TANZANIA: TRACKING POVERTY-REDUCING SPENDING: COUNTRY ASSESSMENT AND ACTION PLAN (AAP) December 3, 2001

TANZANIA: TRACKING POVERTY-REDUCING SPENDING: COUNTRY ASSESSMENT AND ACTION PLAN (AAP) Prepared by Fund staff 1 In collaboration with World Bank staff and the Tanzanian authorities December 3, 2001 1 The mission comprised Sanjeev Gupta (head), Toshihide Endo, Annalisa Fedelino, and Pokar Khemani (FAD). #1330128 v6 - Tanzania (ENG) FINAL 2001 AAP

- 2 - ABBREVIATIONS AND ACRONYMS AAP ACGEN C&AG GFS HIPC IDA IFMS LPO MDAs MDF MoF MTEF PEM PMG PORALG PRBS PRSP RAS Assessment and Action Plan Accountant General Controller and Auditor General Government Finance Statistics Heavily-Indebted Poor Country International Development Association Integrated Financial Management System Local Purchase Order Ministries, Departments, and Agencies Multilateral Debt Fund Ministry of Finance Medium-Term Expenditure Framework Public Expenditure Management Paymaster General Account President Office, Regional Administration, and Local Governments Poverty Reduction Budget Support Fund Poverty Reduction Strategy Paper Regional Administration Secretariat

- 3 - Tracking Poverty-Reducing Spending in Tanzania Assessment and Action Plan (AAP) A. Introduction 1. In February 2001, Boards of the Bank and Fund discussed the issue of tracking poverty-reducing spending in HIPCs in the context of budgetary savings stemming from the HIPC Initiative. The discussion was based on a preliminary assessment by Fund/Bank staff of public expenditure management (PEM) systems of 25 HIPCs. 2 2. The two Boards concluded that HIPCs will need to track all poverty-reducing spending and not just that financed from HIPC inflows. They further requested staffs of the two institutions to review preliminary assessments of PEM systems with concerned HIPCs. This exercise is meant to assist HIPCs in identifying key weaknesses in their PEM systems as well as in formulating action plans for strengthening these systems in the short and medium term. The resulting country-owned action plans are expected to be incorporated into Poverty Reduction Strategy Papers (PRSPs) and Bank s Country Assistance Strategy (CAS) so that financial and technical support from various development partners could be mobilized. In this regard, the Boards also requested that a list of technical assistance extended in recent years and planned in the future by various donors be prepared. 3. The purpose of this mission is to carry out such a review of Tanzania s PEM systems and assist the authorities in the preparation of an action plan. Why should all poverty-reducing expenditure be tracked? 4. There are three reasons why all poverty-reducing expenditure in Tanzania will need to be tracked: A mere focus on HIPC inflows will not ensure additionality of spending on poverty. This is because resources are fungible, and a country can offset HIPC assistance earmarked for poverty-reducing programs by lowering its own spending in those areas. The composition of public spending is envisaged to become increasingly propoor over time. The use of HIPC resources for poverty reduction would not necessarily ensure that the overall composition of public spending has tilted in favor of poverty programs. The latter is a major objective of the PRSP process and a key feature of PRGF. 2 Tracking of Poverty-Reducing Public Spending in Heavily Indebted Poor Countries (HIPCs) (Revised SM/01/16, March 2001) and Summing Up (BUFF/01/21, February 12, 2001). Both documents are available on the IMF website, at www.imf.org/external.

- 4 - It is also critical that the overall PEM systems in HIPCs are strengthened and limited technical skills not diverted to manage just HIPC inflows. Furthermore, efforts currently underway to strengthen overall PEMs should not be undermined. Channeling of debt relief 5. In 2000/01, Tanzania received debt relief under the Enhanced HIPC Initiative amounting to US$160 million (1.9 percent of GDP). Until recently, Tanzania maintained a donor-funded Multilateral Debt Fund (MDF) to service debt obligations to multilaterals (Box 1). Accrual of debt relief and debt service payments are now channeled through the government s main bank account at the central bank. Box 1. Tanzania: Accounting for Debt Relief Flows In July 1998, a Multilateral Debt Fund (MDF) was established by donors for servicing debt obligations to multilateral financial institutions. Funds were disbursed according to debt obligations falling due. The resulting budgetary savings were channeled to expenditures in priority sectors. Coordination with donors was promoted through quarterly meetings, where the government reported on its fiscal performance and, more specifically, on poverty-reducing spending. Disbursements into the MDF reached US$185 million in 1999/2000, more than twice the initial contributions in 1998/99. As Tanzania reached the decision point under the Enhanced HIPC Initiative in March 2000 and began receiving interim relief, funds in MDF quickly exceeded the amounts needed for multilateral debt service. Recognizing the fungibility of resources, the authorities began to monitor increases in total expenditures on priority sectors in order to provide assurance to donors and other stakeholders on the effective use of debt relief. At the same time, donors decided to extend general budgetary support for financing poverty-reducing programs. This led to delinking of contributions to MDF and cash releases to priority sectors from the timing of debt-service payments. Contributions to MDF were discontinued, and a new Poverty Reduction Budget Support (PRBS) fund became operational in December 2000. When debt service payments to multilateral institutions are due, the corresponding amounts are debited from the Paymaster General Account (PMG) at the Bank of Tanzania (BoT). The BoT pays to multilaterals only the debt service net of the relief; the latter is then credited back to the PMG account. In case of debt service to bilaterals, payments are debited from the PMG account when these fall due, but net of interim relief. 1 Originally including Denmark, Finland, Ireland, the Netherlands, Norway, Sweden, Switzerland, and the United Kingdom. What should be tracked? 6. The starting point for tracking poverty-reducing spending is the full-fledged PRSP prepared by the authorities with input from various stakeholders. The Tanzania PRSP endorsed by the IDA and Fund Boards on November 30 and December 1, 2000, respectively identifies interventions in seven areas as poverty reducing (Box 2).

- 5 - Box 2. Tanzania: Poverty-reducing spending identified in the PRSP The PRSP, guided by the government s overarching National Poverty Eradication Strategy (NPES), seeks to reduce income poverty, improve human capabilities, both survival and social well-being, and mitigate extreme vulnerability among the poor. 1 In order to meet these goals, the PRSP identifies the following interventions for reducing poverty: (1) education (mostly at primary school level); (2) health (primary health care); (3) agriculture (research and extension); (4) rural roads; (5) water; (6) judiciary; and (7) HIV/AIDS. The costing of these priority interventions has been carried out in the context of ongoing Public Expenditure Review (PER) and Medium-Term Expenditure Framework (MTEF) exercises, as well as in sector-specific studies. With the objective of improving key social services, expenditure allocations for those priority sectors are being increased relative to outlays for other sectors starting in FY 2000/01. The PRSP also lists the following interventions: (1) abolishing primary school fees in order to ensure that children, especially from poor families, have access to primary school education; (2) catalyzing communities and other stakeholders to play a major role in poverty reduction through self-help schemes that entail construction of classrooms, teachers houses, health centers, dispensaries, water facilities, and rural roads; and (3) creating employment opportunities for the vulnerable. 1 The NPES is drawn from Vision 2025, which lays out the long-term development goals and perspectives of the government in the area of poverty reduction. 7. Prior to the formulation of the PRSP, Tanzania was already reorienting its spending toward pro-poor activities. This was done under the government-led Public Expenditure Review (PER) process, involving various stakeholders. 8. The PRSP Progress Report completed by the authorities in August 2001 reinforces the emphasis on programs identified as poverty-reducing in the PRSP. The 2001/02 budget includes the costs of programs in education, health, water, and HIV/AIDS. Priority programs in roads and judicial system could not be fully funded in the budget. The costing of the new agricultural strategy will be completed in the coming year. Tanzania has put in place a framework for monitoring the implementation of the PRSP. 9. The critical issue is whether existing budget classification can be mapped with poverty-reducing spending identified in the PRSP. It is equally important for PEM systems to facilitate tracking and reporting of public spending on these programs, as well as of changes in the overall public expenditure composition. What did the preliminary assessment for Tanzania show? 10. The preliminary assessment of Tanzania and other HIPCs reported to the two Boards earlier this year was based on 15 benchmarks. These benchmarks describe the basic requirements for a PEM system to track poverty-reducing spending; they are meant to

- 6 - identify both weaknesses in the systems that hinder tracking as well as PEM priority areas that need upgrading. Seven benchmarks relate to budget formulation, four each relate to budget execution and reporting. 11. In the preliminary assessment, Tanzania was identified as requiring substantial upgrading in its PEM capacity (Table 1). Relative to the benchmarks, the coverage of central government was viewed by the two staffs as reasonably comprehensive; however, difficulties in consolidating general government accounts remained. Differences between the original budget and actual outlays were noted. Data on expenditures financed by donors were incomplete. The existing budget classification was assessed as reasonably good. As far as budget execution is concerned, it was noted that Tanzania relies primarily on cash budgeting, although the country is developing an Integrated Financial Management System (IFMS). The internal audit system lacked trained staff. Public Expenditure Tracking Surveys had been carried out for education and health care spending. Fiscal and monetary accounts were not fully reconciled. On reporting, spending units had responded more than four weeks after the end of the period. The audited accounts had been presented to the National Assembly with a delay of more than six months. B. Assessment of Capacity to Track Poverty-Reducing Expenditures 12. Over the last three years, the Tanzanian government has implemented (and is continuing to do so) a comprehensive financial management reform program. Its aim is to strengthen the budget execution process, expenditure control, accountability and transparency. The implementation of IFMS, the introduction of a GFS-compliant chart of accounts for recurrent expenditures, and issuance of Local Purchase Orders (LPO) from IFMS to control expenditure commitments, are notable achievements. 13. Tanzania has increasingly devolved spending to local governments. This adds another dimension to tracking poverty-reducing spending in the country. About 70 percent of the recurrent budget and 17 percent of development budget for the seven priority sectors are transferred to local governments as block grants. However, about 73 percent of the block grants subvented for recurrent expenditures are for personal emoluments, which are centrally managed by the Department of Establishments. In addition, a part of the remaining block grants for other charges is managed centrally by the respective sector ministries. This is the case, for example, for procurement of drugs by the Ministry of Health for supply to local governments. Therefore, the magnitude of block grants to local governments per se overstates the difficulty in tracking poverty-reducing expenditure at the local level. 14. The mission assessed the public expenditure management capacity for tracking poverty-reducing spending against the fifteen benchmarks used in last year s deskbased assessment for HIPCs. For each of these benchmarks, an assessment and a corresponding action plan for the short and medium term are presented (Table 2).

- 7 - Table 1: Tanzania: Assessment of the Capacity to Track Poverty-Reducing Spending ASSESSMENT BUDGET MANAGEMENT Benchmark 1/ Desk Assessment Agreed Assessment (preliminary) COMPREHENSIVENESS 1. Budget reporting follows GFS definition of consolidated general government. A B B 2. Government activities are not funded through extrabudgetary sources to a significant degree. A A A 3. Budget outturn data (levels, functional allocation) are quite close to that of the original budget. B C B 4. Budget includes capital and current expenditure financed by donors. A B B CLASSIFICATION 5. Budget classified on an administrative, economic, functional basis. B B C 6. Poverty-related expenditure clearly identified in the budget. A A B PROJECTION 7. Multi-year expenditure projections integrated into the budget cycle. A B A INTERNAL CONTROL 8. Small stock of expenditure arrears; little accumulation of new arrears over past year. A B A 9. Internal audit is active. A B B 10. Tracking surveys supplement internal control. B A B RECONCILIATION 11. Fiscal and banking reconciliation undertaken routinely. A B A REPORTING 12. Internal budget reports from line ministries/treasury received within four weeks of the end of the relevant period. B C B 13. Functional classification is reflected in the in-year budget reports. A A B FINAL AUDITED ACCOUNTS 14. Closure of the accounts occurs within two months after the end of the fiscal year. A A A Reporting Execution Formulation 15. Audited account presented to the legislature within 12 months of the end of the fiscal year. B C C Source: Based on preliminary desk-assessment; and on discussions with the authorities. Notes: Shading: Meets benchmark 1/ A: fully met; B: almost fully met; C: not met. For some criteria, the benchmark is met with an "A "when full adherence is required; for other, a "B" is sufficient to meet the benchmark. For example, criterion 3 requires that budget outturn data are quite close to that of the original budget. This is met with "B", that is when outturn and budget are reasonably close.

- 8 - Table 2: Tanzania: Action Plan for Enhancing Capacity to Track Poverty-Reducing Spending SHORT-TERM MEASURES TIMING MEDIUM-TERM MEASURES TIMING BRIDGING MECHANISMS FOR TRACKING POVERTY-REDUCING EXPENDITURES Identify poverty-reducing expenditure in the budget: Identify budget lines for povertyreducing expenditures and circulate to respective MDAs. November 2001 1. Improve budget classification and account codes by: (1) completing application of the GFS-based economic classification to recurrent budget by extending it to volume III budget estimates for regions and subventions; and (2) introducing GFS-based economic classification to development budget. Ongoing, to be completed with 2002/03 budget Improve accounting for external disbursements: Set up a project accounting unit at ACGEN and improve accounting of donor disbursements, along the lines suggested in paragraph 13. November 2001- March 2002 2. Apply GFS-compliant functional classification to recurrent and development budget. Ongoing, to be completed with 2002/03 budget Use IFMS database for tracking poverty-reducing spending: Compile quarterly reports for poverty-reducing expenditures, starting with the current fiscal year, based on IFMS database. December 2001 3. Strengthen internal audit by: (1) defining more clearly roles and responsibilities of internal auditors (IAs); (2) including internal audit functions in IFMS and requiring IAs to authorize payments within IFMS; and (3) training IAs in use of computer-assisted audit. 2002-2003 Enforce reporting by local governments: Design a standard reporting format, in cooperation with PORALG, to be used for quarterly reports by LGs. December 2001 Issue a circular requiring submission of quarterly reports by LGs in the prescribed format with a deadline for submission 4. Strengthen external audit by: (1) building skills and capacity at C&AG office; (2) developing a separate IFMS audit module for statutory audit performed by C&AG; (3) training audit officials in computer-assisted audit techniques; and (4) enforcing compliance. December 2001 5. Consolidate operations of local governments and institutional funds with those of the central government; and compile fiscal accounts of general government. 2002-2003 2003-2004 Training and enforcement measures for timely submission of quality quarterly reports by LGs. January-March 2002 6. Strengthen the MTEF process, enhance realism of budget estimates and include projections for two forward fiscal years in budget estimates. Ongoing; to be completed with 2003/04 budget OTHER MEASURES Improve budget reporting system by: (1) transferring electronically the entire accounting database for the remaining five votes to the central database maintained by the ACGEN s office; and (2) designing reporting system that meets the requirements of MoF and spending agencies. December 2001- January 2002 Institute a reporting system on arrears for submission to ACGEN by Accounting Officers. December 2001 Finalize terms of reference for expenditure tracking surveys with the objective of launching them by March 2002. early 2002 Include a table providing budgetary details of poverty-reducing expenditures, starting from 2002/03 budget documents. June 2002

- 9 - Budget Formulation Benchmark 1: Budget reporting follows GFS definition of consolidated general government 15. In Tanzania, the general government consists of the central government, local authorities, executive agencies and other government institutions. Central government operations which are covered in the budget presented by the Minister of Finance are not consolidated with those carried out by local authorities. This consolidation is also impeded by differences in the fiscal year at central and local government levels. 3 Specific fiscal operations are also undertaken by various institutional funds (such as the Road Fund, to which the road toll an earmarked revenue is channeled; provident and pension funds; and the Education Fund) as well as other government agencies. Their operations are also not consolidated with those of the central government. 16. As required by the Public Finance Act 2001, the government presents annual estimates of revenue and expenditure in four volumes. Volume I includes government financial statement and revenue estimates; Volumes II and III contain recurrent expenditure for ministries, departments and agencies (MDAs), and regions, respectively; while estimates of development expenditures are included in Volume IV. Estimates for subventions/transfers to district councils do not follow economic classification as shown in Volume II. Assessment: This benchmark is not met since the budget only partially covers operations of the general government, as defined in the Government Finance Statistics (GFS) Manual (Table 1, last column). Action Plan: Efforts will need to be made over the next three years to consolidate the operations of the local governments and institutional funds with those of the central government. This will allow preparation of fiscal reports covering the entire general government. In the following sections of the Assessment, several other supporting actions have also been identified for strengthening the budget process, enhancing the comprehensiveness of the budget and improving the public expenditure management at the local government level. Benchmark 2: Government activities are not funded through extrabudgetary sources to a significant degree 17. As stated above, the government budget covers almost all activities of the central government. MDAs, which collect non-tax revenue in the form of fees and charges for their services, are required to remit these fees to the treasury. The budget estimates are prepared on a gross basis. 3 The fiscal year of local governments runs from January to December; while that of the central government runs from July to June.

- 10 - Assessment: This benchmark is met. Benchmark 3: Budget outturn data are quite close to the original budget 18. With the implementation of IFMS, the timeliness and quality of budget outturn data has improved considerably. Furthermore, in the past two years, significant improvements have been made in formulating a macroeconomic framework and in enhancing the reliability of budget estimates; areas that in the past have been prone to accumulation of arrears due to insufficient funding have been adequately provided for in the 2001/02 budget. Although authorized by law, supplementary budgets were not presented in the preceding two budget years. The reappropriations within and between different votes are small. Owing in part to a better control of expenditure commitments, accumulation of new domestic arrears during the last year has been small. With these improvements, the budget outturn for recurrent expenditures is now reasonably close to the original budget. Assessment: This benchmark is met. Action Plan: There is scope for further enhancement of the realism of budget estimates by: (1) improving the predictability of revenues and other inflows, particularly disbursements by donors; (2) assessing expenditure requirements based on proper costing; and (3) limiting the use of contingencies and disclosing anticipated use of contingency provision. Benchmark 4: Budget includes capital and current expenditure financed by donors 19. The coverage of donor-financed projects has considerably improved during the last year and is almost complete in the formulation of the budget. However, difficulties continue to be faced in accounting for donor disbursements related to these projects. The reason is that most donor disbursements are not received in the Treasury account at the central bank and released through Exchequer Issues. Furthermore, data on project expenditure financed by donors are not even recorded in the central expenditure database maintained by the Accountant General (ACGEN), in part due to separate project accounts with commercial banks. However, this does not apply to foreign loans disbursed through the budget. Grants provided in kind such as drugs and vehicles supplied by donors are not budgeted and accounted for. Donors have responded hesitantly to government s repeated pleas over the last three years to channel their support through the exchequer system. The provision of donor support through the Poverty Reduction Budget Support facility represents an important step towards better integration of donor funds into the government s budget management. Assessment: This benchmark is not met because a significant proportion of projects financed by donors are outside the exchequer and IFMS systems. Action Plan: In October 2001, the ACGEN sent a circular requesting donors to channel project funds through the exchequer system in order to facilitate recording of all expenditures. The circular also suggested alternative procedures for accounting for all project transactions.

- 11 - In addition to the steps envisaged in the ACGEN s circular, a separate project accounting unit could be set up in the office of ACGEN. This unit could: 4 (1) undertake an inventory of all donor-financed projects by liaising with donors as well as MDAs, reconcile information from the two sources, establish a project database and update it regularly; (2) liaise with the MDAs and Budget Department to ensure that all projects are included in the development budget estimates; (3) request donors to provide quarterly reports on project disbursements to ACGEN with copies to MDAs as well as project managers; (4) ensure that all transactions have been properly recorded and accounted for as described in the ACGEN s circular; (5) input manually or electronically into the FILMS central database accounting project data not on IFMS; and (6) institute quarterly reports on project expenditures; the project unit should reconcile these reports with the quarterly disbursement reports provided by donors and ensure that disbursements have been fully accounted for. Benchmark 5: Budget is classified on an administrative, economic, and functional basis 20. The budget classification and coding structure for the recurrent and development budgets are not uniform. GFS-based economic classification has been adopted for the recurrent expenditure estimates commencing from FY 2000/01. However, the development budget does not follow this classification. In addition, both the recurrent and development budget follow different classifications: the recurrent budget is classified in terms of programs, sub-votes and object/economic type, while the development budget is classified in terms of sub-votes and projects. GFS-compliant functional classification has been developed, but has not been applied by IFMS. The adoption of different budget classifications and account codes hinders a consolidated analysis of poverty-reducing spending in recurrent and development budgets. Finally, the program classification used in the recurrent budget is not appropriate for effective performance budgeting. One of the constraints to extend GFS-based economic classification to the development budget is the fact that about 90 percent of it is donor funded, with only a small share of this funding passing through the exchequer system; and many donors do not yet provide sufficiently detailed information on their support. Assessment: This benchmark is not met. Action Plan: The budget classification and account codes need improvement by: (1) extending GFS-based economic classification to the recurrent budget in Volume III (estimates for regions and subventions); (2) implementing GFS-based economic classification to development budget; and (3) developing a standard and uniform chart of accounts and account codes for recording and accounting government transactions under IFMS. 4 Some of these tasks should be coordinated with the Budget Division and the External Finance Department at the Ministry of Finance.

- 12 - Benchmark 6: Poverty-reducing expenditures are clearly identified in the budget Broad details of poverty-reducing spending identified in the PRSP are outlined in Box 1. It is difficult to identify budget lines relating to this spending in the 2001/02 budget. For example, estimates for basic education listed as poverty-reducing are scattered across the budget. Assessment: The benchmark is not met. Action Plan: The first crucial step is to identify poverty-related expenditure in budget estimates consistent with the PRSP. A table indicating votes, subvotes, and item codes in recurrent and development budgets covering seven priority areas (in the PRSP) has been prepared for 2001/02 by MoF officials (Tables 3-5). These tables could be refined in consultation with spending agencies, including project managers. This exercise would also help ensuring full exchequer releases for genuine poverty-reducing spending and for proper accounting, monitoring and reporting. A summary table along these lines can be included in the budget, effective 2002/03. Benchmark 7: Multi-year expenditure projections are integrated into the budget cycle 21. MTEF was introduced in 1997/98 as a part of the PER exercise and covers a threeyear period. The resource envelope established under MTEF is used for setting sectoral budget allocations. The budget guidelines provide the macroeconomic framework; the key spending priorities and programs; and the proposed allocation of resources over the medium term. Draft revenue and expenditure estimates are presented for the current fiscal year, along with the approved estimates for the previous fiscal year and actual results for the fiscal year prior to the previous one. However, the budget documents do not include resource allocations for two forward fiscal years. Assessment: This benchmark is met. Action Plan: The budget process would benefit from: (1) further strengthening of MTEF; (2) including projections for two forward fiscal years in budget estimates, and (3) using the forward year estimates as a basis for drafting annual estimates. Budget Execution Benchmark 8: Small stock of expenditure arrears, little accumulation of new arrears over past year 22. Expenditure arrears overdue unpaid bills used to be a problem until recently. Expenditure control procedures have been strengthened with the introduction of IFMS. A commitment control system (CCS) was introduced effective July 1, 2001. The issuance of a local purchase order (LPO) generated by IFMS is now required for purchasing goods and services. The ACGEN is required to monitor expenditure commitments with the help of IFMS.

- 13 - Table 3. Tanzania: Poverty-Reducing Expenditure, FY 2001/2002

- 14 - Table 4. Tanzania: Disaggregation of Recurrent Poverty-Reducing Expenditure

- 15 - Table 5. Tanzania: Disaggregation of Development Poverty-Reducing Expenditure, FY 2001/2001

- 16-23. The stock of expenditure arrears accumulated during July 1998-December 2000 which has been verified by external auditors amounted to about 1 percent of GDP. 5 However, data for arrears from January 2001 onwards are not available, although the authorities believe only a small amount of fresh arrears was accumulated. Assessment: This benchmark is met. Action Plan: With the introduction of the CCS and LPO-based system, the problem of overexpenditures including arrears is likely to be eliminated. However, it is important to continue to monitor arrears, particular at subtreasury level. Although IFMS system generates details on recorded commitments, it cannot reveal commitments or claims not recorded due to nonavailability of funds. In the absence of a formal system for monitoring payment arrears on a regular basis, it is therefore necessary to prescribe a quarterly report to collect information on accumulated unpaid bills from all Accounting Officers. A circular to this effect could be issued by ACGEN. Benchmark 9: Internal audit is active 24. An internal audit function exists throughout MDAs. At present, internal audit is concentrated on voucher audit, without any focus on procedures and controls. It does not make sufficient use of IFMS, due to lack of training. In general, internal audit has remained weak due to: (1) inadequate technical skills and weak managerial capacity in the finance arm of MDAs; (2) lack of use of internal audit manuals and weak oversight by the ACGEN; and (3) insufficient enforcement of punitive measures. Assessment: The benchmark is not met. Action Plan: The internal audit units in MDAs and the Financial Management unit in ACGEN need to be strengthened by: (1) defining more clearly their roles, responsibilities, and inter-relationship vis-à-vis each other; (2) incorporating internal audit functions into IFMS and requiring internal auditors to authorize payments within IFMS; (3) training internal auditors in use of computer-assisted audit techniques; (4) requiring a monthly performance report from Internal Auditors and Financial Controllers to Accounting Officers, with copy to MoF, ACGEN, and Controller and Auditor General (C&AG). Such a report should also include cases where payments were authorized despite serious objections from internal audit, or when high value vouchers were not shown to audit; and (5) following-up actions on reports of internal auditors and financial controllers. 5 The European Union has partly financed the clearance of pre-june 1998 domestic arrears. Arrears were temporarily accumulated during the period July-December 2000 (equivalent to about 0.3 percent of GDP). These were mostly related to financial difficulties of the utility sector, where the electricity company was not paid by the water company and thus was unable to service its external debt which was guaranteed by the government. However, these arrears were settled during the fiscal year 2000/01 (clearance of arrears amounted to 0.5 percent of GDP).

- 17 - Benchmark 10: Expenditure tracking surveys supplement internal control 25. Two expenditure tracking surveys on health and education were conducted last fiscal year in the context of PER. These surveys have demonstrated that resources did not flow in full to their intended end-users. The MoF plans to mainstream the use of expenditure tracking surveys in monitoring budget execution; draft terms of reference for consultants for undertaking the surveys are being prepared. The authorities intend to launch these surveys in early 2002. Meanwhile, some donors are also supporting similar surveys for example, the European Union has been financing quarterly tracking surveys on the provision of educational material since July 2000 (Table 6). Assessment: This benchmark is met. Action Plan: It is important that consultants be engaged by early 2002 to implement expenditure tracking surveys, particularly in areas identified as poverty reducing. Budget Reporting Benchmark 11: Fiscal and banking reconciliation is undertaken routinely 26. Since the government s consolidation of its cash balances into a single bank account at the central bank, checks are drawn on this account and printed centrally by ACGEN for most MDAs. However, a significant number of project accounts are maintained at commercial banks and donors funds are not channeled through the central bank. The reconciliation of bank accounts with the Bank of Tanzania which was subject to long delays until last year is now performed on a routine basis by ACGEN, via computerized records kept under the IFMS. Project accounts with commercial banks are reconciled monthly. Differences between the above-the-line budget records and below-the-line financing of the deficit have been contained significantly. Assessment: This benchmark is met. Benchmark 12: Internal budget reports from line ministries/treasury are received within four weeks of the end of the relevant period 27. With the introduction of IFMS, internal budget reports on exchequer issues, commitments and payments can be generated in real time for MDAs, which are online and connected to the central database maintained by ACGEN. However, five Votes (State House and Defense) are using IFMS on a stand-alone basis and off-line. The data for these votes is provided in a diskette. This data is then merged in a spreadsheet with data for the other MDAs. Assessment: This benchmark is met.

- 18 - Table 6. Tanzania: Existing and Planned Assistance in Public Expenditure Management Donor Project Title Objective Scope Implementing Agency Sweden Integrated Financial Management and Accountability Project (IFMAP). IMF Budget classification and management Sweden Controller and Auditor General s Office UK DFID Switzerland (SECO) Training of officials on the Public Finance Act. Strengthening macrofiscal analysis at the Ministry of Finance. Norway MACMOD (Macroeconomic Planning Model). Third phase EU/EDF PRBS-related technical assistance EU/EDF Strengthening of External Resource Management Develop integrated budget systems for planning and accounting. Develop GFS-based classification Provide training and computerization Improve understanding of GoT officials of the newly promulgated Public Finance Act and pursuant regulations. Strengthen the capacity of Ministry of Finance to conduct macro-fiscal analysis for fiscal management. Establish and operate a macroeconomic model. Facilitate channeling of budget support through PRBS Establish IFMS system to capture external resource flows to the Government Central government, sub-treasuries, and local governments. Accountant General s Office Start date Duration Since 1997; September 2001 28 months Central government MoF 1998 24 months Central government Controller and Auditor General s Training of 1300 officials in all departments, regions and districts. Training as well as establishment of a database for analysis. Formulating projections for budget preparation and improving capacity to generate them. Assist in management and monitoring of PRBS funds. Central government, sub-treasuries, and local governments. Office Accountant General s Office International Monetary Fund (IMF) President's Office, Planning Commission Ministry of Finance and others MOF External Finance Division 2002 Under discussion July 2001 4 months May 2000 24 months January 2000 September 2001 24 months 24 months July 2001 12 months

- 19 - Table 6. Tanzania (continued): Existing and Planned Assistance in Public Expenditure Management Donor Project Title Objective Scope Implementing Agency EU/EDF Strengthening Government s Financial Management and Internal Audit Capacity EU/EDF Interim Textbook Program EU/EDF Local Government Reform Program Ireland Aid District Development Programs: Leadership and Administration Netherlands, GTZ, UNDP, and other donors World Bank and other donors Strengthening of financial management at the local level Public Expenditure Review To improve the Treasury s monitoring of financial management in sector ministries, particularly expenditure commitments. Provision of educational materials to primary schools in all LGAs through centralized and (pilot) decentralized procurement. Establishment of sector-based financial management units Monitoring of program expenditure and physical distribution of education materials at 56 LGAs and 112 Accountant General s Office MOF with Ministry of Education, and PORALG Start date Duration February 2000 29 months June 2000 18 months schools. Improved delivery of basic Studies on fiscal PORALG, February Ongoing public services via LGAs under decentralization, staff assisted by 1998-June decentralized management. and finance PriceWater 2000. management HouseCooper January 2001 To improve district Introduction of Four District 2000 Ongoing administration and IFMS. Introduction Councils (first phase accountability. of performance 3 years) management systems. Staff training. Support for internal audit. Strengthening of Public Expenditure Management and Annual Review of Expenditure Trends and Public Expenditure Management Issues Ministry of Finance 1998 Ongoing (annual exercise)

- 20 - Donor Project Title Objective Scope Implementing Agency DFID and World Bank Country Financial Accountability Assessment World Bank Accountability, Transparency, and Integrity Program World Bank IDF grants for Policy Planning, Implementation, and Coordination Word Bank Consultancy to National Board of Accountants and Auditors Word Bank Urban Sector Rehabilitation Project World Bank Financial and Legal Management Upgrading Project Assessment and action plan for strengthening financial accountability Support Tanzania s Program of Good Governance, including strengthening financial accountability Ministry of Finance Office of the Chief Secretary/ Head of Public Service Start date Duration February 2001 August 2001 2002 2007 Capacity building for local treasuries; technical assistance and training to valuation departments Upgrade Accounting and Auditing Capacities in Private and Public Sector Urban Councils President s Office Regional Administration and Local Government OCAG, NBBA, IFM, IAA 1997 2004 1992 2000 Sources: European Union and Fund staff.

- 21 - Action Plan: The internal budget reporting system could be made more effective by: (1) transferring electronically the entire accounting database for the remaining five votes to the central database maintained by the ACGEN s office; (2) improving the reporting system that meets the requirements of MoF and spending agencies; and (3) promoting maximum use of IFMS-generated reports for budget management. Benchmark 13: Functional classification is reflected in the in-year budget reports 28. Functional classification has not been adopted by IFMS for disaggregating expenditure data. Therefore, it has not been possible to compile in-year budget reports on a functional basis. Assessment: This benchmark is not met. Action Plan: As noted earlier, the GFS-compliant functional classification although developed is not being used. The application of IFMS could be made more effective by adopting GFS-based functional classification and by generating relevant reports for economic analysis. Benchmark 14: Closure of the accounts is within two months after the end of the fiscal year 29. In accordance with the Public Finance Act, each Accounting Officer is required to submit prescribed accounts and financial statements to ACGEN within four months after the end of each financial year. The ACGEN is required to submit annual accounts to the C&AG within six months after the end of each financial year. Most Accounting Officers are able to submit their accounts within the limits prescribed by law. Assessment: This benchmark can be viewed as having been met, since the accounts are submitted within the period prescribed by law. Action Plan: With the introduction of IFMS, the formal closure of accounts can be expedited and the lag in their submission reduced to three to four months. Accounting Officers and ACGEN should be able to expedite submission of annual accounts and financial statements to the C&AG so that the latter can complete audit and report to the National Assembly on time. Benchmark 15: Audited accounts are presented to the legislature within 12 months of the end of the fiscal year. 30. The 2001 Public Finance Act has reaffirmed the principle that audited annual accounts be presented to the National Assembly within nine months from the end of the financial year. Over the past three years, the Office of the Controller and Auditor General has been strengthened and it has received increased budget allocations. This has led to a significant improvement in the quality of audit as well as reduction in the delays in the preparation of the audited accounts of the central government. There have also been some improvements in the timeliness of auditing the accounts of local authorities. The last Audit

- 22 - Report, for FY 1998/99, was published in October 2000, that is, 16 months after the end of the fiscal year. Assessment: This benchmark is not met. Action Plan: The statutory audit needs to be further strengthened by: (1) providing resources to the C&AG for capacity building, through additional staff, skill upgrading and acquisition of equipment; (2) training audit personnel in computer use, including application of computer-assisted audit techniques; (3) developing a separate audit module for IFMS for statutory audit performed by C&AG; (4) improving the timeliness and quality of audit reports by applying modern audit techniques, as well as reporting to the National Assembly the results of value-for-money assessments; (5) empowering the oversight of parliamentary committees over the executive; (6) implementing the recommendations of the Public Accounts Committee and the Local Government Accounts Committee; and (7) initiating disciplinary proceedings against those responsible for non-compliance and financial misconduct. Summary of the assessment of the expenditure tracking system 31. As noted earlier, the authorities have implemented a number of measures for improving PEM during the last three years. A continuation of these measures, as well as others identified below, would enable the authorities to track and report poverty-reducing expenditures. 32. This evaluation of Tanzania s PEM system is better than the preliminary assessment carried out by the staffs of the World Bank and IMF at headquarters. The new assessment suggests that Budget formulation is quite good; however, further improvements are required in budget classification and account codes and in accounting donors disbursements; Budget execution procedures meet most standards except in case of internal audit; Budget reporting is relatively satisfactory; however, improvements are required in external audit and application of GFS-compliant functional classification. Bridging mechanism for tracking poverty-reducing expenditures 33. This assessment lists various actions for strengthening Tanzania s PEM system (Table 2). Some of these actions could be implemented in the near term. In particular, a set of short-term actions are referred to as bridging mechanisms as they bridge a country s evolution toward more advanced PEM systems. In the case of Tanzania, four bridging mechanisms could be implemented. Identify poverty reducing expenditure in the budget. These expenditures are not clearly identifiable in budget estimates. The MoF is making efforts to relate priority

- 23 - sector spending in the PRSP with budget estimates. The first critical bridging mechanism is to refine these tables in consultation with concerned MDAs and agencies (Tables 3-5). The budget documents should include a table providing details of poverty-reducing expenditures; such a table can be included in the 2002/03 budget documents. Improve accounting of donor disbursements. It is important that all donor-financed projects be accounted for as suggested earlier, and on this basis, the MoF should begin preparing quarterly reports of project expenditures reconciled with donor disbursements. This will allow a comprehensive recording of donor-funded expenditures, including those in poverty-reducing activities. Use IFMS database for tracking poverty-reducing spending. Poverty-reducing expenditures are included in: (1) estimates of concerned MDAs, and (2) subventions to local authorities. Expenditures incurred by MDAs are already available under the IFMS database. Subventions to local authorities could also be monitored through cash releases recorded by IFMS. This would provide an interim mechanism for tracking poverty-reducing spending until quarterly expenditure reports are submitted by local governments. Enforce the quarterly reporting by local governments. The government has recently launched an initiative requiring local governments to submit quarterly reports on revenue and expenditure to MoF. However, a circular detailing the required reporting format is yet to be issued. MoF could design a simple reporting format in consultation with PORALG. Local governments should submit quarterly reports within fifteen days after the end of the quarter to respective Regional Administration Secretariats (RAS), with a copy to ACGEN and PORALG. RAS should also be made responsible for ensuring the timeliness and quality of these reports. The office of the ACGEN will then be able to record expenditure data against subventions on IFMS. This information should also be released to the public. C. Ongoing and Planned Assistance to Upgrade Capacity to Track Poverty-Reducing Expenditures 34. Donors as well as multilateral institutions have provided substantial technical assistance in public expenditure management to Tanzania. Many programs are ongoing and new ones are planned (Table 6). The annual Public Expenditure Review process, led by government with broad donor and civil society participation, provides the overall framework for monitoring progress in the implementation of public expenditure management reforms and serves as the principal coordination mechanisms for technical assistance in these areas. In addition, during FY 2000/2001, a participatory Country Financial Accountability Assessment (CFAA) was conducted with the support of the World Bank and DFID. It provides a comprehensive assessment of the financial accountability processes in both public and private sectors. DFID and World Bank are currently assisting the authorities in

- 24 - formulating an action plan for implementing the recommendations stemming from CFAA. It is hoped that some of the above-noted actions will be incorporated in this plan. 35. The review of the ongoing and planned technical assistance compared to the Action Plan reveals no major areas of assistance remain uncovered by donors. In particular: The implementation of bridging mechanisms for tracking poverty-reducing expenditures is already underway. This can be successfully completed by the authorities on their own. On the other short-term measures, technical assistance is being provided to strengthen the IFMS; expenditure tracking surveys are already part of the authorities and donors work plan. Most medium-term measures proposed in the Action Plan are receiving or are planned to receive technical assistance from various donors. The areas not covered by donors are internal audit and consolidation of local government operations with those of central government. The authorities may wish to include these areas in the action plan for the implementation of CFAA.