ifr_in_lac_by_bird

7/29/2019 IFR_in_LAC_by_Bird

1/67

Intergovernmental Fiscal Relations in Latin America:

Policy Design and Policy Outcomes

Richard Bird

Inter-American Development Bank

Washington, D.C.

Sustainable Development Department


2/67

Foreword

The Sustainable Development Department commissioned Richard Bird to prepare a paper on intergov-

ernmental relationships as part of the activities leading to the preparation of a Subnational DevelopmentStrategy for the IDB. Using his vast academic and consulting experience, Bird produced a document ofsignificant breath and depth, touching on the most important aspects that determine the complex relation-ship that exist between central and subnational government, with a particular reference to Latin Americaand the Caribbean. His analysis provided key inputs to the work of the team drafting the strategy paperand portions of it were included in the background paper for the strategy. However, given the many in-sights that this work provides on issues surrounding intergovernment relations, the Social DevelopmentDivision considered worth publishing the study in its entirety.

The paper develops recommendations concerning a desirable structure of intergovernmental relations thatcreate the correct incentives for subnational governments to become efficient and equitable providers ofservices. However, as Bird highlights in his conclusions, the discussion and recommendations contained

in the paper, are of a general nature and point towards an optimal structure of intergovernmental relations.They need to be adapted to the realities of each individual country and gain their full significance whenfull consideration is given to the legal, political and socioeconomic determinants of each individual case.

We are pleased to publish the full text of Birds paper as a guide for policymakers and practitioners inter-ested in the topic and as a contribution to promote in-depth analysis of this complex area of government.

Mayra BuvinicChief, Social Development Division


3/67

Contents

Intergovernmental Fiscal Relations 1

The Context of Intergovernmental Fiscal Relations 2Models of Local Government FinanceCentral Government Policy

The Big Questions 8Expenditure AssignmentRevenue AssignmentClosing the GapOther Issues

Government Spending 11

Assigning ExpendituresManaging Expenditures

The Revenue Side 16Tax Assignment in PrincipleFinancing Regional GovernmentsFinancing Local ServicesThe Future of Tax Assignment

Designing Intergovernmental Transfers 25Key Design OptionsFiscal Capacity and Fiscal Effort

Transfers and PovertyConclusion

Improving Local Capacity 32

Macroeconomic Aspects of Decentralization 35Transfers and DeficitsSubnational Borrowing

Information and Accountability 40Institutionalizing DecentralizationAchieving Fiscal Transparency

Conclusion 45

References 47

Annexes 55


4/67

Intergovernmental Fiscal Relations

Decentralization is an important phenomenon in

an increasing number of Latin American countries(See IDB, 1994, 1997; Burki et al., 1999;ECLAC/GTZ, 1996, 1997; Lopez Murphy, 1995;Fukasaku and Hausmann, 1998). Subnationalfinance and intergovernmental fiscal relationshave long been matters of concern in Brazil and,to a lesser extent in Argentina. These issues arenow becoming increasingly important in otherlarger countries, whether formally federal, such asVenezuela and Mexico, or unitary, such as Co-lombia. Even smaller countries such as Paraguay,Uruguay, Ecuador, and Guatemala are increas-

ingly considering decentralization. Decentraliza-tion is not only in the air, it is already on theground and influencing policy outcomes through-out the continent in a major way. It is thereforecritical to understand the actual and potential im-

pact of intergovernmental relationships on suchkey questions as the ability of central govern-ments to conduct sound macroeconomic policies,the incentives for subnational governments to be-have in a fiscally responsible way, and the effi-cient and accountable delivery of public services.

This paper sets out some of the main factors de-termining the effects of intergovernmental fiscalrelations on such matters. It represents not newresearch but rather an attempt to digest and reflectupon what we have learned to date about some ofthese complex questions.1 The balance of the

paper is organized as follows. First, since thestructure and to some extent the effects of inter-governmental fiscal relations depend largely upon

1

the objectives and constraints of the country in

question, Section 2 discusses some core generalquestions that arise with respect to decentraliza-

tion issues specifically, the objectives of de-centralization, the related but distinct question ofthe underlying model of subnational finance, andthe key role played by the central government indecentralization. Against this background, Sec-tion 3 briefly introduces the three key components

of intergovernmental fiscal relations expendi-tures, revenues, and transfers. Sections 4 through6 review, under each of these headings, the impli-cations of alternative arrangements for the effi-

cient delivery of public services, emphasizing theimportance of both enabling and requiring hardbudget constraints at all levels of government.Some have argued that the outcome of recent de-centralization in many countries will be unsatis-factory owing to local incapacity to handle thenew tasks. Others have asserted that decentrali-zation will result in macroeconomic problems.Section 7 considers briefly the question of localcapacity, while Section 8 discusses subnational

borrowing and the implications of decentralizationfor macroeconomic management. Section 9 sets

out the critical need to develop both an adequateinformation base and an appropriate institutionalforum if the dynamics of decentralization are toevolve in a constructive direction in any country.Section 10 summarizes some of the key conclu-sions of the paper.

1Much of the paper draws freely on related prior work.

See, for example, on Colombia (Bird, 1984; Bird and

Fiszbein, 1998; and World Bank, 1996a) and on A r-gentina (World Bank, 1996b). I have also made use of

additional unpublished material (initially preparedlargely for the World Bank at various times) dealingwith these countries and with Mexico, Venezuela,

Chile, and Uruguay, as well as on more general pub-lished materials such as Bird (1993), Bird and Vaillan-court (1998), and Litvack, Ahmad, and Bird (1998).

Some of the issues discussed in this paper are treated inmore detail in Bird (1999a, 1999b, and 1999c).


5/67

The Context of Intergovernmental Fiscal Relations

Intergovernmental fiscal relations are only one

aspect of a complex political, economic, and ad-ministrative system. Their design, role, and ef-fects can only be understood in the specific insti-tutional context in which they operate. How oneviews many issues, for example, depends in partupon whether one views decentralization as a pro-cess of delegation or devolution.Delegation refersto a situation in which local governments act asagents for the central government, executing cer-tain functions on its behalf. The appropriate per-spective is top-down, that is, from the viewpointof a central government2 whose objective in de-

centralizing is often assumed, in theory, to be theachievement of a more efficient allocation of re-sources through the delegation of authority to lo-cal governments. The top-down approach impliesthat the main criterion for evaluating fiscal de-centralization should be how well it serves na-tional policy objectives, whatever they may be.3

In contrast, devolution refers to a situation inwhich not only implementation but also theauthority to decide what is done is in the hands of

local governments or, to put it another way,

local autonomy prevails. The appropriate per-spective is then bottom-up, from the viewpointof local citizens. Those who take this approachoften stress such political values as improved

2

governance (through increased local political par-

ticipation, for example) as well as such economicvalues as allocative efficiency (through increasedresponsiveness to local preferences or, perhaps,the increased scope for dynamic innovation thatmay be afforded by a variety of competing localgovernments). If this is the relevant perspective,the appropriate criteria for appraising intergov-ernmental fiscal relations may differ sharply fromthose under the top-down approach.

For example, is a good outcome one which bestachieves the goals of the central government or

one which frees local governments most fromcentral dictates? Decentralization is unlikely toproduce precisely the expenditure pattern that thecentral government would itself choose to imple-ment unless, improbably, the goals of central andlocal governments coincide and all decisionmak-ers face exactly the same incentives as in a cen-tralized system. Since conflicts between centraland local governments as to what should be doneare inevitable on many issues, the appropriatechoice of perspective is essential in analyzing andinterpreting issues of fiscal decentralization in any

particular jurisdiction.

Similarly, how one evaluates any particular aspectof intergovernmental relations depends in partupon whether one focuses on its intrinsic or itsinstrumentalaspects. Devolution may, for exam-

2In some federal countries, the central government ple, be considered to be an intrinsically worth-from the perspective of local governments might be the while objective. Whatever outcomes emerge fromstate or provincial government. Although for the most a decentralized system of decision-making must,part the terms subnational and local are used more from this perspective, be right. Local people mayor less interchangeably in the present paper, as Bird make wrong decisions from the perspective of the(1995a) sresses, there may be important differences central government or of an outside observer, butbetween regional and local governments, particularly in

ifthey make them, the decisions must, by defini-formally federal countries (see also Sections 5 and 10 tion, be assumed to be right for them. The resultsbelow).3The final qualifying phrase is important, since decen- of a good process must themselves be good.

tralization in the real world has seldom been driven byefficiency objectives. Even those concerned solely For such arguments to be persuasive, however,

with such objectives must take explicitly into account several stringent conditions must be satisfied.the inevitably political context of most decentraliza- First, local decision-making processes must betion: see, for example, Salegh and Tomassi (1999) and fully democratic. Electoral democracy alone isHaggard (1999).


6/67

not enough to satisfy this criterion. The costs and Administrative Condition: Such a system requiresbenefits of decisions must be transparent, and eve- an appropriate institutional framework, including,ryone affected must have an equal opportunity to for example, a clear set of framework laws (oninfluence the decision. Few, if any, countries can local budgeting, financial reporting, taxation,meet this high standard. Second, the costs of local contracting, dispute settlement, rules to be fol-decisions must be fully borne by those who make lowed in designing user charges, etc.), as well as

the decisions. There must be no tax exporting adequate institutional support to enable subna-and no funding at the margin via transfers from tional governments to operate effectively in thisother levels of government. Third, the benefits environment.(like the costs) of local decisions must not spillover jurisdictional boundaries without explicit It is of course much easier to lay down such gen-and transparent compensation. In principle, the eral prescriptions than to satisfy them in the verylast two conditions, unlike the first one, can be diverse situations found in the real world. None-met to a substantial extent by good institutional theless, if these conditions are not met, the per-design, as discussed later in this paper. verse incentives that too often distort public sector

outcomes in many countries may be exacerbatedAgain in principle, even if decentralization is less by further decentralization.4

than perfect it may yield a more efficient and eq-

uitable pattern of public services than the over- The recent wave of decentralization in Latincentralized and unresponsive public sectors that America has largely followed, and accompanied,currently exists in many developing countries. an increase in the level of democratization (IDB,But it will do so in practice only if it is properly 1997). As already emphasized, democracy is aimplemented. The key to ensuring that incentives particularly important mechanism to promote ac-are conducive to good decisions is to ensure that countability. At the same time, however, ac-those who make the decisions bear the financial countabilityin the sense of transparent decision-(and political) consequences. Thus, the three makingis an essential ingredient of democracy.conditions can be extended and restated as polit i- The ultimate assurance of good government is, oncal, economic and administrative requirements. one hand, the ability of citizens to compare gov-

ernments in terms of the services they provide andPolitical Condition: Leaders at all levels should the tax prices they charge and, on the other, the

be responsive and responsible to their constitu- ability of citizens to affect and alter the decisionsents, and those constituents should be as fully in- of government (Breton, 1996). Thus, democracyformed as possible about the consequences of without good information is not enough.their decisions and those of their leaders. Making

politicians bear the consequences of their ownmistakes is as close as one can get to a hard po-

litical budget constraint.

Economic Condition: It should be difficult forlocal residents to shift costs to nonresidents whodo not receive benefits, and local decisionmakersshould be appropriately accountable to three dis-

tinct, though overlapping, groups: (1) their cit i-zens for the use they make of revenues collectedfrom them (through local taxes), (2) users of local

public services for the use made of the revenuesthey contribute (through user charges of varioussorts), and (3) taxpayers in general for the usemade of any transfers (or subsidized loans) theyreceive from the central government.

3

4Along similar lines, Bahl and Linn (1994) noted that

...the situation in a developing country that could pro-

vide maximum gains from a more decentralized localgovernment structure would include: (1) enough skilledlabor, access to materials, and capital plant to expand

public service delivery when desired; (2) an efficient

tax administration; (3) a taxing power able to capturesignificant portions of community income incremen-

tally; (4) an income-elastic demand for public services;(5) popularly elected local officials; and (6) some localdiscretion in shaping the budget and setting the tax

rate. They might have added: (7) a tradition of localdecision-making. All these points seem correct, andare discussed in this paper, albeit with somewhat dif-

ferent emphasis.


7/67

Even in countries without well-established demo- work, decentralization is judged largely by its ef-cratic institutions, in which decentralization is fects on such distributional goals as poverty alle-simply an instrumentality of the central govern- viation. The benefit model, in contrast, viewsment, good information is essential to ensuring subnational governments primarily as agenciesenhanced service outcomes. Regardless of the providing identifiable services to identifiable localrationale and circumstances surrounding the de- residents (Break, 1992).

centralization process, its outcomes will improveas more information is made publicly available. The Ability Model. The ability model dominatesThe extent to which information relevant to un- public discussion in most countries. However,derstanding and evaluating the impacts of decen- while attention must be paid to the redistributivetralization is available is thus critical to improved aspects of public policy at all levels of govern-outcomes, almost regardless of the status of politi- ment, the allocative costs of distorting local taxingcal democracy. From this perspective, will be and spending decisions in the name of equity mayseen below, it is important that intergovernmental often exceed any conceivable equity gain. Pur-fiscal relations in decentralized countries be as suing active distributive polices at the local fi-

simple, transparent, comprehensible , and predict- nance level, though to some exent perhaps inevi-able as possible. table in the context of democratic politics, is

nonetheless generally inadvisable from a strictly

When both democracy and good information sys- economic point of view.5

tems are in place, devolution (in the sense definedearlier) makes sense. When they are not, it may In developing countries, most productive taxnot. Even if one or more of these conditions does bases are invariably taken by the central govern-not hold, however, the delegation of implementa- ment. Thus, in practice, the ability approach totion responsibilities to local bodies may make in- local government usually amounts to a system instrumental sense provided that the incentives fac- which central transfers finance most local servicesing local decisionmakers are properly structured. even though it is often far from clear that thoseIn these circumstances, this means that they who benefit from such central largesse are lessshould be structured to produce the results desired able to pay for what they get than those whose

by the central government. Indeed, whether the incomes are reduced by the taxes that finance theunderlying policy intention is devolution or dele- transfers. If local governments attempt to imple-

gation, central governments must carefully con- ment differential policies (such as progressivesider the effects of alternative intergovernmental taxes or pro-poor pricing policies), the resultfiscal relations upon the outcomes in which it is over time may be an outflow of tax base and, con-interested. The instrumental character of inter- sequently, a smaller total package of local servicesgovernmental fiscal relations their effects on rele- than if the benefit approach had been followed.vant outcomes, is thus critical viewed from any

perspective. Moreover, wealthier local governments, whichcan apply lower tax rates and obtain similar

Models of Local Government Finance yields, may attract tax base from poorer ones, thusaccentuating initial interjurisdictional disparities.

Another important dimension of intergovernmen- Although intergovernmental fiscal relationstal finance is the underlying model of local gov-

ernment finance. Two quite different models im-plicitly underlie much of the discussion of inter-governmental fiscal relations. The first of thesemodels views subnational government in an abil-ity-to-pay framework. This ability model focuseson the effects of governmental action, whether bycentral or local governments, on the interpersonaldistribution of income and wealth. In this frame-

4

5

Governments are not simply providers of services.They are also institutions which reconcile and manage

conflicting interests in society, and distributional policyis one of the major policy instruments used for thispurpose. Even if, as is often the case, subnational at-

tempts at redistribution are vitiated by market forces,the attemptto redistribute may still be politically es-sential. For an extended discussion of this point, see

Bird (1980).


8/67

should presumably not accentuate market-driven nance local services should thus fall primarily oninequalities, it is generally a mistake to design tax local residents. Moreover, as discussed further inand expenditure policy specifically to alleviate Section 5, if such governments are to behave in asuch disparities. As discussed further in Section fiscally responsible way, they must be made po-6, attempts to incorporate regional developmental litically responsible for determining the rates of atgoals into the design of intergovernmental trans- least some major taxes. Subnational governments

fers often result in complex and confused policies may have large revenues from what appear to bethat not only fail to achieve these objectives but local taxes, but if they cannot set tax rates theyalso fail to foster the efficient and effective deliv- cannot be made accountable to their constituentsery of desired local public services (Bird, 1982). at the margin, as both democracy and efficiencyWhether ability concerns are motivated by con- require.siderations of interpersonal distribution orinterregional disparities, they are better addressed Finally, even if local services are largely paid fordirectly rather than indirectly through distortions through locally-determined (and locally-paid) feesin the design of intergovernmental fiscal relations. and taxes, intergovernmental transfers must still

be carefully designed to ensure that, at the mar-The Benefit Model. The benefit model of local gin, the costs and benefits of local fiscal decisionsfinance is more conducive to effective, efficient, are borne locally, while taking adequately into

and accountable local government. In this model, account such interjurisdictional spillovers as arelocal governments are essentially viewed as deemed relevant. Consistent application of theseequivalent to firms that provide services. Resi- rules will impose what is called a hard budgetdents are willing to pay a price or charge for the constraint on local decisionmakers and, hence,last (marginal) units of government services that make them fully accountable for the consequencesthey receive that is just equal to the benefit they of their decisions.6

derive. This approach to local finance is appeal-ing because the results are not only allocatively Central Government Policyefficient but also equitable in the sense that no one

pays less (or more) than he or she would be will- The institutional setting within which local gov-ing to pay in a free market. ernments in most developing countries must func-

tion was recently characterized as falling into one

The benefit model is not without its problems, of three categories: (1) the over-controlled localhowever. First, it is often surprisingly difficult to public sector, (2) the under-controlled local publicdesign an appropriate pricing policy for many lo- sector, and (3) the perversely regulated local pub-cal public services (Annex 3). Second, even if lic sector (World Bank, 1995a). The first of thesesuch prices can be designed, implementing them situations seems most common in developingis seldom politically appealing, particularly if, as countries. Central governments sometimes con-in most countries, the introduction of financing trol all the details of local government (who theyvia user charges means a sharp change in historic hire, what they pay, where and when the buses

practice. Understandably, citizens seldom appre- run, etc.) and leave little or no freedom for localciate being asked to pay for services that were initiative.7 Central governments also often financesupplied free of charge in the past. Despite these

problems, in principle, services that flow to identi-

fiable individuals (or firms) should be priced ap-propriately to the extent feasible.

Where such pricing is not possible or desirable,local expenditures and revenues should be linkedthrough matching service benefit areas to the spa-tial dimension of the financing sources. Anytaxes levied by subnational governments to fi-

5

6Of course, neither local decisionmakers nor, in most

instances, their constituents may be happy to be subjectto such a budget constraint. It is always easier andmore pleasant to spend, as it were, other people'smoney in an unaccountable (and hence inevitably

somewhat irresponsible) fashion, which is no doubtone reason for the continued political attractiveness ofthe ability model of government.7Rojas (1999) stresses the dominance of controlled,

monitored decentralization in Latin America. Of


9/67

local services directly, or earmark and restrict lo-cal financing to such an extent that they might aswell do so (Annex 2). In such circumstances, lo-cal citizens look to the central government to fix

potholes, and they are right to do so. Just as chil-dren who are never given responsibility seldom

act like responsible adults, this type of micro con-trol generally produces weak and incapable localgovernments.

While less common, the opposite ill of undercontrol has also emerged in some countries as aresult of inappropriate decentralization strategies.For example, a number of East European coun-tries in transition have given local governmentslarge shares in national revenues as well as re-sponsibility for important public service functions,without at the same time setting up an adequate

institutional structure to ensure that central fundsare being properly spent in, say, maintainingminimum standards of service in such areas ofoverriding national concern as education or health(Bird, Ebel, and Wallich, 1995). Similar concernshave frequently been expressed in some LatinAmerican countries.

Finally, local governments frequently receive per-verse signals from central governments. If, forexample, the amount of transfers a governmentreceives depends upon the size of its budget defi-

cit, incentives are clearly perverse. As anotherexample, national funding is often available forinfrastructure investment but no funds are avail-able for operation and maintenance (Annex 7).This creates a perverse preference for new in-vestment and rewards localities for not maintain-ing existing facilities (which they would have todo out of their own funds) in order to strengthentheir apparent need for new ones (for which thecentral government will pay).

Even when local government face perverse incen-

tives, local efforts and policies may sometimesmake a real difference. In Brazil, for example,

6

some cities are well-run and provide efficientservices. Others, superficially similar in charac-ter and resources, are badly-run and poorlyequipped. In Colombia, some departments pro-vide superior health services compared with oth-ers whose resources are similar. Almost every-

where, some local governments do much betterthan others. The reason may simply relate to his-torical circumstance. For one reason or anotheran area started to do something well some timeago, and it continues to do so. Or it may be be-cause of a caring and charismatic local leader orsome other chance circumstance. Whatever thecause, such experiences highlight two important

points. First, even in the perverse situation inwhich many local governments are placed by in-appropriate central policies, there is usually somescope for local initiative. Second, such local ini-

tiatives may make a real difference in the lives oflocal people. One way to make decentralizationeffective and efficient is to make it easier for goodexamples to occur by giving subnational govern-ments some room maneuver, and, in part throughfacilitating an adequate flow of information, tomake it easier to emulate their success.8

Many Latin American countries fail to recognizeexplicitly the diversity of local governments.Local government is a term that covers a widerange of realities. Cities of 10 million people,

villages with 200 inhabitants, densely populatedrural and urban areas, sparsely populated territo-

ries all are, as a rule, organized in one form oflocal government or another. Some localities, arerich, some are poor. Some, have strong localcommunity spirit, some have none. Some are run

by well-intentioned, well-trained people ; others byincompetent and corrupt officials. Unfortunately,the diversity of local government reality is seldommatched by equal diversity in central governmentrules. Even though some areas may manage to

break out of the inappropriate mold into which

they have been put, most will not. Successful de-centralization must recognize the diversity and

course, as Carlsen (1998) and Schweger (1999) analyze

rigorously, and as is discussed later in the present pa-per, there are often good theoretical and practical ar-guments for maintaining some degree of central con-

trol.

8As Rojas (1999) notes, dissemination of best practices

and horizontal exchanges among subnational govern-ments are probably more effective in improving localcapacity than formal technical assistance: see also the

discussion in Section 7.


10/67

heterogeneity of local government and allow for ally agreed fashion). To provide a particularan equally diverse and heterogeneous set of re- service, as has to some extent been done in Co-sponses to particular decentralization initiatives, lombia (World Bank, 1996a). At its best, this ap-including accommodating home-brewed solutions proach focuses pragmatically on what may work,to particular local problems. rather than on trying to force everyone into the

same centrally-determined Annex. Given the re-

A useful approach in some circumstances may be ality of diversity and the apparent political need inthrough contracts or the making of specific many countries to have nominally uniform laws,agreements with different areas in accordance perhaps only something like a contract approachwith their capacities and interests. The central may be able to provide the necessarily non-government may, for example, contract with a uniform conditions needed to secure more or less

particular local government (preferably for a pe- uniform outcomes at least cost.riod of years, and preferably in an open and mutu-

7


11/67

The Big Questions

Four big questions must be answered with respect thus requires substantial institutional as well as

to intergovernmental finance in any country: analytical knowledge.

(1) Who does what? The question of expendi- Expenditure Assignment

ture assignment.How government functions should be divided

(2) Who levies what taxes? The question of among levels of government is a complicated

revenue assignment. matter, to which each country has its own uniquesolution. Section 4 sets out a few principles and

(3) How is the (virtually inevitable) imbalance alternative approaches with respect to the spend-

between the revenues and expenditures of subna- ing side of public sector budgets. The main point

tional governments that result from the answers to made is that the most important question about

the first two questions to be resolved? the ques- expenditures, from an economic perspective, is

tion ofvertical imbalance. not who does what but rather what is done, andhow well it is done. The details of expenditure

(4) To what extent should fiscal institutions at- assignment are less important in determining out-

tempt to adjust for the differences in needs and comes than the clarity of that assignment, how it

capacities between different governmental units at is determined and implemented, and how well

the same level of government? The question of expenditures are managed at all levels of govern-ment.horizontal imbalance, orequalization.

Revenue AssignmentIdeally, to achieve the relevant policy objectives,these questions should be approached within eachcountrys specific circumstances. These objectives Who should levy what taxes, and how effectively

include not only the normal public finance trio of they can do so, has been a major issue in some

efficiency (allocation), equity (distribution), andcountries. The correct revenue assignment in a

stabilization but also economic growth as well as multi-level government structure is by no meansclear in principle, and is likely to prove highlysuch nebulous but politically resonant goals ascontroversial in practice. The fundamental prob-regional balance and maintaining national integ-lems are two. First, the central government canrity and political stability. There may be conflicts

between these objectives, as well as differences inherently collect most taxes more efficiently than

between local and central perceptions of the subnational governments. Second, the potential

weights to be attached to them. Moreover, like all tax bases available to the latter vary widely from

public policies, intergovernmental fiscal policies jurisdiction to jurisdiction. The first of theseproblems gives rise to vertical imbalance; the sec-must take into account both political constraintsond produces horizontal imbalance.(such as the strength of different regions and

groups in political decisions) and economic con-

straints (such as the development of financialThe first of these problems could be solved to the

markets). Finally, all policy change must start extent that variable surcharges on central taxes are

from the here and now. Since every country has feasible. Given the paucity of such surcharges in

its own history, the current state of fiscal institu- Latin America, however, vertical imbalance may

tions in large part reflects the results of an accre- indeed give rise to concern. In addition, since thesecond problem, horizontal imbalance, would ac-tionary process of policy change over time.tually be worsened by more decentralized reve-Therefore, to understand, let alone to resolve, thenues, even if vertical imbalances were resolved byintergovernmental fiscal puzzle in any country

8


12/67

adjusting revenue assignments, there would stillbe a need for transfers, at least in countries inwhich such interjurisdictional disparities are of

policy concern.9

Closing the Gap

No matter how revenues and expenditures are as-signed, worldwide experience suggests that somevertical imbalance is likely to remain. Even if thetax base of the richest subnational governmentenabled it to balance own revenues and expen-ditures, imbalances would remain for all the rest.Even for the richest region, the differential elas-ticity of expenditures and revenues assigned todifferent government levels may, in any case,soon lead to a re-emergence of a vertical imbal-ance problem. One way or another, structural

gaps are likely to exist in all multi-tier fiscalsystems.

In principle, there are at least four ways in whichany subnational revenue-expenditure gap might beclosed.

Revenues could be increased at the subnationallevel. In many Latin American countries, suchincreases seem both feasible and desirable (Bird,1999b). But even if subnational revenues are in-creased as much as is politically and economically

feasible, it may not be enough in some countriesand will certainly not provide sufficient resourcesto finance adequate service levels in many poorlocalities.

Subnational expenditures could be reduced.While always popular with central governmentsand no doubt sometimes necessary, this approachis unlikely to be advisable if the system has been

9

properly designed and the right expenditures as-signed to the subnational level in the first place.

Some expenditure functions could be transferredup to the level with more revenue (or revenue-raising power) or some revenues transferred down

to the level with more expenditure. While suchshifts would be both unnecessary and unwise ifthe basic structure of the system is correct, this isa big if in many countries.

Some centrally-collected revenues could betransferred to subnational governments. In theend, in virtually every country in the world, eventhose in which subnational governments havemuch greater access to revenues than in LatinAmerica, there invariably remains some need forintergovernmental transfers. As discussed in Sec-

tion 6, however, the design and implementation oftransfers involves many more considerations thansimply closing the subnational revenue-expenditure gap.

In practice, intergovernmental fiscal transfers arethe major source of subnational revenue in mostLatin American countries (IDB, 1997). Thissituation seems unlikely to change significantly inthe near future. Together with a properly designedsubnational tax base, a properly designed transfersystem is, therefore, essential to ensure that ade-

quate incentives for fiscal responsibility are pro-vided. Unless made accountable in this sense,local decisionmakers are unlikely to make effec-tive and efficient resource allocation decisions (tothe extent that any political process can producesuch decisions). In this respect accountability atthe margin, and not simply the size of the amounttransferred is critical. It is possible in principle fora provincial government to be almost totally de-

pendent on central transfers and still be fully ac-9As May (1969) argues, the taste for regional equali- countable, provided the amounts are fixed in ad-

vance and cannot be altered as a result of any (in-zation may vary greatly from country to country.

Compare, for example, the explicit interregional redis- period) action by the recipient. On the other hand,tribution of the unemployment insurance system in a badly designed transfers, even if quantitativelyCanada with the state-based (and hence non- much smaller, can do a lot of damage to the effi-equalizing) system of the United States. Note that re- cient and effective operation of the local publicgionally redistributive transfers must be distinguished sector. A well-designed transfer can ensure that atfrom so-called equalization transfers discussed in the margin, local actions to raise or lower localSection 6 (and also found in Canada, see Annex 5) revenues or expenditures will directly affect out-which are intended to ensure specified efficiency ob-

comes, which is what is needed to ensure politicaljectives.


13/67

accountability. 10 One should not underestimate,however, either the difficulty of implementingsuch a system in the real world of politics or thelikelihood that the transfers currently in place inmost countries fall far short of ideal. 11

Other Issues

Three additional issues have attracted consider-able attention in recent discussions of fiscal de-centralization in Latin America. The first relatesto local capacity. (Are local governments up tothe job?) The second focuses on the need or de-sirability of specific restrictions on subnational

borrowing. And the third relates to the broaderissue of the impact of decentralization on macro-economic stability and policy. Important as theseissues may be in particular context, in the long run

they are secondary to the key questions of who

10

does what and how is it financed. If subnationaltax policy and transfers are appropriately de-signed, the resulting hard budget constraintsshould, over time, curb any tendency to either mi-croeconomic inefficiency or macroeconomic in-stability, hence largely vitiating the need for spe-

cial concern about subnational borrowing. Given,however, that reality in many countries is often farfrom ideal and that there may be substantial po-litical difficulties in altering it (Salegh and Tom-masi, 1999), second-best policies may beneeded areas. The capacity issue too can andshould be addressed best by a combination ofgood initial design and the establishment of anappropriate institutional framework, even though

political interests may block such moves in thenear future in many countries.12 These points arediscussed further in Sections 7 through 9.

10As discussed in Section 6, two special forms of trans-

fers also need to be considered. First, some transfers

are explicitly intended either to ensure the provision ofspecific services at specific levels (as in the case of the

capitation grants discussed in Annex 6, in which thetransfers are essentially payments to agents). Second,transfers may be explicitly intended to induce subna-tional governments to provide more of certain services

than they would otherwise do (as in the discussion of matching grants for infrastructure in Annex 7).11

See, for example, the detailed analyses of transfers in

Colombia and Argentina in World Bank (1996a,

1996b).

12For example, Rojas (1999) notes that no country in

Latin America has thus far managed to create adequatetechnical unit to monitor and foster decentralization:

see Section 9 for further discussion.


14/67

Government Spending

The basic requirement for efficient and effective

subnational government may be thought of as thematching principle. As already discussed, tothe extent possible, benefit areas should bematched with financing areas as in the benefitmodel of local finance. In addition, expenditureresponsibilities should be matched with revenueresources, and, finally, revenue capacities should

be matched with political accountability.

Assigning Expenditures

The basic rule of efficient expenditure assignment

is often taken to be to assign each function to thelowest level of government consistent with its ef-ficient performance.13 So long as there are localvariations in tastes and costs, there are efficiencygains from carrying out public sector activities inas decentralized a fashion as possible.14 Indeed,from this perspective, the only services thatshould be provided centrally are those for whichthere are no differences in demands in differentlocalities, where there are substantial spillovers

between jurisdictions that cannot be handled insome other way (by contracting, by redrawing

boundaries, or by grant design), or those for whichthe additional costs of local administration aresufficiently higher to outweigh its advantages.

Under this approach, apart from the importantissue of distribution, almost all public services(other than national defence, foreign policy, andsurprisingly few others) should in principle bedelivered at the local level. Decisions about theservices provided, to whom they are provided, andin what quantity and quality, should be made lo-cally and local taxpayers should pay for the serv-ices provided. In practice, however, althoughsome functions (such as street maintenance) are

11

local everywhere, the allocation of functions to

subnational governments varies considerably fromcountry to country (IDB, 1997).

This approach to expenditure assignment need notbe consistent with the matching principle, sincethere is nothing that guarantees that the bundle ofservices thus assigned to any particular level ofgovernment will be matched by the set of revenueinstruments assigned to that same level. On thecontrary, as already noted in Section 3, a funda-mental imbalance emerges in the vertical assign-ment of expenditures and revenues in virtually

every country. Hence, there is an apparent needfor intergovernmental fiscal transfers to close thebudgetary gap.

As emphasized in Section 2, the essential eco-nomic role of local government is to provide tolocal residents those public services for whichthey are willing to pay. Local governments are(or at least should be) accountable to their citizensto the extent that those citizens finance those ac-tions. Such accountability is the public sectorequivalent of the bottom line in the private sec-tor. Three conditions need to be satisfied toachieve accountability in this sense in subnationalfinance. First, subnational governments should,whenever possible, charge for the services thatthey provide (Annex 3). Second, if charging isimpracticable, subnational governments shouldfinance such services from taxes borne by localresidents (Section 5), except to the extent that thecentral government is willing to pay for them(Section 6). Third, where the central governmentdoes pay as a rule, subnational governmentsshould be accountable to the central government(to at least some extent) when the central govern-ment pays (Section 9).

13In the context of the European Union, this has come

to be called the subsidiarity principle: see Breton,Cassone and Fraschini (1998) for a useful discussion.14

Oates (1972) has labelled this the decentralization

theorem.


15/67

To apply these rules, an essential first step is toestablish clear lines of responsibility and account-ability.15 But clarity of assignment in terms ofspecifying exactly what services, each govern-mental agency is responsible for, is only part ofthe story (Annex 1). Clarity must be matched by

accountability, in terms of political democracyand transparency of operation, as well as byauthority in terms of both the ability to manageexpenditures and to determine (within limits)revenues. Moreover, even in the best of all possi-

ble worlds full clarity in expenditure assignmentmay never be fully attainable, for several reasons.

First, with respect to many important spheres ofpublic sector activity (for example, education)different jurisdictional levels may play criticalroles. The central government may, for example,

appropriately set national standards for graduatesand for teachers and may also establish the basiccurriculum to be covered. Within this framework, regional governments may develop theirown policy goals (for instance, with respect toschool facilities) and deploy appropriate regula-tory instruments to achieve them. Local govern-ments may be responsible for actually payingteachers and maintaining facilities. Finally, edu-cational services are delivered by local schoolswhich may often, experience suggests, produce

better outcomes if they have a substantial degree

of budgetary autonomy and can react to inputfrom teachers, parents, and the local community(Burki et al., 1999). Three or more levels of gov-ernment may thus play important roles in deliv-ering educational services.

It is not meaningful to consider a particular serv-ice as assigned to the local level, when much ofthe relevant policy and regulatory framework, andindeed much of the financing, comes from higherlevels of government (and delivery of the servicemay take place at a lower institutional level).

Clarity as to whom is responsible for what is in-deed important, but many different public sectoractors may appropriately be involved in the provi-sion of any particular public service. What mat-ters is not so much that each expenditure function,

12

broadly understood, is clearly assigned to onelevel of government or another, but rather that it isclear to all exactly who is responsible for doing

precisely what.

Second, clarity in expenditure responsibility is

sometimes taken to mean that much attentionneeds to be paid to the problem of coordination. Inthis regard, however, it should be recalled that the

principal argumentfor decentralization is that co-ordination (or cartelization, or monopoly, as itmight also be called) is not delivering the goods,or at least not delivering the right goods in theright quantities to the right people. Moreover,what may at first glance appear to be undesirableduplication or overlapping of functions may, insome instances, reflect either useful redundancy ina complex system (Landau, 1969) or desirable

governmental competition (Breton, 1996). On theother hand, in the real world duplication and con-fusion may often imply waste and the need for

better intergovernmental coordinationa. The an-swer, however, is not to abandon decentralization

but rather to design intergovernmental fiscal rela-tions in such a way as to minimize real coordina-tion problems and to continue working at the dif-ficult task of establishing effective coordinatinginstitutions (Annex 8).

However, even the best-decentralized public sec-

tor is undoubtedly far from a perfectly competi-tive market structure and outcomes will never beoptimal in the technical economic sense.16 Themost appropriate approach is to establish budgetconstraints that are as hard as possible for all rele-vant decisionmakers and to make the operation ofthe system as transparent as possible. Letting100 flowers bloom in the form of relatively un-coordinated decentralized public sector suppliersstriving to meet clearly specified and publicly ac-countable mandates may, in the end, provide a

better laboratory for the development of new and

better public sector services than any conceivablecentralized alternative (Oates, 1998).

15This is, for example, one of the key principles set out

in the IMF's recent Code of Good Fiscal Practice.

16Of course, as Goodin (1996) emphasizes, how im-

portant the failure to achieve economic optimality isdepends upon the extent to which policymakers are

actually trying to do so.


16/67

Managing Expenditures

No matter where subnational governments gettheir funds, they are unlikely to have enough to doall they or their citizens want and expect. Success-ful local government requires that scarce public

funds be managed as efficiently and used as ef-fectively as possible. Both financial honesty and

political accountability require that budgeting andfinancial procedures are properly established andimplemented. Budgeting, financial reporting, andauditing should be comprehensive, comprehensi-

ble, comparable, verifiable, and public.

It is equally important, however, to ensure thatbudgeted resources are applied as efficiently andeffectively as possible to achieve desired publicoutcomes. Adequate and appropriate procedural

norms are important in any financial system, butsubstantive outcomes are what really matters.Proper public expenditure management must ade-quately control the total level of revenue and ex-

penditure, appropriately allocate public resourcesamong sectors and programs, and ensure that gov-ernmental institutions operate as efficiently as

possible (World Bank, 1998). It seems particu-larly critical in this respect for subnational gov-ernments to have sufficient authority to manage

both expenditures and revenues sides.

For this reason, it is important to ensure that cor-rect incentives exist on the revenue side. Specifi-cally, effective subnational governments musthave significant revenue sources under their con-trol, for which they are economically and polit i-cally responsible. They must also be able to pre-dict with considerable certainty anticipated inter-governmental transfers in any financial period. 17

In many Latin American countries, decentraliza-tion has been recently characterized by the ear-marking of substantial parts of intergovernmental

transfers to localities to local infrastructure in-vestment. This is the case in Argentina (for

13

housing), Brazil, Colombia, Chile, Ecuador, Gua-temala, and Venezuela. Presumably motivated in

part by the desire to prevent local governmentsfrom wasting transfers on expanding local pay-rolls, earmarking has had the paradoxical effect ofexacerbating local fiscal problems in some in-

stances (Annex 2). Although not fully effective(since money is fungible, there is usually somesubstitution of transfers for own-source revenues)the result has usually been to expand capitalspending to some extent, while making the al-ready difficult problem of funding the operationand maintenance of these investments even moredifficult. Even when it succeeds in fostering localinvestment, earmarking has little to be said in itsfavor. It distorts local preferences, exacerbates

perverse incentives already found in the local fi-nance system, and often connects revenue sources

with expenditures in totally illogical ways.Moreover, excessive earmarking (like the related

process of mandating subnational governmentsto spend in accordance with central preferencesrather than their own) significantly reduces thescope of subnational governments for effectivelymanaging expenditures, even if they had both thewill and the capacity to do so.

Good subnational budgeting should take placewithin the framework of medium-term expendi-ture framework (MTEF) both to ensure the proper

financing of investment projects and to reduce thescope for the short-term political manipulation of

budgets (for example, to expand pre-election pub-lic employment in an unsustainable fashion).18 Anessential first step in this direction is to put sound

budgetary and financial procedures into place,especially in the more important subnational gov-ernments such as states and large cities.

Subnational budgets, like central budgets, shouldbe comprehensive, accurate, periodic, authorita-tive, timely, and transparent. The budget law

17It is thus not appropriate to have exactly the same

budget period for local and central governments, sincethe former cannot determine their budgets until theyknow the size and nature of the transfers they are likely

to receive from the latter.

18Unless there is an adequate MTEF at the central

level, it may be difficult to require one locally. On theother hand, given the growing importance of local gov-

ernments in many countries, it is increasingly difficultto have an adequate MTEF at the central level withoutexplicitly incorporating aggregate local revenues and

expenditures into the budgeting and planning exercise.


17/67

should be uniform and clear, and it should be en-forced. Moreover, expenditures should be subjectto external audits to ensure that the law is fol-lowed. All this would seem to require is a strongcentral hand to ensure compliance with the rules.For example, the central government should es-

tablish a framework budget law and requireadequate external audits (such as by a privatesector firm). It should not, however, subject sub-national budgets to prior approval, or the whole

point of decentralization is lost.

A strong budgeting and financial system alongthese lines will satisfy two essential requirementsof good government. First, it will establish the

basis for financial control. Second, it will providereasonably accurate, uniform, and timely financialinformation. Even the best set of financial proce-

dures does nothing, however, to ensure that scarcepublic resources, even if properly spent and ac-counted for according to law, have been spent inthe best possible way or as efficiently as possible.

Nor, as some countries have seen in recent years,does even the best-enforced set of budgetary pro-cedures ensure that aggregate fiscal discipline will

be adequately maintained. To attain favorableoutcomes in these respects, additional importantfiscal institutions need to be put in place.

For example, with respect to fiscal discipline, the

MTEF mentioned above can provide an essentialbaseline for measuring and monitoring the budg-etary impact of policy changes. Similarly, it iscritical for fiscal discipline that the amount ofrevenue expected from intergovernmental trans-fers is predetermined and is not subject to politicalre-negotiation during the budget year. Moreover,subnational governments must not be able to de-

pend on central government bailouts of imprudentfinancial decisions, such as unsustainable bor-rowing or expenditure increases. On the otherhand, subnational governments should be able to

increase expenditures to the extent that the fullfiscal consequences of such increases are borne bylocal residents and, equally importantly, theyshould be able to reduce expenditures if they sowish.19

14

Similar institutional safeguards should be in placeto ensure allocative and operational efficiencywithin affordable fiscal limits. Allocative effi-ciency requires that managers at all levels receiveadequate and accurate information on the effec-tiveness and social outcomes of the programs for

which they are responsible. This can take placefor example, through the revenues produced by

properly-designed user charges (Annex 3) and/orthrough participatory interaction with clients at

both the budgetary and implementation stages(Vergara, 1999). Moreover, they must havestrong incentives to respond to these signals, forexample, by facing a predetermined spendinglimit which can be altered only if they can sellmore services that their client groups are willingto pay for.

Operational efficiency may best be achieved inmost cases by allowing line managers (like theschool managers mentioned earlier) significantdiscretion, within budgetary cash limits, to reallo-cate funds among inputs or perhaps even across

budgetary periods, subject to the attainment ofpredefined operational (performance) goals aswell as compliance with appropriate internal andexternal financial control and audit systems. Em-

phasis is thus shifted from input controls (hiringso many persons at such and such a wage or rent-ing so many square meters of space) to output

controls (providing health care of a determinedquality to so many persons within a specified time

period or issuing so many marriage licenses).

Shifting the emphasis in public finance from in-puts to outputs in this way is an essential step inimproving policy outcomes at any level of gov-ernment. It carries with it some risks,20 and it is

by no means as yet fully clear how best, or towhat extent, this shift can be accomplished in thedifficult circumstances facing most countries ofthe region. Nonetheless, there is already consid-

erable experience in Latin America with such

19As Rodden (1999) notes, Brazilian states are consti-

tutionally restricted in the extent to which they can

reduce expenditures, even those funded solely from

their own revenues. Such limitations are clearly notconducive to good expenditure management.20

For example, there must still be full accountability by

clearly identifiable decisionmakers with respect to all

expenditure decisions to reduce the possibility of fraud.


18/67

techniques as client surveys, participatory budg- see more clearly some of the directions in which-eting (notably in some Brazilian cities), perform- countries should begin to move in terms of expen-ance budgeting, and user financing (Burki et al., diture management if their leaders are seriously1999). The path to success is by no means clearly interested in making life better for their popula-marked. But at last, it seems, we are beginning to tions.

15


19/67

The Revenue Side

Before discussing the appropriate design of sub-

national taxes, it may be useful to clarify themeaning of a subnational tax. A completely sub-national tax might be defined as one that is as-sessed by subnational governments, at rates de-cided by subnational governments, which is alsocollected by subnational governments, with its

proceeds accruing to subnational governments. Inthe real world, however, many taxes may possessonly one or two of these characteristics, and theownership of the levy may be unclear.

In Argentina, for example, although part of the

proceeds of many taxes accrue to the provinces,the rates (and bases) of these taxes are determinedby the national government, which also assessesand collects them. For most purposes, such taxesare best thought of as central government taxesthat are allocated to the provinces through trans-fers. This interpretation is particularly plausible

because there is little connection between theamount transferred and the amount collected lo-cally.

Nevertheless, what appears as a central tax with arelated transfer program may, from some per-spectives, be a subnational tax. If, for example, a

provincial government decides whether or not toimpose a particular tax, determines the tax base,sets the tax rate, and receives all the revenues,then even if the tax is collected by the centralgovernment, the only role the latter plays is as acollection agent. Presumably, the central govern-ment has a comparative advantage in tax collec-tion, and the subnational government has con-tracted for its services in this respect. In this case,there is no intergovernmental transfer at all, ex-cept in the narrowest accounting sense.21

16

Intermediate cases between these extremes may

easily be found. In Brazil, for example, the statesimpose and collect their own VATs (ICMS), butthe rate of the tax is set centrally and uniformly.Most Canadian provinces do not levy personalincome taxes, but impose surcharges on the fed-eral income tax, which is collected by the federalgovernment and remitted to the provinces. Athird example is the case of Russia where al-though the VAT base and rates are set centrally,

proceeds are shared with the regional govern-ments on a derivation basis. In fact, the VAT isactually collected locally and the (variable) fed-

eral share is remitted (not always in a timelyfashion) to Moscow. In which of these three casesis subnational tax power the strongest? Theoryand experience suggests that it is in Canada, theonly one of the three in which the subnationalgovernments do not, in fact, collect the tax inquestion. Why? Because the most critical aspectof subnational taxing power is who is politicallyresponsible for setting the tax rate.

Tax Assignment in Principle

The traditional theory of fiscal federalism pre-scribes a very limited tax base for subnationalgovernments. The only good local taxes are saidto be those that are easy to administer locally, thatare imposed solely (or mainly) on local residents,and that does not raise problems of harmoniza-tion or competition between subnational gov-ernments or between subnational and nationalgovernments.22 The only major revenue sourcethat passes these stringent tests is the residential

property tax. Taxes on vehicles and user fees mayalso fall under this category (Annex 3). Since this

21

Argentinas revenue-sharing system has sometimes

been interpreted along these lines because the pro v-inces have delegated much of their revenue-raising

power to the center and are, so to speak, simply beingcompensated through the so-called coparticipacintransfer. As already noted, however, this interpretation

does not appear to be economically meaningful both

because the revenues are not distributed on a derivationbasis and (especially) because the relevant tax rates andbases are determined by the national government.

22The classic tax assignment arguments are set out in

Musgrave (1983). Recent restatements may be foundin Oates (1998) and McLure (1999a). For a review of

this literature, see Bird (1999a).


20/67

conclusion accords with the reluctance of most finance local services through property taxes andcentral governments to provide subnational gov- user fees on residents, regional governments thaternments access to their lucrative sales and in- are responsible for such social services as a rulecome tax bases, it is not surprising that it has be- cannot. The conventional approach, while notcome conventional wisdom. Subnational gov- firmly grounded in anything except expediency,ernments almost everywhere are urged to make has been to suggest that such governments are, in

more use of property taxes, and criticized when principle, best financed by retail sales taxes, Suchthey do not do so enthusiast ically. taxes are usually assumed to fall mainly on resi-

dents.23 As the experience of the United StatesUp to a point, there is much to be said for this ar- and Canada have long demonstrated, retail salesgument. Unfortunately, that point falls far short taxes can be administered at the regional level.of the task facing many subnational governments However, it seems unlikely that this experiencefor a number of reasons. First, the conventional can be applied to developing countries, whichcase for property taxes is to some extent flawed. have universally found it impossible to administerSuch taxes are costly and difficult to administer such taxes even at the national level.well, and these problems are greatly exacerbatedas the tax burden increases. Moreover, in prac- Given the recent move towards decentralization intice, political reality means that increases in prop- Latin America and the concern frequently ex-

erty taxes are often concentrated primarily on pressed about the resulting strain on intergovern-those nonresidential properties which most lend mental fiscal relations and the possibility of irre-themselves to tax exporting, thus undercutting one sponsible behavior by some subnational govern-of the principal arguments for local use of this tax ments (Tanzi, 1996), the issue of subnational

base. revenues requires re-examination. Internationalexperience suggests that subnational governments

Second, even a well-administered local property carrying out important expenditure functions aretax cannot finance major social expenditures more likely to do so responsibly the more respon-(education, health, social assistance) except per- sible they are for raising the revenues they spend.haps in the richest (and usually largest) communi- While there will obviously always remain an im-ties. To the extent that it is desirable for govern- portant role for intergovernmental transfers, espe-ments to finance from their own revenues the cially in countries with wide regional economic

services they provide, either local governments disparities, there seems to be no reason whyfinanced by property taxes are confined to pro- wealthier regions (including metropolitan areas)viding such local services as street cleaning and should not be able to raise and spend most of theirrefuse removal, or they are heavily dependent on budgets themselves.24

transfers from higher levels of government. Thisis the pattern in most developed countries, in-

cluding the relatively few in which the propertytax is the mainstay of local finance. In the OECDcountries, to the extent that local governments arenot dependent on national transfers, they invaria-

bly impose significant direct taxes on businessesor levy surcharges on national income taxes (Bird

and Slack, 1991).

Third, the conventional argument does not takeadequately into account the existence, in mostlarger countries, of important regional (intermedi-ate) levels of government which often play a ma-

jor role in financing social expenditures. Even iflocal governments, can to a considerable extent,

17

23In reality, most retail sales taxes fall to a considerable

extent on business inputs even in developed countries.One Canadian study, for example, found that betweenone-third to one-half of the retail sales tax base in dif-

ferent provinces consisted of such inputs (Kuo,McGirr, and Poddar, 1988).24

For an earlier statement of this argument, see Bird(1993). Note that an important implication of strength-

ening subnational revenues is that the resources accru-ing to different states or provinces will differ greatly,depending upon their access to the tax base in question.

While transfer systems could, in principle, be adjustedas desired to prevent unduly penalizing poorer regions,as noted earlier, the extent to which such adjustments

will actually be made is country-specific.


21/67

Two basic principles of assigning revenues to Not everyone would agree that all these charac-subnational governments may be suggested. First, teristics are necessarily or equally desirable. Forown-source revenues should ideally be suffi- example, is it unequivocably good that subna-cient to enable at least the richest subnational tional governments should be insulated from ei-governments to finance from their own resources ther the tax base consequences of their tax rateall locally-provided services primarily benefitting choices or from inflation? More importantly, the

local residents. Second, to the extent possible, characteristics that may be sought in an ideal localsubnational revenues should be collected only tax from the point of view of local and centralfrom local residents, preferably in relation to the governments are not necessarily compatible.

perceived benefits they receive from local serv- Both levels might agree that the tax base shouldices. This approach is in turn based on three sim- be immobile, and perhaps also that the tax yield

ple principles: First, more attention should be paid should be stable and adequate to meet local needs.to matching expenditure and revenue needs at dif- Yet, while central governments should be con-ferent levels of government. Second, greater effort cerned about ensuring that the tax burden cannotshould be made to ensure that all governments be exported to nonresidents and that the local tax

bear significant responsibility at the margin for base is visible to ensure accountability, such at-financing the expenditures for which they are po- tributes are less likely to be appealing to locallitically responsible. And third, subnational taxes governments.

should not unduly distort the allocation of re-sources. The bottom line, however, is that unless local

governments have some significant degree ofThe characteristics that might be sought in a sub- freedom to alter the level and composition of theirnational tax satisfying these requirements may revenues, neither local autonomy nor local ac-include the following: countability are meaningful concepts. In particu-

lar, as noted earlier, subnational governments

The tax base should be relatively immobile, to should be able to set tax rates (albeit perhapsallow local authorities some leeway in varying within limits). Though this condition is seldom

rates without losing most of their tax base. satisfied in developing countries, such rate flexi-bility is essential if a tax is to be adequately re-

The tax yield should be adequate to meet local sponsive to local needs and decisions, while re-needs and sufficiently buoyant over time (that maining politically accountable.is, it should expand at least as fast as expendi-tures). Local governments should not only have access to

those revenue sources that they are best equipped

The tax yield should be relatively stable and to exploit (such as residential property taxes andpredictable over time. user charges for local services) but they should be

both encouraged and permitted to exploit these

It should not be possible to export much, if any, sources. A potential danger in permitting localof the tax burden to nonresidents. governments even limited freedom to tax is that

they will not utilize fully all the revenue sourcesopen to them, thus allowing the level and quality

The tax base should be visible to ensure ac-of public services to deteriorate. If intergovern-countability.

mental fiscal structures are properly designed, thisshould not be a real problem. If the service in

Taxpayers should perceive the tax to be rea-question is one of national importance (such assonably fair.research, for example) or one in which there is astrong national interest in maintaining standards

The tax should be relatively easy to administer(for example, poverty alleviation), it should pre-efficiently and effectively.sumably be funded and monitored by the centralgovernment. If it is not a matter of national inter-

18


22/67

est, why should the central government be con- These general points are discussed further withcerned? Of course, this argument assumes that the respect to two distinct, and important problems incentral government has both adequate information subnational taxation. The first is how to financeand a predetermined transfer system, as discussed regional governments. This problem is especiallyfurther in Sections 6 and 9 below. important in larger and federal countries, but it is

by no means confined to them. The second prob-

If local electors do not like what their local gov- lem, found everywhere, is how to finance localernment does, or does not do, they can (try to) services more narrowly understood.throw the rascals out at the next election. Thefreedom to make mistakes, and to bear the conse- Financing Regional Governmentsquences of one's mistakes, is an important compo-nent of local autonomy in any country. Indeed, The present assignment of taxes in most countriesunless local governments are given some degree with important regional levels of government,of freedom with respect to local revenues, in- such as Brazil and Argentina, seems particularlycluding the freedom to make mistakes (for which deficient. One problem is that there is a signifi-they are accountable to voters), the development cant vertical imbalance between expenditures andof responsible and responsive local government is revenues, with consequent implications for auton-likely to remain unattainable. Of course, if the omy, efficiency, and accountability. Another

conditions of effective democracy and adequate problem is that the present confused and confus-information are not satisfied, or if those who fail ing system results in significant costsCcosts ofto collect local taxes, or to spend revenues effi- administration, costs of compliance, and costsciently are bailed out by discretionary transfers, arising from tax-induced inefficiencies in the allo-the rascals may not be thrown out but rather re- cation of scarce resources.25

elected for their success in obtaining a larger shareof other people's money. Countries that, for what- In principle, multitiered governments work bestever reason, fail to set up an appropriate intergov- when taxes and the benefits of public spending areernmental fiscal structure are likely to have both as closely related as possibleCwhen citizen-voter-more problems in managing decentralization and consumers residing in a particular political juris-less satisfactory policy outcomes. diction pay for what they get from the public sec-

tor and get what they pay for (that is, benefit from

Another danger is that local governments may the expenditures financed by the taxes they pay).attempt to extract revenues from sources, for Obviously, when citizens reside in several over-which they are not accountable, thus obviating the lapping jurisdictions (local-state-nation) this so-

basic efficiency argument for their existence. To called principle of fiscal equivalence (Olson,avoid this problem, it may be desirable to limit 1969) suggests that they should pay taxes to eachlocal government access to taxes that fall mainly

on nonresidents (such as most natural resourcelevies, pre-retail stage sales taxes and, to someextent, nonresidential real property taxes). Oneway to deal with this problem may be to establisha uniform set of tax bases for local governments(perhaps different for different categories such as

big cities, small towns, and rural areas), with alimited amount of rate flexibility permitted in or-der to provide room for local effort while re-straining unproductive competition and unwar-ranted exploitation. If inappropriate tax bases areassigned to subnational governments, wastefulcompetition and undesirable tax exporting arelikely to result.

19

25For a detailed discussion of the lat ter problem, see

Bird and Mintz (1999). A particular problem arises in

some countries because of the uneven geographicaldistribution of natural resources and the resulting sev-erance of the link between local taxes and benefits

when subnational governments are able to tax such

resources, as is often the case. The ideal solution is toprevent them from doing so (Mieszkowski, 1983), but

if this is not possible, considerable care must be takenin designing other aspects of intergovernmental fi-nance, particularly transfer systems, in order to offset

the resulting distortion as much as possible. Unfortu-nately, although this problem is important in a numberof countries in the region, it cannot be discussed further

here.


23/67

level corresponding to the benefits they receivefrom each jurisdiction. In this framework, inter-governmental transfers would exist to restore thisequivalence when, for example, some benefitsflow from one jurisdiction to another or (nega-tively) when some taxes levied by one jurisdiction

are paid by persons residing in another.26

Considerations of administrative efficiency andfeasibility may dictate that higher (or lower) lev-els of government impose certain taxes or carryout certain expenditures even when it would not

be strictly appropriate to do so on equivalencegrounds. Vertical fiscal flows, like those thatdominate the intergovernmental fiscal scene inmost countries, are motivated largely by this con-sideration, at least with respect to those flowing toricher jurisdications. In contrast, if more adequate

subnational taxes are made available, this fiscalgap (Boadway and Hobson, 1993) argument fortransfers disappears. There is then no case foruniversal intergovernmental fiscal transfers, sincein this system the richest units of government atsubnational levels should be essentially self-sufficient (Bird, 1993). Any grants from higherlevels of government made for reasons of regionalequalization in this system should then be clearlyinframarginal, so that, as McLure (1999a) notes,all subnationalgovernments face the full marginaltax price of the spending decisions for which they

are responsible, thus yielding the hard budget con-straint emphasized by such authors as Tanzi(1996).

Good subnational taxes (at both regional and locallevels) should thus satisfy two main criteria. First,they should provide sufficient revenue for therichest subnational units to be essentially fiscallyautonomous.27 Second, they should clearly im-

pose fiscal responsibility at the margin on subna-

20

tional governments. As mentioned earlier, thesimplest and probably best way to achieve thelatter goal is by allowing them to establish theirown tax rates with respect to, at least, some majortaxes.

The most immediately important subnationalrevenue issue facing large countries in LatinAmerica is the development of a satisfactoryrevenue base for regional governments, that is,one for which they are politically responsible.While more can be done in the form of regionalexcise taxes, especially on vehicles and fuels, inmost countries (Bird, 1999b), if regional govern-ments have significant expenditure responsibili-ties, there are really only two important possibili-ties: a surcharge on the central personal incometax (PIT) or a surcharge on the central value-

added tax (VAT). If local governments are to beboth large spenders andless dependent on grants,they must have access to national tax bases. Pig-gybacking through surcharges seems to be theonly viable way to do this while retaining an im-

portant element of political accountability.

The possibility of local income tax surchargesnow seems to be broadly accepted. Unfortu-nately, as Shome (1999) has recently emphasized,few Latin American countries have sufficientlyrobust central income taxes to offer much hope

that subnational governments will soon be able toderive much revenue from this source. A poten-tially more promising alternative for subnationalrevenues may thus turn out to be a surcharge onthe VAT. Such a tax already exists and workswell in Canada, and its implementation nowseems feasible even in countries with less well-developed tax administrations (Annex 4). At leastin the larger Latin American countries with fed-eral features, such as Brazil, Argentina, or Mex-ico, this path to increased regional taxation seemsto warrant further exploration.

26

Note that such transfers would be horizontal, betweenprovinces or municipalities, and not between levels of Financing Local Servicesgovernment. An example of such a system (at least in

principle the practice is quite deficient from this per- Turning to local taxes, apart from user chargesspective) is theFondo Comn in Chile.27 (Annex 3), there seem to be only two major possi-

This does not preclude intergovernmental fiscal trans-bilitiesa revised, and revived, property tax and

fers to achieve the usual spillover objectives but also,an improved form of local business taxation.

as noted in Section 6, to ensure the adequate provision

of certain services to national standards


24/67

The Property Tax. Almost without exception, of the size of their tax bill than those whose take-much can and should be done to strengthen the home pay is reduced by weekly or monthly taxdeficient property taxes now in place in most deductions. The need to make large periodicLatin American countries. The tax should be sim- payments may well add to the accountability and

plified and applied uniformly. Cadastral maps responsibility of governments, but it also in-should be updated and valuations made more con- creases the sensitivity of taxpayers to even nomi-

sistently and currently. Improved use should be nal increases in taxes.made of flows of information from property reg-istries, local building license authorities, public The inelasticity of the property tax has a similarutilities, etc. In addition, as Dillinger (1991) has effect. Since the base of this tax does not increaseargued, from a revenue perspective, more atten- automatically over time, the periodic nominal in-tion should be paid to improving the sharp end creases in property tax bills needed to maintain(collection and enforcement) rather than to the real revenues when price levels rise require in-technically more costly (and less immediately creased tax rates. In terms of political account-

productive in terms of revenue) mapping and sur- ability, the need to confront the people with theveying of the tradtional cadastral approach. cost of government is a virtue of the property tax.

However, the downside, at least from the govern-Nonetheless, property taxes are not easy to ad- ment's point of view, is the heightened visibility

minister, particularly in countries where inflation of nominal tax increases and the accompanyingis endemic, and they are never politically popular political resistance.owing to their visibility and to certain inherentadministrative difficulties. Even in the most so- Finally, property taxes best finance such services

phisticated countries, local property taxes seldom as roads and garbage collection. The quantity andyield enough to finance local services. No devel- quality of these services (or their absence) is thusoped country which depends significantly upon readily linked to the property tax. When potholes

property taxes for local fiscal resources have a develop in their street, taxpayers are understanda-local government sector that accounts for more bly quick to question the taxes that supposedlythan 10 percent of total public spending (Bird and finance street repair. Again, the very feature thatSlack, 1991). Similarly, property taxes seldom makes the property tax a good source of localaccount for more than 20 percent of local current government revenue also makes it especially vul-

revenues (or less than 1 percent of total public nerable to political resistance.spending) in developing countries (UNCHS,1996). The property tax is a useful, even neces- Other problems result from property tax admini-sary, source of local revenue, but it cannot easily stration. As a rule, for example, property is as-

provide sufficient resources to finance a signif i- sessed on the basis of its market value, usuallycant expansion of local public services in most defined as the price struck between a willingcountries. Indeed, many countries have been buyer and a willing seller in an arm's length trans-hard-pressed even to maintain the present low action. Even in countries with well-developedrelative importance of property tax revenues in the property tax systems, political and technical issuesface of varying price levels and political difficul- lead to discrepancies between assessed values andties. market values within classes of property, between

classes of property, and across municipalities

These facts reflect both the political and the ad- (Bird and Slack, 1993). Since taxpayers can eas-ministrative realities of property taxation. One ily compare their property taxes with those ofreason for the commonly observed phenomenon similar properties in their neighborhood, such dis-of widespread resistance to the property tax, for crepancies lead both to specific assessment ap-example, is simply its visibility. The tax generally peals and to general pressure for tax relief.has to be paid directly by taxpayers in periodiclump sum payments. Taxpayers who pay taxes For such reasons, experience around the worlddirectly to the government tend to be more aware suggests that the political cost of reliance on resi-

21


25/67

dential property taxes is so high that no govern-ment with access to cheaper sources of financewill willingly do so. Both intergovernmentaltransfers, which can be spent as local governmentswish, and local taxes on business, which canlargely be exported, must, as a rule, be curtailed if

local citizens are to be confronted more openly(through their local tax bills) with the true eco-nomic (and political) costs of the decisions made

by those whom they have elected.

In particular, the temptation to indulge in polit i-cally painless but economically inefficient taxexporting suggests that constraints should be

placed on the local taxation of nonresidentialproperty.28 As discussed below, some local busi-ness taxation may be justified on benefit (effi-ciency) grounds, but it should always be strictly

constrained in order to preclude localities fromattempting to shift the costs of services to outsid-ers.

Other policy reforms are needed to turn the prop-erty tax into a responsive instrument of local fiscal

policy. First and most importantly, as emphasizedearlier, local governments must be allowed to settheir own tax rates. Few countries currently givelocal governments much freedom in this respect.Secondly, the tax base must be maintained ade-quately. In countries with much inflation, some

form of index adjustment may be advisable(World Bank, 1989). National assessing agenciesmust, if necessary, be provided direct financialincentives to keep the tax base up to date. Finally,

procedural reforms are often needed to improvecollection efficiency, valuation accuracy, and thecoverage of the potential tax base. None of thesesteps is easy, but countries that want to have localgovernments that are both responsive and respon-sible must follow this difficult road. There are noshort cuts to successful local property taxation.

22

Local Business Taxes. Another critical problem inmany countries is the reform of various unsatis-factory subnational taxes on business. While theability to distort market conditions through suchtaxes must be restrained (for example, by estab-lishing a uniform national base for local business

taxation, with a minimum and maximum rate29)there is both an economic (benefit) case for someregional and local taxation of business and, itseems,

ifr_in_lac_by_bird

Documents