A. Case for Decentralization

1. In Mali, decentralization is viewed as critical to addressing the root causes of the recent political and security crisis. Specifically, it should help restore national integrity; increase the effectiveness of public services by bringing the spending decisions to the intended beneficiaries. This, in turn, would help strengthen local autonomy and public accountability; promote local development; and reduce poverty and inequalities (main objective).

2. The immediate spur for decentralization has been the desire to secure a lasting peace on the entirety of Mali’s territory. A sense of disenfranchisement in the North caused by decades of benign neglect by the central government has been one of the root causes of social and political tensions. Recognizing the North’s aspirations for a measure of control over public services delivered there is a necessary first step in national reconciliation.

3. However, the case for decentralization goes deeper. It has to do with particularly rapid demographic changes in urban centers that have strained the local governments’ capacity to provide adequate basic services.² It has been observed that the fiscal structure for the allocation of resources to urban local governments does not support sound investment planning and service delivery.³ There is a disconnect between the functional mandate given to cities and the amount of resources allocated to them to carry out this mandate. The resources are inadequate, but also unpredictable—in terms of quantity and frequency—and most of them are earmarked according to priorities established by the center. This leaves cities with insufficient discretionary resources that would allow them to determine priorities based on local needs and respond to them in a systemic, sustained, predictable manner.

4. The Peace Accord is not the first time decentralization has been on the policy agenda. The beginnings of the decentralization reform date back to Mali’s 1992 constitution. In 2005, an attempt was made at devolution of some central government function to lower levels of government, but it has left some ambiguities in the transfer of functional assignments.⁴ Today, more clarity is needed to free up the development of local capacity and improve the effectiveness of delivery of local services.

5. Mali’s decentralization efforts reflect a trend within SSA, where other countries have undertaken decentralization for much the same reasons as Mali (boxes 1 and 2). The trend toward devolution of spending—and to a lesser degree, devolution of revenue-raising responsibilities—to lower (sub-national) governments reflects, in part, the political evolution toward more democratic and participatory forms of government. At the same time, the economic efficiency argument has been used to advance this process—namely, ensuring a closer alignment between quantity, quality and composition of public services and the preferences of their beneficiaries.

6. In the remainder of this note we consider the steps to achieve a greater degree of fiscal decentralization and discuss some related practical considerations. First, there has to be agreement on the quantity and quality of the services to be provided by each government level. This, in turn, has implications for the amount of total financing required. The difference between the needed financing and the available (own) financing at the subnational government level is the implied amount of transfers from the center.

B. Assign Spending Responsibilities to Different Levels of Government

7. Identification of functions to be decentralized and associated transfers is likely to be a complex issue that requires careful deliberation. The alignment of spending responsibilities can entail welfare gains if done “right”. There is broad consensus that efficiency in the allocation of resources is best served by promoting a closer correspondence of expenditure priorities with the preferences of affected population by assigning responsibility for each type of public spending to the level of government that most closely represents the beneficiaries of these outlays.

8. However, the design of intergovernmental fiscal relations cannot be determined exclusively by allocative (economic) efficiency considerations. Noneconomic factors—political, social, and cultural—will also play a role. Within economic factors, allocative efficiency needs to be weighed against the objectives of income distribution and macroeconomic management.⁶ The design of intergovernmental relations will have implications for public financial management (PFM). Decentralization entails additional cost for lower levels of government; it complicates budgetary coordination (in particular where the number of local government entities is large); and it usually requires special treatment of large cities, which have their specific problems.

9. Efficiency gains expected from decentralization can turn out to be more modest in the presence of institutional constraints. ⁷ For example, administrative capacity of regional and local governments may be weak. Development of modern and transparent public expenditure management systems—including mechanisms of financial control, reporting and accounting—has proven to be a considerable challenge at the national government level. It is likely that it will be more so at lower government levels. It is also generally the case that the degree of decentralization is not an outcome of some optimizing decision-making, but rather an outcome of historical developments or political factors. For these reasons the size spending by subnational governments varies greatly from country to country.⁸

10. Central government spending has the advantage of assuring uniformity in the provision of public services. In many cases, where decentralization creates differences in policy and the level of provision between jurisdictions, undesirable population and capital movements follow. The case for central spending could be made for provision of national public goods—defense, foreign affairs, trade, interstate transport and communications. In these cases decentralization would lead to inefficient outcomes. Decentralization could also be inefficient in cases where significant economies of scale are present.

C. Provide Subnational Governments with Adequate Resources

11. Assignment of spending responsibilities needs to be accompanied by assignment of commensurate own sources of revenue for regional and local governments. This raises question how the powers to tax should be distributed across government levels. At one end of the spectrum, most or all taxation powers could be assigned to subnational governments. This would not be desirable—both distributional and macroeconomic management considerations argue against such an arrangement, as this would deprive the central government of necessary policy instruments. At the other end of the spectrum, all or most taxing powers could be assigned to the central government. This would also be undesirable: by separating spending authority from revenue-mobilization responsibilities, such arrangements obscure the link between the benefits of public expenditures and the taxes levied to finance them. They do little to promote fiscal responsibility by subnational politicians and their electorate.

12. Reliance on own sources of revenue tends to promote greater fiscal responsibility by regional and local governments, and political accountability of government officials to their electorate. Revenues associated with local governments comprise three main types:

• Charges linked directly to the service provided by the local government (parking, school fees, medical fees, rental of space at local markets, etc.).

• Licenses (professional licenses, business licenses)

• Taxes; property taxation (typically in cities, where a land register is available).

13. Taxes that have proven to fit well the regional and local government requirements are those characterized by low mobility and fairly-even distribution of the base over the national territory and relative stability over the cycle. These considerations favor assignment (full or partial) of the personal income tax, general retail sales taxes and some excises to subnational governments. It is also desirable, although not strictly required, to harmonize the definition of the tax base across the national territory. This will tend to minimize distortions and tax-induced movements of labor and capital.

14. An inexperienced tax administration at the subnational government level will initially constrain the effectiveness of revenue decentralization. One way to safeguard revenue performance is by providing technical training to the new tax administrations. Alternatively, the sub-national governments could continue to rely on the central tax administration for tax revenue collection and receive a set percentage from the national take of a particular tax.

D. Intergovernmental Transfers

15. Subnational governments’ “own resources” are typically complemented by transfers from the central government. There are two possible reasons why a system of transfers is put in place:

• the need to correct vertical imbalances arising from a mismatch between the large expenditure responsibilities of subnational governments and the assignment of major taxes to the central government; and

• with the need to reduce horizontal imbalances (among regions) arising from uneven capacities of subnational governments to raise their own revenues. Horizontal imbalances may also arise because different regions face different costs and demand pressures in the areas of their responsibilities.

16. A higher degree of decentralization tends to complicate achievement of distributional objectives, particularly if regional differences are large. Reducing the regional disparities becomes critical if a country strives to maintain adequate economic and social cohesion. In countries where large disparities exist, it is important that the central government retain sufficient resources to undertake some form of “equalization” transfers. Last, but not least, a system of transfers should be designed in a way that will not discourage subnational governments’ own tax effort and their cost effectiveness.

17. The most common way of topping up subnational governments’ own revenue is through revenue-sharing arrangements. These arrangements can be structured in different ways:

• Some are designed to address vertical imbalances only; others address both vertical and horizontal imbalances.

• Distribution of shared revenues can be based on a derivation principle, or it can utilize formulas based on redistributive criteria.⁹

• Sharing can be done on a tax-by-tax basis, or on the entire pool of central government tax revenues.

18. Distribution of shared revenues on a derivation principle means that each jurisdiction receives a part of the revenue in proportion to the amount collected on its territory. Such a system would do little to reduce regional disparities, as it would only reflect the existing taxing capacity of the regions. Per-capita income or other variables can introduce an element of redistribution into the revenue-sharing formula.

19. Revenue sharing done on a tax-by-tax basis (with each tax having its own set of distribution coefficients) tends to be rather complex and may distort the central government’s incentives in collection and enforcement. Thus it may be preferable to apply revenue sharing on the entire pool of government revenues. But even here, care needs to be taken to define tightly the revenues to be included.¹⁰ A sharing arrangement for a single tax—the VAT, for example—might be a clean solution, assuming it is relatively stable, generates sufficient revenue and can be clearly delineated. Such an arrangement would be following successful examples of countries like Cameroon and Senegal.

20. Revenue-sharing parameters can be codified in a law, or sometimes even in the Constitution. However, there is a trade-off between predictability (which is aided by a law that fixes the coefficients) and flexibility, which may be needed in some situations. For example, fixed coefficients may make the tax revenue pro-cyclical and force subnational governments to make pro-cyclical spending adjustments. For example, a tax revenue shortfall will lead to spending cuts if the government doesn’t have other financing options.¹¹

21. In a number of countries, the revenue-sharing arrangements are monitored by specialized independent bodies. In some cases, these bodies are also tasked with determination of the sharing coefficients.¹² To make the revenue-sharing schemes credible, it is important to develop reliable, transparent and timely statistics.

22. Allocation of national revenues in favor of subnational governments can be also effected via grants. There are three types of grants: general purpose (non-targeted), specific purpose (targeted), and sectoral (block) grants.

• General purpose grants represent unconditional transfers of resources from the central government to subnational governments. They typically aim to address vertical and horizontal imbalances.

• Specific purpose grants come with conditionality—they are meant to be used for a specific purpose and therefore are not fungible. They can be recurrent and finance some type of current expenditures, or one-off, in which case they can finance investment projects. Their advantage, from the central government’s point of view, is control over local governments’ spending. The strings that are attached to these grants represent a drawback from the point of view of the beneficiary level of government. That is because they reduce the subnational government’s room for maneuver and its power to choose. As such, they seem to go against the main (efficiency) argument in support of decentralization.

• Sectoral grants fit somewhere between the two preceding types of grants. They are not narrowly specific, but they are destined for some broad area of expenditure (health, education, etc.).

23. It is important that the system of intergovernmental grants be clearly defined. This facilitates planning and budget execution, as each level of government can better anticipate the amount and timing of these transfers. From this point of view, general purpose grants are superior to narrow-purpose grants, which are subject to uncertainty for both donor and recipient governments.

E. Subnational Government Borrowing

24. There is great diversity in the way in which countries control borrowing by subnational governments. These ways can be divided among four broad categories: (i) autonomy in borrowing decisions (countries rely on market discipline to provide the necessary incentive for restraint); (ii) cooperation between different levels of government in developing controls over borrowing; (iii) rules-based controls; and (iv) administrative controls.

• Examples of rules-based controls include the rule by which borrowing can only be for investment purposes (classical golden rule); or the amount of annual debt service cannot exceed a certain percentage of government own revenue. A particular rule is the one which prohibits borrowing by subnational governments altogether. For example, in Ethiopia, regions can only borrow for the treasury management purposes. Of course, where local government borrowing is prohibited, the central government needs to ensure that local governments have sufficient resources to carry out necessary investment.

• Administrative controls can include setting an annual ceiling for government borrowing. Often, this ceiling is set (approved) by the national legislature at the same time as the national budget. The central government can also exert control over borrowing by subnational governments by selective issuance of guarantees.

25. For Mali, at least initially, borrowing by subnational governments should be forbidden. This arrangement could be revisited if (and when) these governments develop adequate debt management capacity. Administrative controls, involving, for example, approval of individual loan operations, would allow tight control over government borrowing. Nevertheless, such a regime might sit uncomfortably alongside the trend toward greater decentralization and autonomy. Thus some of the rules-based approaches could be explored. Linking borrowing decisions to the debt-servicing capacity of subnational jurisdictions may be an effective way of keeping overall debt within sustainable bounds.¹³

F. Macroeconomic Policy Coordination and Overall Policy Coherence

26. Decentralization needs to be accompanied by the center’s stronger capacity to coordinate the country’s overall fiscal policy in order to safeguard macroeconomic stability. In Mali, which belongs to a monetary union and thus cannot use monetary and exchange policy for macroeconomic stabilization, national fiscal discipline becomes critical. This requires some degree of policy coordination between the central and subnational governments in the area of budgetary and borrowing policy.

27. The center needs to develop a set of rules to guide spending and borrowing decisions of subnational governments, as well as policy levers to influence those decisions. At the same time, it is necessary to develop a monitoring capability to track the finances of lower levels of government.

G. Selected Practical Aspects of Decentralization

28. Fiscal decentralization raises both policy and technical issues. Devolution of revenue mobilization and spending responsibilities will pose new challenges for tax administration and for public expenditure management. In Mali, prudence argues in favor of a gradual approach. The central government may need to continue maintain some measure of control over tax administration and most of the spending for some time. The speed of the transition to a more devolved model will depend on how fast the subnational governments develop their own systems and procedures, technology and human resources.

29. To minimize the risk of disruption, tax administration could remain in the hands of the central government even for taxes that have been assigned to subnational governments in the process of decentralization. The advantages would include continuity, consistency of treatment of taxpayers across the country, lower compliance costs, and economies of scale. On the other hand, a decentralized administration would promote greater ownership (responsibility and accountability) in subnational governments for tax revenue performance.

30. Transition is likely to involve incremental costs. It will be important that, once the transition phase is over, the overall cost of the provision of public goods and services remains comparable to the overall cost of the centralized system. That will require, among other things, a redeployment of personnel from the center to the regions—a process that will require long advanced planning.¹⁴ Transfer of personnel is expected to be accompanied by a transfer of relevant assets (real estate, vehicles, equipment). That will require careful registration of this property as well as development by subnational governments of the capability to carry out necessary maintenance of these assets. Some practical questions that are likely to arise include the following (shown with possible indicative responses):

• How are government services to be shared between levels of government? The Malian law provides for the following services to be shared: education, health, water (rural and urban hydraulics) public transportation, road infrastructure, housing and urban development, sports and culture.

• Which government is responsible for human resource management? (In the sectors of health and education in particular). Given the Malian administrative architecture and the lack of capacity (outside the capital city, Bamako), the initial devolution of responsibilities should not go beyond the regional level. In the first stage of the reform, the responsibility for human resource management should remain central, or be indirectly monitored through earmarked transfers.

• How to coordinate policies in areas that are shared by two (or more) levels of government? The central government should be responsible for maintenance of national policy coherence and application of nation-wide uniform standards.

• Budgetary impact of decentralization should be neutral (reduction in the center and augmentation in subnational government levels). Who will assure that? (Kenya had an independent commission to steer the transition.) As in Rwanda and South Africa too, local jurisdictions should have reporting obligations of monthly and annual financial/ fiscal accounts. To enforce compliance, central government transfers should be conditional the credibility of the local budgets and local (fiscal) discipline.

H. In Conclusion

31. Mali’s project of fiscal decentralization is supported by economic and political considerations. At the same time, it is fraught with risks. This note identified a number of practical issues that need to be addressed to ensure that fiscal and macroeconomic stability are maintained. The reform entails a radical transformation of intergovernmental relations consisting of a number of interlocked steps. Policy coordination will be critical, and each step needs to be carefully prepared. Where feasible, a gradual approach is preferable—it will allow time for development of relevant policy and administrative capacity at subnational government levels, and for the central government to strengthen its monitoring capability and establish its policy coordination credentials.