) than intergovernmental transfers for achieving this goal.43 If there are specific public investment issues, like a lack of access to drinking water or sewerage services in specific regions, then the issue may be resolved through regulation (or contract design) of private water companies, or via specific purpose matching grants (or loans). 59. Derivation and Correspondence. An additional element in this discussion is the high degree of provincial dependence upon transfers. This implies a low degree of correspondence, which breaks down the link between the individual who finances expenditures and the individual who receives public services. In principle, economic theory states that the overall level of expenditures will not be affected by a low degree of correspondence, so long as no discretionary transfers from the center are forthcoming to finance marginal expenditures." In terms of accountability and political economy, the degree of correspondence may be critical in disciplining expenditure decisions. 60. Without advancing on the next chapter, an additional element that should be considered is the decentralization of tax power. Transfers represent well over 50 percent of revenues in most Argentine provinces, so the degree of re-distribution inherent in the transfer system will have an enormous impact on the distribution of total revenues across provinces - much more so than in a country with a higher degree of provincial financial independence. This fact provides a strong argument for the derivation component of the government proposal outlined above. In the absence of further decentralization of tax authority, re-distributive transfers perhaps should be tempered by simply returning a particular share of revenues collected to the source of collection - as a kind of proxy for a system in which provinces raise a greater share of revenues from taxing their own local populations.45 On the other hand, if greater decentralization of tax powers occurs, then a derivation component to revenue-sharing might not be necessary. The derivation idea is broadly consistent with a common theme stated by provincial officials in Argentina: that "coparticipation" belongs to them, since it was the provinces that ceded tax authority to the center over the years. A logical corollary to this statement is that the revenues collected through the federal government's "borrowed" power to tax should flow back to the provinces according to where the taxes are collected. This corollary is rarely stated in the interior of the country. A final consideration with regards to the derivation 43 See Pauly (1973) and Brown and Oates (1987) for some of the debate on federal versus local roles in income re-distribution. See Wildasin (1991) for the implications of local redistribution on optimal federal grant design, and some discussion of centralization of this function. 44 Inter-temporal expenditure decisions might be affected by the degree of correspondence even if extra transfers (in the short run) are not forthcoming. Over time, with economic growth, revenues will grow and if those revenues are not collected from the local jurisdiction there will be less political incentive to save any of those increments, or return them to the local population via tax reductions. In addition, there is the empirical evidence in favor of a "flypaper effect": that subnational governments will spend more of an intergovernmental transfer than of an increase in local income. See Turnbull (1998) for an innovative theoretical approach to the subject and new empirical evidence. 45 See Chapter 3 of The Fiscal Dimension of the Convertibility Plan. -26 - component is that this component should then enter the equalization formula (see Courchene). These would be revenues collected by the federal government but delivered to the provinces according to their relative tax bases. These then could enter the equalization formula in the same way as provincial revenues collected by the provinces. 61. Another issue is the treatment of the special sectoral investment programs in housing, electricity development, etc. The government's proposal is to separate these programs from the general revenue-sharing pool and make them annual discretionary line-items in the federal budget. One important advantage46 to this approach is that it would allow for the re-evaluation of the justification of these programs, clarifying the purpose of these transfers and the criteria for their distribution (along the lines of recommendations in Shah (1994B)). These special purpose programs have focussed on investment in infrastructure: housing, electricity and the like. It is not clear that the areas financed have strong extra-provincial externalities justifying federal financing. In general, these programs seem to have a two-fold purpose: (1) directed assistance to the poor (although sometimes with very poor targeting, e.g. FONAVI housing fund47); or (2) help finance investment in smaller provinces that may have limited access to borrowing. With regards to (1), it is not clear that an intergovernmental transfer is the appropriate instrument. With regards to (2), the financial flows, if justified, could be in the form of a loan, rather than a grant. In any case, these programs should be re-evaluated and a careful analysis conducted as to whether matching or non-matching, conditional or unconditional sectoral transfers make sense - especially, in the context of a new revenue- sharing law. 62. The revision of the sectoral programs also opens the possibility of the federal government exercising greater influence in setting national priorities for these programs. For example, standards of service could be imposed as conditions for receiving these funds. The government is considering calling the new sectoral programs a "Human Development Fund." Per capita transfers, or a fixed transfer per "needy" person, could be used for financing specific social programs. Once again, this could be packaged with federal conditions on the performance of these programs in each jurisdiction. 63. The federal government has already conducted much of the technical analysis required to understand and evaluate the options that they face. Now the difficult consensus building process will begin, in conjunction with provincial authorities. According to the Constitution the new coparticipation law would have to be a law- agreement (convenio-ley) ratified by both houses of the national legislature and approved by all 23 provinces (Article 75, item 2). There must be at least a dozen secondary distribution rules that would generally satisfy the needs for horizontal balance in the Argentine federation. Unfortunately, there are infinitely many bad secondary distribution rules. (Perhaps the best example would be a component that allocates resources according to which jurisdiction spends the most, or hires the most workers.) The value of the comprehensive analytical work conducted by the federal government is to present 46 Another advantage would be the increased federal discretion over fiscal policy (paragraph 28) 47 See The Fiscal Dimension of the convertibility Plan, Chapter 3, Section B for an evaluation of FONAVI. - 27 - some of the advantages and disadvantages of the sensible criteria. Hopefully this will assist them in reaching an agreement with the provincial authorities. E. TRANSITIONAL ISSUES 64. One major issue is what to do about provinces that would suffer dramatic declines in total revenues from implementing a new formula. One approach described above is to agree ahead of time that each province will lose nothing in the short run. One problem with this approach is that the transitional transfers required could consume a high share of total transfers, as was discussed in the case of implementing a "stylized Canadian" transfer system. This fact leads to the danger that one will seek out components to add to the formula, so as to make the distribution more similar to the old distribution - but then locking in place forever a distribution that resembles the old system. If this is a non- negotiable issue, then there are two approaches to assuring no individual province loses in the short run: (1) freeze the total transfer, forcing a quicker transition to the new system as the value of transitional transfers approach zero; or (2) freeze the size of the transitional transfers, allowing them to shrink only as a share of total transfers, as revenues grow over time. Figure 5: Transition Rules Fixed Total Transfer Fixed Transitional Transfer (with 5% annual growth of copa. pool) 160 140 , Mr 7 .7 -- 1 2 36 120 100 100 80 80 60 60 40 40 20 20 0 0 1 2 3 4 5 6 7 8 9 1 2 3 4 5 6 7 8 9 Year Year 65. Within this context of assuring no-one loses in the short-run, there is the technical problem of calculating the distribution of transfers. The only way to iteratively solve for the 24 transitional values is to fix the value of the total provincial share of the revenue- sharing pool. Would one then fix it equal to last year, and then distribute the excess according to the same shares if federal revenues grow? Or would one try to project the value, and have the provinces return excess transfers if federal revenues turn out to be larger than expected? Another possibility would be to use the approach discussed earlier of fixing the size of the pool as a moving average of several recent years. Government technical teams have a full understanding of the issue. If the "no-one receives less than last year" rule is non-negotiable, then at some point a practical solution satisfactory to both the federal government and the provinces will need to be negotiated. -28 - 66. An Alternative Approach. A different approach would be to agree that in the first year only 10 percent would be allocated according to the new formula and 90 percent according to the old shares. 48 This would eliminate the need for the "transitional" transfers and the technical difficulties described above. It would require delicate political negotiations, but the phasing could proceed at an agreed upon pace: perhaps 10 percentage points per year, moving to the new system in ten years. Also, if one is lucky and federal revenues grow well during the first couple of years of the transition, no province may end up losing much in the short-run. 67. A final transition issue is the treatment of special cases, in particular the City of Buenos Aires. In recent years, it has become much more like a province, in terms of expenditure responsibilities. The one remaining standard provincial service area yet to be decentralized is the public safety function. As the federal capital, this jurisdiction receives extraordinary benefits from the federal level: federal police protection, indirect employment and expenditure impacts, maintenance of federal parks, etc. On the other, hand the federal public sector, as well as the diplomatic community, are exempt from a variety of local taxes. Finally, this jurisdiction is clearly the richest in Argentina in both output per capita and personal income per capita. One issue, then, would be whether this "privileged" jurisdiction should receive the same basic transfer component (derivation or equal per capita) as every other jurisdiction - in terms of fairness. Once again, the answer depends on the societal values embodied in the system of transfers. Another issue is with regards to equalization: does the inclusion of this jurisdiction distort this component or make it excessively large? The initial calculations indicate that this is not the case. 3. PROVINCIAL TAX REFORM A. CURRENT STRUCTURE OF TAXES 68. The main source of provincial tax revenues in Argentina is the "gross receipts" business turnover tax. It is a multi-stage sales tax with well-known problems associated with such a tax: the accumulation of tax burden across the stages of production, leading to high final effective rates for particular goods. This can lead to potentially large differences in effective rates across different types of final goods and even across producers of the same good (depending upon the individual producers' ability to vertically integrate). The second most important tax is the property tax on land and buildings. Although often reserved for municipal rather than intermediate level governments in international experience, the fairly high degree of centralization at the provincial level probably justifies the use of this tax instrument by the provinces at this 48 Thanks to Guillermo Perry, Chief Economist (LAC), for this suggestion. - 29 - stage in the evolution of Argentine federalism. The third most important tax is the tax on automobiles, levied on the value of automobiles registered in the particular province. The fourth most important tax is a somewhat archaic stamp tax. It is levied on the transfer of property and a variety of other financial transactions. The fifth most important tax takes the form of natural resource royalties. Up until the early 1990s, royalties were collected by the federal government and shared with the producer provinces. Today the revenues go directly to the provinces themselves; however, the provinces have no autonomy on the setting of royalty rates or the definition of the base.49 These are set by federal law. Table.3: Structure of Provincial Taxes, 1997 Consolidated 24 Jurisdictions, Tax Collection in 1997 ($ millions) % of Total % of GDP Gross Receipts Tax 6,001.5 51.7% 1.8% Property Tax 1,813.2 15.6% 0.6% Automobile Tax 912.9 7.9% 0.3% Stamp Tax 927.1 8.0% 0.3% Royalties 836.9 7.2% 0.3% Others 1,121.0 9.7% 0.3% Total 11,612.6 100.0% 3.6 Source: Ministry of Economy, Informe Economico Regional, November, 1998. B. TAx ASSIGNMENT PRINCIPLES50 69. This section begins with an overview of the general principles that should guide tax assignment in any federal country. These principles will result in a specific set of recommendations on tax assignment for the Argentine provinces. The first principle is: "Decentralization works best when taxes and the benefits of public spending (or costs incurred by governments) are closely related."5' This idea is similar to the idea of fiscal 49 To date, most royalties are collected on petroleum products; however, mineral extraction is poised to be a growth sector in various provinces. The "royalty" on oil extraction is levied on the gross value of production at the point of extraction (12 percent rate for wells that are in the production stage and 15 percent for wells that are still in the exploration stage). In addition, oil-producing provinces levy the gross receipts tax on oil sales. 5o This section draws heavily on the work of Charles McLure, who worked as consultant for this report and engaged the Government in numerous discussions on these issues in Buenos Aires. His complete background paper is included in annex 2 of this report. 51 McLure, 1998 (See Annex). -30- correspondence, described above, but it also would apply across levels of government - for example, the level of government that incurs the costs of the health effects of alcohol and tobacco consumption should receive the revenues from alcohol and tobacco excise taxes. The inherent nature of public or quasi-public goods implies that there are "generalized benefits" of public services. This is relevant in terms of fiscal correspondence across provinces, in that one would like to identify the location of benefits and match that location to the source of revenues. In this way, the beneficiary pays, rather than exporting his or her tax burden to someone else. 70. McLure divides the question into two parts: (1) do the benefits of public services accrue mostly to business or individuals? and (2) when they accrue to individuals, is it based more on the individual's residence or location of consumption, or production (i.e. source of income)? The answers to these questions would determine whether consumption taxes should be based on the origin of the sale (e.g., if public services primarily impact businesses) or on the destination of the sale (e.g., if individuals consume public services where they reside). One should recall that the overriding principle is to align the service recipient with the tax payer. McLure (1998) admits to a lack of clear evidence to answer questions (1) and (2) above, but he argues that "most public services are provided to individuals, and not to businesses, and that most services provided to individuals are more closely related to consumption or the residence of the taxpayer than to production or the source of income." This appears reasonable for the case of the Argentine provinces, given that social sectors comprise the majority of their expenditures - and even police protection is probably as important to individuals as it is to businesses. This will have implications for the specific recommendations for Argentine tax assignment below, but in general, this argument hints at destination-based consumption taxation and residence-based income taxation. 71. A second principle is that sub-national governments should have a source of marginal revenues. In order to have true fiscal autonomy, sub-national governments that wish to raise (or lower) their expenditures at the margin should have the authority to raise taxes (or lower the tax burden) from their own tax sources. Or conversely, they should not be able to finance increased expenditures by securing additional discretionary transfers. Moving ahead to the recommendations, this implies that provinces should have the authority to set rates on provincial taxes. 72. A third principle is that provincial taxes should be designed to limit distortions on the location of economic activity. McLure argues that taxes "levied at the origin of production or the source of income are much more likely to distort locational decisions than are taxes levied at the point of consumption or the residence of households, unless they reflect benefits of services provided to businesses or to employees at their place of employment." As in paragraph 62 above, this principle leads one towards destination- based consumption taxation and residence-based income taxation. 73. A fourth principle is that tax exporting should be avoided. This principle to some extent is a corollary of the first principle above. The recipients of public services in a particular jurisdiction will have the ability to pass off the costs to non-residents ("tax exporting") if they have access to particular types of tax instruments. One example -31 - would be taxes that are borne by non-resident owners of businesses (that cannot be fully passed on to consumers). Once again, McLure argues that this principle leads one towards destination-based consumption taxation and residence-based income taxation, as opposed to production-based sales taxes or source-based income taxes.52 74. McLure discusses a number of other principles, summarized briefly here. One is that to a large extent, tax competition (in terms of rates) can have the positive effect of limiting the use of sub-national taxes on business.53 He argues for a form of subsidiarity for tax assignment similar to subsidiarity in expenditure assignment. Since a number of taxes are clearly inappropriate (or impossible to administer) at the sub-national level, then all taxes that are appropriate should be assigned to that level, in order to limit vertical imbalance. In a similar vein, since the re-distributive function is usually considered to be a federal role (Musgrave, 1959), McLure argues that the re-distributive role of progressive tax rates on the income tax should be left to the federal government, and sub- national income taxes (if there any) should have a flat rate. Finally, he argues that natural resource taxation should be at the federal, not the sub-national level.54 He then states that if natural resource taxation is at the provincial level, then it should certainly enter the calculation of the equalization component of federal transfers (see chapter 2). 75. Summary assessment. McLure concludes with the following. "First, user charges, fees, and taxes closely related to benefits of services provided by governments (at the provincial level) are fair and they do not distort the location of economic activity. Second, taxes on production and the earning of income that do not reflect benefits of public services are likely to distort the location of economic activity and they may be exported in part to non-residents; they are to be avoided (at the provincial level). Finally, taxes on consumption and residence-based income taxes are arguably most likely to reflect the benefits of public services; even if they do not, they are less likely than other taxes to distort the location of economic activity. This suggests that the last type of taxes are the most appropriate for the finance of most generalized public services (at the provincial level)." 76. Autonomy in setting tax rates and the practicality of surcharges. Of the four components of sub-national tax policy - the decision of which taxes are assigned to the sub-national level, the definition of the base, the setting of the rate, and the actual administration (collection) of the taxes - McLure argues that the setting of the tax rate is the most important for establishing fiscal autonomy. In this way, one moves in the 52 Inman and Rubinfeld (1996) are even more emphatic on the issue of residence-based taxation: "Constitutional constraints also have a role to play. Most promising is to assign resident-based taxation as the only constitutionally allowed state and local tax... Source-based taxation would be limited to the central government." 53 Also see Oates and Schwab (1988) or Frey and Eichenberger (1996) for more on these issues. 54 In the literature, special taxes on natural resource extraction are justified as a charge on the economic rents accruing to these special goods (Boadway and Flatters, 1983) and on the externalities associated with environmental losses, or the loss of natural capital. The theory advocates using the economic rents as the tax base; however, in practice, this can be difficult to calculate. As stated earlier, in Argentina the provincial royalties are really gross production taxes levied at the source of extraction. -32- direction of assuring the principle of fiscal correspondence and creating a source of marginal revenues. McLure then argues that provincial surcharges on federally collected taxes (as used in for Canadian personal income taxes) are an ideal way of giving provinces fiscal autonomy, but avoiding administrative duplication and complexity. Surcharges could be used on excise taxes, the residence-based income tax and the VAT (destination-based). 77. Recommendation. McLure recommends provincial surcharges on the value- added tax and excise taxes for Argentina. In principle, a provincial surcharge on the individual income tax could work; however, the current low yield on this tax in Argentina limits much progress in this area. McLure does not recommend provincial surcharges or provincial administration of a company income tax. More details on each tax, and in particular the provincial VAT, are given below. Some participation by the provinces in the administration of the dual VAT is a possibility - rather than a pure surcharge on the centrally administered tax. Box 2: Tax Assignment for Municipal Governments in Argentina. There are approximately 2,150 municipalities, towns and local councils in Argentina. In general, their expenditure responsibilities involve traditional urban functions: local street maintenance and drainage; solid waste disposal; local parks and street lighting. Their total expenditures constitute about 2 percent of GDP or about 7 percent of total government expenditures. They collect themselves about half of their revenues - mostly from a variety of fees and charges, but also through taxes. Of the latter, the taxes used are often on the same base as provincial taxes: real estate or cars, for example. Nine provinces have decentralized the automobile tax to the municipal level, five have decentralized the urban property tax, and one province (Chubut) has handed over the gross receipts tax to local governments. Most transfers received by local governments are transferred from the provincial level; however, in recent years, there has been an increase in discretionary federal to municipal transfers. Most provinces conduct a large share of municipal transfers via revenue- sharing schemes that take on a variety of forms. (Source: Informe Economico Regional, July and November 1998 issues). Until the provinces are able to implement a stronger provincial-level general tax instrument, it is unlikely that they will be able to decentralize more tax-raising powers to the municipalities. It does seem, though, that this should be part of later stages in fiscal-federal reforms in Argentina. If the provinces can strengthen their-own revenue-raising capacity, through the implementation of a provincial VAT, then it would be feasible for them to decentralize traditional local tax instruments like the urban property tax or the automobile tax. At some stage, fiscally strong and fiscally independent governments could be established in all three tiers. 78. Individual income tax. In principle, a provincial flat rate surcharge on the federal individual income tax would make sense in Argentina, and this instrument has been used successfully in Canada. The problem is the current low yields on individual income taxes in Argentina, both due to policy (e.g. high minimum income thresholds for exemption from the tax) and administrative difficulties (evasion).55 One possibility in the longer run might be that if payroll taxes and income taxes are merged, the provinces might apply a surcharge to the new unified tax. This might create problems, however, in terms of the principles outlined above, since the taxes would be collected according to where a person works. Depending upon whether the public services consumed by the individual is on the basis of the location of his/her residence or employment would be a critical question. 55 See the Fiscal Dimension,of the Convertibility Plan, chapter 3. -33- The problem of large cities at jurisdictional boundaries (e.g. Buenos Aires) would further blur the issue of allocating the revenues to where the tax payer pays his or her taxes and consumes his or her public services. In brief, McLure believes that a surcharge on the individual income tax is not a promising alternative for Argentine provinces. 79. Excise taxes. McLure argues that excise taxes on tobacco, alcohol and fuels would be ideal candidates for provincial surcharges.56 The taxes could be levied on behalf of the provinces by the federal government, according to where consumption occurs, applying the destination province's rate, with the revenues then delivered to that province. Distinct tax stamps could be used to identify the destination of each individual product. The possibility of cross-border smuggling probably would lead to some convergence of tax rates across provinces. 80. Company income tax. McLure does not recommend provincial surcharges or provincial administration of a company income tax. There are two main reasons. One is that it is not clear how the benefits of local public services match the location of the business. The other reason is the difficulty of assigning a company's income to a particular jurisdiction, (for those companies that do business in more than one province). This can lead to distortions in the location of economic activity and the breakdown of the principles described above. 81. Provincial VAT surcharge: insurmountable problems?. Recall that one of McLure's principles above was based on the idea that most individuals consume provincial services in the location where they consume private goods. As a result, general sales or consumption taxes at the provincial level should be based on the destination of the sale. For international trade, this principle is observed among most VAT countries via zero-rating of exports. The final consumption of the good is then taxed in the country of destination by that country's own tax administration and at the prevailing VAT rate in that country ("deferred payment"). In practice, it seems that most countries zero-rate exports as a way of not imposing additional costs on the export sector, undermining competitiveness. But on a global scale, the practice actually follows McLure's principles for taxation in a federal country: avoid exporting tax burdens. 82. One problem that occurs when the good crosses the border is that the chain of credits57 -- the center-piece of VAT compliance - is broken. In the context of a federal country, one would want to somehow maintain the chain of credits (or a proxy incentive for those credits), in order to improve tax compliance, while still taxing on a destination 56 Actually, McLure recommends that the federal government "vacate" this tax, reducing the federal rate to zero, to allow maximum room for the provincial rate. The tax would still be collected by the federal government, since it is collected at the source of production (which is usually limited to a few specific locations). 57 For readers not familiar with a VAT, the key feature to induce voluntary compliance with the tax is that there is an opposition of interests between the vendor and the buyer. The vendor may not want to report the tax charged, but the buyer wants to report the tax charged since he receives a tax credit for that amount (unless it is a final sale to the consumer). - 34 - basis. In addition, one does not want to create fiscalfrontiers, i.e. border controls, between provinces that would destroy the internal economic union. 83. Another problem that occurs in VAT countries that zero-rate exports (like Argentina) is that firms that largely specialize in exports accumulate credits on domestic purchases that surpass by a wide margin their tax liability on sales. This creates an administrative problem for the tax administration to make cash payments to these firms, and it also creates the possibility of fraud. This type of problem of excess credits would occur at the provincial level if there is a provincial VAT with zero-rating of sales to registered traders (intermediate producers rather than final consumers) in other provinces. Another possible problem with zero-rating sales to registered traders across provinces is evasion due to the diversion of goods to final consumers. 84. One might think then that taxing all sales based on the origin of production would avoid these problems, but then revenues would go to the province where production occurs rather than consumption - violating one of McLure's key principles.58 A possible solution via some form of clearinghouse to allocate the revenues across provinces would be an administratively complicated approach. Below is a description of a proposal that solves most of these problems, and is gaining a variety of adherents. C. ALTERNATIVES FOR A GENERAL PROVINCIAL CONSUMPTION TAX 85. Provincial tax reform efforts in Argentina in recent years have focussed on two highly distortionary taxes: the gross receipts tax, and the stamp tax (sellos). During the early 1990s, the federal government and the provinces reached agreement on a variety of measures to address particular problems in provincial finances and regional economies. These agreements were legalized via two "Fiscal Pacts" (Pactos Fiscales). These pacts promoted a variety of reforms: marginal changes in the revenue-sharing system; completion of the decentralization of health and education, federal guarantees on the minimum aggregate level of revenue-sharing and deregulation of a variety of local economic activities. In addition, the pacts initiated a gradual process of elimination of the gross receipts and stamp taxes. A timetable for the elimination of the gross receipts tax was established, starting with exemptions on primary production as a way of eventually eliminating the tax on all sales prior to the retail stage. The new general sales tax would be a retail sales tax. Many provinces dutifully initiated this process, but the process has stalled, and some provinces have backtracked. The main reason for the inability to finish the process is that it has become increasingly clear that provinces will not be able to administer a retail sales tax, and processes to resolve this problem have not been forthcoming. A similar situation has occurred with the progress in eliminating the stamp tax: there appears to be a demand for the creation of a strong comprehensive tax to replace the traditional tax instruments prior to their final elimination. 58 The Brazilian ICMS is primarily origin-based, generating a variety of problems in interstate transactions (Varsano, 1999). - 35 - 86. Provincial VAT or VAT surcharge: the federal "compensating" VAT as a solution. Active academic debate", originating primarily in Latin America, now regards a provincial VAT or VAT surcharge as a promising alternative for broad-based intermediate-level government taxation. The general features of the provincial VAT that have emerged from the discussion are: (1) destination based; (2) zero-rating of inter- provincial sales by the province of origin with deferred payment in the province of destination (ala Europe); (3) provinces set their own rate; (4) a new federal "compensating VAT" on inter-provincial trade to reduce the incentives for evading the provincial VAT; and (5) continuation of the current regular federal VAT on all transactions (w/ reimbursement of foreign exports). This idea draws upon proposals that have come out of both Brazil and Argentina for an IVA compartido or IVA partilhado, as well as the dual VAT currently functioning in the province of Quebec, Canada. 87. On sales within a province, the provincial VAT would be charged along with the traditional federal VAT. On sales across jurisdictions, no provincial VAT would be charged in the province of origin (zero-rating), but when a final sale (or next stage of production sale) occurs in the province of destination, then the destination province's VAT rate would apply to that sale (deferred payment). To eliminate the incentive for households (final consumers) in other provinces to pose as registered traders, a federal "compensating VAT" (CVAT) rate would be applied to all inter-provincial sales - whether they be to households or registered traders. The "compensating VAT" not only would discourage evasion, but it would establish a chain of credits when goods move across boundaries between registered traders, thus maintaining one of the most desirable features of the VAT.6o The "compensating VAT" seems to resolve almost all of the problems described in the preceding section. In Quebec, no CVAT is used, and the power of strong enforcement of the concurrent federal VAT seems to be sufficient to solve this compliance problem. This Canadian institutional feature is probably not in place in Argentina. 88. Below is a simple numerical example from McLure (1998, See annex). Recall that the CVAT yields no net revenues on sales between registered traders. Actual collection of revenues only would occur in sales to non-registered traders (e.g., mail- order sales to final consumers in other jurisdictions). In this example, 100 pesos worth of value added is produced in each of three stages. The inter-jurisdictional sale occurs from stages 2 to 3. The federal VAT rate is 20 percent, province A's rate is 4 percent, B's rate is 8 percent, and the CVAT rate is 6 percent. 59 Varsano (1995, 1999); Fenochietto (1998A,B,C, 1999); FIEL (1998); Libonatti and Salinardi (1994); McLure (1998, 1999), Bird (1999), Bird and Gendron (1998); Piffano (1997); Gonzalez Cano (1996); Silvani and dos Santos (1995); and Poddar (1990). 60 Varsano (1999) refers to this quality of the CVAT as the "little boat" (o barquinho) in that it carries the good across the river (border), allowing it to arrive at the next jurisdiction free from provincial taxes, but with appropriate crediting for the "compenasating" federal tax (CVAT credit). -36- Table 4: Example of Provincial VAT/CVAT Calculation of Tax Transactions (VAT on sales; credit for input VAT; net VAT liability) (Purchases, One-province (B) Two Province Example Sales, and ale add) Example Prov'1 Federal Tax Prov'l Federal Tax: . Compen- Ordinary Total Tax Tax (A) (B) sating Federal Federal Stage 1: {All stages in B} {This stage occures in Province A} a. Purchases/Credits 0 0 0 0 n.a. n.a. 0 0 b. Sales/Tax 100 8 20 4 n.a. n.a. 20 20 c. Value added/ 100 Net tax this stage 8 20 4 n.a. n.a. 20 20 (c=b-a) Stage 2: a. Purchases/Credits 100 8 20 4 n.a. n.a. 20 20 (d=b) b. Sales/Tax 200 16 40 {Interstate sale from A to B } 0 n.a. 12 40 52 c. Value added/ 100 Net tax this stage 8 20 -4 n.a. 12 20 32 (f&e-d) Stage 3: {This stage occurs in B} a. Purchases/Credits 200 16 40 n.a. 0 12 40 52 (g=e) b. Sales/Tax 300 24 60 n.a. 24 n.a. 60 60 c. Value added/ 100 Net tax this stage 8 20 n.a. 24 -12 20 8 (i=h-g) * Total Tax 24 60 0 24 0 60 60 (j c+f+i) Source: McLure (1998). 89. Another example is presented below. Here a producer of cheap bulk wine purchases grapes from within the province, sells wine to retailers both within the province and in another province, and final consumer sales occur in both jurisdictions. The various assumptions on tax rates appear in the graphic below. These rates are only assumptions used simply to illustrate how the system would work. They should not be viewed as estimates of what the provincial rate would be for the provinces mentioned below. -37- Figure 6: Another provincial VAT/CVAT Example STAGE I Grape farmer in Mendoza sells $100 of grapes to bulk wine producer m Mendoza F VAT VAT" 6e Federal VAT 21 Provincial VAT 6 STAGE 2 Bulk wine producer makes $200 of wine, and sells $100 to a local Mendoza store, and $100 to a store in the Province of Buenos Aires To Buenos Aires store To 100 Mendoza store Tax on Sale Tax on Sale Federal VAT 21 Federal VAT 21 Provincial VAT 6 Provicial VAT 0 Credits Federal CVAT 7 Federal VAT 21 Provincial VAT 6 NET TAX for STAGE 2 Federal VAT 42 - Credits 21 = 21 Provincial VAT 6 - Credit 6 - 0 Federal CVAT 7 = 7 Stage 3 Wine sold to Consumer Retail price of wine is $300, excluding tax (half sold in Mendoza, half sold in Buenos Aires) In Mendoza store, wine sold In Buenos Aires, wine sold to large family for $150 to another large family for $150 i STax on Final Sale to Consumer Tax on Final Sale to Consumer Federal VAT (21%) 31.5 Federal VAT (2 1%) 31 5 Provincial VAT (8%) 120 Provincial VAT (6%) 9 0 Credits Credits Federal VAT 21 0 Federal VAT 21 0 Federal CVAT 7 0 Provincial VAT 6 0 Provincial VAT 0 0 Net Tax This Stage. Net Tax This Stage Federal VAT 10 5 Federal VAT 10 5 Provincial VAT 3 0 Federal CVAT -7 0 Provincial VAT 120 -38- SUMMING UP: Total Tax Collected on $300 of "bulk" wine Federal VAT: Federal Government receives $63 Net Tax in Stage 1 21 (=21 % of $300), as would Net Tax in Stage 2 21 be the situation pre-reform Net Tax in Stage 3 21 Total $63 Federal CVAT: Net Tax in Stage 2 7 Net Tax in Stage 3 -7 Total $ 0 MendozaProvince of Mendoza receives $9 Net Tax in Stage 1 6 (= 6% of the $150 worth of wine Net Tax in Stage 2 6 that is consumed in Mendoza) N et Tax in Stage 3 Total $9 Buenos Aires Provincial VAT Net Tax in Stage 3 12 Province of Buenos Aires receives $12 Total $12 (= 8% of the $150 worth of wine that is consumed in BuenosArs 90. The problem of excess credits. One problem that the above scheme does not resolve is the problem of excess credits. A firm that sells most of its outputs to other provinces, but purchases most of its inputs from within the province will accumulate credits against its home province for the purchases, but it will generate little or no tax liability to be paid to the province (due to zero-rating). It would, of course, have a tax liability to the federal government for both regular federal VAT and the compensating VAT. One solution (which would be facilitated if there were a unique federal VAT administration) is that the firm could use its provincial credits to cancel its tax liability with the federal government. Some form of clearing mechanism would be required then for the federal government to compensate the province for these credits, but this should not be too complicated. It should be noted that the issue of excess credits in the Argentine case would be significantly less important than in the Brazilian case, for example, since the provincial rate would probably be around one-third as high as the federal rate, while in Brazil the state rate is much higher than the federal rate (making the solution described above less feasible). 91. Compensating VAT: rate and distribution of revenues. McLure recommends setting the rate of the compensating VAT at the average of the provincial VAT rates. Setting the CVAT rate at the maximum provincial rate would give the maximum impact in terms of reducing evasion; however, it would burden inter-provincial commerce to low VAT-rate provinces. Setting the CVAT rate at the minimum provincial rate would open -39- up possible evasion, in particular for high provincial VAT rate provinces. McLure concludes that the average provincial VAT rate might be a reasonable CVAT rate. Another detail that McLure discusses is the distribution of CVAT revenues. Inter- provincial sales to registered traders would produce zero net revenues for the federal tax collector, since credits would cancel with liabilities. Inter-provincial final sales to households, however, would generate revenues, and the question arises of how to distribute these revenues. McLure suggests that the revenues could be distributed across provinces according to the net provincial revenues collected on intra-provincial sales. Varsano (1999) agrees with this approach. Box3: 'Gross Receipts Tax:and.theiMultilateral Agreement (ConvenidMultilateriAl). The:gross receipts:tax 'isilevied:on:gross sales,declared by ifir?ngAs aresult, a that'doeijbusines invaious!jubshictions,- isld liv't%tlgoi ilitieiiiltlse junrsdici s fs 46oidisulinimultipleitaxation could hIive;its'total- gr6ss,ls-txb:I dit'4s4ii' flle 'Oeyond-the problem of cumilativeitaxationof this"turnover"style tax)An greement;was reahed in the mid- I 950s to establish a nilelfordistiibutingithe gross receipt of a pitticuiarmenterprise across the jurisdictions.in which it operates.?Theagreement was'modified over the years, ,but'thirbasic ruleexisting todayis that the,assignmenofth tax base4si made fifty percent iiorioii\to.thelocation ofcmpany expeniditures andfiftypercent in proportion'to the receipts,;generated inieach jurisaiction..The:former is fairly straightforward to calailate; however, the latter calculation tan bebased onseveraFldifferentciiteria: '(i).the-location, where the sale:is agreed to, generating the.right to receive income; (2) thelocation of the deliverloft (3)itheflocationidentifiedbyithe?buyenas theIo ationlwhere economic-use,or final consumption of the godd-occdrs:4hi generihtist critenon h ln lidohie theisite has inspired,Iengthyand frequenilitigation before the Arbitration .Commissioh of theMidfilaterab[Agreement 'It remains a subject for interpretation and debate. Source;' FiEI(1998),'Chapter 10 -`byxQscar Libonatti 92. Administrative Issues. McLure also discusses some of the administrative complications that might.arise in such a system, as well as the question of a possible provincial role in the administration of the tax (instead of a pure surcharge on the federal tax). Both federal and provincial VATs, for example, could be paid using the same tax declaration form, with the provincial obligation passing directly to a provincial account and the federal obligation to the federal account. There are advantages and disadvantages to including provincial tax administrations in the enforcement of one (or both) VATs. One advantage of a provincial administration, or provincial participation in enforcement, is that it would help establish greater sense of autonomy and ownership for the provinces.61 One clear advantage with a single federal administration is to avoid duplication of efforts and to relieve the tax payer from having to respond to two different sets of institutions. It may also simplify the resolution of the issue of net credits. One compromise that McLure proposes is the possibility of converting the federal government's tax agency into more of a federalist agency, in terms of involving the provincial level of the federation in the supervision of this agency. 93. Under the system described above, businesses would need to keep track of only three types of transactions: (1) within province sales, (2) extra-provincial sales to 61 In Quebec, it is the province that collects both federal and provincial VATs (Bird and Gendron, 1998) - perhaps for political reasons, while in Germany, it is the lander that collect the federal VAT as is the case for Mexican states (Poddar, 1990). -40 - registered traders or non-registered traders (consumers), and (3) sales to foreign countries. For the first type of transaction, federal VAT and provincial VAT rates apply. For the second type of transaction, federal VAT and federal CVAT apply, and for the third type of transaction, everything is zero-rated. For inputs, federal VAT and CVAT credits are registered for all purchases from other provinces, and federal VAT and local provincial VAT credits are registered for purchases of inputs from within the province. Federal and provincial VAT tax forms could be on the same sheet.62 All in all the administrative burden on business would be relatively light - certainly when compared to the numerous forms that would have to be filled out if there were a clearinghouse, or if businesses had to charge each destination province's VAT on inter-provincial sales, rather than the unique CVAT. In addition, the system would be much less complicated than the current arrangements for gross receipts tax on multi-jurisdiction businesses (see Box 3). 94. The combined dual VAT rate would be quite high, given that the current federal VAT is 21 percent. Adding another 5 to 7 percent results in a striking total burden -- a total rate that could encourage increased evasion. It should be noted that the overall tax pressure on the economy would not change, since the provincial VAT would replace the current gross receipts tax (and stamp tax), and government simulations of the new provincial VAT rate are based on the need to replace only those revenues. Future tax decentralization (as mentioned below) would only occur in conjunction with the lowering of the federal rate in order to make room for a higher provincial rate (and the provinces' giving back points of coparticipation to compensate the federal level). One potential political problem might be the greater visibility of the provincial VAT, with respect to the gross receipts tax. 95. Another issue to consider is whether upper and lower limits should be placed on the rates that provinces choose. The lower limit might be important to avoid tax rate "wars" among the provinces to attract businesses. The fact that the IVA compartido would be destination-based, rather than origin-based, would already limit the scope for a fiscal "war." Varsano (1999) makes the point that fiscal battles among the states in Brazil were to some extent "financed" by exporting tax burdens to the rest of the country via the origin-based VAT currently in place in Brazil. For prudence sake, however, a lower bound on the rate could be considered. 96. Evaluating the Alternatives. Any final evaluation of the prospects for the provincial VAT should be taken in the context of all the possible alternatives. There are five main alternatives: (1) Continue with the gross receipts tax (and stamp tax); (2) Implement the provincial VAT/ CVAT idea as a substitute for the gross receipts tax; (3) Implement a retail sales tax as a substitute for the gross receipts tax; (4) Eliminate the gross receipts tax and make the provinces even more dependent on federal revenue- sharing (with federal VAT or some other federal tax increasing to make up for the lost revenues, overall); (5) Abandon general sales taxation at the provincial level and implement a provincial surcharge on the income tax (or payroll taxes), or some other tax. 62 See Silvani and dos Santos (1996). -41- The first option implies the economic distortions from this tax would continue to hamper the productivity of Argentine businesses. At some point, an empirical investigation, estimating the economic costs of this tax in Argentina should be conducted, including reduced inter-provincial trade from the complicated administrative procedures involved. With regards to the retail sales tax, a consensus view is emerging that for practical administrative reasons, it would be difficult to implement in Argentina. Eliminating the gross receipts tax and making the provinces more dependent upon revenue-sharing would cause a further deterioration in the low degree of correspondence and fiscal autonomy of the provinces. Option (5) does not appear promising, since the VAT is the only tax in Argentina (other than payroll taxes) that generates a significant share of GDP in revenues. If there are no clear reasons why the provincial/federal dual VAT would not work in Argentina, the government should move ahead with plans for implementing it as the most viable option. D. DECENTRALIZATION OF ADDITIONAL TAX POWERS 97. Two potentially strong tax instruments were discussed above as candidates for decentralizing greater tax authority to the provinces. One was a surcharge on federal excise taxes: tobacco, fuels and alcohol. The practical issue is one of how to allocate the revenues. Once again, given that these revenues would finance public services primarily in the location where the tax payer resides, it seems that the best solution would be to allocate the revenues according to where the goods are consumed. At the location of production, an identifier as to the location where goods are shipped would then identify where the revenues from that sale should go. The second strong tax instrument would be the provincial VAT described above. If successful as a substitute for the gross receipts tax, it could be used as a way to increase provincial fiscal autonomy with the federal government vacating tax room to the provinces, in exchange for a smaller share of revenue-sharing. It should be noted that a third instrument that in principle could be successful is a surcharge on the federal personal income tax. The main reason for limited enthusiasm for this instrument is that it continues to be underused at the national level: via a combination of exemptions and poor compliance. Despite this short-run problem, a provincial surcharge on the federal personal income tax should remain an interesting option for the future reform agenda. 98. The basic trade would be a new tax instrument for the provinces in exchange for a reduced percentage of the primary distribution of the revenue-sharing pool. The latter would be necessary so as not to flood the provinces with revenues, and also to compensate the federal government, since it would have to lower its own tax rate in order to make room for provincial surcharges. The negotiations could be difficult. The decentralization process probably would have to be coordinated so that all provinces agree to the process. Otherwise, it would become difficult to calculate how the provincial-federal shares of the revenue-sharing pool should be adjusted to make up the difference. Even with all the provinces agreeing, one would need to calculate the expected yield on the decentralized tax, assuming ex ante an average tax rate, and using an historical compliance rate (at the federal level). Then one could negotiate the appropriate percentage reduction in the primary share, so that neither level of government -42- loses and the current vertical balance is maintained. Note that, over time, the agreement could be subject to dispute if the decentralized instrument is less buoyant than the taxes feeding the revenue-sharing pool. This is not likely to be an issue for the VAT, but it could be an issue for excise taxes. 99. Sequencing. The sequencing of reforms is another critical element. Experimenting with the provincial VAT as a substitute for the gross receipts and stamp taxes may be the key to later decentralization of tax powers. In addition, decentralization of tax powers may only be viable if a new system of revenue-sharing is in place - to assist in a reasonable way those provinces that will not be able to collect taxes at a level comparable to national norms. There is a strong argument, then, for proceeding with the replacement of the gross receipts tax and the creation of new criteria for revenue-sharing, to be followed thereafter with greater decentralization of tax power. 100. With decentralization of greater revenue-raising capacity -- via excise taxes, the IVA Compartido and eventually, the personal income tax - the problem of fiscal correspondence would be reduced. Simultaneously, the accountability of provincial governments to their local tax payers would increase. At this stage, unconditional block transfers to the provinces (via revenue-sharing) may become irrelevant - with the exception of the fiscal equalization transfers discussed in Chapter Two. These would continue to be important for provinces to be assured that tax decentralization does not leave behind the low per capita tax base provinces. At this later stage, with fiscally strong provinces, the central government may still choose to enact specific, conditional transfers for supporting additional provincial actions in those specific areas where there is a clear national interest. 4. CONCLUDING COMMENTS 101. The last revision of the revenue-sharing law occurred in 1988. The Argentine economy and the Argentine public sector have undergone dramatic transformations since the late 1980s. The new Argentine economy is no longer dominated by state-owned enterprises in key sectors, and significant trade liberalization and deregulation have 63 occurred. Through this process, the public sector can now focus on its core functions -- assuring basic social services and infrastructure for the poor, attending to pure public goods of security and public safety, and regulating markets following economic principles. The severe macroeconomic instability of previous decades led to a consensus for this more focussed public sector. Within the more focussed public sector, the provinces have a key role in protecting property rights and in providing the kinds of 63 See World Bank, The Convertibility Plan- Assessment and Potential Prospects. -43 - services - education, health, and the most basic infrastructure - that will assure a broader participation in the gains from the new Argentine economy. 102. A consensus also has been achieved for a key commitment mechanism for the federal government to refrain from excessive intervention in the economy: the convertibility law establishing a currency board arrangement. This arrangement severely curtailed the government's ability to finance fiscal deficits from money creation. It also severely restrained the monetary authority from using expansionary policy for short-term growth, or for shifting the exchange rate to improve international competitiveness. These factors shifted the focus of policy debate to improving economic productivity by minimizing distortions created by the tax system, removing distortions in factor markets, and maximizing the long-run development impact of public expenditures. In this context, the country can ill afford to have a system of transfers that sends excessive sums to particular areas of the country, where the funds are spent inefficiently. 103. Without digressing too far, the provinces and the system of fiscal federalism have a key role to play in the process of establishing sustainable growth and "market- preserving" federalism; that is, a federal system in which all levels of government are committed to "honoring economic and political rights."64 The Convertibility Law has helped establish some of the conditions: for example, limiting the ability of the Central Bank from saving bankrupt provincially owned banks.65 This has already had an impact on establishing firmer budget constraints in the provinces. In Weingast's (1995) analysis, a key element to assuring limited intervention by sub-national governments is competition. The current system of financing provincial governments in Argentina fails in this regard. With many provinces dependent upon transfers for 70 to 90 percent of their revenues, they are not required to compete for the economic base that can be taxed to finance local services. Greater decentralization of tax authority might provide part of the incentive to compete on the basis of efficient local economic regulation, or on the basis of quality public services provided at reasonable cost. 104. It would be virtually impossible, however, to build a consensus for moving to a more competitive environment, without providing assurances of a more equal starting point for all the provinces. In addition, it will probably be impossible to completely close the vertical imbalance through tax decentralization. Here enters the role of a well- designed and predictable system of intergovernmental transfers. The proposed reforms under consideration by the government follow reasonable criteria for determining the structure of these transfers. The technical evaluation of alternatives is far advanced: it will now be a question of building the consensus for a final structure. 105. Finally, no matter what structure is reached today, it may be necessary to review the system and adjust it according to changing circumstances. The creation of the federal fiscal organism could be a critical institutional arrangement for reviewing the rules for revenue-sharing and provincial tax assignment on a regular basis in the future. 64 Weingast (1995). 65 See Dillinger and Webb (1999) and Saiegh and Tommasi (1998) -44- 5. REFERENCES Alemann, R.T., 1992, Breve Historia de la Politica Econ6mica Argentina: 1500-1989, (Editorial Heliasta: Buenos Aires). Bird, R.M., editor, 1980, Fiscal Dimensions of Canadian Federalism, Financing Canadian Federation: 4, Canadian Tax Foundation: Toronto. Bird, R.M., 1999, "Rethinking Tax Assignment: The need for Better Subnational Taxes," Seminar presented at the World Bank, April 27, 1999, Washington D.C., processed. Bird, R.M., and Pierre Paul Gendron, 1998, "Dual VATs and Cross-Border Trade: Two Problems, One Solution?", International Tax and Public Finance, vol. 5, pp. 429- 442. Boadway, R.W. and Paul A.R. Hobson, 1993, "Intergovernmental Fiscal Relations in Canada," Canadian Tax Paper No. 96, Canadian Tax Foundation: Toronto. Boadway, R., Sandra Roberts, S., and Anwar Shah, 1994, "Fiscal Federalism Dimensions of Tax Reform in Developing Countries," Policy Research Department Working Paper, No. 1385, The World Bank, Washington D.C. Brown, C. and Wallace Oates, 1987, "Assistance to the Poor in a Federal System," Journal ofPublic Economics, Vol. 32, pp. 307-330. Courchene, T.J., 1998A, "Fiscal Federalism in Argentina: Towards a Nation-Province Fiscal Covenant," Background paper prepared for the World Bank (see Annex). ---------------, 1998B, "Renegotiating Equalization: National Polity, Federal State, International Economy," Commentary No. 113, C.D. Howe Institute, Toronto. ---------------,1983, "Canada's New Equalization Program: Description and Evaluation," Canadian Public Policy, Vol. 9, No. 4, pp. 458-75. Dahlby, B. and L.S. Wilson, 1994, "Fiscal capacity, tax effort, and optimal equalization grants," Canadian Journal of Economics, Vol. 27, No. 3, pp. 657-72. Dillinger, W. and S. Webb, 1999, "Fiscal Management in Federal Democracies: Argentina and Brazil," Paper to be delivered at the Fifth Annual Bank Conference on Development in Latin America, Valdivia, Chile. -45- Fenochietto, R., 1998A, "El IVA compartido, su funcionamiento y la autonomia provincial, Chief of Cabinet's Office, Buenos Aires, processed. ----------------, 1998B, "Proponen la creaci6n de IVA "compartido," Commentary in the newspaper Ambito Financiero, June 2, 1998. ----------------, 1998C, "The shared value added tax: A useful tool to replace the gross receipts tax and the decentralization of taxes," Chief of Cabinet's Office, Buenos Aires, processed. FIEL (Fundaci6n de Investigaciones Economicas Latinoamericanas), 1998, La Reforma Tributaria en la Argentina, Buenos Aires. ---------, 1993, Hacia una nueva organizaci6n del federalismo fiscal en la Argentina, Buenos Aires. Frey, B. and R. Eichenberger, 1996, "To harmonize or compete? That's not the question," Journal ofPublic Economics, Vol. 60, pp. 335-349. Gonzalez Cano, H., 1996, "La reforma tributaria de Brasil y posible aplicaci6n del nuevo IVA federal y estadual para el reemplazo del impuesto a los ingresos brutos," Boletin de la DGI, No. 513, pp. 1391-97, Buenos Aires. Inman, R.P. and Daniel L. Rubinfeld, 1996, "Designing tax policy in federalist economies: An overview," Journal of Public Economics, Vol. 60, pp. 307-334. McLure, C.E., 1998, "Tax Assignment in Argentina: Conceptual and Practical Issues," Paper prepared for the World Bank (see Annex). ---------, forthcoming, "Protecting Dual VATs from Evasion on Cross-Border Trade: An Addendum to Bird and Gendron," International Tax..... Ministry of Economy, various issues, Boletin Economico, Buenos Aires. ----------, various issues, Boletin Fiscal, Buenos Aires. ----------, 1998, Informe Economico Regional, July and November issues, Buenos Aires. Oates, W., 1972, Fiscal Federalism, (Harcourt Brace Javonovich: New York). Oates, W. and Robert M. Schwab, 1988, "Economic competition among jurisdictions: efficiency enhancing or distortion inducing?," Journal ofPublic Economics, Vol. 35, pp. 333-54. Pauly, M., 1973, "Income redistribution as a local public good," Journal ofPublic Economics, Vol. 2, pp. 35-58. -46 - Piffano, H., 1997, "El nuevo sistema tributario provincial en un escenario de descentralizacion tributaria," Paper presented at the Second International Seminar on Federalism and Local Governments, National University of La Plata. Piffano, H., 1998, "Un ejercicio de simulaci6n de la propuesta pare el acuerdo fiscal federal," Paper presented at the Third International Seminar on Federalism and Local Governments, National University of La Plata. Poddar, S., 1990, "Options for a VAT at the State Level," in Malcolm Gillis, Carl S. Shoup, and Gerardo P. Sicat, Value Added Taxation in Developing Countries, (The World Bank: Washington D.C.). Saiegh, S and M. Tommasi, 1998, "Argentina's Federal Fiscal Institutions: A case study in the transaction-cost theory of politics," Paper delivered at the UNDP conference on "Modernization and Institutional Development in Argentina," Buenos Aires. Sanguinetti, P., 1994, "Intergovernmental transfers and public sector expenditures: a game-theoretic approach," Estudios de Economia, Vol. 21, No. 2. Sanguinetti, J. A. Zentner and R. Rotztein, 1997, "Disciplina Fiscal y Endeudamiento Provincial, Se Puede Confiar en el Mercado como Disciplinador?, Ministry of Economy Discussion Paper, Buenos Aires. Schwartz, G. and C. Liuksila, 1997, "Argentina," in T. Ter-Minassian, ed., Fiscal Federalism in Theory and Practice, (International Monetary Fund: Washington D.C.). Shah, A., 1994A, "A Fiscal Needs Approach to Equalization Transfers in a Decentralized Federation," World Bank Policy Research Working Paper No. 1289. Shah, A., 1994B, "The reform of intergovernmental fiscal relations in developing and emerging market economies," Policy and Research Series, No. 23, The World Bank, Washington D.C. Silvani, C. and P. dos Santos, 1996, "Administrative aspects of Brazil's consumption tax reform," International VAT Monitor, Vol. 7, pp. 123-32. Skidmore, T. E. and P.H. Smith, 1984, Modern Latin America, (Oxford University Press: New York). Ter-Minassian, T. and Jon Craig, 1997, "Control of Subnational Government Borrowing," in T. Ter-Minassian, ed., Fiscal Federalism in Theory and Practice, (International Monetary Fund: Washington D.C.). Turnbull, G.K., 1998, "The Overspending and Flypaper Effects of Fiscal Illusion: Theory and Empirical Evidence," Journal of Urban Economics, Vol. 44, pp. 1-26. -47 - Varsano, R., 1999, "Subnational Taxation and the Treatment of Interstate Trade in Brazil: Problems and a Proposed Solutionl," Paper to be delivered at the Fifth Annual Bank Conference on Development Economics in Latin America, Valdivia, Chile. Varsano, R., 1995, "A Tributagdo do Com6rcio Interestadual: ICMS versus ICMS Partilhado," Texto para Discussdo No. 382, Instituto de Pesquisa Econ6mica Aplicada (IPEA), Rio de Janeiro. Weingast, B.R., 1995, "The Economic Role of Political Institutions: Market-Preserving Federalism and Economic Development," The Journal ofLaw, Economics, and Organization, Vol. 11, No. 1. Wildasin, D., 1991, "Income Redistribution in a Common Labor Market," American Economic Review, Vol. 81, No. 4, pp.757-74. World Bank, 1998, Argentina: The Fiscal Dimension of the Convertibility Plan, Report No. 16996-AR. World Bank, 1996, Argentina: Provincial Finances Study - Selected Issues in Fiscal Federalism, Report No. 15487-AR. World Bank, 1996, Argentina: Reforming Provincial Utilities - Issues, Challenges and Best Practice, Report No. 15063-AR. World Bank, 1996, Argentina: Transport Privatization and Regulation - The Next Wave of Challenges, Report No. 14469-AR. World Bank, 199, Argentina: Second Generation Reforms, Report No. World Bank, 1996, Argentina: The Convertibility Plan - Assessment and Potential Prospects, Report No. 15402-AR. Zapata, J.A., 1997, "Sistema de Coparticipaci6n Federal: Cambios para una mayor eficiencia fiscal," Background Paper for the World Bank Report, The Fiscal Dimension of the Convertibility Plan. -48- Annex 1: Calculation of Equalization Transfers Tax Bases Per Capita MCBA B. Aires Catamarca Cordoba Corrientes Chaco Chubut Entre Rios Formosa Jujuy La Pampa La Rioja Mendoza Mislones Automobile 29680 15493 7275 1702.8 7707 7198 18016 14671 5347 660.3 20991 929.2 15841 8619 Gross Receipts 290077 9127.9 5561 2 9692 7 4656 5 43985 118385 73284 36250 4871 5 10008.6 72994 9462.6 49695 Property 12454.4 98536 8091 5 97316 77976 78562 94247 9120.6 75185 73827 104268 75126 82600 77191 Royalties 00 00 0.0 00 103 00 2198 5.8 6.4 05 377 00 531 58 Neuquen R. Negro Salta S. Juan S. Luis Sta. Cruz Santa Fe S. del Estero T. del Fuego Tucuman Total Automobile 1306.1 1364.3 654.9 1238.1 11964 1897.6 16727 540.3 2007.8 757.4 1517.7 Gross Receipts 142667 8607.7 5126.7 6235.0 15148.6 16465.9 9943.9 3427.0 32775.8 5450.3 10236.2 Property 8605.8 9346.1 6912.8 7533.6 8466.0 8958.6 9068.8 7659.3 12365.1 7336.2 93586 Royalties 689.8 69.6 32 4 0.0 0.0 879.8 0.0 0.0 407.6 0.0 23 5 - 49 - Equalization Transfers (Arg$ per capita) MCBA B. Aires Catamarca Cordoba Corrientes Chaco Chubut Entre Rios Formosa Jujuy La Pampa La Rioja Mendoza Misiones Automobiles -239 -05 130 -31 123 132 -47 08 162 141 -96 97 -1 1 108 Gross Receipts -2535 150 631 73 753 788 -216 393 893 724 31 397 104 71 1 Property -157 -25 64 -1 9 79 76 -03 12 93 100 -54 94 56 83 Royalties 235 235 235 235 132 235 -1963 177 171 229 -142 235 -297 176 Total Transfers per Capita 0 354 1061 259 1088 1231 00 590 1319 1196 00 822 00 1079 Total Transfers (Arg$ millions) 00 4853 31 9 774 961 1131 00 641 619 687 00 214 00 1001 Neuquen R. Negro Salta S. Juan S. Luis Sta. Cruz Santa Fe S. del Estero T. del Fuego Tucuman Total Automobiles 3.5 25 14.2 46 53 -63 -26 161 -81 12.5 00 Gross Receipts -544 220 69.0 540 -663 -841 39 919 -3044 646 00 Property 38 01 124 93 45 2.0 15 86 -15.3 10.3 0.0 Royalties -666.3 -462 -8.9 23 5 23.5 -856.3 23.5 23 5 -384.1 23.5 00 Total Transfers per Capita 00 00 86.7 91.4 0.0 00 263 140.2 00 110.9 Total Transfers (Arg$ millions) 0.0 00 868 51.6 00 00 792 996 00 138.3 15757 - 50 - ANNEX 2: CONSULTANT'S REPORTS (1) Thomas J. Courchene (2) Charles E. McLure, Jr. FISCAL FEDERALISM IN ARGENTINA: TOWARDS A NATION-PROVINCE FISCAL COVENANT by Thomas J. Courchene Queen's University Kingston, Canada A Report Prepared For the World Bank September 1998 I: INTRODUCTION With roughly 50 percent of overall public sector spending occurring at the provincial level, Argentina is the most decentralized country in Latin American and would rank high in terms of federations elsewhere (e.g. the United States).' On the own-source revenue side, however, the picture is very different. The Argentinean provinces, on average, collect taxes equal to only 35% of their expenditure requirements, with 65% coming from transfers from the nation. Moreover, this 35 percent average masks a widely variable experience across provinces. Setting aside the city of Buenos Aires, the share of own-source revenues as a percent of total revenues ranges from over 50% for the province of Buenos Aires to less than 10 percent for La Rioja. In addition, those states that do raise a significant proportion of their revenues frequently do so via taxes, such as Ingresos Brutos, that incorporating significant "tax exporting" to other provinces. As McLure (1998) and other fiscal federalism scholars have repeatedly noted, such origin-based taxes are not appropriate for sub- national governments. Beyond this issue, the Argentinean tax-sharing agreement (Coparticipati6n Federal de Impuestos) embodies a distribution rule that provides per capita revenues for some provinces that are several times the value of per capital revenue for other provinces. Not only is there no equity rationale for the status quo but, as well, the allocative and incentive implications are serving neither the provinces nor the nation well. In light of these and other dysfunctional features of the operations of Argentinean fiscal federalism, the purpose of the present paper is to work towards the design and implementation of a new nation-province Fiscal Covenant. As elaborated below, the key features of this Covenant are a) reducing the degree of vertical fiscal balance via the devolution of taxing powers; b) addressing 1 the horizontal fiscal balance across provinces via a comprehensive equalization program; c) embedding an internal economic union and a set of national social policy principles as an integral part of the Covenant; and d) providing generous transitional measures to ensure that, in absolute terms, no province will suffer vis-i-vis the status quo. In more detail, the paper proceeds as follows. Part II presents a series of basic principles or precepts, drawn from the fiscal-federalism literature, that represent- "best-practice" in terms of financing federations. Part III then presents the "guiding principles" underlying the Covenant. Part IV elaborates on the broad design and the key components of the Covenant, with Part V providing some additional detail. A short conclusion completes the paper. II: PUBLIC FINANCE PRINCIPLES AND PRECEPTS There are several sets of principles applicable to financing a federation. I shall deal, in turn, with four of these - principles for allocating taxes and expenditures; principles relating to what has come to be referred to as "market-preserving federalism;" basic principles relating to all public finances, and, finally, a set of principles relating to intergovernmental transfers. 1. Principles for Allocating Expenditures and Taxes Expenditures Increasingly, the guideline evoked in terms of which expenditures might be assigned to sub- national government is the principle of subsidiarity. The principle of subsidiary states that government should be as close as possible to citizens: powers or competences should be delegated to the lowest level of government where they can be effectively exercised. Of and by itself, this 2 imparts a bias toward decentralization. However, if the nature of the service or the activity means that it cannot be carried out efficiently at the local level, then a higher level of government should assume responsibility. The presence of cross-provincial policy "spillovers", or "expenditure externalities", for example, would imply the need for an upward shift of the policy area. But upward need not mean central: it could also mean interprovincial or federal-provincial. Given that the degree of decentralization varies quite dramatically across federations, many other factors (history, culture, homogeneity of regions, etc.) influence the choice where a federation will locate itself along the centralization-decentralization spectrum in terms of expenditures. As noted earlier, however, Argentina is already the most decentralized nation in South America, so it may wish to carry the principle of subsidiarity to significant lengths. Taxes: In terms of the allocation of taxes to sub-national governments, the four general principles enunciated by McLure (1998) have considerable merit. - the tax base should be relatively immobile, in order to allow provincial authorities some leeway in varying tax rates without losing their tax base, and it should be visible to ensure accountability; - the tax yield should be adequate to meet local needs, i.e. it should expand roughly in line with expenditure responsibilities; it should not be possible to export the tax burden to non-residents; it should be perceived by taxpayers to be reasonably fair and should be easy to administer. Transferred to the real world, this means that broad-based resident taxation is appropriate. This would include general sales taxes (at the retail or destination level), as well as a share of 3 personal income taxatioi along, say, Canadian tax-sharing lines. Recently, however, based on the experience of Quebec and recent research by South American tax experts (summarized by McLure (1998)), the values added tax (VAT) may well be an appropriate sub-national tax. What these principles rule out are taxes like corporate income taxation (where the tax base is highly mobile) and the various "origin" taxes (which allow for tax exporting across provinces). 2. Market-Preserving Federalism Drawing from research by Weingast (1995), Stanford's Ron McKinnon (1998) outlines four principles that will ensure that he calls "market-preserving federalism": * monetary separation; * fiscal separation; * freedom of interstate commerce; * unrestricted public choice. These principles are presented in the EU context, so that in their original form they are not fully appropriate for Argentine federalism.2 Hence, they will be modified somewhat to meet the needs of federal systems. With respect to monetary separation, the key issue is that the provincial governments not own or control commercial banks that allow it to discount freely with the bank. With the privatization of several provincial banks, the situation in Argentina is improving considerably, although the activities of the Banco de la Naci6n still remain offside of this principle. Fiscal separation means that access of provincial governments to "discretionary" or "additional" central government financing (to cover provincial deficits) must be eliminated. In tandem, these two principles will ensure that there are no "bail-outs"; that is, provincial 4 governments will face hard budget constraints. Provinces can still borrow, but financial markets will be more vigilant if they realize that the provinces now face hard budget constraints. McKinnon's third principle is the freedom of interstate [interprovincial] trade, i.e., no provincial restraints on the movement of goods, business firms, people or capital across provinces. I interpret to this principle as preserving and promoting the internal economic union. It would preclude, for example, the assigning of origin-based taxes to the provinces. With these three principles in place, McKinnon argues for unrestricted public choice, by which he means that provinces can, within the above framework, design and deliver alternative bundles of public services and finance them by alternative approaches to taxation. This will ensure that the horizontal competition (in either public goods provision or taxation) will be efficient and presumably welfare enhancing. 3. Basic Public Financing Principles The third set of principles relates to precepts that ought to inform taxation and spending at all levels of government. These would include: - Accountability: Governments must be accountable to citizens for the prudent use of public monies and the programs that incorporate these funds. In part, enhanced accountability means a clarifying of the expenditure roles of the two levels of government. Suppose that expenditure category X is assigned to the provinces. Citizens unhappy with the way that a given province approaches this expenditure should not be able to appeal to the central government to "rectify" the situation, or else there will be no meaningful accountability. It is, of course, appropriate for the national government to impose some principles in terms of how selected provincial expenditures are allocated. But within this framework, provincial governments have to be accountable to their own citizens for their spending and taxing decisions. - Transparency: This principle is closely related to accountability. If programs are not transparent, accountability will become blurred. In any new fiscal Covenant, it is important that provisions are transparent so that citizens and government alike can recognize 5 where accountability resides. * Efficiency: Given the fiscal burden at all levels of government, efficiency becomes a virtue. Beyond the obvious, namely ensuring that no more than the essential amounts of funds are expended in terms of achieving a particular policy goal, there are other facets to efficiency. One is the appropriate matching of policy instruments to policy objectives: if the objective is distributional, then it should be delivered via a distributional instrument (e.g., the tax-transfer system), not via an allocative instrument. Another is that where intergovernmental transfers are involved, the incentives should not be such as to encourage what elsewhere (Courchene, 1994) 1 have labelled "intergovernmental gaming" (for example, the incentives under existing legislation for provincial governments in Canada to create make-work projects to transfer citizens from provincial welfare to federal unemployment insurance). Finally, the elimination of duplication and overlap has an obvious efficiency component and in the process it probably also contributes to enhanced transparency and accountability. . Fiscal Correspondence or Fiscal Coincidence: Other things equal, the jurisdiction responsible for spending funds should be the one responsible for raising these funds in the first place. The existence of intergovernmental transfers is an obvious exception to this principle, an exception that can be rationalized in a variety of ways (e.g., the economies of scale in tax collection; transfers designed to offset horizontal fiscal imbalance (equalization) and transfers designed to ensure that provinces supply the appropriate level of what the society views as "merit" public goods and services). However, in these circumstances, the incentives within the transfer system should be as consistent as possible in terms of isolating the locus of responsibility and, therefore, accountability. Phrased differently, the incentive should ensure that province X cannot pass off the costs of its spending decisions to other provinces or to the nation. The Federal Principle: While this principle is inherent in McKinnon's market- preserving federalism, it merits re-emphasis here. Subject to adhering the provisions of the Covenant, the provinces must have the flexibility to design and deliver their own vision and version of the socio-economic envelope. Economists typically refer to this as competitive federalism. In Canada, the most cited exemplar here is the experimentation in Saskatchewan which led to the national Medicare (health care) program. Recently, this province has substituted free drugs to the elderly with a system based on ability to pay. Other provinces are following suit. The more general point is that the on-going blossoming of provincial experimentation across a range of fronts is absolutely critical to creating an efficient and viable social infrastructure. Winning strategies in one province will be adopted by other provinces. The policy challenge here is to ensure that this experimentation takes place within a framework of "national" (federal, federal-provincial or interprovincial) norms or principles. In other words, it must be consistent with market-preserving federalism. 6 4. Intergovernmental Transfer Principles Finally, it is important to enunciate a set of principles relating to intergovernmental transfers. The first of these relates to fiscal equity. All provinces should have access to a level of overall revenues such that they can provide reasonably comparable levels of public services at reasonably comparable tax rates. This wording is embodied in the Canadian Constitution. Other federations employ a similar rhetoric. For example, the German Basic Law refers to the "uniformity of living conditions" which is now being altered to "equivalency of living conditions." In most federations, addressing this horizontal fiscal imbalance across provinces is the role of the equalization program. Several further implications derive from this overarching principle: (1) Equalization need not be limited, as it is in Canada, to equalizing revenues. Both the German and Australian federations also equalize on the expenditure side, i.e. they incorporate expenditure needs as well as revenue capacity. However, they do so in opposite ways. Australia's needs equalization puts heavy emphasis on low-population density, presumably on the argument that it is more expensive to deliver a given bundle of goods and services to a sparsely-populated region. In contrast, in the German model, needs are deemed to be high in densely populated areas, presumably reflecting the fact that wages, rents, etc., are higher in densely populated regions and, therefore it is more costly to deliver public goods since most provincial expenditures are related to either wages or living conditions. Thus, there is no "ideal" model to draw upon when equalizing expenditures. (2) Fiscal equity does not focus on the distribution of transfers, per se. Rather, the objective is to ensure that total provincial revenues (i.e. own revenues plus transfers) conform to the equity criterion. Fourth, the transfer literature tends to be cast in a unitary state framework. Implicitly, and often explicitly, it argues that the role of intergovernmental transfers is to "convert" a federation into a version of a unitary state, i.e. ensuring that net fiscal benefits from all governments are identical 7 across citizens. This is a complex issue, too complex to detail here. Accordingly, I have appended a recent paper (the appended C.D. Howe Institute Commentary) that focusses in detail on the questions that are at issue here. Drawing from this appended paper, there are two federal rationales for equalization payments, and intergovernmental transfers generally: - The constitutional rationale. For a federal system to be meaningful, each level of government must have access to revenues sufficient for it to carry out its constitutional responsibilities. In most federations, this requires transfers from the central government to the sub-national governments. By their nature, these transfers should be unconditional. This is the constitutional or federal rationale for intergovernmental transfers. - The citizenship rationale. In modem economies, citizens ought to have rights to basic levels of certain services (education, health, etc.). If these services fall under provincial jurisdiction, there is then a citizenship rationale for intergovernmental transfers. And until these "merit" goods become well established, there is a strong case for making the transfers conditional on the delivery of these "merit" goods and services. The final transfer principle relates further to these "merit goods". Most federations do not attempt to fully eliminate vertical fiscal imbalances. Rather they hold back a portion of transfers to be allocated on conditional terms (either in terms of where these monies are spent (specific purpose transfers) or how they are spent (a set of principles relating to spending). For example, Australia's specific purpose payments exceed its unconditional payments. Canada has now eliminated all of its specific purpose transfers but still maintains a set of social policy principles relating to how provincial monies are to be spent in terms of health and welfare. For example, the principle associated with welfare (social assistance) in Canada states that there shall be no residency requirement for out-of-province citizens in terms of accessing a province's welfare system. This is more of an "internal market principle", rather than a social policy principle per se. In any event, 8 most federations attach some conditions on portions of their intergovernmental transfers. It is probably important to ensure that these principles are not overly restrictive. For example, they should allow for equivalent approaches to satisfying the principles in order not to frustrate the advantages that will arise from the process of competitive federalism. 5. Summary While not intended to be exhaustive, the above four sets of principles relating to the operations of fiscal federalism are sufficient to inform the design of a new nation-province Fiscal Covenant for Argentina. The remainder of this paper is devoted to articulating a vision and version of such a Covenant, beginning with the overall guiding principles. III: A FISCAL COVENANT FOR ARGENTINA: UNDERLYING OBJECTIVES While there is no single best way to design a system of intergovernmental transfers, in this section I shall spell out the broad objectives of one such model that accords well with the fiscal- federalism precepts articulated in the previous section. The following section will elaborate on the manner in which these guiding principles or objectives might be implemented. Consistent with the earlier-enunciated principles, the following six objectives constitute an effective approach to designing an intergovernmental Fiscal Covenant: 1. DEVOLUTION OF TAXNG POWER TO THE PROVINCES ONA DERIVATION BASIS: Among the rationales for this is to ensure a greater degree of fiscal coincidence, namely the jurisdiction responsible for spending should also bear more responsibility for raising the required revenues. Apart from being fully compatible with the federal or constitutional rationale, tax 9 devolution will also result in enhanced accountability and transparency, in a decrease in the degree of vertical fiscal imbalance, and in an increase in the fiscal autonomy of the Argentine provinces. While some provincial taxes may require provincially run tax collection agencies, the spirt of this Covenant is that there would be an overarching tax collection system for those taxes where this is feasible and efficient. Note that this overarching tax collection system need not be run by the naci6n: it could be, and perhaps should be, a joint naci6n-province agency. For example, a VAT with variable provincial rates would be much more feasible, let along effective in terms of collection and compliance costs, under the rubric of a national collection agency. 2. FISCAL EQUITY: A key objective of the fiscal Covenant is that all provinces should have access to a level of total revenues (own revenues plus transfers) sufficient to ensure that all provinces can provide reasonably comparable levels of public goods and services at reasonably comparable levels of taxation. This is also consistent with the federal principle in that the provinces will have revenues adequate to discharge their constitutional responsibilities. Phrased differently, because tax devolution will be run on a "derivation" concept, different provinces will end up with different levels of per capita revenues. Hence, the achievement of fiscal equity will require the existence of an equalization program. Moreover, it may also require that a portion of any intergovernmental transfers be devoted to taking account of different expenditure needs across the provinces, i.e. it may take more revenues, per capita, for certain provinces to deliver comparable levels of public goods and services. 3. A CODE OF FISCAL CONDUCT: The objective here is to preserve and promote the internal economic union and in particular the internal fiscal union. Provinces will sign on to an 10 agreement that limits the degree to which their fiscal initiatives can fragment or mount barriers to the free flow of goods and services and capital across their borders. Ideally, the naci6n should also become a signatory to this internal economic union, with abrogation powers limited to certain emergencies (natural disasters, etc.). 4. A SET OF PRINCIPLES FOR MINIMUM STANDARDS RELA TING TO THE SOCIAL UNION: These principles will be articulated by the naci6n in consultation with the provinces, and their enforcement will be monitored by a federal-provincial agency. Agreement to these principles will be a requirement for receiving federal transfers. In effect, this is the quid pro quo for those transfers that relate to the expenditure needs of the provinces. S. A REVENUE GUARANTEE:. For the duration of this Covenant, no province will be in receipt of a smaller amount of overall revenues (own revenues plus transfers) than it currently enjoys. Ideally, the Covenant should "sunset" (be subject to renegotiation) after five years. Some "bridging" revenue guarantee (across the five-year agreements) may also-be considered in order to ensure full opt-in by the provinces. 6. A NACI6N-PROVINCIAL MONITORING AGENCY: The final component of the Covenant would be joint nacidn-province commission or technical committee. The Canadian version of this is a joint committee of senior executives from the central and provincial governments with a mandate for updating the technical aspects of the Covenant-defining tax bases, for reworking the principles relating to the social union, for monitoring the conformity to the economic union, and the like. The Australian version (The Commonwealth Grants Commission) is a more formal and more powerful variant, which would be charged with the design of the overall transfer system. 11 Either version would be acceptable, although it may be appropriate to begin with the former, more ad hoc, variant. These objectives are rather general and could accommodate a variety of intergovernmental transfer systems. In the following section, I shall focus on one approach (with a few options) to implementing these underlying objectives. This approach is designed to convert the existing Argentinian transfer system to a model that incorporates the above objectives. Alternative models are, of course, possible and these are left to readers or analysts who are more familiar with the priorities and constraints of Argentinian fiscal federalism. Presumably, however, these alternatives would differ from the proposed model largely to terms of detail (emphasis or weighting of the components) and not in terms of substance. IV: A FISCAL COVENANT FOR ARGENTINA: IMPLEMENTATION The details relating to the implementation of the proposed model take as the status quo the current allocation across provinces of the provincial share of the Masa Coparticipada, henceforth MC. (Readers are assumed to be familiar with the status quo in this regard). 1. Allocating the Provincial Share of MC Consistent with guiding principles or objectives in part III above, the provinces' share of the MC would tnow have four components: * A tax devolution component; * An equalization component; * A formula-based transfer (based either on equal per capita and needs criteria or solely on needs - see later discussion); * A revenue guarantee component. 12 We shall deal with each of these in turn. A. The Tax Devolution Component Eventually, decentralization of taxes will comprise a very significant portion of the provinces' share of the MC. Ideally, this would include a broad-based tax (such as the VAT) and would replace current original-based taxes like Ingresos Brutos. What is required, operationally, with respect to any devolution is to deduct the value of the devolved taxation from the value of the MC. There are two polar ways to do this - one on a once- and-for-all basis and the other on an on-going basis. The once-and-for-all basis might work as follows. From the table Distribucion Primaria Nuevo Esquema De Coparticipacion (Annex A to this Report), the data for 1999 reveal that the value of the MC equals $44.748 billion and the provinces' share is $16.442 billion, or 36.7%. Consider the devolution of the fuels tax. Suppose, only for illustrative purposes, that at some agreed upon equilibrium tax rate (the existing rate?), the fuels tax yields $1 billion. One would then deduct this $1 billion from the $16.442 billion figure to obtain $15.442 billion. (The returns from this tax will accrue to the various provinces in line with what is raised in the provinces (i.e. the allocation is on a derivation principle). As a percent of the MC, this now represents 34.5%. Henceforth, therefore, the provincial share of the MC will be 34.5% rather than the current 36.7%, since the provinces will already have the difference in the form of the decentralized fuels tax. (Note that this implies that if the provinces then increase their taxes above this equilibrium rate they will pocket the additional revenues. This is what tax decentralization is all about.) Further devolution of taxes will be treated 13 in the same manner. The second approach would be to make an annual deduction from the MC equal to the then current value of the devolved tax. Presumably, this would vary in line with the volume of fuel sales. While there are pros and cons relating to both approaches, simplicity and the federal principle would appear to favour the former. B. The Equalization Component A Canadian-style equalization formula would apply to all provincial revenues - existing provincial revenues as well as any newly devolved taxes. This would ensure that all provinces would have access to a per capita level of own revenues equal to what they would obtain if they were to apply the all-province average tax rate (defined on the population-weighted average of all provinces actual tax rates) to the all-province per capita average tax bases. Let us call this the "all- provinces standard" (APS). Provinces with per capita revenues above the APS would keep these revenues, as in the Canadian model. The rationale for this is that richer provinces will have higher wages and rents, that is the revenues will be capitalized into wages, rents and the like so that the costs (as distinct from the needs) of providing public goods and services are more expensive in these richer provinces. Readers wishing more detail on the Canadian equalization system can refer to the appendix of the appended C.D. Howe Commentary. Arguably, equalization is the critical instrument that allows for tax decentralization. Without equalization, the poorer provinces would never support a significant decentralization of taxation. The value of these equalization payments to the fiscally poor provinces would also be 14 deducted, on an annual basis, from the provinces' share of the MC. To recapitulate a bit, what the system (devolution and equalization) does thus far is ensure that the combination of provinces' existing revenues and any devolution of new taxes leaves all the provinces with at least the APS per capita value of own-source revenues. Moreover, both new devolutions of taxes and the equalization payments come out of the provinces' share of the MC, so that this is revenue neutral from the naci6n vantage point. We now turn to the formula-based component. C. Formula-Based Cash Transfers This is the component that the economics research group allocated among transferencias devolutivas, transferencias de equiparaci6n territorial, etc. This represents one approach to this component. However, we have the following comments on this approach. First, with respect to the transferencias devolutivas, if one proceeds along these lines, namely allocating transfers in some relation to taxable capacity, then it seems logical that one would then want to "equalize" these transfers. This follows because they are similar to a tax devolution in the sense that they give more per capita to the richer provinces. With respect to the second component, our view is that the correction for need is too "rich". This is not meant to downplay needs, only to express a view about the large variance in the needs index. As an aside, if there are one or two provinces that have very special needs (e.g., Tierra Del Fiego), would it not be better to treat them as special cases and, therefore, outside the formula-based transfer rather than distort the formula to suit the needs of these few provinces? (For example, in the Canadian case, the two northern territories have a quite 15 different set of fiscal transfers than do the ten provinces). On reflection, my preference would be for two components - an equal per capita component and a needs component. The former will serve as a holding tank, as it were, since future tax devolutions should probably come out of this component. Phrased differently, once there is full decentralization, the only component left for this formula-based transfer would be the needs-based component. It is important that some cash transfers will remain in order to ensure that the federal government retains some leverage to mount, monitor, and enforce the social policy principles as well as to monitor the internal fiscal union. Ideally, the needs component (which is addressed in more detail under Social Policy Principles later) should provide some monies for all provinces, so that the social policy principles can be enforced. As an aside, there may be a problem here. If, as a result of tax devolution, one or two provinces end up with per capita revenues way above the APS, then consideration may be given to offsetting (in whole or in part) the needs-based allocation from their own revenues, which in turn may reduce the needs payment to zero for these provinces. In this case, the nacidn will have to find some other way to enforce the social policy principles for these rich provinces. One has to "do the numbers", as it were, to see if this situation is likely to arise. D. The Revenue Guarantee This component is similar to, but not identical with, what the background simulations have termed transferencias compensatorias. The differences are two fold. First, we assume that this component will disappear over time, hopefully rather quickly. Second, the guarantee relates to total 16 revenues (own revenues plus transfers) and not just transfers. For the provinces affected, this is a residual. If, as a result of own revenues and the above three components, a province finds that its overall revenues are below existing levels, the role for this component is to fill the gap. As provincial revenues grow over time, this gap will narrow and, therefore, so will this component. It may be important to ensure that this component is not fully confiscatory. For example, if a province's overall revenues grow by $10 million (say, from its own revenues plus equalization), it may be difficult politically (and incentive-wise) to decrease the revenue guarantee component dollar-for-dollar. One could conceive of, say, an 80% tax-back rather than a 100% tax-back rate. This will ensure that all provinces will benefit from increasing their tax rates. Yet it will still ensure that the revenue guarantee will disappear over time. The last point probably deserves more emphasis. For provinces whose current revenues are way above the all-province average (because, say, of special transfer provisions), it may be difficult politically for them to sign on to the pact if the result is that their total revenues are frozen at existing levels until the growth of the economy eventually dominates their current favourable transfer provisions. For these provinces, the ability to share partially in the growth of the economy (via an 80% rather than a 100% tax-back rate, as elaborated above) may be the political price for their support. In general, this would not be a very expensive proposition since the provinces affected are not that numerous and their populations tend to be small. The more important point is that the growth of the economy will ensure that most provinces will be better off fairly quickly under the new program - and all provinces will eventually be better off. And of course in the process this Fiscal Covenant has other benefits for the provinces, such as 17 enhanced fiscal autonomy. The fact that this component is a residual will complicate the overall calculations. For example, one may have to scale back component three in order that the whole scheme fits within the overall provincial portion of the MC. 2. A Code of Fiscal Conduct Argentina has made great strides on the monetary side of the macro equation. Decentralizing the tax system is an important component toward making similar headway on the fiscal side. The role for the code of fiscal conduct is to carry this progress further and, in particular, to ensure that budgetary decisions of provincial governments do not serve to erode the internal economic union, i.e., are consistent with "market-preserving federalism." Examples of relevant provisions for such a code are easy to come by: . prohibiting fiscal actions that discriminate against citizens of other provinces. Perhaps one could refer to this as "provincial treatment" - provinces will be free to conduct fiscal measures within broad parameters provided that they treat-non-province residents in their jurisdiction the same way that they treat their own residents; . moving toward a system that minimizes or prohibits origin-based taxes at the provincial level; * prohibiting the mounting of fiscal barriers to the free flow of goods, services and capital across provincial boundaries; - at a more general level, this code of fiscal conduct could become the vehicle for coordinating and even harmonizing fiscal policy between the two levels of government, i.e., some move toward the Maastricht-type arrangements. For its part, the federal government might be wise to become a full partner in such a code. More importantly, a quid quo pro for the provinces might be that adherence to such a code would 18 mean that, as in Canada, the federal government would collect the provincial share of any joint taxes (e.g., the VAT) free of charge. I recognize that this concept of a code of fiscal conduct has not surfaced in our earlier discussions. Indeed, it is not an essential component of the fiscal pact in the sense that the pact could go forward without it. However, it does seem to be a valuable addition to the pact and it is important for increasing the efficiency of the overall tax system. 3. Social Policy Principles Once one distributes transfers on the basis of expenditure needs, the issue of accountability arises. There are several ways to handle this. Many federations simply make these grants conditional - they must be spent on specific expenditure categories. This is an option for the federal government. In this case one might take these cash transfers right out of the arrangements and put them on their own footing. It is worth elaborating on this alternative. There are many other transfers to the provinces coming from a variety of federal ministries/departments. Many of these are designed to cater to various aspects of provincial "needs". This alternative might fit well into this general set of provincial transfers. While not ruling this out,4 my solution is along differient lines. Specifically, I propose a set of social policy principles (e.g., no residency requirements for new in-migrants in order to access a province's social programs, equality of opportunity in accessing essential public services). These principles can be in the form of national standards or in terms of equivalencies. The latter are closer to what might be considered performance standards or outcome standards. They would allow 19 equivalent approaches to social policy challenges provided that the outcomes are equivalent. This allows for provincial experimentation in these areas and, therefore, the approach fits well in terms of the economic theory of federalism. One way to enforce these principles is to make their acceptance a requirement for receiving the cash transfers, or variants of this approach. It would probably be important to bring the provinces in on the design of these principles. And perhaps they should also have some say in terms of how compliance is monitored. While this may not be wildly popular among the provinces, it does have the potential for being very well received by ordinary citizens. This is so because the principles will amount to a sort of set of "citizen rights" on the social policy front that will be national in nature - they will apply no matter which province citizens choose to reside in. It may be important to ensure that these are not "rights" in the legal sense. In other words, the monitoring is via the transfer system, not via the courts. Nonetheless, in the political environment where unanimity of the provinces is required for passage of this sort of fiscal pact, incorporating some principles relating to the "social union" may find enormous sympathy across citizens and, in the process, they may pressure their respective provinces to sign on to the pact. 4. Equalization The operational principles underpinning equalization are essentially those that are in place in Canada. One major difference is that for Argentina we are talking about very few revenue sources in contrast to the more than 30 that exist in Canada. This is not the place to detail the procedures relating to calculating equalization. (See the appended C.D. Howe Commentary.) For present 20 purposes, only a few points merit highlight. First, the all-provinces average tax rate should reflect the actual taxing practices of the provinces. In principle, it should be a (population-weighted) average of existing provincial tax rates. In some sectors where this may be difficult to derive or even define, the best approach is to divide total revenues by the total base - this may be a useful approach in terms of energy royalties, for example. Second, ideally one would like a tax base that relates directly to what is being taxed. In the initial years this may be difficult to derive. Hence, some proxy measures may have to be used until more appropriate data become available. We do not believe that this should be a major stumbling block. The silver lining here is that the existence of the Revenue Guarantee will ensure that there are no losers, so that this provides the needed flexibility to move boldly forward. And moving boldly forward is probably critically important. Very tentative initiatives will more likely be open to stalling. Moreover, the fiscal pact as outlined here is an omnibus package that has many elements that will benefit the provinces. Most important of all, issues such as the calculation of tax bases are technical issues and addressing these will fall to the federal-provincial committee on the fiscal pact (outlined below). There is another very important consideration, namely how to begin the equalization process. Our approach is, again, to move boldly forward and to include not only the existing major provincial revenues (perhaps with a catch-all category to capture the remaining revenues where PBG could be used as a base) as well as Royalties. It is important to include these Royalties from the outset because the resulting revenues have a very high variance across provinces. 5. A Federal-Provincial Committee on the Fiscal Pact The final operational principle relates to the necessity of having a federal-provincial 21 committee that deals with the Fiscal Covenant. Actually, there will probably be several committees - e.g., one of the relevant Ministers and one of officials. There are many roles for these committees. Focussing only on the committee of officials (senior bureaucrats), one of their key roles is to work toward ensuring that the tax bases are appropriately defined, to file requests for data requirements, and generally to iron out the many problems and challenges that are sure to arise. One enormous advantage of such a committee is that, without really realizing it, the members will all buy into the overall conception. Moreover, if the Canadian experience is any guide, the members will also provide valuable insight into the myriad of problems that have to arise in an undertaking such as this. Whether or not this is the committee that should also be responsible for defining the social policy principles is not for me to decide. In operational terms, the committee might have co-chairs - one from the federal government and one on a rotating basis from the provinces, with meetings held alternatively in Buenos Aires and the capital of the province of the current co-chair. While this is "democraticizing" the process relating to the Fiscal Covenant, it is important to note that any legislation relating to the fiscal pact will be naci6n legislation, so that in the final analysis the nacidn holds the upper hand. Further detail on the structure, role and value of such committees appears in Robert McLarty's report. 22 V: UNFINISHED BUSINESS Thus far, this report focusses only on some generalities relating to implementing the fiscal covenant. Much more research is necessary. For example: - One needs to focus on the details relating to the design of the decentralized VAT, if the VAT is the instrument of choice for tax devolution. Charles McLure is the expert here, along with Argentine analysts. - We need to have access to comprehensive simulations to assess the nature of any complications that will arise. * One has to devise a more acceptable approach to fiscal needs. The existing variant is, in my view, far too generous. Above all, the looming imperative is to make the Fiscal Covenant politically acceptable to the provinces. This challenge is well beyond my ability to address. VI: CONCLUSION The model for the Fiscal Covenant presented above accords well with the principles enunciated in Part II above. Among other features: - it accords well with the existing theory of allocating taxes and expenditures to sub- national governments; * it enhances accountability and transparency. it moves strongly in the direction of achieving fiscal correspondence or fiscal coincidence; - it accords well with "market preserving" federalism" in that provincial budget constraints will henceforth be "hard". - it embodies internal free trade (via the code of fiscal conduct) which will increase the efficiency of the government sector; 23 - via equalization, it achieves fiscal equity within a decentralized federation; * it redirects the focus away from an ideal set of transfers and toward the more appropriate ideal set of overall revenues (transfers plus own revenues); - it embodies both the potential differences in costs and needs of implementing provincial programs; * it sets aside a portion of the overall grants so as to ensure that selected key "merit goods" are accessible for all citizens. This is accomplished either by a set of social policy principles or by making a portion of the grants conditional; - it brings the provinces into the implementation process (and perhaps into the design process) by means of establishing a naci6n-province committee to monitor the process and progress; - it embodies revenue guarantees as a key component of the transition to the new system. * finally, it should also appeal to Argentine citizens, since they are guaranteed a set of (non-justiciable) citizens rights. This may be a critically important element in terms of the overall political acceptance of the Fiscal Covenant since, in the final analysis, the rationale for a federal system is not what it does for governments, but what it does for citizens. Postscript On a personal note, I wish to thank the World Bank (and, in particular, David Rosenblatt) as well as the Argentine officials (and, in particular, Carola Pessino) for allowing me this wonderful opportunity to become more familiar with the Argentine economy and society. However, it is critically important to note that this report reflects my own biases and my own experience with federal systems. Whether it is useful for Argentina in the 21st century is for others to decide. 24 Endnotes 1. The data in this sentence and the next are drawn from Saiegh and Tommasi (1998). 2. For example, under fiscal separation, McKinnon would preclude equalization payments. While perhaps appropriate for member nations of the EU, this is not appropriate for federal nations. Indeed, in all federations except the United States, equalization is a core element of sub-national financing. See the discussion in 4 below. 3. As a conceptual and technical aside, if the introduction of a VAT is going to dramatically skew provincial per capita revenues, then one might contemplate a version of the German approach to VAT revenue-sharing. This would involve allocating a part of the provincial VAT revenues on an equal-per-capita basis. By way of an example, suppose that 75% of the provincial share of the VAT is allocated on a derivation principle and the other 25% is allocated equal per capita across the provinces. This will reduce the per capita variance across provinces (and reduce equalization as well since equalization will take account of the 25% share). This may be a moot issue - there may be no dramatic variance across provinces. Again, one needs to do the simulations. 4. One problem with the above alternative is that the very high transfers to some provinces will likely be viewed as an entitlement if they are transferred outside of revenue sharing framework. Among other things, this will likely mean that it will be very difficult to subject these transfers to any phase-out, as proposed above in connection with the Revenue Guarantee. 25 Tax Assignment in Argentina: Conceptual and Practical Issues Charles E. McLure, Jr. Hoover Institution Stanford University I. Introduction There seems to be substantial agreement that the present method of financing the subnational governments of Argentina, especially the provinces, is unsatisfactory. It is complex, it is not transparent, and it impedes accountability and efficiency in government.' The existing system should be replaced with one that a) assigns to the provinces revenues from particular taxes that, at the margin, are under the control of the provinces and b) establishes a coherent system of grants; such a system would include equalization grants designed to offset the horizontal disparities among provinces that would remain following decentralization of taxing power.2 This paper describes and analyzes the possibilities for assigning taxes to the provinces. It considers both conceptual and practical issues in tax assignment. It does not, however, consider such basic issues as the case for decentralization of government, the assignment of expenditure responsibilities between levels of government, or the need to assure that subnational governments face a hard budget constraint; these are taken as given.' Nor does it consider the structure of grants, which is the subject of a companion piece.4 The paper focuses primarily on economic and administrative considerations. It does not consider in detail the potentially important constitutional and political constraints that would be encountered in implementing the proposed assignments. Section II discusses the economic objectives of federalism and their implications for tax assignment. Section III describes four methods of revenue assignment and the advantages and disadvantages of each. Section IV discusses issues in the design of provincial surcharges on taxes of the federal government. Section V describes a method of implementing provincial surcharges on the federal VAT and Section VI discusses whether provincial surcharges on the 'See World Bank (1996 a, b), Boadway (1997a), Oates (1997). Saeigh and Tommasi (1998) argue that, "the argentine political actors made a collection of choices that, in the view of all experts is clearly inefficient." These authors provide an intriguing description and analysis of the political processes that led to the present arrangements for coparticipation. 2And, of course, provincial governments should face a hard budget constraint. They should not be able to spend more than the revenues they receive from own sources or from the central government; the latter amount should be determined ex ante by application of objective and sensible criteria, and not ex post or by negotiation. To the extent that subnational governments are allowed to borrow, it should be subject to strict controls that ensure accountability and fiscal responsibility and avoid the need for bail-outs by the central government. 'See, generally, Oates (1972) and (1994). 4See Courchene (1998). VAT should be administered solely by the federal tax administration or by a dual federal/provincial tax administration. The paper's conclusion is dramatic and potentially quite important: It appears that, contrary to the conventional wisdom, there is a method that would allow provincial imposition of surcharges on the VAT imposed by the federal government. If this is true, the provinces of Argentina need not be fiscally weak. (Given the well-known and unfortunate experience of Brazil, where the states impose a very complicated origin-based VAT, perhaps it should be noted here that the system being proposed in not the Brazilian system.' It is a destination-based system of provincial surcharges that requires no fiscal frontiers between provinces, yet protects revenues from attempts to disguise household purchases as purchases by registered taxpayers; see Section V for a complete description of the proposed system.) II. Tax Assignment and the Economic Objectives of Federalism The economic case for federalism rests on the view that decentralization allows citizens of subnational governments to choose the levels and kinds of public services they wish. This view has several important implications for tax assignment. Relation of taxes to benefits. Decentralization works best when taxes and the benefits of public spending (or costs incurred by governments) are closely related. In the extreme case of user charges and fees, payments act almost like prices for private goods, in that citizens get what they pay for and pay for what they get. The implication for tax assignment is that revenues from sThe proposed system did, however, apparently originate in Brazil. See Varsano (1995) and Gonzalez Cano (1996). 'World Bank (1996a, p. 23-24) provides the following principles, which are broadly consistent with what follows: The key to a sound, decentralized fiscal structure lies in the revenue structure of subnational governments. If the provinces are responsible for the provision of important public sector activities such as health and education, sound policy suggests that they should be able to levy sufficient taxes to meet the demands for such services--provided that those who make the demands are also those who pay the taxes. While this may not always be possible, it is often possible to approximate this goal for at least the richest subnational governments. Theory provides three guidelines on local financing: (i) the principle of benefit taxation-- benefit taxes (user charges) should be used to the maximum extent possible to finance local public services; (ii) the principle of equity: tax bases that are unevenly distributed among jurisdictions are not appropriate for decentralization as they promote dislocation of population and activities; and (iii) the principle of efficiency: taxes that induce the tax base from high rate areas to low rate areas (e.g., the corporate income tax) are not good candidates for decentralization as they may lead to misallocation of resources. The combination of these principles leads to some basic rules. First, user fees should be used to the maximum extent by local governments. Second, income taxes and sales or property taxes may be used as proxies for benefit taxation insofar as the burden of taxation corresponds to the benefits of local public services. Third, add-on or shared taxes should be used to minimize administrative and compliance costs--the most usual is a surtax on the federal personal income tax base allocated to the provinces. The characteristics of a good provincial tax include: (i) the tax base should be relatively immobile, to allow local authorities some leeway in varying rates without losing their tax base, and it should be visible to ensure accountability; (ii) the tax yield should be adequate to meet local needs, buoyant over time (i.e., it should expand as fast as expenditures), predictable over time and stable; (iii) it should not be possible to export the tar burden to non-residents; and (iv) it should be perceived to be reasonably fair by taxpayers and be easy to administer. 2 taxes that are closely related to identifiable benefits (or public expenditures) should go to the level of government that provides the benefits (or incurs the costs). Thus, for example, revenues from taxes on motor fuels intended to reflect benefits of the use of roads and highways should go to the governments that finance the construction and maintenance of roads and highways, and revenues from taxes on alcoholic beverages and tobacco products should go the governments that incur the health-related costs occasioned by consumption of those products. Generalized benefits of public services. User charges, fees, and the use of taxes related closely to benefits of public services are most likely to be feasible for the finance of services that, although provided publicly, exhibit an important characteristic of private goods, the feasibility of exclusion of those who do not pay.' Inherent in the nature of public goods is the inability to charge those who benefit from them. Moreover, for reasons of social policy charges are often not applied even when they would be feasible, as in the case of health and education. Thus much of public spending provides generalized benefits that must be financed by taxes that are only loosely related to the benefits of public services, in particular, by general sales taxes, income taxes, and property taxes. When production and consumption do not occur in the same jurisdiction or individuals do not earn income where they live (for example, because they commute between jurisdictions or have investments in other jurisdictions), an important two-stage question arises in the attempt to identify beneficiaries of generalized public services: first, whether the benefits of public services are provided primarily to businesses or to individuals, and second, when such benefits are provided to individuals, whether they are more closely related to a) consumption or the residence of taxpayers or b) production or the source of income. If generalized benefits are provided primarily to business (and are thus more closely related to production than to consumption) an origin-based sales tax or a property tax on non-residential property tax might be appropriate.! If such benefits are provided to individuals and are more closely related to consumption or the residence of taxpayers, a destination-based sales tax, a residence-based individual income tax, or a property tax on residential property might closely reflect benefits of public services. But if such benefits are related more closely to the earning of income, a source-based individual income tax would be more appropriate. It is difficult to marshal evidence on either side of this argument. Yet it seems reasonable to believe that most public services are provided to individuals, and not to businesses, and that "The two characteristics of public goods are jointness (the characteristic that once provided, the good can be made available to all at no additional cost) and the inability to exclude those who do not pay. Those who do not pay motor fuel taxes cannot legally use roads and highways, even if access to them is not restricted. In theory it would be possible to deny access to education and health care, but this is commonly not done for social reasons (especially in the case of primary education and basic health care). Given this, taxes may be levied on alcoholic beverages and tobacco products in order to charge future users of such services.. "In general it seems that an origin-based sales tax would be more appropriate than a tax on non-residential property, because firms providing services generally have relatively little tangible property. But public services may be closely related to the existence of tangible property, especially real property, the base of most successful property taxation.. It may appear that a source-based corporate income tax would also be appropriate in this case. But there are several problems with this reasoning. Services are presumably not provided only to corporations, and the value of such services is not likely to be closely related to profitability. 3 most services provided to individuals are more closely related to consumption or the residence of the taxpayer than to production or the source of income. Section IV discusses the implications of this view in greater detail." The role of marginal revenues. According to the theory of fiscal federalism underlying this discussion, by levying higher or lower taxes on themselves, citizens of subnational governments can have more or less public services. For this scenario of subnational fiscal autonomy to be realized, it is necessary that subnational governments control potentially important marginal sources of revenues (additional revenues). This rule has several corollaries. First, it must be possible for subnational governments to vary the amount of tax revenue they colleci. Revenues from subnational taxes that can only be levied at fixed rates are inframarginal and thus do not provide subnational autonomy. Taxes that cannot possibly yield significant amounts of revenue do not provide fiscal autonomy, even if there is no legal limit on tax rates. Finally, it must be possible for subnational governments to reduce taxes at the margin, as well as to raise them. For purposes of this discussion, limits on tax rates can be the result of either legal restrictions or tax competition. Legal restrictions on rates of subnational taxation are likely to involve encroachment on subnational autonomy that is undesirable (except under certain conditions, as when subnational governments attempt to "milk" taxes that might better be centralized). By comparison, restrictions that result from tax competition are more likely to be benign, as they prevent subnational governments from engaging in inappropriate policies. Tax competition is more likely to constrain tax rates in the case of production-based taxes that do not match benefits of publically provided services than in the case of destination-based sales taxes, residence-based income taxes, or production-based taxes that do reflect benefits. (See also the discussion of the role of tax competition below.) Second, except where spillovers of benefits justify matching provisions, grants from higher levels of government should generally be inframarginal, so as not to compromise the fiscal autonomy of subnational governments by altering the tax prices they face. Taxation and the location of economic activity. If the benefits of free markets are to be realized, subnational governments should not impose taxes that distort the location of economic activity. Taxes levied at the origin of production or the source of income are much more likely to distort locational decisions than are taxes levied at the point of consumption or the residence of households, unless they reflect benefits of services provided to businesses or to employees at their place of employment. Avoidance of tax exporting. Some taxes can be exported to non-residents, if imposed by (or for) subnational governments. The best examples of exported taxes are taxes on business that cannot be shifted to consumers or to labor, and thus are borne by owners of business, many of whom may be non-residents. (The view that taxes on business will automatically be shifted to consumers, many of whom are non-residents, is generally incorrect. Exporting to non-resident consumers is likely to occur only when the taxing jurisdiction dominates the national market for the taxed product. In the absence of competition, excises imposed on alcoholic beverages or 'The fact that a destination-based VAT reflects the generalized benefits of public services better than the gross receipts tax does not mean that a shift from the latter to the former will not have important and wrenching effects on provincial revenues. Whereas revenues from the GRT are split evenly between provinces of origin and destination, those from the VAT would flow entirely to provinces where consumption occurs. 4 tobacco products by the jurisdiction where production occurs could be exported to consumers throughout a nation.) Tax exporting is generally undesirable. Not only does tax exporting impose burdens on non-residents that are likely to be seen to be unfair; it cheapens the provision of public services in the taxing jurisdiction, thereby encouraging their over-consumption. An important exception involves taxation that is closely related to benefits of public services (e.g., taxes on motor fuels that finance roads used by non-residents); in this case failure to export taxes to users of the public services would be undesirable. Again, this points to the superiority of destination-based sales taxes and residence-based income taxes over production-based sales taxes and source-based income taxes. The former are much less likely to be exported than the latter. The role of tax competition. Governments often wish to levy taxes on business that exceed the value of benefits provided to (or costs incurred on behalf of) business. This may be done because taxing business is more popular than taxing people (or the products they buy) or because it is thought that taxes on business can be exported to non-residents of the taxing jurisdiction. Taxes on business that exceed the value of benefits provided are likely to distort the location of economic activity, unless levied at uniform rates throughout the country. But taxes that must be levied at uniform rates do not provide a marginal source of revenue for any one subnational government, and thus do not further fiscal autonomy. Moreover, agreements between jurisdictions to fix tax rates (or imposition of uniform rates by a higher level of government) represents a form of cartelization that is not likely to be more benign than cartelization in the private sector, as it encourages over-expansion of the public sector and allows politicians and bureaucrats to be inefficient. Competition among subnational governments has the salutary effect of both preventing excess taxation of business and impeding inefficiency in government. Vertical fiscal imbalance and subsidiarity in taxation. It is generally accepted that public spending should be lodged at the lowest possible level of government, consistent with the objectives of avoiding spillovers of benefits between jurisdictions and realizing economies of scale, in order to maximize the fiscal autonomy of those affected by public spending. This principle of subsidiarity has a counterpart in the field of tax assignment, because of the seemingly inevitable existence of vertical fiscal imbalance. Higher levels of government can effectively administer virtually any tax that lower levels can administer, but the converse is not true; some forms of taxation do not lend themselves to implementation by or for lower levels of government.'o Moreover, some taxes would not be appropriate for local governments, even if administratively feasible; customs duties are perhaps the best example of a tax that is not appropriate for a subnational government. In order to minimize vertical fiscal imbalance, it is thus desirable to assign to subnational governments all taxes that are administratively feasible and are not inappropriate for them. In judging administrative feasibility and appropriateness, one should consider the use of subnational surcharges on the base of a higher level of government, as "oWorld Bank (1 996a, pp. 22-23) notes, "No matter how revenues and expenditures are assigned, a problem of vertical imbalance occurs so long as provinces have major expenditure responsibilities because few taxes are suited to local application. ... To fill this gap the common practice is to transfer some of the centrally-collected revenues to provincial governments." It is important how these revenues are transferred: as grants or shared revenues or as tax base made available to the subnational governments to be taxed at their discretion. Only the latter provides marginal subnational revenue. 5 well as independent legislation and administration by the subnational government; see also the next section. Horizontal fiscal disparities and the need for equalization. Except in rare cases, subnational governments within a nation commonly differ substantially in their inherent fiscal capacity." Most obviously, some provinces are simply richer or poorer than others, as measured, for example, by per capita income or per capita GDP. But there may be other explanations. For example, some activities (e.g., tourism) are easier to tax than others. Particularly important is the presence of geographically concentrated, high-value natural resources that can be taxed by subnational governments. (Whether subnational taxation is appropriate is considered below.) Where horizontal fiscal disparities are important, equalization grants may be appropriate to allow all provinces to provide comparable levels of services with comparable levels of tax effort.12 Income distribution and macroeconomic stabilization. This section began with the following assertion: "The economic case for federalism rests on the view that decentralization allows citizens of subnational governments to choose the levels and kinds of public services they wish." In terms of the conceptual model of public finance advanced by Musgrave (1959), this is a statement about the "allocation branch" of government. It is widely accepted that the allocation branch is the natural realm of subnational government and that subnational governments may not be able to implement the functions of either the distribution or stabilization branches, and perhaps should not try." This view has important implications for tax assignment. Unfortunately, these implications are often misunderstood. Because the progressive individual income tax is the tax mechanism of choice for both income redistribution and endogenous macroeconomic stabilization, it should be reserved primarily for use by the central government. This does not mean, however, that subnational governments should not also rely on the individual income tax. A flat-rate individual income tax may provide a satisfactory surrogate for benefit-related taxation. Under certain circumstances a flat-rate subnational tax can be imposed on the same base as the graduated-rate national tax; see also the discussion of surcharges on the individual income tax in Section IV.'" Corporation income taxation also can be used for income redistribution and stabilization. It is thus also an appropriate source of revenue for the central government. It is less appropriate "The following words were omitted from the quotation in *note 10 above: "...provincial tax bases are not uniformly distributed (e..g., in proportion to the population). This creates a fiscal imbalance across jurisdictions." Fiscal capacity may be measured by application of a representative tax system--that is, by calculating how much revenue each province would receive, per capita, if it levied the taxes typical of all the provinces. While such measurement depends on the structure of the typical provincial tax system, it seems unlikely to be so sensitive as to be very misleading. "Tax effort can be measured by comparing actual tax collections with those under the representative tax system. Those that collect substantially more than under the representative tax system are exerting high tax effort; those that are collecting substantially less are exerting low tax effort. "See, for example, Oates (1972). 'Note that the result under a flat-rate surcharge on the base of the central government is very different from that under the Canadian system of imposing provincial surcharges on the tax liability to the national government. Since the national tax in Canada is levied at graduated rates, the provincial surcharge is also progressive. 6 for use by subnational governments, because it is not a good surrogate for benefit-related taxation, it is likely to distort the location of economic activity, and identification of income originating in a given subnational jurisdiction is difficult. (Section IV explains these points further.) Taxation of natural resources. The assignment of taxes on natural resources raises difficult philosophical and political questions, as well as economic issues." For the most part economic arguments favor assignment of taxes on economically important natural resources to the national government, especially if the resources are geographically concentrated in only some taxing jurisdictions." Because revenues tend to be unstable, taxes on natural resources are ill- suited for use by subnational governments, which generally need stable sources of revenue. Moreover, subnational governments may not be in as good a position as national governments to impose taxes on economic rents, generally agreed to be the most satisfactory form of tax on natural resources." If effective ownership of natural resources is in private hands, subnational taxes may be exported to non-resident owners." Finally, if important natural resources are distributed unequally across a nation, subnational taxation may produce or aggravate horizontal fiscal disparities. Besides raising questions of equity, such disparities may induce economically inefficient allocation of private resources to jurisdictions with low tax rates and or high levels of public services." Philosophical arguments are more ambiguous. Advocates of taxation by subnational governments commonly argue that natural resources and the right to tax them are part of the "heritage" of such jurisdictions. But it can equally be argued that the resources and taxes are the heritage of the entire nation. The resolution of this issue calls into question the nature of nationhood, as it is understood in particular countries.20 The emphasis on solidarity in the Constitution of Argentina suggests that taxes on important natural resources might best be "For a more complete statement of this discussion, see McLure (1994). "Taxes on common low-value and ubiquitous resources such as sand and gravel can be assigned to subnational governments under the principle of subsidiarity in taxation. "In theory, taxes on economic rent, being a tax on surplus, do not affect production decisions; the pattern of production that maximizes before-tax income also maximizes after-tax income. Taxes on production encourage inefficient exploitation (high-grading), by discouraging production that would cover marginal costs, but not the tax. Property taxes on deposits of natural resources are even worse, as they encourage uneconomically rapid exploitation, in order to avoid future taxes. "Note that effective ownership may be in private hands, even if the state owns natural resources. This may occur, for example, when resources are exploited under concessions. "What little empirical evidence exists suggests that these inefficiencies are likely to be relatively small, except in extreme cases, such as Alaska, where the combination of vast natural resources and a small population allows the financing of generous public services (including annual per capita grants) with low tax rates. See Mieszkowski and Toder (1983). "'This is no more clearly illustrated than in Russia, where subnational governments have threatened to break away if not allowed access to taxes on "their" natural resources. See McLure (1994). In some cases separatist tendencies also reflect ethnic differences. See McLure, Wallich, and Litvack (1995). 7 assigned to the federal government. On the other hand, resource-rich provinces are likely to defend their existing right to levy such taxes. However this issue is settled, one thing is clear. Revenues from provincial taxes on natural resources should be considered along with revenues from other sources in determining the need for equalization grants. (Indeed, if the provinces have important non-tax revenues from natural resources, those should also be considered.) A bottom-line assessment. Much of the previous discussion points in the same direction. First, user charges, fees, and taxes closely related to benefits of services provided by governments are fair and they do not distort the location of economic activity. Second, taxes on production and the earning of income that do not reflect benefits of public services are likely to distort the location of economic activity and they may be exported in part to non-residents; they are to be avoided. Finally, taxes on consumption and residence-based income taxes are arguably most likely to reflect the benefits of public services; even if they do not, they are less likely than other taxes to distort the location of economic activity. This suggests that the last type of taxes are the most appropriate for the finance of most generalized public services. III. Four Methods of Revenue Assignment Revenue assignment, a concept that encompasses tax assignment, involves four aspects of subnational tax policy, in addition to which level of government gets the money: 1) which taxes subnational governments are allowed to levy, 2) the bases of subnational taxes, 3) the rates of subnational taxes, and 4) the administration of subnational taxes.21 Depending on how these four questions are answered, subnational governments are likely to be fiscally weak or fiscally strong. A. Techniques of Fiscally Weak Subnational Governments The system of coparticipation employed in Argentina results in provinces that are fiscally weak, in the sense of lacking control over the four decisions listed above, and thus the ability to raise marginal revenues. In this it resembles the grant-based system used in Australia and the system based on tax sharing employed in much of the former Soviet Union. Revenue sharing and grants: the Australian approach. The Australian states have relatively little independent taxing powers. As a result, they raise far less revenue than they spend. The difference in revenues and expenditures of state governments consists of transfers from the central government.22 This system of tax assignment is relatively simple and economically neutral with regard to locational decisions. But the Australian states are fiscally weak, as they lack substantial autonomy over the four issues identified above: the choice of taxes, the definition of the tax base, the setting of tax rates, and tax administration. Tax sharing: the prototypical system of the former Soviet Union. Most of the newly independent nations of the former Soviet Union rely on tax sharing for the finance of subnational governments (commonly called oblasts). That is, the central government collects virtually all taxes, directs given fractions of revenues from certain taxes to the various oblasts, and provides 21Revenue assignment includes the power of subnational governments to borrow. Strong subnational governments have the power to borrow, whereas weak ones do not. But the problem of assuring that governments face a hard budget constraint is more important, as well as more difficult, if such governments are allowed to borrow. Further discussion of this important topic is beyond the scope of this paper. 'For more on the Australian system, see the papers in Collins (1993), Boothe (1996), and Warren (1997). 8 additional grants to some oblasts. Thus, as in the approach used in Australia, the oblasts have essentially no control over the choice of taxes, tax base, tax rates, or tax administration; they are fiscally weak." Though different in appearance, tax sharing is best seen as merely a form of revenue sharing or grant. B. Two Techniques of Fiscally Strong Subnational Governments The United States and Canada both have systems of tax assignment that allow strong state and provincial governments. While the U.S. system does not provide a suitable model for reform of Argentina's system, the approach used in Canada (but not the extant Canadian system) does. Independent legislation and administration: the US. system. The U.S. constitution allows the states virtually unlimited sovereignty in the tax field, subject only to the very general restrictions that state taxes cannot interfere with interstate or international commerce or violate due process. Thus, subject to these restrictions, the states choose the taxes they levy, they define their tax bases, they set their tax rates, and they administer their own taxes. There is no requirement that the states answer any of these questions in the same way--or that there be codrdination among states or between states and the federal government.24 The U.S. system provides maximum state sovereignty, but this sovereignty is bought at a high price--a price that Argentina can ill-afford to pay and need not pay. The inconsistency of state laws and administrative practices creates substantial complexity, and thus high costs of compliance and administration, as well as inequity. Use of origin-based taxes such as the corporation income tax distorts the geographic allocation of resources and may result in tax exporting. Even the retail sales tax, which is imposed by 45 states and the District of Columbia, contains an origin-based component that distorts locational decisions, since a substantial amount of revenue (one estimate is 40 percent) is derived from taxes levied on sales of business inputs, instead of sales to households.25 Taxes on natural resources generally take the form of state severance (production) and income taxes and local property taxes, all of which are distortionary. (Distortion is mitigated by the fact that the corporate income tax and sales taxes on business "For more detailed descriptions, see McLure (1995), Martinez-Vasquez, McLure, and Wallace (1995), and Asian Development Bank (forthcoming). As explained there, the lack of oblast fiscal autonomy is much worse than suggested by the description in the text. (The tax sharing rates commonly vary across taxes and between oblasts and are set to provide the revenue needed to finance a given level of expenditures. Where shared taxes do not provide adequate revenues, the central government provides subventions. Expenditure levels reflect a combination of norms, historical patterns, inflation adjustment, and negotiation.) The German laender also rely heavily on shared taxes, particularly the VAT. In Canada the federal government collects the Harmonized Sales Tax for itself and several of the poor maritime provinces, dividing the provincial share on the basis of estimated consumption in the various provinces. "For a more complete description, see Duncan and McLure (1997). "In addition, like the gross receipts tax levied by the provinces of Argentina, the retail sales tax provides some incentive for vertical integration. This is reduced by exemption of goods bought for resale or direct use in production, for fuels, and for utilities; such exemption vary from state to state. 9 inputs are deductible in calculating liabilities under the federal tax.) There is substantial tax exporting, especially by resource-rich states." Surcharges: the Canadian approach. Canada relies heavily on provincial surcharges on the tax or tax base of the federal government. The Canadian approach is not easily summarized, because, like the U.S. system, it exhibits considerable diversity; yet some generalizations are possible." The federal government collects the income tax for those provinces that choose this option, which requires adoption of the federal base and a uniform formula for apportionment of taxable income of multi-provincial firms among the provinces. Provinces are allowed to choose their own surcharge rate, which is applied to the federal liability, rather than to the province's part of the federal base. Although not all provinces choose to join the federal collection program, those that do not participate (Alberta, Ontario, and the French-speaking province of Quebec) generally follow the federal tax base and apportionment formula quite closely. In short, the income tax provides provinces with fiscal sovereignty, albeit within the limits imposed by tax competition, and does so in a relatively simple manner. Canadian sales taxation is more diverse.28 Several provinces levy retail sales taxes (or no sales tax at all), with little coordination with the federal value added tax (VAT). While these taxes provide provincial fiscal autonomy, they suffer from many of the defects of the state retail sales taxes employed in the United States, including a substantial burden on business purchases." Three provinces (essentially the poorer Maritime provinces) participate in a Harmonized Sales Tax, in which the federal government shares VAT revenues with the participating provinces on the basis of statistics on provincial consumption patterns. This approach provides participating provinces with no fiscal autonomy, since no province can control the tax rate applied to its consumption. Of particular relevance for present purposes is the approach used in Quebec, to be described in greater detail below, in which the province collects both the federal and provincial VATs. Since the province retains control of the tax rate, it retains fiscal sovereignty, as do the provinces that impose retail sales taxes. Surcharges: Ideal Practice. The extant Canadian system avoids some of the complexities of the U.S. system, while maintaining substantial fiscal sovereignty for the provinces. Yet the Canadian system does not provide a suitable model for emulation by Argentina, without substantial modification; this is especially true of the sales tax, which exhibits essentially three variants of federal/provincial tax assignment. The remainder of this section describes in broad 26By far the most important cause of tax exporting in the United States is the deduction for state and local income and property taxes that is allowed in computing liability for federal income tax--a provision that is hard to justify. Taxes are also exported to owners of firms that cannot shift taxes to others (customers, suppliers, or labor). Exporting to out-of-state customers is not likely to be important, because of pressures from untaxed competitors. "For a more detailed description, see Boadway (1997b). 21See Bird and Gendron (1998) and literature cited there. 2'An additional burden not found in the U.S. system is created by the combination of the federal VAT and the provincial RST. These two systems operate in quite different ways to eliminate or ameliorate the burden on business purchases. Under the VAT most sales are taxed and business taxpayers claim credits for tax paid on purchased inputs. Under the RST some sales to business are exempt. Compliance with either system is complex and costly and the combination is especially burdensome. 10 outline a system based on surcharges that might better form the basis for reform of the Argentinian system of revenue assignment. The next section provides further details on how this approach might be implemented for individual taxes. Of the four choices involved in tax assignment (which taxes, tax bases, tax rates, and tax administration), the choice of tax rates is clearly the most important for subnational fiscal sovereignty (assuming subnational governments are assigned adequate sources of marginal revenue of some type). The choice of tax rates is what allows the constituents of subnational governments to choose the level of public spending they want. Subnational discretion in the other three areas are much less important for fiscal autonomy and can actually be counterproductive. As the experience of the United States shows, depending on how choices are made, there can be complexities, inequities, distortions of the location of economic activity, and tax exporting. This suggests that an ideal system for many countries (including Argentina) would consist of subnational surcharges levied on a tax base (or tax bases) defined by the central government and administered by the central government, perhaps with input from the provinces. Ideally surcharges would be imposed on some combination of excises, a residence-based income tax, and the VAT (at destination). In this scheme, subnational governments would exercise the all-important choice of tax rates, but most of the complexity, inequities, tax exporting, and locational distortions inherent in subnational choice of tax bases and subnational tax administration would be avoided. In fact, resort to provincial surcharges on source-based payroll taxes and the corporation income tax may be required in order to raise adequate revenues. C. Transition To implement the scheme envisaged here, the central government would a) reduce the tax rates it imposes on various taxes chosen for devolution, in order to make "tax room" for the provincial surcharges, b) calculate the revenues each province would receive from fully utilizing the tax room, c) reduce provincial shares in coparticipation by this amount, and d) allow provinces to levy the chosen taxes--at the rates needed to absorb the tax room or at other rates. As a transition device, it might be desirable to move in stages toward the ultimate objective, by a) initially providing less tax room (and reducing coparticipation by less) than is envisaged as the ultimate target and b) initially requiring that all provincial surcharges be the same (or nearly the same). As experience is gained, these constraints could be relaxed. IV. Designing Surcharges on Individual Taxes This section discusses techniques for imposing subnational surcharges on four taxes: the individual income tax, excises, the company income tax, and the value added tax. It does not discuss subnational taxation of natural resources, because of the belief that such taxes should not be assigned to subnational governments. It does not discuss property taxes, as they are most appropriately assigned to local governments. The discussion of VAT in this section is limited to brief descriptions of well-known problems of alternative methods of implementing "stand-alone" provincial-level VATs. The next section provides a more complete discussion of techniques for implementing a provincial VAT in a country where there is a federal VAT. A. Individual Income/Payroll Tax In theory, the individual income tax might be an attractive source of revenue for the provinces of Argentina. There could be a flat-rate provincial surcharge and a graduated-rate 11 federal tax; both would be imposed on the base defined by the central government. (Actually, it might 'be desirable to have a lower tax threshold for the provincial surcharges than for the federal tax, given the differences in objectives of the two taxes: benefit-related taxation in the case of the provincial surcharge and progressive taxation in the case of the federal tax.30 This would complicate compliance and administration somewhat, as the two thresholds would need to be reflected in withholding tables, as well as rate tables used in completing tax returns.") In fact, under present conditions, allowing the provinces to levy surcharges on the individual income tax would not seem to achieve much. Because the tax threshold is quite high and tax administration is, by all accounts, quite weak, only a small percentage of the (economically active) population currently pays the tax. Thus the individual income tax is not currently the kind of "mass tax" that might be used to provide a source of significant marginal revenues to pay for provincial services. If the individual income tax is to serve as a benefit- related provincial tax, it should be paid by a high percentage of the population.32 Moreover, it would be important to improve tax administration. A better alternative might be to allocate to the provinces the payroll tax, over and above the amount needed to finance pensions under the new system of individual accounts." The primary objection to doing this would seem to be the presence of several cities on or near provincial boundaries. Since payroll taxes are commonly collected at the place of employment, instead of the place of residence, revenues from taxes paid by commuters would flow to the "wrong' provinces.34 A long-run objective might be to merge the payroll and individual income tax systems, allocating to the provinces the capacity to levy flat rate income tax surcharges, while leaving the progressive element at the central level, as described earlier. This would necessitate unification of the two tax bases and the administration of the two taxes--a step that would presumably allow some savings in costs of compliance and administration. Whether this is a sensible objective from the point of tax assignment depends in part on ones view of the nature of the services provided by the provinces and the prevalence of commuting across provincial boundaries. If labor income is thought to be a reasonable proxy for the consumption of such services and if a) most such services are provided where people work, rather than where they live, or b) cross- boundary commuting is not overly important, it would be better to employ the (source-based) "oOne might think of the flat-rate provincial tax as being levied at the "basic" rate on virtually all income, with the graduated rates of the federal tax being levied only on income above a certain level. "Depending on the structure of personal deductions and credits, deductions and credits for personal expenditures, and other features of the individual income tax, the situation could be even worse. This cannot be assessed without a thorough examination of the structure of the individual income tax, which is beyond the scope of this paper. "A tax limited to a small portion of the population invites "tyranny of he minority." Whether this is more problematic for a federal or provincial tax is unclear. "It might be appropriate to remove contributions to individual accounts from the tax statistics, if this has not already been done. Pensions under the residual (old) pay-as-you-go system should be financed from general revenues. There is no principled reason to continue to use the payroll tax to finance those pensions. 34Local input is needed. How much cross-border commuting between provinces is there? 12 payroll tax to finance them, even in the long run. If total income (including non-labor income) were thought to be a better proxy, if most services are provided where people live, or if cross- boundary commuting is important, unification of the income and payroll taxes into a single residence-based tax would be more appropriate, if it can be achieved. But residence-based taxation requires that most taxpayers file income tax declarations, an objective not to be realized- -or even desirable--soon in Argentina.35 For reasons stated above, it does not appear that the individual income ad payroll taxes are likely to be serious candidates for decentralization in the immediate future. B. Excises Excises, for example those on alcoholic beverages and tobacco products, would seem to be an ideal candidate for assignment to the provinces. They may be related to health care services and to highways and roads provided by the provinces. Even if they are not, they are reasonably assigned to the provinces under the principle of subsidiarity in taxation. They would probably be politically acceptable as a means of financing provincial expenditures. Because they are relatively visible, their use by the provinces would help ensure accountability of provincial officials. Provincial excises should be levied by (or on behalf of) the provinces where consumption occurs, not the provinces where production or importation occurs. The ordinary way to implement provincial excises on alcoholic beverages and tobacco products would be to attach distinct tax stamps to excisable products destined for various provinces. The primary problem that might be encountered would be smuggling from provinces with low excises to provinces with high excises. This would tend to limit the differentials between excise rates that could be sustained. Smuggling is likely to be particularly problematic for products with high ratios of value to weight and volume, such as alcoholic beverages and tobacco products. In the case of motor fuels the problem is more likely to take the form primarily of households residing in high- tax provinces filling their cars in neighboring low-tax provinces. This abuse is difficult to prevent, but seems unlikely to be quantitatively significant. Excises will provide only a limited amount of provincial revenues, even if levied at high rates. To maximize the latitude of the provinces to utilize this revenue source, the central government should essentially vacate its use. C. Company Income Tax The company income tax is a not a particularly good way to finance subnational governments, for reasons explained below. It should be seen as a vehicle of last resort, to be used, if necessary, in order to reduce horizontal vertical imbalance. Economic arguments. There is little reason to believe that company income taxes closely reflect the cost of services provided to business. (This is especially true if tax is levied only on "If residence-based taxation were feasible, it might make more sense to reserve surcharges on the individual income tax for use by municipal governments. This is especially true if, as suggested below, provincial surcharges on the VAT are feasible. "The ease of smuggling from Uruguay and other neighboring countries effectively limits the level of excises. Above a certain level, increases in excises are likely to produce less revenue, not more. 13 corporations.) There is no reason to believe that only profitable companies consume public services or that consumption of public services is closely related to company profits. Being origin-based taxes, company income taxes tend to distort the location of economic activity and to accentuate horizontal fiscal disparities. Finally, they are likely to be exported in part to non- resident owners of companies. Administrative arguments. It is inherently difficult to determine the geographic origin of the income of companies operating in more than one province." Firms may manipulate transfer prices to shift income to the provinces where tax rates are lowest. Moreover, it is ordinarily conceptually impossible to divide the income of multijurisdictional companies accurately, because of economic interdependence between activities in various jurisdictions. For this reason, it is common to use formulas to apportion the tax base among jurisdictions. The "apportionment factors" that are commonly used in the United State are payroll, property, and sales; the Canadian provinces use payroll and sales. But using a formula to apportion income is equivalent to taxing whatever is in the apportionment formula at a rate that depends on the nation-wide profitability of the firm, relative to the apportionment factors." This being the case, it may be more appropriate simply to tax the apportionment factors (e.g., payroll, property, and sales) directly. In either event, the tax will have economic effects, including distortion of the location of economic activity, that resemble those of taxes on the factors. Designing surcharges. The only sensible way to implement a provincial tax on company income would be as a surcharge on the tax of the central government. Because of the need for uniformity in tax bases, apportionment formulas, and administrative practices, it would be a mistake to allow the provinces to define their own tax bases, choose their own apportionment formulas, or administer their own taxes. The central government would apply the relevant apportionment formula to the federal tax base to determine the income subject to the tax of the various provinces and then apply the provincial tax rates to determine liability for provincial taxes. D. Value Added Tax The conventional wisdom has been that the value added tax does not lend itself to use by subnational governments, either as a tax administered by such governments or as a subnational surcharge on the VAT of a higher level government." The basic problems involve trade crossing borders between taxing jurisdictions within the nation (hereafter "internal borders" between "It is also generally inefficient to use "geographic separate accounting" for such a purpose, since such accounts generally are not needed for any other reason. "See McLure (1980). "See, for example, Boadway (1 997a) and McLure (1980; 1993). Indeed, even sharing of revenues from a national tax has been thought to involve similar problems; see McLure (1995). Neither the well-known defects of a gross- receipts tax nor the disadvantages of the specific type of gross receipts tax employed by the provinces of Argentina are discussed her. See, however, World Bank (*). Nor is the combination of a federal VAT with provincial retail sales taxes seen to be a viable option, because of the complexity of subjecting taxpayers to two very different tax regimes for relieving business purchases and exports from tax: credits and zero-rating under the VAT, vs. exemption under the retail sales tax. 14 "provinces"),40 especially avoidance of multiple (cascading) taxation of business purchases and the proper (destination basis) treatment of sales to households and unregistered traders." The rest of this section indicates why a "stand-alone" provincial VAT has generally been thought problematic, if not infeasible, whether imposed on the origin or destination basis. Following Poddar (1990, p. 105) these systems are judged on the basis of the following four criteria: lack of interference with the location of economic activity, which is inherent in the destination principle, lack of internal border controls, provincial autonomy over tax rates, and simplicity. Section V explains the operation of two "dual" systems of destination-based federal/provincial VATs; one that is especially attractive for Argentina relies on imposition of a "compensating VAT" on trade between provinces; this tax could be imposed either by the central government or by a consortium of the provinces. Origin-based taxation. Unless imposed at the same rate by all provinces, origin-based taxation distorts the location of economic activity and is difficult to administer (because all transactions must be valued, to prevent artificial attribution of value added to low-tax jurisdictions).42 But limiting all provinces to a single rate undercuts one of the primary reasons for assigning taxing powers to subnational governments, allowing subnational choice of tax rates. Moreover, there is a risk that triangular trade will produce undesirable patterns of fiscal flows, as has happened in Brazil. (There the more affluent states of the south tax imports and the poorer states of the Northeast rebate tax on exports. See Longo, 1982. In order to ameliorate the resulting inequities, lower rates are applied to interstate sales than to intrastate sales.) Destination-based taxation. By comparison, destination-based taxation has been thought to involve one of several undesirable features: internal fiscal frontiers that would interfere with trade between the provinces, the risk that revenues will be lost (if cross-border sales to registered "oMuch of this discussion is based on the excellent paper by Poddar (1990). It is assumed that international trade would be taxed under the destination principle, as is the virtually universal practice. Thus, exports would be zero- rated, and the full value of imported products would be subject to tax. "'It is assumed that unregistered traders are to be treated like households, since they are "outside the system," and thus not eligible for credits for tax paid on purchased inputs. This discussion pertains directly only to tangible products sent to the customer across provincial lines by mail of common carrier or delivered by the vendor or a contract carrier, and not to tangible products that are sold "over the counter" that the buyer then transports to another province for use there. It seems virtually inevitable that the latter transactions will bear tax at the rate prevailing in the province where the sale is made (the province of origin), and not that of the province of destination, except in rare cases, for example, where goods must be registered in the province to be used there, as in the case of automobiles (and perhaps boats and planes). While such origin-based taxation may provide undesirable incentives for "cross-border-shopping" that is problematic in some cases (especially within the Buenos Aires metropolitan area), on the whole it seems not to be a major problem. This issue may deserve further study. It appears that interprovincial sales of digital content over the Internet ("electronic commerce") can be handled in the same way as interprovincial sales of tangible products. The most difficult problems in this area involve sales originating outside the country. It is extremely difficult to tax a transaction in digital content between an unknown customer and a foreign seller (perhaps located in a tax haven), especially if payment is made in untraceable money; See McLure (1997) and (forthcoming). "'Even with uniform rates the value of goods crossing internal borders may be contentious. Values attached to such trade are a matter of indifference to taxpayers, but subnational governments would not be indifferent, as the valuation at the border would affect their tax bases. Thus taxpayers could be caught between tax authorities in the two jurisdictions. 15 businesses are zero-rated and diverted to households and non-registered traders without payment of tax), onerous burdens of compliance and administration, or complicated clearing-house arrangements between provinces of origin and destination. These potential problems are described more fully. Fiscal frontiers would impede the operation of a single market within a country, if utilized to implement destination-based taxation. Destination-based taxation is achieved for international trade through the use of "border tax adjustments." That is, imports are subject to VAT and exports are zero-rated (and tax collected on previous stages of the production- distribution process are refunded to exporters, if they exceed liability for VAT on domestic sales). Implementation of this system requires fiscal frontiers, in order to collect VAT on imports and verify that reported exportation actually occurs. This is ordinarily not a problem for international trade, since goods commonly stop at the border for other reasons, including collection of customs duties and inspections for health and safety. (Table 1 illustrates the application of border-tax adjustments to trade between provinces. The illustration is explained in footnote *51.) Zero-rating/deferred payment provides a conceptually attractive approach; the European Union employs it as a "transitional" scheme, and Libonatti and Salinardi (1994) have proposed it use in Argentina.43 Zero-rating of sales to registered vendors in other provinces would eliminate tax in the province of origin, and use of deferred payment by the province of destination would avoid the need for fiscal frontiers between provinces. (Since interprovincial sales would not be taxed, registered vendors who import from another province would have no VAT to claim as a credit against tax on sales. Tax is "deferred" from the time of importation to the time tax is paid on sales.) Its operation is illustrated in Table 1 and is described further in the next section. Sales to households and non-registered traders in other provinces could be taxed at the rate in the province of destination (with revenue submitted to that province) or, to simplify matters, at the rate in the province of origin (which would receive the revenue). The primary risk is that goods that have been zero-rated as sales to a registered trader in another province will be diverted to use by households or unregistered traders, perhaps even in the province of origin. Taxation by the vendor at the tax rate of the province of destination (with submission of revenue to that province) also achieves the conceptually correct result, again without border controls." Moreover, it avoids the need to treat cross-border sales to registered traders and to households and non-registered traders differently. The primary drawback of this approach is its enormous complexity. It would be necessary a) for every invoice to show the tax calculated using the tax rate of the province of destination and b) for each vendor to file a tax return in each province to which it makes sales. It does not seem worthy of serious consideration. Taxation by the vendor at the rate of the jurisdiction of origin, with a tax credit clearinghouse seems also to be inordinately complicated. (Moreover, sales to households and non-registered traders would be taxed at the "wrong" rate and tax revenue would go the "wrong" province, if such sales were accorded the same treatment.) Registered purchasers would be allowed credits for all tax on inputs, but credits would need to be identified by provinces "'Keen and Smith (1996) describe the system used in the European Union and discuss alternative arrangement. "Poddar (1990, pp. I10-111) calls this a joint national-state VAT when imposed in the context of a two-tiered nationallsubnational VAT. 16 collecting the input tax--and thus identified by vendors claiming the credits, so that destination provinces granting credits could be reimbursed by provinces of origin that had originally collected tax for which credit is granted. Since provinces of destination are to be reimbursed for the credits they grant for taxes paid to provinces of origin, they have little incentive to verify the validity of credits claimed for tax paid on interprovincial sales. Again, this option does not seem worthy of further consideration. V. The Compensating VAT on Interprovincial Trade It appears that the above analyses, which have commonly been based on the (usually implicit) assumption of taxation by a single level of government, may overstate the difficulty of implementing a destination-based subnational tax in the context of a national VAT. If this is so, and a subnational VAT is possible, the provincial VAT could be a major source of revenues for the provinces of Argentina. Moreover, it would provide increased provincial autonomy to raise revenue at the margin, as well as a source of "own" revenue, via control over tax rates. In recent years there have been several proposals in Argentina and Brazil for a dual provincial/central VAT. In Argentina Libontti and Salinardi (1994) have proposed provincial use of zero-rating and deferred payment, and Gonzalez Cano (1996) and Fenochietto (no date) have proposed an "IVA compartido" or shared VAT that would combine zero rating/deferred payment with an additional "compensating" VAT on interprovincial sales to registered traders.45 The remainder of this section discusses these alternative ways to implement a provincial VAT as part of a dual federal/provincial system, focusing primarily on the purpose and function of the compensating VAT on trade between provinces. The next section discusses whether and how the provinces should participate in administration of the dual VAT. Zero-rating/deferred payment in a dual system. The dual system used in Canada by the federal government and the province of Quebec (hereafter "the Quebec system") relies on zero- rating and deferred payment for the taxation of trade between provinces.6 Bird and Gendron (1998) assert that the control provided by the presence of the federal tax is enough to prevent unacceptable abuse.47 This argument may not be totally convincing, especially if tax administration is not as good as in Quebec.48 Primary reliance on provincial administration would accentuate the problem, due to the risk that communication and coordination between 45Cano (1996), and Fenochietto (no date), apparently draw on a proposal originally made for Brazil; see (Version, 1995). 46Bird and Gendron (1998) discuss the use of zero-rating/deferral by the governments of Canada and Quebec. 47At the risk of excessive proliferation of labels, one might refer to the Quebec system as employing "unprotected zero-rating/deferred payment" and the proposed system as involving "protected zero-rating/deferred payment." 48It is worth noting that in this case the province administers both the federal and provincial VATs. Being explained by the ethnic situation that is unique to Quebec (the only French-speaking province in Canada), this precedent seems to have little or no relevance for Argentina. While the provinces of Argentina may desire autonomy, this desire is not motivated by linguistic differences, as in Quebec. The federal component of any dual system employed in Argentina--and perhaps the provincial component--would presumably be administered by the federal tax authorities. In some countries (but not Canada) there is a further issue: whether the provinces trust the central government to deliver the revenue collected on their behalf; see the discussion of direct deposit of provincial revenues in the next section. 17 provincial administrations would not be as good as within a centralized administration. Finally, it would appear that this system would work best in countries where relatively little commerce crosses provincial boundaries. Where there is substantial interprovincial trade, the administrative and compliance burden might make the system unworkable. Decomposing the IVA Compartido (shared VAT). An IVA compartido could contain any of several separable elements: a) a common tax base for federal and provincial VATs; b) either a uniform provincial rate or provincial autonomy over tax rates; c) centralized administration, joint federal/provincial administration, or parallel federal and provincial administration; d) separate deposit of the provincial share of tax payments directly in provincial bank accounts; and e) collection of an additional VAT on interprovincial sales to registered traders--what this paper calls a "compensating VAT"' on such sales (perhaps usefully translated into Spanish as "perception" on interprovincial sales, by analogy to the existing "perception" on sales to small business). It should be axiomatic that the bases of the federal and provincial VATs are identical, for reasons stated in Section III above. The case for direct deposit of provincial shares of tax liability is also persuasive; there is no principled reason revenues should pass through the bank account of the federal government before reaching provincial accounts. The remaining issues are more controversial. The remainder of this section is devoted to an explanation of the operation of the compensating VAT on interprovincial trade. The next section discusses the choice between centralized administration, joint federal/provincial administration, and parallel federal and provincial administration. It appears that most proposals for an IVA compartido anticipate a uniform provincial VAT rate. This seems to impose an undesirable and unnecessary limitation on provincial fiscal autonomy. First, while the importance of giving the provinces a source of "own" revenue should not be underestimated, it is also important to let the provinces choose their own tax rates. Second, even though a wide divergence of provincial rates is potentially troublesome and perhaps unsustainable, because of the incentives for (and against) cross-border shopping, this should not be allowed to force the provinces into a legal strait-jacket of uniform rates. Finally, there seems to be no compelling administrative reason that provincial rates should be uniform. The remainder of this section assumes provincial autonomy over rates to be an important and feasible objective. A compensating VA Ton interprovincial sales to registered business would avoid most of the problems associated with stand-alone VATs and with Quebec's system of "unprotected" zero-rating/deferred payment. Under this scheme ("the proposed system") cross-border sales to registered traders would be zero-rated by the province of origin and subject to deferred payment of VAT by registered businesses in the province of destination, as in the Quebec system. The risk that zero-rated sales would be diverted to households or to non-registered traders would be avoided (or reduced) by the collection of a federal "compensating VAT" on interprovincial sales to registered traders. (For argument's sake, it is assumed initially that the compensating VAT would be collected by the federal government; this assumption is examined below.) Likewise, registered taxpayers would be allowed to claim credits for the compensating VAT, as well as the 18 "ordinary" federal VAT in computing federal tax liability; in principle, the two taxes need not be differentiated in calculating liability to the federal government. Assuming that the same compensating VAT rate would be applied to interprovincial sales to households and non-registered traders, as suggested below, vendors would need to deal with only three rates (plus zero, for exports), 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 rate of the local province on sales within the province, and the compensating VAT rate on sales to other provinces. There would be no need to deal with the tax rates of any other provinces or to engage in interprovincial clearing of tax credits. As with the Quebec system (the dual system based on zero-rating and deferred payment, without the compensating VAT), the proposed system would work best if there were not large amounts of interprovincial trade. Even so, it seems less vulnerable than the "unprotected" Quebec system if such trade is substantial. An illustration. Table 1 illustrates this method. It assumes a three-stage production- distribution process, with value added of 100 in each stage and total value of sales to households of 300; the first column summarizes these transactions. It is assumed that province A imposes a VAT of 4 percent and province B imposes a VAT of 8 percent." The federal government imposes two taxes: an "ordinary" VAT of 20 percent and a compensating VAT of 6 percent on interprovincial sales to registered traders. (This rate of the compensating VAT is assumed for convenience to lie midway between the two provincial rates. Alternatively, it could equal the highest provincial rate, the lowest provincial rate, or any rate between, without affecting the basic conclusions presented here. The choice or this rate is discussed further below.) Gross 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. The second set of columns shows the result if all three stages occur in province B; it describes destination-based taxation and thus serves as a useful benchmark." The province collects VAT of 24 (8 percent of total value of sales to ultimate consumers of 300) and the federal government collects VAT of 60 (20 percent of 300). The third set of columns illustrates the situation in which the first two stages occur in province A and only the third occurs in province B.52 Thus, Stage 1 occurs in province A, Stage 491t should be noted that international trade poses no conceptual problem in this case. Provincial tax is collected on imports and credit is allowed for the tax paid on imports. Similarly, exports are zero-rated for the provincial tax, as well as the federal tax. soTable 2 illustrates a system involving a uniform provincial VAT rate of 6 percent and a compensating VAT on interprovincial trade of 6 percent. It is not featured in the text because doing so might obscure the way the compensating VAT operates. "Table 3 provides another benchmark; it illustrates the use of border tax adjustments to implement destination- based taxation in the case described by the third set of columns in Table 1. In it, province A zero-rates exports, as in the example of Table 1. But province B collects VAT of 16 on imports of 200 from province A, instead of employing deferred payment, as in Table 1. Since the producer at Stage 3 can claim credit for the tax paid on imports, the result is the same as in Table 1. "One could easily add columns showing the results if the first two stages occur in province B and the last in province A or if each of the three stages occurs in different provinces. In all cases, the federal government collects 60, the province of destination collects an amount equal to the product of its tax rate and total sales to consumers 19 2 involves an interprovincial sale by producers in province A to registered traders in province B, and Stage 3 involves sales to consumers in province B. Province A collects VAT of 4 on the production of 100 in Stage 1. Since it zero-rates the interprovincial sale that occurs at the second stage, it refunds to the producer at Stage 2 the VAT of 4 collected at the first stage. Province B collects VAT of 24 on the entire sale of 300 to households (allowing no credit for tax on prior stages), replicating the result in the second set of columns, as is appropriate under a destination- based VAT. This is exactly the result that occurs under the system of zero-rating and deferred payment employed in the European Union and proposed for Argentina in Libontti and Salinardi (1994) 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 compensating VAT of 12 on the interprovincial sale of 200 occurring at the second stage, but allows a credit for it at stage 3, thereby eliminating any net liability when the two stages are consolidated. In short, the final result for all three governments (A, B, and federal) is the same as if a) all production occurred in province B or b) zero-rating/deferred payment had been used without the compensating VAT. Choosing the rate of the compensating VAT. Several considerations would seem to be relevant to the choice of the rate of the compensating VAT. Setting this rate at the level of the lowest provincial rate would leave some latitude for diversion of products to household use and to non-registered traders. Moreover, if this rate were also applied to sales to households and non- registered traders, as suggested below, there would be some discrimination against local merchants, who would be subject to higher rates. Finally, there would be an incentive to use mail-order purchases from low-tax provinces as a surrogate for cross-border shopping in high-tax jurisdictions. These problems would seem to be serious primarily if the lowest provincial rate were well below the typical rate." Setting the rate equal to the highest provincial rate would prevent diversion of products to untaxed uses, but would burden interprovincial commerce directed at provinces 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 provinces. On balance a rate that is near the (weighted) average of provincial rates would seem appropriate." (e.g., 4 percent of 300 if A is the province of ultimate destination), and other provinces collect nothing, as is required for destination-based taxation.. "It appears that choices of where to make purchases is dictated more by availability and convenience than by small differences in tax rates. Even so, the effects of such differences should not cavalierly be dismissed. 'To simplify compliance, the rate of the compensating VAT should be a whole number. For the same reason, it should be applied to sales prices exclusive of the ordinary federal VAT. (There would, by assumption, be no provincial VAT on interprovincial sales.) Indeed, both the federal and provincial VAT rates should be applied to this basd, in order that aggregate rates can be determined by adding the various rates. (Under this approach a federal rate of 20 percent and a provincial rate of 8 percent produce an aggregate rate of 28 percent, whereas if one of the taxes is included in the base of the other, the aggregate rate is 29.6. The point is not merely that the aggregate rate is higher; this can be overcome be adjusting the rates; the point is that whole numbers are easier to use in calculations of tax than are decimal fractions. 20 Taxation of interprovincial sales to households and unregistered traders. The treatment of interprovincial sales to households and unregistered traders poses a dilemma. It would be possible, but probably not desirable, for vendors to apply the tax rate of either the province of origin or the province of destination to such sales (and submit revenues to that province). The first option gives the wrong answer (origin-based taxation for such sales) and the second would be needlessly complicated. It seems preferable, especially if provincial VAT rates cluster fairly close together, simply to levy the compensating VAT at the same rate to these sales as to sales to registered businesses. As noted above, this solution is not perfect. But it is probably acceptable, since interprovincial sales to households and unregistered traders are not likely to be important, except in a few instances." Moreover, it seems likely that the threat of cross-border shopping will prevent substantial divergence of tax rates between provinces where this problem would otherwise be the greatest. If provincial VAT rates were to diverge substantially, there would be an incentive for mail-order purchasing from low-tax jurisdictions, if such sales were taxed at the rate of the province of origin of sales. Imposing the compensating VAT on these sales would reduce this incentive. Under this alternative vendors would need to distinguish between only three types of sales: intraprovincial sales (taxed at the federal and provincial VAT rates), sales to customers (whether registered traders, households, or unregistered traders) in other provinces (taxed at the ordinary federal rate and the compensating VAT rate and zero-rated for purposes of the provincial VAT), and sales for exports from the nation (zero-rated). This seems somewhat simpler than applying the compensating VAT only to interprovincial sales to registered traders and levying the provincial VAT to all other domestic sales. Under that alternative it would be necessary to differentiate between interprovincial sales to registered traders and other interprovincial sales. Since revenues from the compensating VAT on interprovincial sales to households and unregistered traders would initially flow to the federal government under the proposed scheme, it would be necessary to divide them among the provinces. A natural division would be in proportion to net provincial VAT revenues from other sales. This would combine the simplicity of a single rate on all interprovincial sales with a fair division of revenues. Moreover, it would not undermine incentives to set tax rates with regard to the value of public services, because provincial revenues from the compensating VAT on interprovincial sales would be linked directly to provincial revenues from intra-provincial sales, and thus to tax rates. The problems of excess credits/refunds. One potentially troublesome feature of the scheme proposed here is the likelihood that many firms will have excess credits for, and thus deserve refunds of, provincial VAT. For example, in the example of Table 1, the firm operating at Stage 2 should be able to receive refund of the VAT of 4 it has paid to province A when it makes sales to province B. Before examining the likely importance of this problem and seeking ways to reduce its severity, it is worthwhile to place the problem in perspective, by noting that exactly the same problem exists in the case of the "unprotected" zero-rating and deferred ssThis solution is likely to be problematic only where large cities straddle provincial boundaries and VAT rates in contiguous provinces differ substantially. If experience under the proposed system were to prove to be unacceptable, vendors who make interprovincial sales to households and unregistered traders in excess of some minimum amount could be required to collect the tax of the province of residence of the customer, as indicated by shipping or delivery addresses, and remit it to that province. This issue deserves further study. 21 payment systems found in the European Union and Quebec and proposed for Argentina in Libontti and Salinardi (1994). Moreover, it would be vastly less important than in Brazil, where for historical reasons the state VAT rate would almost certainly be far higher than in Argentina. The problem of excess credits and refunds is likely to be most important in sectors that produce predominantly for export to other provinces (or for export from the nation--a problem that exists in any destination-based VAT), using purchased inputs that have not been zero-rated for purposes of the provincial VAT (because they are bought from other provinces)." Products that come readily to mind are primary products: agricultural products and petroleum and other products of mining. To the extent agricultural products are tax-exempt, they pose little problem of this type. Petroleum and other mining are more problematic, but it would be possible to alleviate the problem by zero-rating (for federal as well as provincial purposes) the purchase of all specialized oil-field, mining, and refining equipment. Given the location of vineyards, breweries, cigarette factories (primarily in the province of Buenos Aires), and perhaps other production facilities, it might be desirable to extend the same type of treatment to them. Another way to deal with excess credits--or perhaps a supplementary way, to be employed where the approach described above is not sufficient to eliminate excess credits--is to allow excess credits/refund claims against the provinces to be offset against liabilities for the federal VAT. While there would be a clearing of net liabilities between the federal government and the provincial governments, this would be relatively limited in purpose and scope and would not approach the pervasiveness and complexity of a clearinghouse system in which provincial credits are netted (described in Section IV). Summary appraisal. It appears that the use of a compensating VAT on interprovincial sales would cut the Gordian knot that has stymied assignment of the VAT to subnational governments throughout the world." If so, Argentina could probably have fiscally strong provinces. s"Zero-rating of interprovincial sales under the provincial tax results in liability for the compensating VAT. This tax can be claimed as a credit against the compensating VAT on sales. "Thus the following statement from World Bank (1 996a, p. 24) may not be totally persuasive for two reasons: "International experience suggests that in developed countries, if local governments are to be both large spenders and less dependent on grants, subnational government must gain access to the personal income tax...." First, the individual income tax may not provide adequate revenue to provide financial independence of subnational governments in some developing countries, including Argentina in the foreseeable future. Second, despite conventional wisdom to the contrary, it may be possible to use the VAT to provide this financial independence. Whereas subnational governments in most of the developed countries of Western Europe rely heavily on income taxation, those in Austria, Canada, Japan, Spain, and the United States rely heavily on sales taxes. South Africa has fiscally weak provinces for political reasons, despite having included in its constitution tax assignments that follow quite closely the prescriptions presented here (except those in the present discussion of the VAT), including allowing provincial surcharges on the individual income tax; the ruling African National Congress does not want strong provinces, since two of the nine provinces would be controlled by opposition parties. Given the constitutional assignment of taxing powers, South Africa would inevitably have fiscally weak provinces, because the taxes assigned to the provinces do not provide adequate revenues. At the time the post-apartheid constitution was written, provincial surcharges on the VAT were thought to be unworkable (because the possibility of a compensating VAT on sales between provinces had not been considered) and provincial surcharges on the company income tax were deemed to be undesirable. 22 VI. Administrative Issues It is common for surcharges of subnational governments to be administered by national governments; in some countries subnational governments administer national taxes, but this is less common. (Administration of the VAT by the German laender and the Canadian province of Quebec are notable examples.) There appears to be concern that the typical arrangement--federal administration of provincial VAT--would not be optimal for Argentina." This section addresses this issue, focusing on administration of the VAT, which Section IV argues is the most appropriate potential source of substantial provincial revenue, assuming inclusion of the federal VAT on trade between provinces, described in Section V. It describes the concerns of provincial officials, some of the potential costs of dual administration, and possible solutions to the problems identified. Several types of potential concerns can be readily identified: a) that the federal government would not pay to the provincial governments the revenue collected on their behalf, b) that the federal tax administration is weak, and c) that the federal and provincial governments might have different priorities in the assignment of scarce administrative resources. On the other hand, dual administration (by both federal and provincial authorities) raises both the specter of costly duplication of administrative efforts and overly burdensome compliance and the possibility that administration would not be consistent between provinces--and not consistent between them and the federal government. Collection. It seems to be appropriate to distinguish between two types of functions: the collection of taxes and the auditing of taxpayers." The risk that the federal government would not pay to the provincial governments the revenue collected on their behalf can easily be addressed by having taxpayers deposit taxes directly in the bank accouits of each government. A greater risk is that taxpayers might meet their obligations to the federal government, but not to provincial governments, and that the federal government would condone this behavior. It would be desirable to examine ways to prevent this.o Weakness offederal administration. Some provincial officials appear to be unwilling to rely on the federal tax administration, which they claim to be weak.' Thus they may want to "It is assumed that the alternative of provincial administration of the federal VAT would not be a realistic alternative for Argentina. Certainly, it does not seem worthy of consideration, as it would entail many of the weaknesses of dual (federal/provincial) administration, without its strengths. "This two-way division of activities is not intended to be fully adequate; the purpose is to distinguish between collection and other activities. It might be useful to distinguish further between collection of funds deposited normally" or "voluntarily" by the taxpayer, essentially on a timely basis, and collection of taxes paid "involuntarily" as a result of enforcement procedures employed by the tax administration. 'If Argentina uses a system of certificates of "paz y salvo" or its equivalent, there might be a requirement that the taxpayer fulfill obligations to pay provincial taxes, as well as federal taxes, in order to obtain the certificate. There appears to be no workable way (not to mention no legal way) of requiring that bank deposits to discharge tax liabilities be split in fixed proportions between provincial and federal governments (or that a mmimum fraction be paid to provincial governments); in the case of the VAT the proper proportions will vary, depending on the relative importance of the taxpayer's interprovincial sales. "Provincial officials may prefer federal administration if the federal tax administration is more effective than the provincial administration, both because of the revenue implications and because the federal government is likely to be blamed for tough administration. They may prefer provincial administration if it is more effective. In either 23 retain the right to audit taxpayers, at least in the short run. Because of the high cost of dual administration--and the implied high cost of compliance, discussed below, this is not an attractive solution in the long run. Indeed, given the likelihood that provincial administrations will become entrenched, it is troublesome, even in the short run. Divergent priorities. Federal and provincial tax administrations may have quite different priorities. First, the federal tax administration may prefer to concentrate its resources on the taxes that have the greatest marginal revenue yield for the federal government; these may not be the taxes that yield the most marginal provincial revenues.62 Second, in the administration of a particular tax the federal government will presumably prefer to concentrate its activities where the marginal federal revenue yield is highest. Since economic activity is not homogeneous throughout the country, such a policy might imply that the federal tax administration would pay less attention than provincial tax administrations to taxpayers in some provinces. (What might be a "big fish" for the province might be "small fry" for the federal tax administration.) To the extent the federal administration ignored taxpayers in the poorer provinces, horizontal fiscal disparities would be accentuated. Third, the assumption that the federal tax administration allocates its resources rationally may not be totally valid. Thus the administration's coverage of the taxpayer population in various parts of the country may vary widely for reasons that have no rational explanation. (It may be difficult to hire trained auditors to work in the hinterland; alternatively, coverage in out-lying some parts of the country may be "denser" than in the major cities.) Incentives under deferred payment systems. Provincial administration suffers from a distinct disadvantage; provincial tax administrations may lack the incentives to work together as closely as the different provincial offices of the federal administration should. (This is not to say that they will.) Exporting provinces have considerable interest in certifying that goods said to be sent to other provinces actually leave their territory, since they are to be zero-rated. In the case of exports from the country, a province can rely on the federal tax administration to protect its interests. In the case of exports to other countries this source of information is unavailable under a system of provincial administration; at best, exporting provinces must rely on the cooperation of importing provinces. But importing provinces have relatively little reason to pay attention to the tax treatment of business purchases acquired from another province, since such purchases carry no tax credits. (Indeed, the government of the importing province might have an incentive to assist its resident firms evade taxes on purchases from other provinces.) Conversely, an exporting province might have relatively little interest in knowing whether a given sale to a purchaser in another province was for a legitimate business purpose, since it would not gain any revenue, in any event. (For example, auditors in the exporting state might not care that a car dealer in another province bought a truck-load of wine; the problem of determining the legitimacy of the purchase as a business expense would be the problem of the tax administration case, provincial administration is likely to be preferred, all else equal, because of the patronage and local control involved. 'The existence of pervasive "coparticipation" conditions the applicability of this statement, since it implies that the federal government does not keep the revenues from any tax that enters the coparticipation pool. Yet, it is possible that, following reorganization of intergovernmental fiscal relations, the federal and provincial government will have different priorities. 24 in the importing province. (This is one reason it might be better to tax interprovincial sales to consumers on an origin basis; it would give tax administration more incentive to verify the purpose of such sales.) By comparison, the federal tax administration would have an incentive to pay attention to both sides of interprovincial transactions if it were collecting the compensating VAT, since it would want to be sure it was not allowing credits for compensating VAT that it had not collected. Duplication of administration and compliance. Under the present gross receipts tax, which provides for revenues to be split evenly between the provinces of origin and destination of taxable sales, a given taxpayer may be required to file a tax return in (and be subject to audit by) all 24 provinces (as well as by the federal government, for VAT). (Both origin and destination provinces would have the right to require a return and conduct an audit.) Under the proposal for federal administration of provincial VAT (with the compensating VAT on interprovincial sales), this system would be replaced by a single return filed with the federal government and an audit conducted by the federal government on behalf of itself and all the provinces where the taxpayer makes intraprovincial sales. By comparison, under a system of dual administration of the VAT, the taxpayer would need to deal with the federal administration and with the administration of each province where it has production or makes intraprovincial sales." Instead of filing a single tax return with the federal administration containing information for all relevant provinces, the taxpayer would need to file a return with each province where there was an actual or potential tax liability. Similarly, it would be potentially liable to audit by the tax authorities of all these governments. Potential inconsistencies. Tax administrators in all provinces must apply the law consistently if there are not to be gaps and overlaps in tax bases. Among the important issues that might be subject to differing interpretations by tax authorities in different provinces are the definition of taxable supply, the place of supply, and the time of supply. While all these would be defined in a single law, they might not be interpreted uniformly by provincial tax administrations. Rulings by auditors, appeals officers, and courts would presumably dominate rulings at the provincial level, but inconsistencies could occur in the absence of federal rulings. The treatment of interprovincial transactions is especially important; Section V proposes that such transactions should be zero-rated by the province of origin and subject to the compensating VAT.' (For sales to households and unregistered traders the compensating VAT would be a final tax. Purchases by registered traders would be taxed by the province of destination on a deferred payment basis.) With federal tax administration, the flow of information on interprovincial sales would be internal to the federal tax administration. By comparison, in a dual system of tax administration, the information on interprovincial transactions would need to flow between provincial administrations. "Production would be enough to subject a taxpayer to audit by a province, even if all output were sold to other provinces, and thus zero-rated by the province, since it would be necessary to determine whether any (and how much) was sold to purchasers in the province. "It appears that the problem would be vastly more difficult in the absence of the compensating VAT. The province of destination would need to rely on the federal tax administration to see that the province of origin was not making zero-rated sales to households in its jurisdiction. 25 Joint federal-provincial oversight. Many of the potential concerns of the provinces could be addressed by giving them a role in the supervision of the federal tax administration-- essentially making it a national agency that is accountable to both the federal and provincial governments, instead of a federal agency that is accountable only to the federal government. This would facilitate combining the benefits of uniformity and the avoidance of duplication of effort and the benefits of respect for provincial priorities. Staffing. Substitution of federally administered provincial surcharges on the VAT for the provincial gross receipts tax creates both the danger and the opportunity of redundant provincial employees. Among the dangers are increased unemployment and loss of patronage. The opportunities include making available to the federal tax administration a pool of trained and competent tax administrators and the possibility of weeding out poorly trained and incompetent provincial tax administrators. On balance it appears that the advantages far outweigh the disadvantages. Care must be taken, however, not to guarantee that the federal tax administration will not hire all existing provincial tax administrators who are displaced, regardless of their qualifications and competence. Jurisdiction over the compensating VAT The discussion above assumes that the compensating VAT would be administered by the federal government. Under a system of dual administration of the two VATs (federal and provincial), one could imagine' an alternative arrangement: joint administration by all the provinces. This arrangement seems to have several disadvantages, including the need to channel information on transactions to yet another administrative agency and the possibility of audit by yet another organization. Moreover, since the compensating VAT yields no net revenue (except for that on sales to households and unregistered traders, which is to be divided among the provinces on a pro-rata basis) this arrangement could not be used to solve one of the most pressing problems, that of excess credits/refunds of provincial VAT. On balance, this does not seem to be a good idea. 26 REFERENCES Asian Development Bank, Fiscal Transition in Kazakhstan (Manilla: Asian Development Bank, forthcoming). Bird, Richard M., and Pierre Pascal Gendron, "Dual VATs and Cross-Border Trade: Two Problems, One Solution?" International Tax and Public Finance, Vol. 5 (1998), pp. 429- 42. Boadway, Robin, "Reforming the Fiscal Arrangements in Argentina: Lessons from Industrialized Federations," xeroxed, March 4, 1997.(a) Boadway, Robin, "Tax Assignment in the Canadian Federal System," in Neil A. Warren, editor, Reshaping Fiscal Federalism in Australia (Sydney: Australian Tax Research Foundation, 1997), pp. 61-90.(b) Boothe, Paul, editor, Reforming Fiscal Federalism for Global Competition: A Canada-Australia Comparison (University of Alberta Press: Edmonton, 1996). Collins, D. J., editor, Vertical Fiscal Imbalance and the Allocation of Taxing Powers (Sydney: Australian Tax Research Foundation, 1993). Cnossen, Sijbren, "Interjurisdictional Coordination of Sales Taxes," in Malcolm Gillis, Carl S. Shoup, and Gerardo P. Sicat, Value Added Taxation in Developing Countries (Washington: The World Bank, 1990), pp. 43-57. Courchene, Thomas, "Towards a Federal-Provincial Fiscal Pact," 1998. Duncan, Harley T., and Charles E. McLure, Jr., "Tax Administration in the United States of America: A Decentralized System," Bulletin for International Fiscal Documentation, Vol. 51, No. 2 (February 1997), pp. 74-85; also forthcoming in Spanish translation in Hacienda Publica Espailol. Fenochietto, Ricardo, "El IVA Compartido: Una Herramienta LOtil para el Reemplazo del Impuesto sobre los Ingresos Brutos y la Descentralizaci6n de Tributos, xeroxed, no date. Gonzalez Cano, Hugo, "La Reforma Tributaria de Brasil y Posible Aplicaci6n del Nuevo IVA Federal y Estadual para el Reemplazo del Impuesto a los Ingresos Brutos," Boletin de la DGI, No. 513 (September 1996), pp. 1391-97. James, Denis, Intergovernmental Financial Relations in Australia (Sydney: Australian Tax Research Foundation, 1992). Keen, Michael, and Stephen Smith, "The Future of the Value Added Tax in the European Union," Economic Policy, Vol. , No. (October 1996), pp. 373-420. Libonatti, Oscar, and Mario Salinardi, "IVA Provincial: Una Propuesta para Su Implementaci6n," Buenos Aires, 1994. Longo, Carlos, "Restricted Origin Principle under Triangular Trade Flows: Implications for Trade and Tax Revenues," Journal of Development Economics, Vol. 10 (1982), pp. 103- 12. Martinez-Vasquez, Jorge, Charles E. McLure, Jr., and Sally Wallace, "Subnational Fiscal Decentralization in Ukraine," in Decentralization of the Socialist State: Intergovemmental Finance in Transition Economies, edited by Richard M. Bird, Robert D. Ebel, and Christine I. Wallich (Washington: World Bank, 1995), pp. 281-319. 27 McLure, Jr., Charles E., "State and Federal Relations in the Taxation of Value Added," Journal of Corporation Law, Vol. 6 (1980), pp. 127-39. , "The Brazilian Tax Assignment Problem: Ends, Means, and Constraints," A Reforma Fiscal no Brasil, the proceedings of the International Symposium on Fiscal Reform, Sao Paulo, Brazil, September 6-10, 1993, pp. 45-71. McLure, Jr., Charles E., "The Sharing of Taxes on Natural Resources and the Future of the Russian Federation," in Christine Wallich, editor, Russia and the Challenge of Fiscal Federalism (Washington: The World Bank, 1994), pp. 181-217. McLure, Charles E., Jr., "Revenue Assignment and Intergovernmental Fiscal Relations in Russia," in Edward Lazear, editor, Economic Reform in Eastern Europe and Russia: Realities ofReform (Palo Alto, CA:, Hoover Institution Press, 1995), pp. 199-246.(b) McLure, Charles E., Jr., Christine Wallich, and Jennie I. Litvack, "Special Issues in Russian Federal Finance: Ethnic Separatism and Natural Resources," in Richard M. Bird, Robert D. Ebel, and Christine I. Wallich, editors, Decentralization of the Socialist State: Intergovernmental Finance in Transition Economies (Washington: World Bank, 1995), pp. 379-404. Mieszkowski, Peter, and Eric Toder, "Taxation of Energy Resources," in Charles E. McLure, Jr., and Peter Mieszkowski, editors, Fiscal Federalism and the Taxation of Natural Resources (Cambridge: Lexington Books, 1983), pp. 65-91. Musgrave, Richard A., The Theory ofPublic Finance (New York: McGraw Hill, 1959). Musgrave, Richard A., "Who Should Tax, Where, and What?," in Charles E. McLure, Jr., editor, Tax Assignment in Federal Countries (Canberra: Centre for Research on Federal Financial Relations, 1983), pp. 2-19. Oates, Wallace E., Fiscal Federalism (New York: Harcourt Brace Jovanovich, Inc., 1972). Oates, Wallace E., "Federalism and Government Finance," in John M. Quigley and Eugene Smolensky, editors, Modern Public Finance (Cambridge, Mass.: Harvard University Press, 1994), pp. 126-5 1. Oates, "On the Reform of Fiscal Federalism in Argentina," xeroxed, March 1997. Poddar, Satya, "Value-Added Tax at the State Level," in Malcolm Gillis, Carl S. Shoup, and Gerardo P. Sicat, Value Added Taxation in Developing Countries (Washington: The World Bank, 1990), pp. 104-12. Saeigh, Sebastian, and Mariano Tommasi, "Argentina's Fiscal Institutions: A case study if the transactions-cost theory of politics," prepared for a conference on Modernization and Institutional Development in Argentina," Buenos Aires, May 20-21, 1998. Varsano, Ricardo, "A Tributagio do Com6rcio Interestadual: ICMS versus ICMS Partilhado," Texto para Discussdo No. 382, Instituto de Pesquisa Econ6mica Aplicada, Brasilia, Setembro de 1995. Warren, Neil A., editor, Reshaping Fiscal Federalism in Australia (Sydney: Australian Tax Research Foundation, 1997). World Bank, "Argentina Provincial Finances Study: Selected Issues in Fiscal Federalism," Vol. I; The Main Report, July 12, 1996.(a) World Bank, "Argentina Provincial Finances Study: Selected Issues in Fiscal Federalism," Vol. II; Technical Annexes, July 12, 1996.(b) 28 Table 1 Illustration of Compensating VAT on Interprovincial Sales to Business: Unequal Provincial VAT Rates Tax rates: "Ordinary" Federal VAT Rate = 20%; Provincial VAT Rates: 4% in A and 8% in B Compensating VAT Rate = 6%; Assumptions: Column 2, All Production in Province B Column 3: Two Stages in Province of Origin; One Stage in Province of Destination Transactions Production: All Production: Origin (Stages 1 & 2): A (Purchases, Stages in Destination (Stage 3): B Sales, and Province B Provincial Federal Value Prov. Federal Tax: Compen- Ordinary Total Added) Tax: Tax (A) (B) sating Federal Federal Stage 1: {This stage occurs in province A} a. Purchases/Credits 0 0 0 0 n.a. n.a. 0 0 b. Sales/Tax 100 8 20 4 n.a. n.a. 20 20 c. Value added/ 100 Net tax (c= b-a) 8 20 4 n.a. n.a. 20 20 Stage 2: d. Purchases/Credits 100 8 20 4 n.a. n.a. 20 20 (d=b) e. Sales/Tax 200 16 40 {Interprovincial sale from A to B} 0 n.a 12 40 52 f. Value added/ 100 Net tax (f-e-d) 8 20 -4 n.a. 12 20 32 Stage 3: {This stage occurs in province B} g. Purchases/Credits 200 16 40 n.a 0 12 40 52 (g=e) h. Sales/Tax 300 24 60 n.a. 24 n.a. 60 60 i. Value added/ 100 Net tax 8 20 n.a. 24 -12 20 8 (i=h-g) j. Total tax 24 60 0 24 0 60 60 (j=c+f+i) n.a. : not applicable 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. 29 Table 2 Illustration of Compensating VAT on Interprovincial Sales to Business: Equal Provincial Rates Tax rates: "Ordinary" Federal VAT Rate= 20%; Uniform Provincial VAT Rate = 6%; Compensating VAT Rate = 6%; Assumptions: Column 2, All Production in Province B Column 3: Two Stages in Province of Origin; One Stage in Province of Destination Transactions Production: All Production: Origin (Stages 1 & 2): A (Purchases, Stages in Destination (Stage 3): B Sales, and Province B Provincial Federal Value Prov. Federal Tax: Compen- Ordinary Total Added) Tax: Tax (A) (B) sating Federal Federal Stage 1: {This stage occurs in province A} a. Purchases/Credits 0 0 0 0 n.a. n.a. 0 0 b. Sales/Tax 100 6 20 6 n.a. n.a. 20 20 c. Value added/ 100 Net tax (c= b-a) 6 20 6 n.a. n.a. 20 20 Stage 2: d. Purchases/Credits 100 6 20 6 n.a. n.a. 20 20 (d=b) e. Sales/Tax 200 12 40 {Interprovincial sale from A to B} 0 n.a 12 40 52 f. Value added/ 100 Net tax (f=e-d) 6 20 -6 n.a. 12 20 32 Stage 3: {This stage occurs in province B} g. Purchases/Credits 200 12 40 n.a 0 12 40 52 (g=e) h. Sales/Tax 300 18 60 n.a. 18 n.a. 60 60 i. Value added/ 100 Net tax 6 20 n.a. 18 -12 20 8 (i=h-g) j. Total tax 18 60 0 18 0 60 60 (j=c+f+i) n.a. : not applicable 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. 30 Table 3 Illustration of VAT with Border Tax Adjustments on Interprovincial Sales to Business Tax rates: Federal VAT Rate= 20%; Provincial VAT Rates: 4% in A and 8% in B Assumptions: Column 2, All Production in Province B Column 3: Two Stages in Province of Origin; One Stage in Province of Destination Transactions Production: All Production: Origin (Stages 1 & 2): A (Purchases, Stages in Destination (Stage 3): B Sales, and Province B Provincial Federal Tax Value Prov. Federal (A) (B) Added) Tax: Tax Stage 1: {This stage occurs in province A} a. Purchases/Credits 0 0 0 0 n.a. 0 b. Sales/Tax 100 8 20 4 n.a. 20 c. Value added/ 100 Net tax (c= b-a) 8 20 4 n.a. 20 Stage 2: d. Purchases/Credits 100 8 20 4 n.a. 20 (d=b) e. Sales/Tax 200 {Interprovincial sale from A to B} 16 40 0 16 40 f. Value added/ 100 Net tax (f=e-d) 8 20 -4 16 20 Stage 3: {This stage occurs in province B} g. Purchases/Credits 200 16 40 n.a 16 40 (g=e) h. Sales/Tax 300 24 60 n.a. 24 60 i. Value added/ 100 Net tax 8 20 n.a. 8 20 (i=h-g) j. Total tax 24 60 0 24 60 (j=c+f+i) n.a. : not applicable 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. 31 N_ � � (� � � � � � � � � � е- � � д � 4 1�пП U "'� � � � о ®� о $° 'Q � � � � ® ��