Lessons for Effective Fiscal Decentralization in Sub-Saharan Africa
  • 1 https://isni.org/isni/0000000404811396, International Monetary Fund

Fiscal decentralization is becoming a pressing issue in a number of countries in sub-Saharan Africa, reflecting demands for a greater local voice in spending decisions and efforts to strengthen social cohesion. Against this backdrop, this paper seeks to distill the lessons for an effective fiscal decentralization reform, focusing on the macroeconomic aspects. The main findings for sub-Saharan African countries that have decentralized, based on an empirical analysis and four case studies (Kenya, Nigeria, South Africa, Uganda), are as follows: • Determinants and effectiveness: Empirical results suggest that (1) the major driving forces behind fiscal decentralization in sub-Saharan Africa include efforts to defuse ethnic conflicts, the initial level of income, and the urban-ization rate, whereas strength of democracy is not an important determi-nant for decentralization; and (2) decentralization in sub-Saharan Africa is associated with higher growth in the presence of stronger institutions. • Spending assignments: The allocation of spending across levels of gov-ernment in the four case studies is broadly consistent with best practice. However, in Uganda, unlike in the other three case studies, subnational governments have little flexibility to make spending decisions as a result of a deconcentrated rather than a devolved system of government. • Own revenue: The assignment of taxing powers is broadly in line with best practice in the four case studies, with the bulk of subnational revenue coming from property taxes and from fees for local services. However, own revenues are a very small fraction of subnational spending, reflecting weak cadaster systems and a high level of informality in the economy.

Abstract

Fiscal decentralization is becoming a pressing issue in a number of countries in sub-Saharan Africa, reflecting demands for a greater local voice in spending decisions and efforts to strengthen social cohesion. Against this backdrop, this paper seeks to distill the lessons for an effective fiscal decentralization reform, focusing on the macroeconomic aspects. The main findings for sub-Saharan African countries that have decentralized, based on an empirical analysis and four case studies (Kenya, Nigeria, South Africa, Uganda), are as follows: • Determinants and effectiveness: Empirical results suggest that (1) the major driving forces behind fiscal decentralization in sub-Saharan Africa include efforts to defuse ethnic conflicts, the initial level of income, and the urban-ization rate, whereas strength of democracy is not an important determi-nant for decentralization; and (2) decentralization in sub-Saharan Africa is associated with higher growth in the presence of stronger institutions. • Spending assignments: The allocation of spending across levels of gov-ernment in the four case studies is broadly consistent with best practice. However, in Uganda, unlike in the other three case studies, subnational governments have little flexibility to make spending decisions as a result of a deconcentrated rather than a devolved system of government. • Own revenue: The assignment of taxing powers is broadly in line with best practice in the four case studies, with the bulk of subnational revenue coming from property taxes and from fees for local services. However, own revenues are a very small fraction of subnational spending, reflecting weak cadaster systems and a high level of informality in the economy.

Chapter 1 Context

At independence,1 sub-Saharan African countries inherited a highly centralized model of territorial government and fiscal arrangements. Most post-colonial regimes promised a swift decentralization of power, resources, and responsibilities to subnational levels of government. However, not much changed across the continent for quite some time. Through the early 1990s the combination of administrative centralization with a nondemocratic political framework maintained an unaccountable and inefficient system in most sub-Saharan African countries. Owing to the chronic scarcity of resources, service provision and maintenance of basic infrastructure were neglected, particularly in the rural areas, and the very few resources devolved to local governments were concentrated in urban areas, especially the capital city (Brosio 2000).

Fiscal decentralization reforms started gaining momentum in sub-Saharan Africa in the early 1990s. Fiscal decentralization is most advanced in three sub-Saharan African countries—Ethiopia, Nigeria, and South Africa—where spending at the subnational government level represents about half of total general government spending (Figure 1 and IMF 2006). Only in a handful of other sub-Saharan African countries is spending at the subnational level significant—notably in Kenya, Rwanda, Tanzania, and Uganda, where it amounts to about 15–20 percent of general government spending. Subnational government spending in these countries is in line with the level in other emerging markets, although well below that in a typical Organization for Economic Co-operation and Development (OECD) country (Figure 2).

Figure 1.
Figure 1.

Subnational Government Spending

(Percent of general government spending)

Citation: Departmental Papers 2018, 009; 10.5089/9781484358269.087.A001

Sources: Organization for Economic Co-operation and Developments database; and IMF staff calculations.
Figure 2.
Figure 2.

Subnational Government Spending

(Percent of GDP)

Citation: Departmental Papers 2018, 009; 10.5089/9781484358269.087.A001

Sources: Organization for Economic Co-operation and Development (OECD) and IMF Government Finance Statistics databases; and IMF staff calculations.1 Includes Bolivia, Colombia, Honduras, Indonesia, Jordan, Paraguay, Peru, Russia, Thailand, Tunisia, Turkey, and Ukraine.

Decentralization has recently come to the forefront of the policy agenda in many sub-Saharan African countries. A number of sub-Saharan African countries have announced major decentralization reforms, including Lesotho, Liberia, Madagascar, Mali, and Zambia. This seems to reflect two driving forces: (1) a political evolution toward more democratic and participatory forms of government, which is creating demand for greater local voice in spending decisions, and (2) efforts to increase the participation of various ethnic groups or of former warring factions in the governance of the country, with a view to increasing social cohesion and reducing risks of secession.

Against this background, the paper seeks to distill the lessons of decentralization in sub-Saharan Africa and elsewhere, focusing on the macroeconomic aspects. Recent decades have seen a push toward fiscal decentralization around the world. For example, Aldasoro and Seiferling (2014) find an upward trend in expenditure decentralization since the mid-1990s in a number of advanced and emerging market economies. Dziobek and others (2011) on the other hand find that the level of decentralization has been relatively stable since the early 1990s, except the the transtion economies in the Eastern and Central Europe and several countries of the Former Soviet Union. This paper does not take a position for or against fiscal decentralization. Rather, given that decisions in this area reflect political preferences, the paper uses lessons from international experience to suggest how to minimize macroeconomic risks while strengthening the quality of public services when embarking on fiscal decentralization reforms. Section 2 covers the pros and cons of fiscal decentralization; Section 3 summarizes best practices for the main design elements of fiscal decentralization (focusing on the macro-economic aspects); Section 4 takes stock of fiscal decentralization so far in sub-Saharan Africa; and Section 5 concludes. The experience in Kenya, Nigeria, South Africa, and Uganda is summarized in Annexes 14.

Chapter 2 Pros and Cons of Fiscal Decentralization

The classic argument in favor of decentralization is that it can increase economic efficiency and reduce regional income disparity (Table 1). The traditional economic rationale for decentralization relies mainly on efficiency arguments related to the purported information advantage of local politicians (Musgrave 1959). This advantage is expected to increase both allocative efficiency, by better matching policies with citizens’ preferences (Oates 1972), and productive efficiency; that is, more output for the same input (Ahmad, Brosio, and Tanzi, 2008). In addition, fiscal decentralization can help reduce regional income differences and can become a positive force in efforts to alleviate poverty. Fiscal decentralization also offers the potential to address perceived ethnic and political bias by giving local communities greater control over resources and decisions about service delivery, especially if the control is accompanied by greater transparency and accountability.

But fiscal decentralization can also have drawbacks (Table 1).

  • Macro-fiscal risks: One of the main risks with fiscal decentralization is that it can undermine fiscal discipline. This risk arises because decentralization of spending is usually financed through a common pool of transfers from the center; as a result, subnational governments do not fully internalize the cost of local expenditure. Combined with soft budget constraints, this can result in overspending and lower tax effort at the subnational level (Escolano and others 2012). A mismatch between spending responsibilities and the ability to collect revenue has also been an issue. Indeed, several studies suggest that a faster decentralization of spending relative to that of revenue collection (resulting in large vertical fiscal imbalances) tends to increase the overall fiscal deficit and result in higher public debt (Eyraud and Lusinyan 2011; Aldasoro and Seiferling 2014) and weaken the discipline-enhancing effect of fiscal rules (Kotia and Lledó 2016).

  • Lower spending efficiency and higher nuisance taxes: Weak administrative or technical capacity and typically higher corruption at the subnational levels may result in less efficient and effective provision of local services. In addition, decentralization of spending without adequate resources from the center can contribute to the imposition of nuisance taxes by subnational governments to close the fiscal gap, undermining the business environment. Decentralization may also contribute to increasing horizontal disparities rather than reducing them in cases of geographical differences in economic endowments (for example, countries that share much of their natural resource revenue with producing regions).

  • Political risks: Devolution could undermine national unity by encouraging fragmentation along ethnic lines or by decentralizing corruption, leaving citizens worse off if local elites are able to capture resources to the detriment of the majority (World Bank 2012).

Table 1.

Fiscal Decentralization Trade-Offs

article image
Source: Fedelino and Ter-Minassian 2009.

The push for fiscal decentralization often reflects political pressures. Economists tend to focus on efficiency arguments, but in many countries fiscal decentralization is driven by political demands for regional autonomy. In some countries (for example, Italy), the economic divide between rich and poor regions has led to demands for autonomy from the former. In some other OECD countries, moves toward fiscal decentralization have reflected ethnic or linguistic divides (for example, Belgium, Spain, and the United Kingdom). The OECD experience shows that the more culturally or ethnically identified the population of a region is, the more likely that it will push for decentralization. As for emerging and low-income countries, pressure for decentralization has come from various sources, including as a reaction to extended periods of centralized rule (Indonesia, Peru) and ethnically motivated secessionist pressures (Nigeria), and following social strife among different ethnic groups (Kenya).

Evidence is inconclusive for the benefits of fiscal decentralization for enhanced service delivery, economic growth, and alleviating regional and ethnic rivalries.

  • Fiscal decentralization has an ambiguous impact on service delivery: For example, Barankay and Lockwood (2006) find that educational attainment in OECD countries is positively and significantly correlated with the degree of decentralization, and Jimenez and Smith (2005) find that infant mortality in Canadian provinces is negatively correlated with the degree of decentralization. In sub-Saharan Africa, findings in Nigeria suggest that higher fiscal decentralization is associated with lower mortality and higher literacy rates (Akpan 2011). However, studies on convergence of service provision across regions suggest that if substantial equalization grants are not provided, decentralization actually increases regional disparities in service delivery. In addition, the degree of fractionalization seems to adversely affect access to public services in ethnically diverse jurisdictions (for example, Ghana) (Aramov and Asante 2009).

  • The impact of fiscal decentralization on growth is also inconclusive: Empirical studies on this topic have yielded contradictory results. Thiessen (2000, 2003) finds a bell-shaped curve for OECD countries; that is, growth accelerates when countries move from low to medium levels of decentralization, but higher levels of decentralization reduce growth. The type of decentralization (decentralization of spending versus that of revenue) also can affect the impact on growth. Gemmell, Kneller, and Sanz (2013) find that spending decentralization in 23 OECD countries was associated with lower economic growth, whereas revenue decentralization has been associated with higher growth. Rodriguez-Pose and Kroijer (2009) reach a similar conclusion in a study covering 16 central and eastern European countries. Decentralization of spending has also been found to result in lower public investment, in part owing to larger public employment following decentralization, which tends to have an adverse impact on growth (Ahmad and Tanzi 2002).

  • Using decentralization to alleviate ethnic rivalries has not always worked: Local elections may catalyze the expression of divisive demands and exacerbate interregional and interethnic competition for central resources. Such problems have tended to become acute in countries with regionally concentrated reserves of natural resources (Brosio 2000). As for the impact of decentralization on the redistribution of resources, there is some evidence from sub-Saharan Africa that decentralization has increased the share of resources directed to poorer regions. This includes South Africa (regarding expenditure on education) and Ethiopia, where centrally provided transfers have benefited the poorest regions, although the impact on educational outcomes has been uneven (Ahmad and Tanzi 2002).

    Conflicting results of fiscal decentralization reforms reflect the great variety of experiences, as well as the political legitimacy and effectiveness, of governments that have undertaken fiscal decentralization. For example, the effectiveness of decentralization depends on what drives it; that is, on whether decentralization is aimed at increasing democracy or at diffusing local separatist movements. In addition, the purported benefits of decentralization are more likely to be achieved under a devolved system (in which decision making on spending is transferred to subnational governments) than under a deconcentrated system (in which subnational governments have very little autonomy with regard to spending decisions). Also, the type of decentralization varies from giving more powers to large units (for example, states in federal countries) or much smaller ones (for example, local governments in unitary states). At the end of the day, the impact of fiscal decentralization depends largely on the design of intergovernmental fiscal relations rather than the ultimate extent of decentralization (Fedelino and Ter-Minassian 2009, and Fedelino 2010). Experience suggests that gradual decentralization is better than a big-bang approach, but it is important that all the pieces of fiscal decentralization are in the plan from the beginning (Bahl and Bird 2018). The following section reviews international experience to identify the key elements of effective fiscal decentralization.

Chapter 3 Design Elements for Effective Fiscal Decentralization

For decentralization to be successful, it must be implemented as part of a comprehensive framework. A typical mistake that governments make is changing certain aspects of the intergovernmental fiscal framework in isolation; for example, modifying expenditure mandates, introducing new revenue-sharing schemes, or changing the transfer system. If they are not part of a comprehensive framework, such isolated changes may eventually create inconsistencies across government levels, undermining the effectiveness of fiscal policy and increasing macroeconomic risks (Fedelino and Ter-Minassian 2009, and Goerl and Seiferling 2014). In addition, there are various tradeoffs involved in designing decentralization, so it must be tailored to the priority goals and circumstances of each country (Prud’Home 2003). Having said that, experience suggests that the following four elements are essential for a comprehensive decentralization framework that maximizes potential benefits while minimizing risks: (1) clarifying spending responsibilities across levels of government; (2) allowing subnational government to raise own revenues to reduce vertical imbalances and increase fiscal responsibility; (3) designing a transfer system that encourages subnational governments to collect own revenue and manage their functions efficiently; and (4) imposing hard budget constraints on subnational governments to foster fiscal discipline (see Ter-Minassian 1997, and IMF 2009a and 2009b for a detailed discussion).

Assignment of Responsibilities

Few clear-cut rules exist for assigning spending responsibilities across levels of government. For some spending functions, responsibility can be easily assigned to different levels of government; for example, those related to the provision of pure public goods such as defense and foreign affairs to the central government, and local services such as garbage collection and street cleaning to the subnational governments. The provision of social protection (such as public pensions, unemployment insurance, and social assistance) is also typically centralized, and the center has expanded the provision of services with large positive network externalities, such as national transportation and energy transmission grids (Escolano and others 2015). However, many spending functions inevitably overlap and are jointly undertaken by all levels of government.

Lack of clarity in spending assignments contributes to the inefficient provision or underprovision of public services. A typical example is the separation between the assignment of responsibility for maintenance and operation of infrastructure facilities to the subnational governments and the assignment of responsibility for capital investment to the central government. This dichotomy tends to reduce the expenditures for both maintenance and capital infrastructure, because each level of government can blame the other for not doing its part, and each level expects that the other will ultimately replace or renovate and maintain the property. For example, as a result of such separation of spending assignments in Mexico’s education system, neither maintenance nor new investment was adequate, resulting in most physical infrastructure being decrepit or poorly maintained (Mendoza and Vazquez 2000).

Failure to identify appropriate resources for spending responsibilities assigned to subnational government can lead to fiscal pressures. The mismatch of spending responsibilities with the necessary financing has been an issue in many countries. In some cases (for example, Brazil and China) the devolution of resources significantly outpaced that of spending, creating fiscal pressures at the central government level. In others (for example, the transition economies in Eastern and Central Europe in the early 1990s), spending mandates were pushed down to the subnational level without an adequate provision of resources, leading to the accumulation of debt or arrears, or to a significant deterioration in the quality of decentralized public services.

Against this background, fiscal decentralization reform should include legislative clarification of the responsibilities of each level of government and identification of appropriate resources for their financing. While there is no single best way to assign expenditure across the various levels of government, it is essential to have a concrete assignment of these responsibilities between the central and subnational governments. Best practice suggests that they should be specified in the law. Some countries list spending responsibilities in their constitutions, while many others do so in lower level legislation (such as the law on the budgetary system or the law on subnational budget). The latter approach is preferred, because tweaks are needed in the intergovernmental relations from time to time and changing the constitution is much harder than changing a law.

Asymmetric decentralization can be considered in cases of strong cultural or linguistic differences or differences in the capacity of subnational governments to effectively carry out the functions assigned to them.

  • Cultural and linguistic differences: Several countries in Europe (for example, Belgium, Italy, and Spain) have given certain regions more spending responsibilities because of a particularly strong historical, cultural, or linguistic identity that differentiates them from the rest of the populace (Ahmad and Tanzi 2002).

  • Different capacity at the local level: Countries that face a disparity in the capacity of local governments can adopt an asymmetric approach by devolving functions and expenditure assignments first to local governments that have sound institutions and sufficient capacity (for example, Macedonia in the 1990s) or by giving more expenditure autonomy and taxing powers to urban centers (for example, in Bangladesh, Brazil, and India) (Bahl and Bird 2018). To operationalize this approach, the central government can make the devolution of spending responsibilities to individual subnational governments subject to compliance with minimum public finance management (PFM) requirements.

Own Revenues

Giving subnational governments revenue-raising powers may increase their fiscal responsibility. In this case, increased spending would require increased taxation, which could make the subnational governments more accountable to local voters. Own-source revenues for subnational governments may be preferable in a more heterogeneous country, because such revenues (and the associated downward accountability) allow for greater diversity in service provision compared with the standardization (and upward accountability) that comes from heavy reliance on transfers from the central government. Such transfers often have conditions attached that are based on central government preferences, which may be very different from those of a particular subnational government (Bahl and Bird 2008). Revenues also need to be decentralized at the same time as expenditures, ensuring that finance follows function (World Bank 1999).

A “good” local tax has an immobile base and a stable and predictable yield (Box 1). These criteria are generally met by property and personal income taxes, and user fees for services provided locally (for example, tolls on local roads and charges for business, vehicle registration, and use of local assets). Indirect taxes such as the value added tax (VAT) or the corporate income tax, which can be built into the price of goods and passed on to consumers outside the taxing jurisdictions, are not appropriate as local taxes (Escolano and others in IMF 2015).

Criteria for “Good” Local Taxes

Bahl and Bird (2008) suggest several criteria for what typically constitutes a good local tax:

  • The tax base should be relatively immobile, to allow local authorities some leeway to vary rates without losing much of their tax base.

  • The tax yield should be relatively stable and predictable over time.

  • It should not be possible to export much, if any, of the tax burden to nonresidents.

  • The tax base should be visible, to ensure accountability.

  • The tax should be perceived to be reasonably fair by taxpayers.

  • The tax should be relatively easy to administer efficiently and effectively.

The yield from subnational taxes and fees is very limited, especially in low income countries (LICs). The property tax, which is usually the main tax source of local revenue, requires good information systems (such as effective cadaster systems), which are not in place in most LICs, especially outside the main cities. In addition, the yield in LICs from other fees (such as business or car registration fees) is very low owing to a high rate of informality in the economy. When major social expenditure is liberalized (such as health and education), the own-source revenue of subnational governments is usually insufficient (Ahmad and Tanzi, 2002). Indeed, own revenue in sub-Saharan African countries with significant fiscal decentralization is only a small fraction of subnational governments’ overall revenue, ranging from about 5 percent in Uganda to about 30 percent in Ethiopia, much lower than in a typical OECD country or comparable emerging and developing market economies (Figure 3). The picture in other LICs is similar to that in sub-Saharan Africa, with the government finance statistics (GFS) database suggesting that own revenue at the subnational level in low-income and developing countries is, on average, about 20 percent of general government revenue. While GFS data are not consolidated and should thus be interpreted with caution, the share of subnational revenue in South East Europe is broadly similar, amounting to about 16 percent, on average, of total general government revenue (NALAS 2016).

Figure 3.
Figure 3.

Own Revenue of Subnational Government

(Relative to total sub-national revenue, in percent)

Citation: Departmental Papers 2018, 009; 10.5089/9781484358269.087.A001

Sources: Organization for Economic Co-operation and Development (OECD); and IMF staff calculations.1 Includes Bolivia, Colombia, Honduras, Indonesia, Jordan, Paraguay, Peru, Russia, Thailand, Tunisia, Turkey, and Ukraine.

Allowing subnational governments to tax freely can lead to a proliferation of low-yielding taxes with high compliance and administration costs. These “nuisance taxes” can end up undermining the business environment. To avoid this risk, countries can adopt national laws governing subnational revenues (over and above those granted to subnational governments in the constitution) or define a “closed list” of allowable revenue sources. Best practice is for the central government to either determine or at least set guidelines on bases and rates of local taxes.

Natural resource revenues are not suitable for sharing with subnational governments. Allowing producing regions to keep their natural resource revenues tends to contribute to significant fluctuations in the provision of public services in these regions, given the relatively volatile prices for natural resources (especially oil), and to exacerbate regional imbalances. However, political realities might dictate some form of regional taxation or tax sharing on natural resources. In any event, some sharing of natural resource revenue with subnational governments may make sense to compensate them for the costs associated with building infrastructure to enable the extraction of natural resources and for related environmental damage.

Transfers

Transfers to subnational governments from the center are inevitable, even in countries with modest fiscal decentralization. In most countries, including in European fiscal federations over the past 20 years, spending decentralization has outpaced revenue decentralization (Escolano and others 2012). As a result, own revenues are not sufficient to cover subnational spending. This is true even in rich countries, which have a greater capacity to collect revenue at the local level. In the OECD, for example, own revenue covers on average just over half of subnational government spending, requiring significant transfers from the center to fill the fiscal gap. The required amount of transfers in sub-Saharan Africa is much higher, ranging from about 75 percent of subnational spending in Ethiopia to about 95 percent in Uganda.

The main lesson on transfers is that resources should be made available to subnational governments at the same rate as the assignment of spending responsibilities. As noted earlier, countries in which the devolution of resources significantly outpaced that of spending (for example, Brazil and China) faced significant fiscal pressures at the central government level. To avoid such problems, countries should first assign spending responsibilities to the subnational governments and then design an appropriate transfer system.

The key to an appropriate transfer system is to design it so that it combines transparency, simplicity, and equity, and builds incentives for mobilizing revenue and managing functions efficiently at the local level (Ahmad and Searle 2005). If the transfer system is not designed well, decentralization may actually reinforce and perpetuate disparities in the provision of public services rather than reducing them. International experience suggests the following principles for a transfer system:

  • Establish a transparent formula based on indicators outside the control of subnational governments: Best practice is setting earmarked grants on minimum standards and setting equalization grants on objective criteria (for example, population, surface area, and relative income and poverty levels). To avoid manipulation, it is important that indicators included in the formula do not reflect discretionary policy choices by the recipient subnational government.

  • Ensure that transfers remain relatively stable to help subnational governments plan their budgets: Good practice is to set transfers as a fixed proportion of central revenue in the context of a multiyear fiscal framework to enable long-term planning, especially for capital spending. Experience suggests that it is better to base transfers on overall central revenue than on particular taxes, as the central government tends to pay more attention to collecting the revenues that will remain with the center, which ends up distorting the structure of the tax system (Ahmad and Tanzi 2002).

  • Align incentives: Design the transfer system to encourage subnational governments to collect own revenue and manage their functions efficiently. Transfers should fill the gap between expenditure needs and the revenue potential for each subnational government. This requires setting the amount of transfers for each subnational government ex ante during the budget process. The alternative (that is, ex post gap-filling transfers) contributes to weak fiscal discipline (Ahmad and Tanzi 2002).

  • Flexibility in the design of transfers can be useful: Enshrining the main principles of a transfer system in the constitution can help ensure stability in the amount of transfers; however, change is inevitable. Several countries have set the formula for the transfer system in legislation, while providing for a periodic reassessment of the main parameters determining the amount of transfers to subnational governments; for example, the constitutions of South Africa and India require quinquennial assessments by dedicated finance commissions.

Hard Budget Constraints on Subnational Governments

Options to control subnational borrowing range from market discipline to administrative controls (Table 2). Experience indicates that sole reliance on market discipline is often not sufficient, unless several strict conditions are met. These include open capital markets, adequate information, responsiveness of the borrower to market signals, and a strict no-bailout policy by the central government. Few or none of these conditions are met in most LICs, creating the risk of excessive borrowing by subnational governments and exposure to contingent fiscal risks (for example, from public private partnerships (PPPs)).

Table 2.

Options to Control Subnational Borrowing

article image
Source: Fedelino and Ter-Minassian 2009.

Adoption of fiscal responsibility laws or limits on the debt and debt service burden of subnational governments helps foster fiscal discipline. Many advanced economies (including Germany, Korea, Spain, and Switzerland) rely on legally binding rules to restrict borrowing by subnational governments. One approach is to legislate debt thresholds that mimic market discipline. Among sub-Saharan African countries, Kenya has recently done this, setting limits on both debt and debt service (Annex 1). A “golden rule” (that is, allowing borrowing only to finance capital spending) is another fiscal rule aimed at preventing excessive borrowing by subnational governments. As noted earlier, however, fiscal rules may not be very effective in ensuring fiscal discipline if decentralization results in large vertical imbalances at the sub-national level (Kotia and Lledó 2016). In addition, golden rules may create incentives to reclassify current spending as capital spending. Dipping into pension funds and using subnational corporations to borrow on behalf of the subnational government are other ways to bypass debt limits. Insolvency mechanisms for subnational governments can help align incentives and could thus be used to complement borrowing limits (Liu and Waibel 2008).

Enforcement of legislative limits or administrative controls on borrowing and on contingent fiscal risks by subnational governments is essential. In Argentina in the 1990s, constitutional limits on borrowing by provinces existed, but they were not enforced, leading to excessive borrowing by large provinces (for example, Buenos Aires). To avoid such problems, it may be useful to have clear penalties for exceeding legislative debt limits. Some countries impose administrative sanctions on officials who violate the rules (such as withholding salaries or subjecting the officials to criminal penalties or fines), while other countries impose sanctions on subnational governments that miss fiscal targets (for example, Germany suspends consolidation payments to states that miss their fiscal targets). Experience suggests that while such sanctions have a disciplinary effect if they are strictly enforced by the central government, they are not a substitute for a properly designed system of intergovernmental fiscal relations (see Eyraud and Sirera in IMF 2015). Establishing transparent mechanisms for enforcing bankruptcies of subnational governments (as is the case in South Africa) helps mitigate the moral hazard risk (see Annex 3).

A strong accountability framework will support the effectiveness of fiscal decentralization to subnational governments and help enforce fiscal rules. Accountability is a prerequisite for good public sector performance, and information is the key to accountability. Regular collection, analysis, and reporting of information on fiscal operations are critical elements of fiscal decentralization reforms, because the information can be used to verify compliance with policy goals and to strengthen citizen oversight.

Chapter 4 Fiscal Decentralization in Sub-Saharan Africa

Trends

Fiscal decentralization in sub-Saharan Africa is much less common than in advanced and emerging market economies. Most sub-Saharan African countries remain centralized, with very limited spending autonomy for subnational governments. A limited number of sub-Saharan African countries have embarked on significant fiscal decentralization over the past two decades, but the decentralization of revenue is comparatively much lower than that of expenditure (Figure 4).

Figure 4.
Figure 4.

Spending versus Revenue Decentralization

Citation: Departmental Papers 2018, 009; 10.5089/9781484358269.087.A001

Sources: OECD; and IMF staff calculations.

Determinants

The determinants of decentralization in sub-Saharan Africa are broadly consistent with findings in the literature. Studies generally find that the main determinants of decentralization are population density, country size, share of urban population, income level, degree of ethnic fractionalization, and level of democracy (see, for example, Bodman 2009). An empirical analysis suggests that these results broadly hold for sub-Saharan Africa. The findings from a logit model (summarized in Box 2) suggest that the probability for a sub-Saharan African country to decentralize increases with the initial values of GDP per capita and the degree of ethnic fractionalization, and decreases as the share of population living in urban areas rises. In contrast to most findings in the literature, however, the strength of democracy (as measured by the polity index) does not seem to be an important determinant for decentralization in sub-Saharan Africa. In addition, population density is statistically significant in only one out of four specifications.

Effectiveness

Fiscal decentralization in sub-Saharan Africa seems to be positively correlated with better growth performance in the presence of better institutions but does not seem to be correlated with better income distribution (Table 3). As mentioned earlier, fiscal decentralization is expected to have two major benefits: (1) by promoting the efficiency of government spending, it eventually contributes to higher economic growth, and (2) it facilitates greater economic equality as resources are spread more evenly across the country. An empirical analysis using an ordinary least squares (OLS) approach suggests that decentralization in sub-Saharan Africa is positively correlated with higher growth in the presence of stronger institutions (as the interactive terms of decentralization and the quality of institutions have a statistically significant positive correlation with the GDP per capita). Decentralization in sub-Saharan Africa does not appear to have a statistically significant correlation with better income distribution; however, as expected, higher GDP per capita is positively correlated with lower income inequality, and a higher level of urbanization seems to be correlated with lower inequality and higher growth.

Table 3.

Does Decentralization Improve Income Distribution or Increase Growth?

article image
Source: Authors’ calculations.Note: Robust standard errors in parentheses. Random effect panel regression.*p , .05, **p , .01, ***p , .001

Likely Determinants of Fiscal Decentralization in Sub-Saharan Africa

In theory, it is generally expected that fiscal decentralization is positively correlated with the level of economic development, geographical size, population density, level of urbanization, extent of ethnic fractionalization, and strength of democracy.

To assess the effect of these determinants in sub-Saharan Africa, a logit analysis was carried out, which estimates the probability of an event (in this case, the probability of a country being decentralized) in response to one or more predictor (or independent) variables.

  • The dependent (binary) variable in Table 2.1 below is decentralization, which takes the value 1 if the share of subnational spending is at least 5 percent of total government spending, and 0 otherwise. Of the 39 sub-Saharan African countries included in the analysis, seven countries (Ethiopia, Kenya, Nigeria, Rwanda, South Africa, Tanzania, Uganda) cross the 5 percent threshold.

  • The predictor variables reported in Table 2.1 below include GDP per capita, population density, share of urban population, the polity index (which measures the extent of democracy, with an increase in the index implying a higher level of democracy), and the degree of ethnic fractionalization.1 Geographic size and a dummy for resource-rich countries were not found significant in any of the specifications and are thus not reported.

Table 2.1

Logit Analysis on the Determinants of Decentralization in sub-Saharan Africa

article image
Note: Robust standard errors are reported in parenthesis. Coefficients are marginal effects with explanatory variables set equal to their median values in the sample.*p , .05, ** p, .01, ***p , .001.
1 The degree of ethnic fractionalization is measured as a Herfindahl concentration index of the various ethnic groups in each country. The indicators of ethnic fractionalization reported in the table have strong pairwise correlations (above 0.7) (see Posner 2004).

Accountability Framework

The sub-Saharan African countries that are most fiscally decentralized tend to have better overall government effectiveness indicators (Figure 5). A strong accountability framework is key for supporting fiscal decentralization. It enables central monitoring and evaluation of local performance, which helps enforce the fiscal rules on subnational governments and improves service delivery at the local level. However, the quality of governance is relatively weak in some of the sub-Saharan African countries that are planning to embark on fiscal decentralization reforms (for example, Madagascar and Mali). This situation suggests the need for a gradual approach to the assignment of responsibilities and of the necessary financing, as the capacity of subnational governments and the quality of PFM system strengthens.

Figure 5.
Figure 5.

Government Effectiveness Index1

(Ranges from 0 lowest to 5 highest)

Citation: Departmental Papers 2018, 009; 10.5089/9781484358269.087.A001

Source: World Bank Development Indicators database.1 The most fiscally decentralized sub-Saharan African countries are highlighted in red.

Lessons Learned about Decentralization from Four Case Studies

A review of fiscal decentralization in four sub-Saharan Africa case studies—Kenya, Nigeria, South Africa, and Uganda—suggests that it was driven mainly by efforts to defuse ethnic conflicts. Table 4 summarizes the main aspects of fiscal decentralization in these case studies. Specifically, in Kenya, a major devolution was launched in 2013–14 following deadly postelection violence in 2008. The violence was precipitated by the systemic exclusion of certain ethnic groups, which had resulted in glaring regional disparities (Annex 1). In Nigeria, fiscal decentralization began in 1967 with the creation of a federal state, in an effort to diffuse strong regional and tribal rivalries reflecting the country’s significant ethnic and cultural diversity (Annex 2). In South Africa, fiscal decentralization was introduced soon after the collapse of the apartheid system in 1994 to address a legacy of severe economic disparity across social groups and regions (Annex 3). In Uganda, fiscal decentralization was initiated following the end of the civil war in 1986 as part of a broader strategy to reduce ethnic tensions (Annex 4).

Table 4.

Main Aspects of Fiscal Decentralization in the Four Case Studies1

article image

Summary of Annexes 1–4.

The rapid pace of fiscal decentralization in Kenya and South Africa created challenges for containing spending and maintaining service delivery. The pace of fiscal decentralization in Kenya and South Africa was very fast, owing to strong political pressures, with the subnational authorities receiving the authority to spend significant amounts over a very short period. This created challenges in both countries for controlling spending and ensuring service delivery, as many of the subnational governments lacked the capacity for budgeting or for monitoring and reporting fiscal developments. As a result, national governments in both countries were forced to intervene to contain spending, including freezing public sector employment. Nevertheless, in

Kenya recurrent spending has risen by about 2 percent of GDP compared with the pre-devolution period, putting pressure on fiscal balances.

Initial problems have been addressed over time. In South Africa, for example, a monthly reporting system by subnational governments was created and, in the late 1990s, the national treasury started helping provinces prepare realistic budgets. The implementation of a multiyear budget framework from 1998 onward also helped provinces prepare more realistic budgets, and benchmarking of spending helped them identify cost drivers. In Kenya, the initial challenges in service delivery, especially in the health sector, have largely been addressed, although difficulties in preparing realistic budgets at the county level persist, requiring frequent expenditure reallocations during the course of the year.

The division of powers and functions at the various levels of government are clearly defined in legislation and broadly consistent with best practice (see Table 3). In all four countries the functions assigned to the national level relate mainly to policy, standard setting, and the provision of public goods such as national security and macroeconomic policy. Subnational governments are responsible for policy implementation and local service delivery, such as health, water, local roads and transportation, most agriculture extension services, and primary education (except for Kenya where the subnational governments provide only preschool education.)

The authority of subnational governments to make independent spending decisions varies. The subnational governments in Kenya, Nigeria, and South Africa have broad authority to make independent decisions for spending financed by own revenue and unconditional grants, which comprise most of their overall funding sources. This reflects the fact that decentralization in these three countries has taken the form of devolution, with decision making on spending at the local level transferred to elected subnational governments. In Uganda, subnational governments have little flexibility to make significant resource allocation decisions, with the majority of the transfers earmarked for the provision of specific services determined by the central government (Ahmad, Brosio, and Gonzalez 2006). This reflects Uganda’s deconcentrated rather than devolved system of government (as noted earlier, subnational governments in a deconcentrated system are upwardly accountable to the central government and have little authority on spending decisions).

The assignment of taxation powers is broadly in line with best practice. Except for Nigeria (where the personal income tax is collected by the states), the national government maintains full control over the major taxes, such as personal and corporate income taxes, VAT, excises, and custom duties. As for the subnational governments, the bulk of their revenue in Kenya, South Africa, and Uganda comes from property taxes and fees for local services. Own revenues are a very small fraction of subnational spending in all four countries: about 5 percent in Uganda, 10 percent in Kenya, 15 percent in Nigeria, and 20 percent in South Africa. This reflects the weak cadaster systems and high level of informality in these countries. Only in the richest areas does the property tax finance the provision of local services, which is usually the responsibility of municipalities and other small-area jurisdictions (for example, the rich urban areas in South Africa) (Ahmad and Tanzi 2002).

The design of the transfer system to subnational governments varies greatly.

  • As noted earlier, most transfers in Kenya, Nigeria, and South Africa are unconditional, with subnational governments having broad spending autonomy. In all three countries, the redistribution element in the formula is quite strong (taking into account primarily population and income distribution). In Kenya and South Africa the formula is reviewed every five years to reflect changes in the main parameters.

  • In Uganda, on the other hand, the transfers to subnational governments are largely conditional; that is, earmarked by the center for the provision of specific services. In addition, the formula used to determine grants for each subnational government is quite complicated (there is a separate formula for each of some 38 different conditional grants), which undermines the transparency of the system.

The case study countries have kept subnational spending aligned with available resources. All four countries have strict fiscal rules aimed at preventing excessive borrowing by subnational governments. Specifically, all four countries limit long-term borrowing by subnational governments to financing development projects (golden rule) and short-term borrowing to liquidity management. In addition, any long-term borrowing in Kenya, Nigeria (for external debt only), and Uganda is subject to approval by the national government (requires a guarantee in the case of Kenya and Nigeria), and there is an overall cap on the stock of debt for subnational governments in these three countries (20 percent of the previous year’s audited revenue in Kenya, 50 percent in Nigeria, and 25 percent in Uganda). In South Africa, legislation prohibits the national government from guaranteeing subnational government debt. To address moral hazard and help ensure that the strict no-bailout commitment is observed, a transparent mechanism for public bankruptcies is in place, complemented by tough sanctions if subnational governments ignore good fiscal practices.

In Nigeria, the sharing of oil proceeds with the oil-producing states has exposed those states to large swings in revenue and has undermined the regional income equalization objective. For example, Ahmad and Singh (2003) find that a 10 percent fall in the oil price caused a 20 percent decline in federal oil revenues but a reduction by more than a third in oil-producing states. This situation has hampered a sound budget process in the oil-producing states and exposed them to large oil price declines. Indeed, after the latest oil price shock, many state and local governments ran large salary arrears in 2015. This forced the federal government to provide a partial bailout to 23 states and to facilitate the restructuring (longer maturities and lower rates) of commercial bank loans (IMF Country Report No. 16/101).

South Africa’s experience shows that a country cannot enjoy the benefits of decentralization without simultaneously reforming its budget process. From the beginning of fiscal decentralization in the mid-1990s, the authorities in South Africa have simultaneously addressed the problems of capacity, information, and financial management, while ensuring that the political system is responsive at the subnational level. To spread best practice, the authorities have showcased the subnational governments that have demonstrated their capacity to perform through a peer learning and mentorship approach, combined with benchmarking to identify and address the main cost drivers. Improving budget formats, introducing a new GFS classification system, improving the accounting standards, and reforming the chart of accounts have been essential elements for reforming the financial management system and improving the collection of information used for benchmarking.

The fiscal decentralization reform in Kenya underscores the importance of implementing devolution gradually in line with implementation capacity at the subnational government level (see Annex 1). Kenya’s recent devolution involved transferring substantial powers and resources (to about 5 percent of GDP beginning in 2013/14 from less than 1 percent previously) to entirely new units of subnational government (47 counties). In line with best practice, this was done as part of a comprehensive framework based on a new constitution and a new PFM framework. However, the big-bang rollout (in one year rather than over three years as originally envisaged) created challenges in service delivery, especially in the health sector. In addition, as discussed earlier, recurrent spending in the post-devolution period has increased considerably (by about 2 percent of GDP), in part reflecting a duplication of functions between the central and county governments. At the same time, Kenya has kept borrowing by counties under control with the adoption of prudent ceilings on their borrowing and debt service (respectively, 20 percent and 15 percent of the previous year’s audited revenue). Excessive borrowing by subnational governments has been one of the pitfalls that has befallen countries that embarked on similar big-bang approaches; for example, Latin America during the 1980s and 1990s (Giugale and others in World Bank 2000).

Chapter 5 Conclusions

Most sub-Saharan African countries retain a highly centralized model of territorial government and fiscal arrangements. Only a handful have embarked on meaningful fiscal decentralization, and even in these countries, subnational spending is well below the levels in a typical OECD country, especially as a share of GDP.

The determinants of fiscal decentralization in the few sub-Saharan African countries where it has taken place are consistent with experience elsewhere. Empirical results presented in this paper suggest that the decision to decentralize increases with the initial values of GDP per capita, population density, and the degree of ethnic fractionalization. The last factor is confirmed by the four case studies—Kenya, Nigeria, South Africa, and Uganda—where fiscal decentralization has been driven mainly by efforts to reduce regional and ethnic tensions.

Fiscal decentralization is correlated with higher growth in the presence of better institutions, but it does not seem to affect income distribution. The combination of decentralization with a better quality of institutions has a statistically significant correlation with GDP per capita. However, decentralization does not appear to have had the expected impact on reducing income inequality in sub-Saharan Africa.

Among the few sub-Saharan African countries that have embarked on significant decentralization, the macroeconomic risks have largely been contained, although in some cases political imperatives have caused the speed of decentralization to be faster than warranted by capacity implementation at the subnational level.

  • Strategy: The recent fiscal decentralization in Kenya underscores the importance of implementing such reforms as part of a comprehensive framework, encompassing spending assignments, own revenue, the transfer system, and subnational borrowing.

  • Spending assignments and fiscal autonomy of subnational governments: The division of the spending function across levels of government in the four case studies is broadly consistent with best practice, although there are differences in spending powers. For example, in Uganda subnational governments have little flexibility to make spending decisions as a result of an effectively deconcentrated rather than a devolved system of government; in the other three countries, the subnational governments have broad spending autonomy, reflecting a political rather than merely administrative decentralization.

  • Own revenue: The assignment of taxing powers in the four case studies is also broadly in line with best practice, with the bulk of subnational revenue coming from property taxes and fees for local services. However, own revenues are a very small fraction of subnational spending, reflecting weak cadaster systems and a high level of economic informality. In Nigeria, the sharing of oil revenue with oil-producing states has undermined the regional income equalization objective and exposed the oil-producing states to large swings in revenue.

  • Transfers: In three of the four case studies (Kenya, Nigeria, South Africa) most transfers are unconditional and based on transparent formulas with strong redistributive elements. In these countries, the subnational governments have broad spending autonomy. But in Uganda, the transfers are largely conditional; that is, earmarked by the center for the provision of specific services, leaving little autonomy to the subnational governments.

  • Hard budget constraints: Governments in the four case studies have succeeded in keeping borrowing by subnational governments under control. This reflects strict fiscal rules aimed at preventing excessive borrowing by subnational governments, including ceilings on debt stock (Kenya, Nigeria, Uganda) or a transparent mechanism for public bankruptcies complemented by tough sanctions if subnational governments ignore good fiscal practices (South Africa).

Against this background and in light of the generally weak PFM systems in sub-Saharan Africa, countries that decide to proceed with fiscal decentralization should do so in phases. The big-bang approach to decentralization has in some cases resulted in excessive accumulation of debt at the subnational level; for example, the abrupt decentralization in a number of Latin American countries in the 1980s and 1990s. While the rapid pace of fiscal decentralization in some sub-Saharan African countries (for example, Kenya and South Africa) has not been accompanied by excessive borrowing at the subnational level, it has nonetheless created challenges for controlling current spending (creating fiscal pressure at the center) and ensuring service delivery, as subnational governments lacked the capacity for budgeting or for monitoring and reporting fiscal developments. In light of the relatively weak quality of governance in some of the sub-Saharan African countries that are planning to embark on fiscal decentralization reforms, a gradual approach would seem appropriate and consistent with the pace of strengthening the PFM systems at the subnational level.

Lessons for Effective Fiscal Decentralization in Sub-Saharan Africa
Author: Mr. Niko A Hobdari, Vina Nguyen, Mr. Salvatore Dell'Erba, and Mr. Edgardo Ruggiero