229 Public Sector Reform and Capacity Building August 2003 Findings reports on ongoing operational, economic, and sector work carried out by the World Bank and its member governments in the Africa Region. It is published periodically by the Knowledge and Learning Center on behalf of the Region. The views expressed in Findings are those of the author/s and should not be attributed to the World Bank Group. http://www.worldbank.org/afr/findings Decentralization in Africa: Emerging Trends and Progress Decentralization, defined broadly a detailed overview of decentrali- as the transfer of public authority, zation in Africa based on the as- resources, and personnel from the sessments of World Bank special- national level to sub-national ju- ists working on each country cov- risdictions, has been a recurrent ered. The report catalogs the sta- theme in African countries since tus of decentralization based on a independence. In the last decade number of indicators for each or so decentralization has gained country as reported in response to prominence as an expressed goal a survey administered in 2002. or as an actual programmatic pur- suit in the context of or as a con- Progress on three fronts sequence of two prominent move- ments affecting the African state. Progress on decentralization was One consists of structural adjust- measured by three indices to re- ment programs that sought to re- flect the three defining aspects of form the public sector starting in decentralization: political, adminis- the 1980s while the other is the trative, and fiscal. Assessing decen- ongoing transition toward more tralization at this disaggregated democratic and competitive poli- level helps better clarify the com- tics. In nearly all African coun- ponent structures underpinning tries, structures of local adminis- claims to decentralization as well tration exist but are often subordi- as point to the limits of such nated in their legal creation, man- claims. date, and operation to the central Thus, a political decentralization state, especially the executive. As index was computed from the mean elsewhere in the developing world, of the following: the number of political and economic liberaliza- elected sub-national tiers, the tion have opened possibilities or at score for the existence of direct least revived claims for greater de- elections for local governments, centralization. Major development and the score for turnout and fair- Findings donors have also pushed decen- ness of such elections. Thus a tralization as a pathway to improv- country was likely to score high ing governance and service deliv- (i.e., 4) if it had more levels of sub- ery in developing countries. national government that were This paper summarizes a larger elected rather than appointed, report, Decentralization in Africa: A where local governments in par- Stocktaking Survey, which provides ticular were elected, and where local government elections were this score, two countries indicated tries that scored high (i.e., where adjudged free and fair. The high- a high degree of administrative de- local governments control 5­10 est score attainable was 4, while centralization (South Africa and percent of public expenditures) are the lowest was 0. Of the 30 coun- Uganda) while ten others indicated Nigeria, Uganda, and Zimbabwe. tries analyzed, eight scored high a moderate degree of administra- Countries with moderate levels of (at least a 3 on the index) indicat- tive decentralization (Nigeria, fiscal responsibilities anchored at ing a high degree of political de- Rwanda, Tanzania, Zimbabwe, the local level (i.e., where 3­5 per- centralization. These were South Kenya, Ghana, Ethiopia, Madagas- cent of public expenditures are Africa, Uganda, and Namibia (all car, Burkina Faso and Eritrea). Of controlled by local authorities) in- above 3.0) and Kenya, Ghana, the sixteen countries that showed cluded Kenya, Rwanda, Ghana, Senegal, Ethiopia and Cote d'Ivoire. very low degrees of administrative Senegal, Burundi, and the Congo A second set of countries indicated decentralization, six had practi- Republic, and Congo DRC. a moderate level of political decen- cally made no effort to decentral- These rankings of fiscal decen- tralization, with scores ranging ize administrative systems. tralization should be approached from 2.0 to 2.9; these included Finally, fiscal decentralization was with care. First, the proportion of seven countries (i.e., Nigeria, measured using two indicators: the public expenditures controlled by Rwanda, Madagascar, Zambia, score given for the arrangements the local authorities is very small Tanzania, Mali and Malawi). The for fiscal transfers from the cen- such that even the "high perform- rest of the countries (15) indicated tral government to localities and ers" are marginal in their perfor- very low levels of political decen- the score corresponding to the pro- mance compared to the global tralization, although four of these portion (offered as a range) of pub- trends. Moreover, the proportion of at least held elections in part of the lic expenditure controlled by the public expenditures controlled by territory. localities. The index was simply the local authorities is only one Administrative decentralization the mean of these two scores. dimension of fiscal decentraliza- was tracked using three indicators Thus, a country with an estab- tion, which ought to include rules whose mean produced the index. lished and often-used formula for of fiscal control and revenue col- The index consisted of the score fiscal transfers and in which locally lection--almost always reported in for the clarity of roles for national controlled expenditures account for interviews to be in favor of the cen- and local governments provided by a high proportion of overall public tral state and excessive. the law, the score indicative of expenditures would have scored Taken as a whole, data indicated where the actual responsibility for high on this index (i.e., highest that the degree to which the vari- service delivery resided, and the score 4). A country whose fiscal ous components of decentralization score indicative of where the re- transfers to localities were ad hoc have been achieved varies mark- sponsibility for (hiring and firing) and whose localities accounted for edly such that one component may civil servants resided. Thus a coun- a miniscule proportion of national be rated very high while another try with a very clear legal frame- public expenditures would score quite low. That political decentrali- work demarcating local and cen- low on the fiscal decentralization zation was the most advanced com- tral government roles and institu- measure (i.e., lowest score 0). ponent of decentralization was not tionalizing the principle of The degree of fiscal decentrali- surprising. First, political decen- subsidiarity, with actual service zation across the continent is very tralization is the local manifesta- delivery delegated to and performed low as revealed by the fact that in tion of the democratic reform that by localities, and with civil servants 19 of the 30 countries analyzed lo- swept the continent in the 1990s. responsible to local authorities, cal governments control less than However, although widespread it is would have scored highest on this 5 percent of the national public not necessarily deeply ensconced. index (i.e., 4). Those with no legal expenditure. Only South Africa was Second, administrative decentraliza- framework, and with persistent assessed to have a very high de- tion appears as often and is often centralization in service delivery gree of fiscal decentralization (i.e., rated high because administrative and in civil service oversight (hir- more than 10 percent of public ex- reform was the preferred pathway ing and firing) would score low on penditures are controlled by sub- to decentralization (in particular, this index (i.e., lowest score 0). On national governments). The coun- deconcentration) before democra- Fig. 11. Overall Decentralization Score 3.5 3 2.5 ero 2 Sc exdnI 1.5 1 0.5 0 Ma inea itrea li . lawiFaso bique of eroon Nigeca r e d lic gola undi nin C r DR Rep on Cha SouthAfricaUganda nya ana nda Ke Gh Rwa NigeriNa a mibiaene l ga a we Be , Le S EthiopiTanzaniabab Zim da Er Ma a An Cote gscard'IvoireZambia Gu pub Bu Ma rkin zam ngoCam Afri Bu Mo Re ngo, Co ntral Sierra Co Ce tization. That fiscal decentraliza- ing the lowest level of decentrali- Niger, Sierra Leone, and Chad), but tion lags behind these other vari- zation and 4 the highest level pos- could conceivably include nearly ables is not surprising. Even where sible, only two countries (South Af- half of the fifteen countries for which it seems much further along it be- rica and Uganda) scored in the top complete data was unavailable. lies the small proportions that are range (3.0-4.0). The next level (2.0­ In relation to overall decentrali- used for comparison to the national 2.9) indicates countries with a zation, it is notable that only four public expenditures rather than moderate degree of decentraliza- of the 13 countries in the upper the advancement of fiscal respon- tion. Eleven countries were in this levels (high and moderate) of over- sibilities at the local level. category: Kenya, Ghana, Nigeria, all decentralization are Rwanda, Namibia, Senegal, Ethio- francophone (Senegal, Rwanda, Overall progress pia, Zimbabwe, Tanzania, Côte Cote d'Ivoire and Madagascar). Of d'Ivoire, and Madagascar. The third these, two (Rwanda and Madagas- A composite index was created to group of countries, those with low car) have overhauled their admin- rank countries on their overall levels of decentralization (1.0­1.9), istrative systems substantially and decentralization, taking into ac- had the largest number of coun- can be considered to have had a count structural and performance tries (13): Zambia, Guinea, Mali, major break with the established factors. The index consisted of Eritrea, Burkina Faso, Malawi, francophone patterns. Indeed, ex- twelve distinct indicators used in Republic of Congo, Mozambique, cept for four countries (Sierra the analysis of components of de- Angola, Burundi, Benin, The Leone, Eritrea, Zambia, and centralization. Democratic Republic of Congo, and Malawi), all the countries scoring Overall, the data indicates a Cameroon. The final group, with in the lower half (low and no de- moderate degree of decentraliza- nominal or no decentralization centralization) of the overall index tion in the Africa region for the (range 0­0.9), included four coun- are either francophone (11) or thirty countries for which data was tries for which data was complete lusophone (2). complete and analyzed (Figure 1). (The Central Africa Republic, On a scale of 0 to 4, with 0 indicat- Conclusion While these findings suggest de- promise in the decentralization of centralized authorities in Africa the African state, although there Our findings indicate decentrali- are short on the requisites for ef- are also many challenges that con- zation in Africa is progressing but fective administration and devel- front the process and, beyond it, ser- unevenly both in terms of regional opment response to local commu- vice delivery and poverty reduction. spread and in terms of the aspects nity needs, it is critical to recog- of decentralization that are in- nize that these authorities are in stalled. Overall, the pace and con- the midst of immense change. This tent of decentralization in the re- change is indeed in the right di- gion can best be described as mod- rection--toward deliberate cre- erate and, where it has effectively ation and enhancement of local been installed, in need of deepen- authority structures and in par- ing. Among the thirty countries ticular responsive to the changed analyzed, a little less than half have context of democratic politics. It is high or moderate levels of overall significant to note that in no coun- decentralization, with the least de- try was the claim to centralization This article was written by centralization occurring in as a preferred organizational model Stephen N. Ndegwa, a Young francophone countries. Consider- made or implied, nor was decen- Professional formerly attached to ing most of the countries excluded tralization considered undesirable, the Africa Public Sector Reform from analysis were anecdotally only difficult to effect and sustain. unit. It is based on Decentralization noted to have minimal or no de- Significantly, political decentrali- in Africa: A Stocktaking Survey centralization, we can assuredly zation, which is often the critical (Africa Working Paper Series #40, say that only about a third of Afri- first step toward effective decen- November 2002). Please direct can countries have effective decen- tralization, has only recently taken comments or requests for the full tralized structures of governance. root. There is therefore legitimate report to sndegwa@worldbank.org. Findings Findings can be accessed via the World Bank Group's website at http://www.worldbank.org/ Click on Publications, then Periodicals. Or, Findings would also be of interest to: click on Countries and Regions, then Africa. Name Institution Address Letters, comments, and requests for publications not available at the World Bank Bookstore should be addressed to: Editor, Findings Operational Quality and Knowledge Services Africa Region, The World Bank 1818 H Street NW, Mailstop J-8-811 Washington, D.C. 20433 e-mail: pmohan@worldbank.org