A. Introduction

6. In December 2001, Comorians overwhelmingly voted in favor of a new constitution. Its principal feature was a radical devolution of authority from the central government to island governments led by their own elected presidents. With this vote, the country moved from a unitary state, with local administrations overseen by island governors appointed by the central government, to a federal structure. At the same time, however, the name of the country was changed from “Federal Islamic Republic of Comoros” to “Union of Comoros.” This apparent paradox can be readily understood by drawing a parallel with the European Union, which, though it has an increasing number of supranational features, remains so far primarily a union of independent states. It is clear from the events leading up to the new constitution’s adoption that framers and citizens alike shared this vision of decentralized government.

7. The three islands that make up the Comoros are home to only 630,000 people, and their per capita income is only around US$500 per year.² The small size and low level of income sharply circumscribe the degree of decentralization that can be considered efficient, at the same time that the country’s pelagic nature creates a degree of natural independence of aspects of public administration on the three islands.

8. There are tensions between the constitutional call for decentralization and farreaching autonomy of the islands on the one hand, and the small size and limited resources of the country, on the other. They are not simple to resolve, and have lain at the heart of much of the political tumult in the country’s fewer than 30 years of independence. Nevertheless, a great deal of progress has been made toward defining a governmental structure that can command a popular consensus. The first serious test of whether these conceptual solutions can be successfully implemented is the agreement on transitional arrangements that was reached in December 2003.³

9. This paper seeks to document the path that the country has followed in defining its approach to intergovernmental relations. It will focus on fiscal relations under the new constitution and draw upon both the existing literature on decentralization and previous Fund technical assistance (many of the issues raised and observations made reiterate advice given by earlier technical assistance missions) to contribute to the internal debate and to establish reference points for the road ahead.

B. Toward New Intergovernmental Fiscal Relations

Implementing the agreement on transitional arrangements

18. As has been the case since the approval of the 2001 constitution, decisions on a significant number of key issues were deferred under the ATAC. Deferral was either to the two committees or to final resolution by the Union Assembly to be installed following the April elections.

19. The Implementation Committee was initially mandated by the ATAC to meet every two weeks but has met somewhat less often. Its principal contribution to the transition process has been threefold: to interpret the ATAC text, to recognize measures taken by the parties—and by the external partners—in accordance with the ATAC, and to encourage, sometimes through the delegation of individual members, required measures to be taken in a timely manner.

20. The operationalization of the ATAC has been principally the task of the Harmonization Committee. It met under the chairmanship of the former Prime Minister and Minister of Finance of Senegal, Mr. A. L. Loum, and focused on the time-consuming task of seeking agreement on the ATAC’s provisions for fiscal management during the transition. Meeting in almost continuous session from January 26 until March 2 (and with the attendance of Fund staff until February 4), the Harmonization Committee submitted agreements for endorsement by the Implementation Committee in seven key areas:

• Promulgation of a revised budget for the first half of 2004 for the Union and island governments (Table II.1). This budget incorporated revised shares for the allocation of revenue based on the centralization of salary administration for the Union, Ngazidja, and Mohéli governments during the transition. The dispute over competencies and revenue was resolved by further adjusting the revised revenue shares agreed for Ngazidja and Mohéli by a fixed percentage of revenues to cover salary payments effected on their behalf by the Union government.

• Establishment of a system of seven bank accounts for shared revenue sources, and daily procedures for directing the flow of revenues to meet the agreement’s provision for an “automatic” transfer of funds: a central account, one for each government, and separate accounts for wages and debt service.

• The establishment of procedures and guidelines for budget execution across governments, to ensure that budget execution conforms to the agreed budgets. This includes the circulation of detailed budgets and monthly cash plans based on the framework below, and commitments not to incur new arrears, to engage new staff, or to raise salaries. Improved monitoring mechanisms were also agreed.

• Both the principles for calculation and the amount of the Agreement’s special allocation for Ngazidja to cover 2003 arrears. Some donor funding has been pledged toward the €1.5 million agreed (against the €2.4 million that had been requested), and the balance is being sought.

• Supervision of the establishment of the unified customs authority, including the harmonization between the Union and Anjouan of customs procedures, which has been completed, and of tariff rates, planned for end-April.

• Delegation to a technical committee of work to strengthen the collection and transparent dissemination of statistics among the parties to the agreement, with a view both to monitoring the transition and to preparing for an eventual staff-monitored program with the Fund. Fund and Bank staff contributed to the initial stages of this work.

• Harmonized staffing levels and salaries for the senior levels of the new island administrations and establishment of terms of reference for a study of the civil service called for by the transition agreement. The study will focus on how to divide staff between union and island governments.

Table II.1.

Comoros: Agreed Budget Framework, January–June 2004 ¹/

(In billions of Comorian francs)

Source: Harmonization Committee.

¹Each budget is a priori balanced, as no domestic financing is envisaged, and foreign assistance channeled through the multidonor trust fund has been excluded from the budget framework.

Table II.1.

Comoros: Agreed Budget Framework, January–June 2004 ¹/

(In billions of Comorian francs)

		Union	Ngazidja	Anjouan	Mohéli	Total
Total revenue		3.6	2.0	2.7	0.6	9.6
	Earmarked for debt service	0.0	0.0	0.0	0.0	0.8
	Shared sources	3.6	1.5	2.4	0.6	8.0
	Dedicated sources	0.0	0.5	0.3	0.03	0.8
Total expenditure		3.6	2.0	2.6	0.6	9.6
	Wages and salaries	2.1	1.3	1.9	0.4	5.7
	Goods, services, and transfers	1.5	0.7	0.7	0.2	3.1
	External debt service	0.8	0.0	0.0	0.0	0.8

Source: Harmonization Committee.

¹Each budget is a priori balanced, as no domestic financing is envisaged, and foreign assistance channeled through the multidonor trust fund has been excluded from the budget framework.

View Table

Table II.1.

Comoros: Agreed Budget Framework, January–June 2004 ¹/

(In billions of Comorian francs)

		Union	Ngazidja	Anjouan	Mohéli	Total
Total revenue		3.6	2.0	2.7	0.6	9.6
	Earmarked for debt service	0.0	0.0	0.0	0.0	0.8
	Shared sources	3.6	1.5	2.4	0.6	8.0
	Dedicated sources	0.0	0.5	0.3	0.03	0.8
Total expenditure		3.6	2.0	2.6	0.6	9.6
	Wages and salaries	2.1	1.3	1.9	0.4	5.7
	Goods, services, and transfers	1.5	0.7	0.7	0.2	3.1
	External debt service	0.8	0.0	0.0	0.0	0.8

Source: Harmonization Committee.

¹Each budget is a priori balanced, as no domestic financing is envisaged, and foreign assistance channeled through the multidonor trust fund has been excluded from the budget framework.

21. Successful implementation of these agreements, while often concerning narrow technical issues, would represent substantial progress. This is particularly so, given the complexity of the issues involved, the high level of mistrust among the various governments, and the lack of confidence expressed by many participants that the transition process would proceed to the envisaged installation of a smoothly functioning national legislature.

C. Key Issues in Decentralization in the Comoros

22. Despite the significant progress achieved on a number of issues, critical questions remain unresolved. While the constitution is clear regarding certain key competencies of the central government, most functions, including the provision of basic services (which represent the dominant share of public expenditure and employment), remain for final resolution. Some progress has been made, in principle, among representatives of the various governments: in education, for example, primary and secondary education was designated as an island competency, whereas higher education was designated a national function (the country’s first university was formally inaugurated in February 2004).

23. It is expected that these tentative agreements, for the most part embodied in draft organic laws to be considered by the Union Assembly once it is in place, will indeed come into effect in time. However, there is likely to remain some uncertainty on the ground for some time to come. In addition, there remain issues on which a common understanding over competencies has not yet been reached. Finally, some of the issues that have been resolved for the transition, such as a unified customs administration between the Union and Anjouan and a unified salary administration between the Union and Grande Comore, are likely to be reopened after the transition is completed.

24. It is, therefore, useful to revisit the basic issues that the country is confronting in the design of its intergovernmental fiscal relations. As decentralization has been pursued by an increasing number of countries both in Africa and elsewhere, a growing literature has developed to analyze and assess the issues involved (see, for example, www.decentralization.org). Indeed, a Fund-sponsored conference on the subject in November 2000 brought together many of the leading specialists in this area. Certain conclusions of relevance to the Comoros emerge quite clearly from the literature.

25. The economics of decentralization dates to the seminal texts on public economics of the 1950s. They emphasize that decentralization can be beneficial by improving the efficiency of public good provision through competition among jurisdictions and through better matching the mix of public goods to the desires of local residents. Tiebout (1956), who argued that jurisdictional competition disciplines governments, wrote that, when firms and citizens can move freely, this should pressure local governments to provide local public goods efficiently. Stigler (1957) made the point that fiscal decentralization brings government closer to the people, and representative government works best when it is closer to the people. Musgrave (1959) argued that local governments should be permitted to tailor their policies to reflect the preferences of their residents. This suggests that the Comorian constitution, which centralizes public goods (like national defense) that cannot be provided differentially at the island level and decentralizes public services which can, begins from a sound conceptual base. However, while the modern literature largely retains these arguments in favor of decentralization (see, for example, Breton (2002)), it adds important nuance to the arguments, most often in the form of cautionary qualifications, as well as looking at the empirical evidence.

26. One question that arises concerns the potential loss of scale efficiencies from decentralization. Tanzi (2002), for example, argues that, if local governments are so much better able to meet the needs of citizens as to outweigh the loss of scale efficiencies in the provision of public services, then maybe they should be separate countries (and that, if not, then decentralization may not provide net benefits). Indeed, Easterley and Kraay (1999) show that, for a sample of 153 countries, poverty and welfare indicators are significantly better in small countries. But their definition of small is less than 1 million people; at the scale of the Comorian islands, the question of lost scale economies remains open, at best. Treisman (2002), for example, might argue against decentralization in the Comoros, as he suggests that the best governments tend to occur in small countries with only limited decentralization.

27. The so-called fiscal illusion problem raised in the recent literature by Ahmed, Hewitt, and Ruggiero (1997) appears highly relevant to the Comoros. Reliance on grants and transfers from the central government to finance sub national government spending creates an incentive for sub national governments to over provide services, inflate expenditures, and engage in perennial negotiations with the central government to attract more grants and transfers.

28. Much of the recent research examines whether decentralization improves public accountability or whether it makes it easier for local elites to effectively control government. While Brosio’s (2002) view of the evidence permits no conclusions either way, papers by von Braun and Grote (2002) and by Gurgur and Shah (2002) conclude that the evidence shows that decentralization is beneficial. They say it supports greater public sector accountability, reduces corruption, and is generally good for the poor. However, von Braun and Grote also note that most of the observed benefits come from political decentralization, while the evidence for fiscal decentralization remains more ambiguous. In the context of the Comoros, with political decentralization already well advanced, this would argue for caution in the design of fiscal mechanisms to support the decentralization. Indeed, de Mello (2000) looks at 30 countries and concludes that coordination failures in intergovernmental fiscal relations are likely to result in a deficit bias in decentralized decision making. Given the tensions between Union and island governments, this is a warning that should be taken seriously in the Comoros.

29. The practical problems that have tended to occur in the implementation of decentralization efforts are looked at by many writers. Ter-Minassian (1997) agrees that decentralizing government operations can improve economic welfare but requires close coordination among levels of government. There are potential efficiency gains, she believes, but they can be easily dissipated in waste. In particular, arrangements that assign most or all taxes to the center are undesirable: own revenue should match expenditure for accountability. While this may not be practical in the case of the Comoros, it does argue for close attention to the revenue-sharing apparatus.

30. Other problems are also flagged in the literature. Unfortunately, as noted by Ebel and Yilmaz (2002), local governments generally don’t have adequate own revenue; local revenues are not responsive enough to changing needs; and local governments lack the legal authority to levy enough taxes to meet their needs. Although these are not immediate issues but medium-term problems, they will nevertheless need to be addressed by the Comoros in the coming years.

31. Ebel and Yilmaz also note that there are large variations in the size and capacity of local governments. Even in the central government, there is little depth in the civil service, with well-trained staff spread thinly around the administration. Indeed, the need to conserve talent is so great that one may find former ministers running departments in their former ministries. To the extent that new administrations are created in the islands, this problem is likely to be even more acute. Attention will need to be focused, in this context, on whether the Mohéli government in particular has the financial and human resources to fulfill its responsibilities, and on how appropriate assistance can be provided to remedy any shortcomings.

32. For Africa, implementation problems have been looked at by Brosio (2002) and by Smoke (2000). They conclude that there are certain prerequisites for successful decentralization, a theme that runs all through the literature. For Smoke, for example, the minimum elements of a decentralized fiscal system include an adequate legislative and institutional enabling environment; assignment of an appropriate set of functions to sub national governments; assignment of an appropriate set of own-resource revenues to sub national governments; establishment of an effective intergovernmental transfer system; and establishment of adequate access to development capital. These are the precisely the questions with which the Comorian authorities are grappling—with mixed success so far—and which will need to be resolved in a satisfactory manner by the incoming Union Assembly.

D. Ensuring Effective and Efficient Economic Management

33. Political instability has been a major factor in the decline in the Comoros’ per capita income from over US$600 in the early 1990s to around US$500 at present. The new constitution with its high degree of political devolution, offers an opportunity for the country to make a new beginning and focus on the economic challenges posed by rising poverty.

34. In order to be able to make the best use of this opportunity, however, it will be important that the country’s scarce resources not be dissipated in the establishment of additional levels of administration with redundant functions. There is no reason why political devolution must be accompanied by the duplication of administrative structures at the island level of functions already performed at the Union level. The civil service in the Comoros is already large, yet it is short of appropriately trained staff.

35. The small size of the country and the even smaller size of the islands imposes restrictions on decentralization. What appears to be needed, to the extent practical, is for island governments to delegate the administration of their policies to a central service not necessarily controlled by the central government, in order to exploit economies of scale, minimize duplication and overlap, and make efficient use of scarce skills.

36. The most obvious domain for the application of this principle is in the area of tax collection. Under the transition, a unified customs administration is collecting customs revenue for the benefit of all, with participation on the Customs Management Board of all three islands and the Union government. This is an approach that could be extended to the entire tax system. It appears to be particularly needed because the administration and collection of taxes earmarked for the islands and those destined to be shared are inextricably intertwined. Indeed, there has been a working agreement between the Tax Department of the Union and that of Ngazidja to work together on the collection of taxes during the transition, whereby each officer will be allowed to pursue his or her duties without regard to where the moneys collected eventually go.

37. The expenditure side of the fiscal administration is also well-suited to collective administration. The Civil Service Administration (fonction publique) is paying salaries to civil servants of the Union, Ngazidja, and Mohéli governments during the transition—mainly through computerized transfers to staff bank accounts—with the information on payments made shared with each government. This is another area in which the compelling logic that has pushed the different entities to work together during the transition should be made a permanent feature of the fiscal system; indeed, it is possibly that economies could be made by integrating Anjouanese staff into the same system.⁶

38. It must be recognized, however, that for such a system to be acceptable to all of the parties, there will need to be changes in macroeconomic policy formulation. It has been not uncommon in the past for the Union government to respond to fiscal pressures by delaying salary payments. It is believed that, in these circumstances, the payments that do get made are made on a discriminatory basis, for example, civil servants on Grande Comore get paid first, and those on the other islands only later.

39. The constitution clearly provides for monetary affairs to be managed at the central level. One issue that arises is the provision for statutory advances to the government from the central bank, ostensibly to smooth intrayear fluctuations in revenue. The limit on these advances is determined as a function of average annual revenue. Over the medium term, the government should return to compliance with the loan condition that stipulates that all such loans be repaid by the end of each year. This measure will be helpful in avoiding new arrears on government payments as a financing item.

40. Finally, it is essential that the line ministries be made to function as efficiently as possible. This will require very clear delineation of competencies. While the issue is to be taken up by the Union Assembly, it is likely that upcoming legislation will not be sufficiently detailed to cover all the questions that might arise. In this event, it will be critical that the various governments work together. The experience over the past few months has demonstrated an enormous capacity to work together when necessary; it is to be hoped that this spirit of cooperation will carry forward, as it must, to the implementation of the new constitution and its new laws.

Box II.1.

Key Dates

August 1997	Anjouan declaration of independence.
December 1997	Addis Ababa International Conference on national reconciliation framework to restore unity (sponsored by the OAU).
April 1999	Antananarivo Agreement on reunification. Anjouan declines to ratify it, prompting violent demonstrations in Moroni.
April 1999	Col. Azali Assoumani takes power in Moroni in a coup d’état, citing a need to restore order.
August 2000	Fomboni I agreement on reunification reached by Comorian and Anjouanese governments.
February 2001	Fomboni II (all-party) agreement signed (see main text); sanctions against Anjouan lifted.
December 2001	New constitution adopted by plebiscite.
April 2002	Col. Azali elected Union President.
May 2002	Island presidents elected.
November 2002	Paris I workshops to prepare drafts of by-laws to the new constitution.
May 2003	Paris II workshops to continue work on draft by-laws.
August 2003	Pretoria Memorandum on Transitional Arrangements is signed by representatives of all four governments (subsequently renounced by the Union President).
October 2003	Paris Declaration by “Partners of Comoros” calls for renewed dialogue and proposes trust fund to support a transition agreement.
December 2003	Moroni Agreement on Transition Arrangements is signed by the Union and island presidents.

View Table

August 1997	Anjouan declaration of independence.
December 1997	Addis Ababa International Conference on national reconciliation framework to restore unity (sponsored by the OAU).
April 1999	Antananarivo Agreement on reunification. Anjouan declines to ratify it, prompting violent demonstrations in Moroni.
April 1999	Col. Azali Assoumani takes power in Moroni in a coup d’état, citing a need to restore order.
August 2000	Fomboni I agreement on reunification reached by Comorian and Anjouanese governments.
February 2001	Fomboni II (all-party) agreement signed (see main text); sanctions against Anjouan lifted.
December 2001	New constitution adopted by plebiscite.
April 2002	Col. Azali elected Union President.
May 2002	Island presidents elected.
November 2002	Paris I workshops to prepare drafts of by-laws to the new constitution.
May 2003	Paris II workshops to continue work on draft by-laws.
August 2003	Pretoria Memorandum on Transitional Arrangements is signed by representatives of all four governments (subsequently renounced by the Union President).
October 2003	Paris Declaration by “Partners of Comoros” calls for renewed dialogue and proposes trust fund to support a transition agreement.
December 2003	Moroni Agreement on Transition Arrangements is signed by the Union and island presidents.