Somalia - Crisis in public expenditure management (Vol. 1 of 3) : Summary of main findings and recommendations

Report No. 8727-SO Somalia Crisis in Public Expenditure Management (in Ihree Volunies) VolIme 1: Summary ot Main Findings and Recommendations March 8, 1991 ( )Lntr' (Operation L Division Lcstehrnt Aria d Department .\Iri( d Region FOR OFFICIAL USE ONLY Document of the World Bank This document has a restricted distribution and may be used by recipients only in the performance of their official duties. Its contents may not otherwise be disclosed without World Bank authorization.  Fiscal Year January I - Dxember 31 Abbreviations and Acronyms ADC - Agricultural Development Corporation ADP - Annual Development Plan CSBS - Commercial and Savings Bank of Somalia CBPP - Contagious Bovine Pleuropneumonia AG - Department of Accountant General DDB - Domestic Development Budget DDD - Domestic Development Department, Ministry of Finance EPI - Expanded Programme of Immunization FYDP - Five-Year Development Plan GSP - Generating Shilling Process LMs - Line Ministries LMH - Livestock Marketing and Health MCH - Maternal and Child Care MOL - Ministry of Labor, Sports and Social Affairs MOA - Ministry of Agriculture MOF - Ministry of Finance and Revenue MFMR - Ministry of Fisheries and Marine Resources MILFR - Ministry of Livestock, Forestry and Range MFA - Ministry of Foreign Affairs MOHEC - Ministry of Higher Education and Culture MOH - Ministry of Health NRA - National Rangeland Agency NAHA - Nomadic Animal Health Assistant NGO's - Non-Governmental Organizations ODB - Ordinary Budget Department, Ministry of Finance PD - Planning Department, Ministry of National Planning MPU - Planning Unit, line ministries PER - Public Expenditure Review PIP - Public Investment Program PHC - Primary Health Care MPU - Planning Unit, line ministries SPA - Special Program of Assistance for Sub-Saharan Africa SHIFCO - Somali High Seas Fishing Company TA - Technical Assistance WFP - World Food Program WHO - World Health Organization  FOR OFFICIAL USE ONLY Foreword This relport is based on the findings of two public expenditure missions which visited Somalia in May/JLne and in November 1989, led by Baran Tuncer, Principal Economist. Baran Tuncer was the Task Manager up to yellow cover and the Principal Author, with the assistance of Gianni Zanini (subsequent Task Manager), and Fayez Omar. The areas of responsibility were: Institutional Aspects Baran Tuncer, Ercis Kurtulus (Consultant) Government Finances Fayez Omar, Gianni Zanini Macroeconomic Framework Gianni Zanini, Fayez Omar Public Investment Program and Planning Young Kimaro Agricultural Sector Mohamed Usman Health and Education Nicholas Burnett, Swadesh Bose Statistical Appendix Aeran Lee, Gianni Zanini Important contributions were received from Bruce Jones (Government Finances) and Peter Gregory (Consultant, Pay and Employment Issues in the Civil Service). Jack van Holst Pellekaan and Harry Walters have provided useful comments on the agricultural sector. Yaw Ansu contributed to the writing of the report at the yellow cover stage. Caroline Milad was responsible for the typing and layout of the report, with assistance from Afsar Nokhostin. Since the time that the green cover draft of this report was completed in August 1990, there has been a serious deterioration in the security and economic situation in Somalia. As a result, the macroeconomic projections presented in Chapters 3 and 4 are unlikely to be realized in the time frame envisaged (1990-93). Therefore, they should be considered only indicative of a possible medium-term scenario after political reconciliation and stability have been achieved. The analysis and data on historical patterns and institutional issues presented in chapters 2, 5, 6, and 7 remain fully relevant. This document has a restricted distribution and may be used by recipients only in the performance of their official duties Its contents may not otherwise be disclosed without World Bank authorization.  Summary of Main Findings and Recommendations A. Introduction 1. Somalia is among the poorest countries in the world. Per capita GDP in 1989 was about US$170. The country is large and sparsely populated by about 6 million people (9.4 million according to the Government, based on preliminary analysis of census returns). Agriculture contributes about 62 percent to GDP in an average year, with the dominant share of production coming from nomadic pastoralism. Indeed, about 50 percent of the population are nomads. The climate, which is arid to semi-arid, is harsh; moderate droughts occur every 3 to 4 years and major ones every 8 to 10 years, inflicting heavy losses in agricultural output, and hence in GDP. Somalia's low level of economic and social development is reflected in high infant mortality, low average life expectancy, and a very low literacy level. 2. Because of the poverty of the country, and the severe constraints to economic development posed by its harsh physical environment, Somalia has attracted a great deal of development assistance. On average, concessional development assistance to the country has beer about 198 percent of exports, and 24 percent of GDP since the mid-1970s. At US$102 per capita (1987), Somalia is a leading recipient of Official Development Assistance. Foreign assistance is channelled predominantly through the Government, and throughout the 1980s around 62 percent of all public expenditures in Somalia, and at least 90 of public development expenditures, were financed by donors. Thanks to donors, investment in the country during the 1980s has averaged about 26 percent of GDP. 3. However, economic performance has not matched the generous level of foreign development assistance. Annual real GDP growth rates during most of the 1980s were little above the population growth rate, and towards the end of the decade they fell considerably, leading to negative per capita growth. For the decade as a whole, per capita GDP was essentially stagnant, physical infrastructure as well as the quality of government services deteriorated, and macroeconomic imbalances became more pronounced. 4. The poor economic performance was in part due to adverse exogenous shocks, such as periodic droughts and the ban placed on Somali cattle in 1983 by Saudi Arabia, which had been the main export market. The increasing level of armed conflicts within the country in recent years has also had its impact on economic performance. 5. The problems with government expenditures, however, go beyond external shocks and the unfortunate fact that the civil strife has led to large amounts of resources being devoted to defense. There are serious problems with the Government's management of its expenditures in terms of its size relative  2 S&mmary of Main Findings and RecommendatiOns to domestic revenues, approprihteness of the allocation, and the efficiency of the institutions and processes by which expenditures are planned, budgeted, allocated, monitored, and accounted for. 6. The Government has become aware of some of the deficiencies, and has repeatedly expressed its desire to address them. In particular, in 1988, within the context of the Agricultural Sector Adjustment Program (ASAP II) supported by the World Bank, and the Policy Framework Paper agreed to with the World Bank and the IMF, the Government gave its commitment to public expenditure restructuring. The purpose of this report is to help translate the Government's desire into action by analyzing government expenditures in Somalia, identifying the deficiencies that exist, and developing a reform agenda for correcting them. Medium-term macroeconomic projections for Somalia, assuming that the recommended actions on expenditure management are implemented together with other policy reforms within the context of a broader economic adjustment program, are also provided. 7. Due to staff and time constraints as well as the severe difficulty imposed by lack of data, it has not been possible to analyze all aspects of government expenditures in Somalia. In particular, a detailed review of public enterprises is not included in this report. Similarly, sector-specific analysis is limited to agriculture, education, and health. These limitations should certainly be corrected through future work. Nevertheless, as it stands, the report is the first extensive and systematic review of government expenditures in Somalia, and through its analysis of macroeconomic and cross-sectoral issues it provides recommendations that would help improve expenditure man3gement even in the sectors not directly analyzed. This summary, Volume I, highlights the main findings and recommendations arising out of the work of the Public Expenditure Review (PER) missions that visited Somalia in June and November 1989, and the analyses conducted subsequently at headquarters in Washington. Detailed analyses are contained in Volume II of the report. B. Problems of Government Expenditure Management 8. Size of Government Expenditures Government expenditures in Somalia between 1981 and 1989 have averaged around 27 percent of GDP a year, with the average over the last five years a little higher, at 32 percent.12 In contrast, the ratio of the Government's domestic revenue to GDP has averaged about 7 percent over the 1980s (6 percent in the last five years). Comparisons with low-income Sub-Saharan African countries, where the average expenditure and revenue ratios over the 1980s have been 30 and 18 percent respectively strongly suggest that while the level of government expenditures in Somalia are comparable to the norms in the Region, revenues are dismally low. 9. Due to the high expenditures and low revenues, budget deficits before grants have averaged around 20 percent of GDP a year. On average, however, domestic financing of the deficit amounted to only about I percent of GDP, or about 6 percent of the deficit. Hence, although the rate of inflation has been high, averaging about 60 percent over the 1980s, domestic bank financing of the deficit has generally not been a primary contributor, except in 1984 and 1987. In the first half of the 1980s, most of the foreign financing was in the form of loans, but in the second half it was largely in the form of grants. The heavy Tbese ratios include Governrent expenditures not reported in the budget, and estimated to be on average about 4 pereent of GDP. This ratio is very much affected by the impact of exchange rate changes on a significant portion of government expenditures, particularly the Public Investment Program (PIP) financed by donors. Excluding the foreign contribution to the PIP, government expenditures have averaged about 13 percent of GDP since 1981.  Summary of Main Findings and Recommendations 3 reliance on financing in either form has not been without serious costs to Somalia. The foreign loans relied on for financing in the earlier years have led now to an external arrears problem, since the expenditures for which they were incurred were not productive enough to provide the resources for repayment. Thle result has been that Somalia's access to additional financial flows has been severely curtailed, with adverse consequences for the economy. Even foreign grants have had their costs. With such massive underwriting of the Government's budget, donors have become increasingly involved directly in originating and executing projects. While this direct involvement may enhance the success of particular donor projects, it does raise serious issues. First, is the issue of the lack of budgetary control over a significant share of public expenditures. Second, is the incentive and morale problems in the civil service arising from the fact that staff who manage to be associated with donor projects benefit from extra allowances paid by donors, which generally are several times larger than the very low basic salaries. Third, of course, is the issue of sustainability of projects once donors hand them over. These issues will be elaborated on in the course of the following discussion. 10. Budget Structure and the Lack of Transparency Government expenditures in Somalia are classified under three sub-accounts: the Ordinary Budget, the Domestic Development Budget, and the Public Investment Program (PIP). The Ordinary Budget contains expenditures on mostly recurrent-type activities. In principle, expenditures in this budget, with the exception of foreign interest payments, are to be financed by revenues raised domestically. Foreign interest payments are supposed to be partly financed by counterpart funds in local currency generated by the sale of foreign exchange or imported commodities provided under various donor programs. In reality, however, a sizable amount of expenditures in the Ordinary Budget, other than foreign interest payments, is also financed by counterpart funds. The Domestic Development Budget contains expenditures on projects that are undertaken by the Government without the participation of donors, and also the government's contribution to the PIP, which is generally less than 10 percent of the total PIP expenditures. However, this is financed entirely from counterpart funds of donor import support. The PIP contains projects with donor participation, and almost all of its financing, except for the small government contribution mentioned above, is directly from donors. 11. Since the Government, to a great extent, controls the allocation of counterpart funds, the Ordinary and the Domestic Development Budgets reflect government-controlled expenditures, while the PIP is for all purposes externally controlled. In fact, as perceived by the Somali authorities, budgeting of public expenditures refers only to expenditures under the Ordinary and Domestic Development Budgets. Hence there is no systematic and integrated accounting anywhere in the Government of annual expenditure flows under the PIP, which over the 1980s have averaged about 44 percent of total expenditures. To make matters worse, there are expenditures from a number of sources, including counterpart funds, donors and foreign Non-Governmental Organizations (NGOs), or suppliers' credits that are not accounted for in either the PIP or the two government budgets. The lack of any systematic and integrated knowledge in the Government of expenditures under the PIP is only one of several serious institutional problems regarding foreign-financed development expenditures that are reviewed below, following the discussion of the problems of government-controlled budgets. 12. The structure of the budget in Somalia renders it essentially useless as a tool for analyzing government expenditures, evaluating their effectiveness, and for planning future expenditures. Recurrent expenditures, expenditures for fixed investment, and financial transfers are all included in both the Ordinary and Domestic Development Budgets. The PIP also includes recurrent as well as fixed investment expenditures. Since there is no clear way of identifying the shares of recurrent, investment, and transfer expenditures in the various budgets, it is very difficult for the Government to readily  4 Summary of Main Findings and Recommendations ascertain, even at the aggregate level, how much money it is spending under any of the broad economic categories. 13. Further, in the Ordinary Budget allocations and expenditures within LMs are not broken down according to functions or programs; only line items such as salaries, material purchases, and travel, which give no indication of the purposes of the expenditures, are given. The situation is even worse for allocations through the Domestic Development Budget. Here, only a line item for the total allocation to be made in the budget year to each project is given; and while the allocation for the previous year is also shown, neither the actual expenditures in the previous year nor total expendit,.res to-date on the project is shown. 14. The two government-controlled budgets (i.e the Ordinary and Domestic Development Budgets) are not integrated to provide a comprehensive picture of government expenditures, and any attempt to do so is hindered by several problems. First, both budgets contain recurrent, capital, and transfer expenditures, but neither budget provides a sufficiently detailed classification for the various types of expenditures to be identified. In particular, large chunks of expenditures are completely unclassified; they are simply labelled "other". For example, about 32 percent of the Ordinary Budget falls in the "other" category. Second, the broad classifications given in the different budgets do not easily match with each other. Thbird, there are frequent changes in classifications from year to year. This lack of transparency creates the serious problem that at no time do the authorities in Somalia have a clear idea of how total budgetary expenditures have been allocated, and hence there is very little basis for evaluating past and current performance, and for planning future expenditures to meet the Government's objectives. 15. Alocation of Expenditures under the Government-ControUed Budgets Despite the problems mentioned above, it has been possible to piece together the outlines of the pattern of government expenditure allocations in Somalia, and the picture that emerges is very disturbing. In sum, recurrent expenditures, both as a share of total expenditures and in real terms, have been falling. Within recurrent expenditures, the share going to economic and social services has fallen drastically, while that going to general services--mostly defense and security-related expenditures--has risen. With the share of general services expenditures increasing within an overall allocation for recurrent expenditures that has been falling in real terms, expenditures for wages and salaries, and for operation and maintenance in the civilian sectors of Government have fallen to levels that are totally inadequate to allow these sectors to carry out their functions with any semblance of effectiveness. The dimensions and effects of this problem are illustrated below. 16. Recurrent versus Development Expenditures The Ordinary Budget, which in theory constitutes recurrent expenditures, decreased from a high of 21 percent of GDP in 1981 to 12 percent in 1988. In real terms, it decreased by 7 percent a year on average over the period. In contrast, expenditures from the Domestic Development Budget rose by 19 percent in real terms, and those from the PIP rose by 85 percent in real dollar terms. This basic picture is not significantly altered even after making allowances for the fact that each of the three budgets include elements of recurrent-type as well as capital or investment expenditures. 17. Sedorni Allocations of Government-ControUed Expenditures Over the 1980s, there has been a drastic shift of expenditure allocations away from economic and social services to general services, which are dominated by defense and security expenditures. Spending on economic services declined from 24 percent of government-controlled expenditures (i.e., Ordinary and Domestic Development Budgets) in 1974, to 15 percent in 1981, and to 11 percent in 1988. Over the same periods, expenditures on social  Summary of Main Findings and Recommendations 5 services declined from 16 percent to 12 percent, and then to 4 percent. More significantly, within the Ordinary Budget the combined share of the economic and social services fell from 36 percent in 1974 to 6 percent in 1988. (See tables 2.1, 2.7). 18. For all practical purposes in recent years, the ordinary budget finances only defense and general services. A detailed exercise, under which expenditures labelled as "other" in the 1987 budget were properly re-classified (see table 2.6), suggests that 49 percent of government-controlled expenditures (and 30 percent of total expenditures including the PIP) in 1987 went to defense. The combination of an increasing share of defense and other general services expenditures and shrinking overall recurrent expenditures has meant a severe compression in expenditures on wages and salaries as well as on operations and maintenance in the civilian sectors of Government. 19. Wages and Salaries The civil service has grown from around 20,000 employees in 1974 to ever 45,000 in 1989, representing an average annual growth of around 6 percent. During the same period, the total civil service wage bill fell in real terms, to the extent that the average real wage in 1989 was only 6 percent of the level in 1974. Budgetary expenditures on wages and salaries have also fallen both as a share of GDP and as a share of the budget. Whereas in 1974, wages and salaries were 22 percent of the government-controlled budget, or 5 percent of GDP, by 1989 they had fallen to as low as 6 percent of the government-controlled budget or 0.6 percent of GDP. By comparison, budgetary wages and salaries in 1988 were 5 percent of GDP in Malawi and 9 percent in Kenya. 20. With the average wage as low as US$3 per month in 1989, administrators tolerate excessive absenteeism to permit the staff to pursue gainful employment outside the civil service. This, coupled with the fact that many of the highly skilled and capable personnel have long been drawn away by alternative employment opportunities--most commonly in connection with donor projects that sometimes yield pay levels as high as 30 times those of the public sector--has resulted in extremely poor quality of government services funded under the Ordinary Budget. 21. To compensate for deficient pay levels, civil service salaries are supplemented informally by provisions from projects in the Domestic Development Budget and in the PIP. These supplementn have added up to about 0.7 percent of GDP in recent years. Hence, in 1989 the supplements were in all likelihood much larger than the expenditures specifically allocated in the budget for wages and salaries. This practice, while providing badly needed salary supplements to Government employees, has, however, resulted in some undesirable consequences. Both the compensation levels and the coverage vary substantially. In some departments, wage supplements are given to many or all employees regardless of their contribution to the implementation of projects, thus exacerbating the salary range compression in favor of unskilled labor. Time and energy that go into the pursuit of Pew projects by government departments in order to benefit from salary supplements deriving from them have greatly compromised the provision of basic services. Furthermore, projects find it difficult to retain the personnel that have been trained, as the latter move continuously in pursuit of new projects with attractive donor-financed supplements. These salary supplements have also added to the opaqueness of government expenditures, making precise estimates on government pay virtually impossible. 22. Operations and Maintenance Expenditures Operations and maintenance expenditures dropped by 67 percer t in real terms between 1974 and 1989. This corresponds to a fall from 3.5 percent of GDP in 1974 to 0.9 percent in 1989 (see table 4.4). In addition, one also has to consider the fact that the size of Government operations, in particular capital formation in the Government sector, has been growing since 1974. In effect, government departments are now required to provide a much larger volume of  6 Summary of Main Findings and Recommendations services than they did in 1974 with only a fraction of the real material resources. The predictable result is that hardly anything gets done. The debilitating effects of the low recurrent cost funding on the operations of government departments are sharply illustrated by the situation in education, health and agriculture. 23. Recurrent Expenditures on Education Somalia's education system, which is almost entirely in the public sector, is one of the least developed in Africa. At independence in 1960, Sonmalia had a grade one primary enrolment of only 6,000 in 233 primary schools. During the 1960s, the system stagnated with marginal increases in enrolment. The picture changed during the 1970s, however, following the adoption of Somali as the language of instruction in 1972, the abolition of school fees, and an intensive literacy campaign. Enrolment expanded at all school levels, as did the proportion of female students. Quality remained relatively low, but sufficient for basic literacy and numeracy. Moreover, the institutional basis for quality improvement was established through teacher training, curriculum development and the production of instructional materials. 24. Unfortunately, most of these gains have been reversed in the 1980s due mainly to a sharp decline in government funding for education. Allocations from the Ordinary and Domestic Development Budgets to education fell from 2.2 percent of GDP in 1975 to 0.3 percent in 1989, a decline of 86 percent in real terms. The number of primary schools fell by 22 percent from 1981 to 1989. School buildings have deteriorated drastically. Teacher salaries and living conditions have been eroded in real terms, causing high attrition rates, with average seniority of only 4.3 years and only about 10 percent of teachers with more than 10 years of experience. Textbooks and supplies have almost disappeared from classrooms. The Ministry of Education is unable to supervise and support schools. As a result of all these deprivations, quality has declined precipitously, and so has school enrolment. Only 22 percent of students now complete the 8-grade primary cycle. Also, the gains made earlier in female education are being reversed. Deterioration has occurred at all levels of the education system except the university, which has been sustained largely with donor financing from Italy. 25. Donor expenditures on education in Somalia is about nine times the amount spent by the Government through the Ordinary and Domestic Development Budgets. Almost all donor expenditures- about 90 percent--go to technical assistance. However, only 10 percent of the technical assistance goes to primary and secondary education. The bulk of technical assistance is from Italy, for the university, and from Egypt, for those few secondary schools which use Arabic as the medium of instruction. In both cases the assistance consists of the funding of expatriate staff. Hence, even though donors provide significant resources to the educational sector, they neither benefit Somali teachers, nor help in operating and maintaining the schools. Thus, for example, despite four increases in the last four years, primary teacher salaries are only half of the depressed civil service levels, and about one-third of per capita GDP, compared to about 5 times per capita GDP elsewhere in East Africa. Also, allocations for primary school textbooks, which as recently as 1982 were about US$6 per student, have now fallen to virtually zero. 26. Recurrent Expenditures on Health Indicators of health in Somalia are among the poorest in Africa, and have not improved as much as in other parts of the continent in recent years. For example, in 1987 life expectancy in Somalia was 47 years, infant mortality 150 per thousand, and child death rate 19 per thousand. The averages for Sub-Saharan Africa in the same year were 51 years, 115 per thousand, and 16 per thousand respectively. Many of the diseases that account for Somalia's high mortality rates are avoidable with known preventive health measures, low cost curative care such as oral rehydration, increased birth intervals, and environmental control (especially sanitation). All of these interventions require effective organization, infrastructure, and resources.  Su.nmar,y of Main Findings and Recommendations 7 27. Modern health care in the country is almost entirely a public sector service, for which the Ministry of Health bears the responsibility. The situation with regard to expenditures in health is almost the same as in education. Expenditures controlled mainly by the Government through the Ordinary and Domestic Development Budgets have fallen steadily in real terms, and in 1989 were only 22 percent of their 1975 level. For example, spending on health funded out of the Ordinary Budget has declined fron 0.8 percent of GDP in 1975 to only 0.1 percent in 1988, or from 7.3 percent of that budget in 1975 to 1.1 percent in 1988. 28. Even when externally-financed PIP and technical assistance projects are included, total budgetary spending on health declined in real terms until 1987, although it now appears to be increasing again, solely due to an increase in donor funding. Donor funding now contributes 90 percent of public expenditures on health, up from 70 percent only four years ago. Moreover, the trend of this donor financing is toward an increased share for technical assistance. 29. Government recurrent expenditures on health was equivalent in 1989 to US$0.11 per person, compared to about US$5.50 in Kenya, US$2.76 in Senegal and US$1.19 in Maii. The effective result of this low level of government recurrent expenditures in Somalia is that almost all services which are not supported by donors have collapsed. Doctors and staff are ill-paid; drugs and supplies are scarce; and maintenance is neglected. Real salaries of government employees in the health sector have fallen along with those of other Somali public servants. This has to some extent been moderated by increased per diem payments and other salary supplements from donors, but the result has been that all employees now wish to be associated with donor projects. 30. Even within the low level of government spending, curative services and the major urban areas get a disproportionate share in contrast to the Government's goals, expressed in the National Health Plans for 1980-85 and 1989-91, of re-orienting services towards the rural areas and to nomads, and placing more emphasis on Primary Health Care (PHC). 31. Recurrent Expenditures in Agriculture The agricultural sector, including crops, livestock, fcrestry, and fishing is the backbone of the Somali economy, contributing 65 percent of GDP, 95 percent of exports, and 80 percent of employment. However, recurrent allocation to the sector from the Government's Ordinary Budget does not reflect the sector's importance, and the neo,. for the Government to do its part to sustair' it by providing it with adequate and efficient public services. Allocations from the Ordinary Budget to the Ministries of Agriculture and Natural Resources (MOA), Livestock Forestry and Range (MLFR), Fisheries and Marine Resources (MFMR) and Juba Valley Development (which has recently become part of the Ministry of National Planning) shrank from 7.4 percent of the Ordinary Budget in 1975 to 3 percent in 1984, and to less than 1 percent in 1988. 32. Not only are salary levels absurdly low, the allocation for operations and maintenance is so low that 70 percent of it is consumed by vehicle maintenance and running costs alone. Hence, very little resources remain for a variety of activities including maintenance of civil works, irrigatio. system maintenance, and other essential operations such as travel costs and overnight allowances. T1he sector's activities have been kept alive by the increasing financing of recurrent expenditures by donors through the Domestic Development Budgets and the PIP. Most of the departments finance their field operations and training programs through their sponsored development projects.  8 Summary of Main Findings and Recommendations Institutions and Processes for Government-Controlled Budgets 33. Organizational Issues There are severe problems of coordination among and within the institutions responsible for formulating, spending, and accounting for budgetary expenditures. The Ministry of Finance and Revenue (MOF) is responsible for budgeting government expenditures as well as revenues. In performing this task, it is expected to coordinate with the Ministry of National Planning (MNP) and the line ministries (LMs). However, there are no formal guidelines that establish the nature of expected coordination and the responsibilities of each of these entities. Currently, the MO- interacts with the MNP only in the selection of development projects to be ir.cluded in the budget (i.e. the Domestic Development Budget). Coordination between the MOF and the LMs is weak, and the same is true of coordination between the LMs and the MNP in the selection of projects to include in the budget. 34. Even though most LMs have a planning department, they are not adequately staffed, nor are they given responsibility for determining the ministerial priorities for budgetary allocations. The budgetary submissions from the LMs to the MOF for the Ordinary Budget are, in essence, prepared by the accountants assigned to each ministry by the MOF, and hence do not necessarily reflect the needs and priorities of the ministries as determined by their technical staff. The submissions are usually drastically reduced at -ie MOF before the budget is finalized, often without consultation with the ministry involved. With respect to expenditures in the Domestic Development Budget, the LMs generally submit a relatively large list of projects, usually with minimal scrutiny, if at all, of a centralized department within the ministry. These proposals carry inflated estimates for costs, in anticipation of their being trimmed down by first the MNP, and then by the MOF. 35. The weakness ir the ability of the LMs to formulate well-articulated budgets reflecting sectoral priorities, and to effectively oversee their implementation are greatly exacerbated by the fragmentation of cxpenditure planning and spending authority within LMs caused by the proliferation of autonomous projects or units. All donors, including the World Bank, have encouraged this trend in the past to the point of making approval of development aid at times conditional to the Government's acceptance of autonomous project management units, in the belief that only by sidestepping the normal ministerial channels they could effectively implement and supervise their projects. For example, in the agricultural sector, the MOA has about 15 autonomous projects and enterprises that do not have any linkage to the Ministry's internal administr? ve framework. Similarly, the MLFR has four autonomous agencies including the National Range Agency, National Tsetse and Trypanosomiases, Central Rangelands and Marketing, and Health. In fact, many recurrent activities in the agricultural sector, such as extension, irrigation network maintenance and rehabilitation, and animal marketing and health services are performed by autonomous units. Each unit has its own administration and service departments supported by donor funds and expatriate staff. The recurrent expenditures of the units far exceed those of the ministries. Each project has a bank account managed by the project manager, and generally -ere is no central unit in the parent ministry that knows the status of the project expenditure accounts. In most cases, management and recording is further complicated by separate expenditure procedures for funds from different donors. In most bilateral projects, the local project manager controls the part of the funds provided by the Government, and the expatriate project manager controls the donor-provided funds, with neither manager having clear information on the amount of expenditures accrued by the other. 36. Budgetary Procedures The budgetary process for government-controlled expenditures does not appear to follow any guidelines. In theory, the legal framework of the budgetary process and organizational issues are set by a number of decrees and standing orders. These documents were  Summary of Main Findings and Recommendations 9 prepared primarily in the early 1960s. Revisions were made, when necessary, until the early 1970s, However, necessary amendments and revisions have since been neglected. More importantly, the written guidelines are often disregarded in practice--in budget formulating, in making supplementary allocations, and in making transfers between expenditure line items. As a result of this administrative looseness, the linkage between the writtern regulations and actual implementation has broken down, so much so that it has been very difficult even to locate these documents in the MOF. 37. Monitoring, control, and auditing of expenditures are also lax, particularly in the case of counterpart funds used to finance expenditures in the Ordinary Budget. In fact, there is no indication in the budget documents or in the closed accounts of the ordinary expenditures that are funded by the counterpart funds. The MOF prefers to leave a large portion of counterpart funds outside the budget to finance extra-ordinary expenditures in the course of the year, subject only to the Minister's discretion. The same problem also applies to extra-budgetary allocations financed by the "excess' revenue realized over and above the ir itial revenue projections. This is clearly a deficient procedure which makes the reconciliation of actual expenditures and budgeted amounts virtually impossible. 38. Problems with the Management of Development Expenditures Given that development expenditures in Som. 'ia are all financed by donors--either directly through the PIP (the predominant portion), or indirectly through counterpart funds for the Domestic Development Budget, and given also that most PIP projects are directly managed by donors through project implementation units, the central questions of Government development expenditure policy then become: (a) does the process through which projects are selected for donor funding ensure that only economically viable projects, or projects that satisfy social objectiv's in the least-cost manner are chosen? (b) does the process provide a mechanism for anticipating and making provision for the recurrent cost requirements of projects in the post-implementation phase, when donors are no longer responsible for financing? and (c) is the project selection process organized in a way that achieves efficiency in the use of government administrative resources? Unfortunately, the answer to each one of these questions is no. These are discussed below, starting with the last one. 39. Institutional Issues in the Planning of Development Expenditures The project selection process, or planning, in Somalia is organized in a way that results in considerable duplication, confusion, and waste of scarce administrative resources. In addition to the Domestic Dev-lopment Budget that caters to the Government's own development projects, which are financed with counterpart funds, there are three plan documents dealing with directly foreign financed development projects: the Five-Year Development Plan (FYDP), the Annual Development Plan (ADP), and the three-year Public Investment Program (PIP). 40. The preparation of each one of these three plans engages substantial resources in the Ministry of National Planning, and also imposes on resources in the line ministries as well as on the time of Cabinet ar.d Parliament. However, only the PIP has any operational relevance, in so far as it is the document used to solicit donor fi:nds. Once the funds are obtained for a project, the actual budgeting of expenditure flows from year to year are done primarily by the donor implementation units and MOF, independently of any of these plans. By law, a project can be included in the PIP only if it has already been included in the FYDP, but this requirement is not adhered to in practice. Even if the requirement were adhered to so that the PIP at any time were a proper sub-set of the FYDP, the question would still have to be faced whether the carrying out of two separate planning exercises, each of which is supposed to produce a list of projects to be funded mostly by donors, but only one of which is presented to donors is an efficient use of administrative resources. The ADP is, in principle, supposed to link planning to  10 Summary of Main Findings and Recommendations government budgeting by providing information on the required government contribution to the implementation of the PIP projects. However, the ADP is prepared concurrently with the government budget (i.e., the Domestic Development Budget), rather than before it. Hence the ADP has not been a source of significant input into budget preparation. 41. Another source of waste and major confusion is the fragmented authority among various ministries %.ith regard to aid coordination, and hence the coordination of development expenditures. Aid- coordination responsibilities are now split four ways: the Prime Minister's office handles emergency relief assistance; MOF oversees foreign loans; grants which constitute about three quarters of all foreign aid to the country are negotiated and administered by the Ministry of Foreign Affairs (MFA); and MNP has the responsibility for planning how the foreign aid is to be used. 42. Under the present set up, coordination between the various institutions associated with external aid is weak, and the flow of information from MFA to MNP is sporadic, at best. Many grant financed investment projects are signed and implemented by MFA without prior consultation with MNP or sector ministries. Donors have expressed frustration over wasteful overlaps of some activities they finance because of this lack of coordination between the two ministries. It also frustrates efforts by the MNP and sector ministries to build a PIP consistent with sectoral and macroeconomic development objectives. This lack of coordinat;on, more than any other, undermines the planning process itself and diminishes the value of the PIP as a tool for . s cating public sector resources. 43. Linkages Between Planning and Budgeting, and the Recurrent Costs Problem As may be surmised from the discussion of planning institutions above, there is no effective link between planning and budgeting. First, the Annual Plan, the linkage that is supposed to provide information to the budget on the Government's contribution to the implementation of projects, is ineffective since it is prepared concur-ently with the budget. Second, and even more importantly, the MNP and MOF fail to confer and assess the future recurrent costs implications of alternative PIP packages, and ensure that the package selected and presented to donors is consistent with projections of future government resources for funding recurrent costs. The result has been that once donors complete a project and hand it over to the Government, the recurrent expenditures are perennially underfunded, and not only do the physical structures then begin to crumble, but also the local personnel who no longer receive the donor-financed pay supplements quickly abandon the project in search of newer projects that are still under the wings of donors. 44. The Project Selection Process for the PIP The formal criteria for a project to be included in the PIP are that: (a) the project is included in the Five Year Development Plan; (b) a feasibility study has been prepared; and (c) external financing has been secured. In principle, priority is given to rehabilitation and projects which make f .11 use of idle capacities, projects with low recurrent expenditures and high rates of return, and projects with high social impact. These criteria and principles are not adhered to in practice. Feasibility studies, even if carried out, in many cases lack a serious rate of return analysis, and the PIP often contains proposals for several projects that are of dubious economic viability. In addition, contrary to the criteria that there should be external financing, PIP does include a few (but not all) projects which are temporarily funded purely from local resources, whose operations are being maintained at minimal levels untlI external financing is secured. However, if this criteria were strictly applied, it would reduce the usefulness of the PIP document for soliciting donors' assistance for project funding.  Summary of Main Findings and Recommendations 11 45. There are also projects outside the PIP (non-PIP projects) which are funded by the domestic development budget. For these there are no set selection criteria, comparable to those for the PIP projects; rather, social and political considerations predominate. 46. Size and Allocation of A' #P penditures On an estimated disbursement basis for three year periods from 1975 to 1989, th - .f the PIP has ranged between 6 percent of GDP in 1981-83 to 14 percent in 1987-89, with donor.- c ntributions converted at the official exchange rate. The rate of implementation has been low; beiow 60 percent for most years. In 1987-89, slightly more than 50 percent of the slippage is attributable to delays in twelve large projects (including the Baardhere dam). These projects accounted for as much as 38 percent of the planned PIP expenditures, but realized only about 7 percent of their planned disbursements. When these projects are excluded, the implementation rate over 1987-89 was about 74 percent. 47. The sectoral allocation of PIP expenditures over 1987-89 (excluding expenditures on the Baardhere Dam and on Technical Assistance) was: 35 percent to agriculture (including livestock, crops, forestry, wildlife, and fisheries), 16 percent for industry and energy, 41 percent for infrastructure, 5 percent for social services (education and health), and 3 percent for urban and rural development. For 1990-92, the proposed PIP allocation roughly doubles the allocations to social services, and to urban and rural development with slight reductions in the allocations to the other sectors. 48. While the broad sectoral allocation in the proposed PIP is generally consistent with the requirements for growth in the economy, and the increase in the allocation to the social sectors is welcome, there are still some problems. First, the allocation to the social sectors is still very low when compared to the requirements. Given this, the problems of implementation of the overall PIP, and the severe recurrent cost constraints in the operation of completed projects, the question needs to be raised to the Government, and particularly to donors, whether the PIP should not be down-sized, in favor of donors providing explicit funding for recurrent costs, especially in the social sectors? In this connection, the shifting of the emphasis of donor assistance to education from the higher levels to the primary and secondary levels is recommended. Second, the proposed PIP includes a number of production oriented activities in agriculture from which the Government needs to disengage, given its inability to fund key recurrent public services in the sector, and its announced intentions of privatizing existing state farms. Elatedly, while petroleum refining may be regarded as a strategic activity worthy of continued government investment, a thorough evaluation of the economic viability of the proposed rehabilitation of the Petroleum Refinery at the cost of US$36 million has shown that, unless radically altered, the project cannot be justified. Thilrd, most new investments in telecommunications, which formn the bulk of the proposed investments in infrastructure, appear overly ambitious, possibly duplicative. and quite beyond the technical and managerial capacity of the Telecommunications Authority. 49. Classification Problems Inconsistencies in the definition of projects in the PIP document make it difficult to categorize the projects by size. While some projects represent a consolidation of several phases of a development project spanning a number of years, others include only the current or proposed new phase from which disbursements are anticipated over the plan period. Some projects have a program approach, treating several donor financed projects for a sub-sector as one; in other cases, components of one large project have been treated as separate and independent projects. In several cases, historical data are not complete. Information provided on individual projects is also inadequate. Financial data, consisting of total project investment cost and sources of financing, are provided in one section whereas project descriptions appear in another. Project descriptions include both technical assistance and PIP  12 Summary of Main Findings and Recommendations projects as well as numeruus others for which no financial information is given in the PIP and technical assistance tables, creating further confusion as to what constitutes the overall investment programn. 50. The rest of this volume is concerned with providing an agenda designed to address the problems of public expenditures in Somalia diagnosed above. The agenda includes expenditure restructuring and institutional reform, both of which will have significant impact on government-controlled expenditures. Hence, before going into the elements of the reform agenda, we discuss the likely evolution of the economy, and of the total financial resources that are likely to be available for public expenditures, particularly the resources potentially available to government-controlled budgets. C. Medium-term Macroeconomic Projections and the Potential Resource Envelope (1990-93) 51. A fundamental pre-condition for macroeconomic viability in the medium term is a quick and peaceful resolution of political differences in order to halt the armed conflict. Without this, an inordinate amount of Government resources would continue to be devoted to defense, and the pursuit of fundamental economic reforms is unlikely to receive the priority it deserves in the Government's agenda. Furthermore, the recent uneasiness on the part of donors about providing financial assistance to Somalia, is likely to be heightened, leading to significant withdrawal of donor support; and for the foreseeable future, sustained economic growth in Somalia in the absence of substantial donor financial support is unlikely. The reader should keep in mind that all estimates and projections were finalized in July 1990 and were predicated on a resolution of the political and military strife and an improvement in macroeconomic performance before the end of the year. As macroeconomic management continued to deteriorate in the second half of the year and as full scale civil war broke out in Mogadishu at the end of December and continued into January 1991, the macroeconomic framework presented in the report should be taken as only indicative of a possible medium-term scenario after political stability has been achieved. 52. Economic Policy Reforns However, peace alone would not be sufficient to engender the levels of donor support needed. The Government will have to seriously implement economic policy reforms, and will also have to undertake comprehensive reforms in public expenditure management. Economic policy reforms would have to include inte.sified efforts by the Government to restrain monetary expansion, with the objective of easing inflationary pressures and arresting the rapid movement in the parallel-market exchange rate. Government policy should also address the weak performance of exports and domestic revenues. Regarding exports, the Government should re-design its retention and surrendor schemes so as to improve export incentives for both traditional and non-traditional exports. As for revenue, while it is recognized that 1ie dismal performance is due in part to the security situation, the Government should, nevertheless, take measures to bring about a significant improvement in performance. Measures needed include: comprehensive application of ad-valorem tax rates, particularly on petroleum; factoring in the high inflation levels of recent years in assessing business and income tax liability; instituting an effective import verification system; and the strengthening of tax administration. These macroeconomic measures are necessary for the economy to realize the full benefits of the structural economic reforms that the Government has already undertaken, and is committed to sustaining. Specifically, the Government must follow through on its intentions regarding the strengthening of policy formulation and implementation capacity, increasing the role of the private sector and reliance on market forces, and reforming parastatals, including autonomy and accountability in pricing, production and staffing decisions for viable ones, and liquidation of those judged to be not viable.  Summary of Main Findings and Recommendations 13 53. Macroeconomic Prospects I With an improved macroeconomic policy environment and political stability, real GDP growth over 1991-1993 could average almost 5 percent a year, compared to 3.3 percent over the past four years (1986-89). Most of the growth would come from agriculture, particularly livestock. Exports would rebound as a result of the higher GDP growth rate, and more importantly because of enhanced incentives, including improved exchangu rate management. In contrast to the real growth rate of -12 percent over 1986-89, exports would grow at an average real rate of 13 percent over 1991-93. The rate of inflation is projected to fall steadily from the 1989 level of 150 percent to 15 percent by 1993. Non-project commodity imports would grow at 4 percent a year, but due to a slow-down in project-related imports in accordance with restraint on the PIP (explained below), overall imports would only grow at an average annual rate of 0.7 percent over the period 1991-93. The current account deficit excluding grants but taking foreign interest (on commitment basis) into account would average around 40 percent a year between 1991 and 1993. With fiscal pressure on the private sector sharply on the rise, private national savings would be expected to improve only gradually reaching 2 percent of GDP by 1993. Private fixed investment would increase correspondingly from around 2.7 percent in 1989 to 6.4 percent by 1993. Government savings performance would also improve from an average of -16.5 percent of GDP over the past four years to around -12 percent a year by the end of 1993. However, the Government would still continue to depend heavily on external finance, as the discussion of the 'resource envelope' below shows. 54. The Resource Envelope As a result of improved tax administration, including a better import verification system, and the introduction of new taxes, the ratio of domestic government revenue to GDP is projected to rise steadily from the low level of 5.3 percent in 1989 to a little over 9 percent by 1993, with an average over the four-year period of 7.6 percent. Foreign project assistance--the PIP-would average 24 percent of GDP, or about US$255 million a year. This would represent a decrease in both real and nominal terms compared to the last four years. This trend reflects the fact that the PIP has become too large relative to implementation capacity in Somalia, the need to increase recurrent cost financing by donors, and the need for donors to also provide substantial financial assistance in the form of debt relief. Around 70 percent of project financing is expected to be in the form of grants. Non- project foreign assistance that generates counterpart funds is expected to average about 10 percent of GDP a year, again, around 70 percent in the form of grants. Total resources potentially available to the Government (not counting debt relief) would, therefore, average about 40 percent of GDP a year over the projection period. However, the resources whose allocation the Government can actually control would be much smaller, as explained below. (See Chapter 3 section C, and table 3.4). 55. Committed Expenditures Project financing, which represents 21 percent of GDP on average, or a little over half the available resource, would come in pre-committed to the PIP. Most of the control over its composition lies with donors, rather than the Government. As in the past, we expect the Government's contribution to be about 8 percent of that of the donors, which translates to about 1.8 percent of GDP a year. Another block of committed expenditures concerns debt payments and the clearance of arrears. In order for the macroeconomic program to attract adequate external funding, Somalia would have to clear its arrears to its foreign creditors, including those to the IMF and other multilaterals, which stood at US$150 million at end-June 1990. We project that in 1991 an amount of 'Te reader should keep in mind that all estimates and projections wore finalized in July 1990 and we-re predicated on a rcsolution of the political and military strife and an improvement in macroeconomic performance before the end of the year. As macroconomic management continued to deteriotate in the second half of the year and as full scale civil war bmke out in Mogadishu at the cad of December and continued into January 1991, the macroeconomic franework presented in the report should be taken as only indicative of a possible medium-term cenario after political stability has been achieved.  14 Summary of Main Findings and Recommendations about 24 percent of GDP would be needed to clear interest and principle arrears to multilaterals. We assume that this amount would be obtained as grants, separate from the grants assumed above in projecting the budgetary resource envelope. In addition, we assume that 6.6 percent of GDP in debt relief would be obtained in 1991 on interest arrears. We further project relief on current interest payments failing due to average about 3.1 percent of GDP from 1991 to 1993. With these assumptions about debt relief and grants to help with clearance of arrears, the net debt payments actually coming out of the projected budgetary resources would be 4.7 percent of GDP a year, from 1991 to 1993. (It is assumed that in 1990 the Government continued to accumulate arrears.) Further, consistent with our projection of falling inflation rates and also with the need to increase credit to the private sector to sustain higher levels of investment, we are projecting a repayment to the banking system from the budget at an average level of 5.4 percent of GDP a year between 1991 and 1993. Summing the foreign PIP expenditure commitment, the required Government contribution to the PIP, the net foreign deb( service requirements, and the repayments to the domestic banking system results in committed expenditures which average about 30 percent of GDP a year, out of the total budgetary resources of around 40 percent of GDP a year. (See table 3.4). 56. Apart from debt and banking system repayments, budgetary resources that could be subjected to allocational reforms over 1990-93 would on average be around 10 percent of GDP. With the pre- committed expenditures accounted for, what remain are non-interest expenditures in the Ordinary Budget, and expenditures on the government's own projects in the Domestic Development Budget. Many of these projects are not economically viable, and in fact given the already very large portfolio of PIP projects, there is little justification for this second channel of project implementation. We, therefore, assume minimal expenditures on these projects, leaving almost all of the non-committed government resources-10 percent of GDP- for non-interest expenditures in the Ordinary Budget. In addition, we expect that donors would increase the share of recurrent cost financing in the PIP from an average of 24 percent of PIP expenditures to around 30 percent up to 1992 before beginning a gradual reduction of recurrent-cost financing. How the resources for non-interest expenditures in the Ordinary Budget are to be allocated to address the problems identified earlier is the subject of the expenditure restructuring part of the reform agenda in the next section. D. Agenda for Improving Public Expenditure Management 57. Summary of Probkms The diagnosis of government expenditures in Section B above clearly brought out six pressing issues: (a) recurrent expenditures are now too low; (b) within recurrent expenditures, the increasing share going to defence and security services have been preempting funding for wages and salaries as well as for operation and maintenance in the civil service. Economic and social services, particularly education and health, have consequently deteriorated to depressingly low levels. (c) the PIP, which is essentially all funded by donors, is much too large for the Government's administrative capacity, leading to a situation where donor-funded autonomous implementation units practically run the whole program; (d) the very large funding of government expenditures by donors- roughly 62 percent of the total-while beneficial in terms of providing resources to Somalia, a very poor country, nevertheless creates serious management and morale problems wi *n the civil service; (e) the structure and formats of the government-controlled budgets lack transparency, and thus deprive the Government of the essential service of the budget as a tool for evaluating and planning government expenditures; and (f) there are serious institutional, procedural, and processing problems of budgeting as well as of investment planning.  Summaty of Main Findings and Recommendations 15 58. Strategy of Reform All of the problems identified above have to be addressed if public expenditures in Somalia are to make productive contributions to the economic and social development of the country. However, clearly the Government cannot tackle all of the problems at once, and expect to make good progress on all fronts. We would, therefore, propose a two-phased approach. The phases could overlap in time, but each would tackle a subset of the problems identified. The first phase would deal with the immediate need to increase recurrent funding, particularly for wages and salaries as well as for operations and maintenance in order to get the civil service functioning at a respectable level of efficiency. This phase would also try to achieve a quick redressing of the very low level of funding for educational and health services, particularly at the primary levels. In addition, it would eliminate projects that are clearly non-viable from the Domestic Development Budget and the PIP. However, some reforms in the budgeting and planning process would also be initiated, focusing on improving transparency. 59. The second overlapping phase would include a comprehensive reform of the civil service, which would have as one of its main objectives adequate remuneration for workers in a formal and transparent manner, without reliance on informal and nebulous supplements financed from either outside the budget or from diversion of allocations from other items in the budget. A key requirement for sustained progress on this front would be reform and improvement of the domestic revenue systems of the Government in order to push the revenue to GDP ratio from the current level of 5.3 percent to a respectable level of around 18 percent by the year 2000. An intermediate target between now and then is around 9 percent by 1993. As the domestic revenue effort improves, donors would gradually phase out of providing recurrent budgetary support. The second phase would also focus on rationalizing the processes of budgeting and project preparation, strengthening the institutions involved, clearly defining and streamlining coordination among them, and reforming the budget structure in order to facilitate analysis and planning of expenditures. This phase is labelled 'second" only in the sense that the actions required under it would take longer to accomplish, but there is no reason why some of the actions under it cannot begin at the same time as those under the first phase. In particular, preparatory work for at least the civil service reform, and also actions to improve the tax system need to start immediately. 60. Phase I of Reform Expenditure Restructuring (i) Raise the level of recurrent expenditures: The 10 to 11 percent of GDP projected above (Section C) for non-interest expenditures in the Ordinary Budget represents a minimum. Given the severe problems that have accumulated for over a decade due to underfunding of recurrent expenditures, the Government should make every effort to surpass the projected levels by increasing its domestic revenue effort. To ensure that all of the resources projected for non-interest expenditures actually go to finance recurrent costs, funding of capital projects in the Ordinary Budget as well as for project foreign funding in the Domestic Development Budget would need to be kept at a minimum, no more than a total of about 0.5 percent of GDP. (ii) Increase the share of allocation in the Ordinary Budget to wages and salakis andfor operation and maintenance in the civil service: To achieve this objective, the share of defense and security related expenditures would need to be reduced steadily from the range of 8 to 10 percent of GDP in recent years (i.e including all the unclassified expenditures) to around 3 percent by 1993. Further, expenditures on foreign representation would have to be held in check, to around 1 percent of GDP a year. In terms of dollars, they should fall to about 50 percent of the annual levels in recent years. In addition, transfers and miscellaneous expenditures should be held to no  16 Summary of Main Findings and Recommendations mt)re that 0.5 percent of GDP per annum. The above measures would allow average expenditures on operations and maintenance to be pushed to above 3 percent of GDP a year, compared to 0.7 percent in the past four years (i.e 1986-89), and for wages and salaries to double to an average of I percent of GDP a year and to increase four-fold by 1993 to above 2 percent, compared to the 0.5 percent in the previous period. The averages for operations and maintenance expenditures and tor wages and salaries given above are for allocation trends that would be rising over time. (iii) Tere should be a S percent cut in the size of the civil serice in 1991 before the increased allocationfor wages and salaries are translated into individualpay increases: The evidence of overemployment in the civil service is so clear that, without even waiting for comprehensive civil service reforms which would tackle all the detailed institutional issues, the Government, for purely expenditure control reasons, can, and should shed some of its labor force. (iv) Within the increased global allocation to wages and salaries, and for operations and maintenance, the share going to the economic and social services should be increased: Our analyses indicate that to get the economic and social services functioning at tolerable level of effectiveness would require operations and maintenance expenditures starting from around 1.1 percent of GDP in 1991, and reaching about 3.1 percent in 1993. Similarly, allocations for wages and salaries in the economic and social services would start from around 0.5 percent in 1991 and reach almost 1 percent by 1993. Essentially, what this boils down to is that between 70 and 75 percent of the allocation in the Ordinary Budget for operations and maintenance, and for wages and salaries in the domestic civil service (i.e. not including security related services, and not including foreign services) should go to the economic and social services. (v) The Government should restrain total project expenditures to within the disbursement levels shown in the 1990-92 PIP, and should seriously consider trimming this down fwther by dropping or down-sizing projects of questionable economic merits: The size of the PIP is now greatly taxing the implementation capacity of the Government. In view of this, the lower level of disbursements projected in the current (1990-92) PIP compared to actual disbursements under the 1987-89 PIP is a move in the right direction. As reviewed by the mission in November 1989, expected disbursements in 1990 were estimated at US$188 million, which is about 20 percent below the actual PIP disbursements for 1989. However, there is reason to believe that the PIP document understates project expenditures that may eventually take place over 1990-92. This is so, because, for some projects, especially in the social sectors, only the level of activity for which external funding has already been secured has been included in the PIP, rather than the ultimate level anticipated for the project. The Government is urged to resist attempts to either expand the levels of activity beyond those shown in the PIP document, or to commit funds to more projects than have been shown in that document. Projects in the PIP which need re-consideration include Jowhar Sugar (US$23 million), the Petroleum Refinery (US$36 million), and Mogadishu Teaching Hospital (US$11 million). The first two are of questionable economic merit, while the objectives of the third could be achieved at a much lower cost by expanding the Mogadishu General Hospital. Further, the portfolio of new telecommunications projects amounting to about US$60 million should be carefully re-examined. In fact we recommend that the Government focus on rehabilitation projects in the sector, and delay all new projects until a complete needs assessment of the sector is performed.  Summary of Main Findings and Recommendations 17 Institutional Reforms in Budgeting (vi) The authorities should ensure that all central government expenditures are covered either by the Ordinary or the Domestic Development Budgets. (vii) The practice of keeping a large "other" item in the MOF budget, which reduces the transparency of expenditure allocation, should be abandoned, and expenditures properly identified. In particular, the Closed Accounts for 1990 and a revised budget for 1991 should reflect the new approach. (viii) Special attention should be paid to the use of counterpart funds, and the uncertainty surrounding the use and control of these funds should be removed. The practice of leaving the authorization of part of the counterpart funds to the MOF without going through the established budgetary procedures should be abandoned. (ix) The Government should make use of supplementary budgets following the same steps as the annual budget and providing the same detail when it needs to allocate any additional resources. (x) Minor improvements may be made in the format of the ordinary budget, while the format of the domestic development budget should be totally restructured. Recommended sample formats are provided in Volu" e II (Chapter VI, Annex 1 and Annex 2) (xi) The budget calls issued by the MOF to guide LMs in budget preparation need to provide sufficient information on objectives, targets, policies, and ceilings that would apply to a given year's budget. (xii) A Civil Service Commission, independent of any government ministry, should be established, preferably in the President's Office of Prime Minister's Office, to deal with civil service management issues. In particular, this commission should prepare for and oversee the civil service reforms in coordination with the MOF. (xiii) The Civil Service Commission should re-institute an "Establishment Register" and a "Staff List' and keep them up to date to help in the monitoring and controlling of the size of employment in the civil service, and also in determining the level and spread of salaries. Once these lists are completed, they should be attached to each ministry's Ordinary Budget when it is submitted to the National Assembly, as had been the practice in earlier years. Institutional Reforms in Investment Planning (xiv) The existence of two medium-term plan documents for externally financed projects-the FYDP and PIP-is both a waste of scarce Government administrative resources, and a source of confusion in planning. There should, therefore, be only one medium-term plan document. Since the three-year PIP is the one that is actually presented to donors for funding, and hence is the only one that is operational, the Government should transform the FYDP into a longer-term (seven to ten years) indicative plan. In addition, the Annual Plan should be dropped. With the administrative and other resources freed by dropping these plans, the Government should then improve quality and comprehensiveness of the PIP. In particular, the PIP should be made into a proper rolling investment plan, thereby serving the objectives of the Annual Plan for annual budgeting of development expenditures in a more operational manner.  18 iummary of Main Findings and Recommendations (xv) The PIP must be a comprehensive document, covering all public sector investments regardless of their source of financing (Goveanment or external) and of the level of assured external funding. It should include all projects for which the Government is actively seeking foreign financing. (xvi) In addition to projects, the PIP should include a clear articulation of the Government's medium- term economic objectives and policies, as well as a medium-term financial plan produced by MNP in close consultation with MOF and the Bank of Somalia. (xvii) There must be a rigorous project selection process, with clearly established economic and social criteria, which is conducted or overseen by the MNP. All projects, regardless of the source of funding (foreign or the Govertnent) or type of funding (loan or grant), must be subjected to the selection process. Sustainability of operations resulting from investments must receive primary consideration with due attention given to future recurrent cost requirements. Grant-financed projects should be subjected to a rigorous selection process because of the opportunity cost associated with it and the current and future claims it would impose on Government resources in terms of required current cost financing. Hence there is need to ascertain whether such costs are warranted either in terms of economic criteria, or key social objectives. This means the practice whereby the Ministry of Foreign Affairs negotiates grant-financed projects without prior technical and economic appraisal by the MNP should be terminated. xviii) Closer attention needs to be paid by all donors to strengthening and working through the existing institutional set-up, instead of sidestepping the normal ministerial channels in the design, implementation and supervision phases of their projects, as in the past. Donors should offer the necessary technical and financial assistance to strengthen the line ministries' planning and accounting functions, as well as their ability to effectively implement and monitor projects. (xix) The link between planning and annual budgeting should be strengthened through better coordination between the Ministries of Finance and Planning. Also, the coordination between the Ministry of Planning and the line ministries must be improved. Line ministries should be encouraged to make a genuine contribution to investment planning in their respective areas of responsibility. 63. Phase II of Reform Civil Service Reform (i) The objective should be to gradually phase out the salary supplements from the Development Budget and the PIP projects, and to fund wages and salaries at adequate levels solely from the Ordinary Budget. Given the constraints to raising domestic revenues, and the other demands on government resources, this program cannot be achieved without reducing employment in the civil service, which at around 45,000, is estimated to be between 40 to 60 percent over-staffed. Taking the lower estimate of 40 percena. would mean civil service employment should be around 27,000 instead of 45,000. Given the complexity of retrenchment programs, we would recommend that the required reductions be phased in over 5 years, starting from 1991 (i.e. 1991-1995). This would imply an average of 10 percent reduction per annum in the number of employees. At the same time, allocations for wages and salaries in the Ordinary Budget would be increasing by about 30 percent a year in real terms. However, only a part of this increase in wages and salaries from  Summary of Main Findings and Recommendations 19 the Ordinary Budget would translate into actual increases to workers, since there would also be some phasing out of the informal salary supplements. There would be financial costs to laying-off workers. The law in Somalia requires payment of one month's salary for each year of service, which should be probably be considered as the lower bound for the purpose of estimating the total financial cost. The total cost is also dependent on the sequencing of wage increases and retrenchment. Granting major wage increases before retrenchment would make the cost of the separation package much higher, and also increase resistance of workers to lay-offs. On the other hand, since the retrenchment program, for practical reasons, has to be stretched over 5 years, the Government cannot delay pay increases till it attains the desired reduced size of employment before granting wage and salary increases. It would seem, therefore, that the best option for the Government would be to offer in each year of the retrenchment period, a separation package that is not too generous, but still attractive enough that despite the increase in wages and salaries, those workers being laid-off would accept their situation without much resistance. Preliminary calculations by the mission indicate that a separation package designed along this principle, given the increases in wages being recommended, would cost around US$15 million, or about 10 percent of the projected aggregate wage bill over l991-1995. However, this estimate needs to be further refined in a more detailed project design. The detailed project design should also include specific programs to provide training and other assistance to retrenched workers to help them re-establish themselves in alternative gainful occupations. It is important that the cost of the separation package be regarded as additional to the normal budgetary expenditures. There are indications that donors may be willing to cover temporary financing for reform of the civil service on the order of US$3 to US$4 million a year for a period of three to five years. Institutional Reforms in the Ordinary and Development Budgets (ii) The accounting system should be strengthened to ensure proper recording and monitoring of actual expenditures in a more precise and timely manner. (iii) The responsibilities of the agencies involved in expenditure planning should be clearly defined so as to improve coordination among them, and properly staffed budget units should be set up in the LMs. (iv) Training programs should be instituted to upgrade the skill levels of those who participate in the budgeting exercise both in MOF and LMs. (v) A better expenditure classification in the budget needs to be adopted to differentiate between the recurrent and capital-type expenditures. A gradual phasing out of the existing system is advisable, since an immediate transfer of all recurrent-type expenditures from the domestic development budget to the ordinary budget may not be feasible. (vi) A system of program budgeting should be adopted for more efficient expenditure planning. This would introduce great.- rationality to decision making for expenditure allocation by allowing the comparison of costs and benefits of various government programs. The system may be initiated by identifying some broad programs in LMs, and be fine tuned gradually.  20 Summay of Maln Findings and Recommendations (vii) The concept of forward budgeting should be introduced, to allow better planning for resource availability and expenditure requirements in the medium term. This would necessitate major improvements in the capability for revenue forecasting and expenditure projections. Integrated View of Public Expenditures (viii) To facilitate analysis and evaluation of past expenditures, and also to provide a better basis for planning future expenditures, the Government should work towards producing, in each year, a comprehensive budget of all recurrent and development expenditures that allows all public expenditures to be classified along both economic and functional lines. Ideally, all the expenditures should be presented in a single budget document. If this cannot be done then at least a common system of classification and format should be used by each of the various budgets to break down expenditures into their economic and functional uses. E. Conclusions 64. The expenditure reform agenda given above is broad, and successful implementation would require full dedication by the highest levels of Government. Fortunately, with the possible exception of the retrenchment under the civil service reform, none of the proposed measures is likely to impose hardships that might engender civil resistance. On the contrary, many &i the measures proposed would actually provide quick benefits to the population, either directly in the form of wage and salary increases, or indirectly in the form of greatly improved government services. Somalia is also fortunate, in that there are donors willing to underwrite the cost of the civil service reform, which would provide resources to help reduce hardships on the workers who are retrenched. However, as pointed out in the discussion in Section C, a fundamental requirement for success of any expenditure reform program in Somalia is a quick and peaceful end to the armed conflicts, for while the conflict continues it would be difficult to effect the necessary reduction in the extremely high defense expenditures in order to redress the serious underfunding of civil administration and essential government services. It must also be reiterated that reform of government expenditures must be pursued within the context of a broader economic adjustment program. This broader program should include, in particular, continuation and broadening of the reforms in incentive policies already begun by the Government, reforms in the parastatal and financial sectors, and the strengthening of the Government's revenue collection system.