Scoping the introduction of a Medium Term Expenditure Framework in Nigerian State Governments. Discussion Draft

Transcription

1 Scoping the introduction of a Medium Term Expenditure Framework in Nigerian State Governments Discussion Draft September 2009

2 The opinions expressed in this report are those of the authors and do not necessarily represent the views of the Department for International Development

3 Content List Abbreviations and Acronyms Executive Summary...1 Should an MTEF be introduced in a weak budget environment?... 1 Types of MTEFs...2 Next steps...2 Section One Background PFM in the Nigerian States...1 The case for State level MTEFs...2 Section Two How far does current State budgetting involve a multi-year dimension?.. 5 Progress by State governments... 5 Good plans may not mean good budgets... 6 Section Three Challenges in introducing State level MTEFs... 9 Predictability of State resource envelopes... 9 Payments from the Federation Account Internally generated State revenues Net State borrowing/lending A sound budget classification Political constraints on introducing an MTEF Lack of budget credibility undermines an MTEF Section Four Using an MTEF to improve credibility of the annual budget Multi-year forward estimates? Possible transition strategy States capacity to forward plan Concluding comment Section Five Approaches to MTEF design Indonesian Option: adding an out year to existing budget preparation processes Making the Indonesian approach work Philippines Option: managing changes in spending in a multi-year context... 22

4 Comparison of the two approaches Next steps Appendix One Implications of the Excess Crude Account (ECA) for State MTEFs MTEF implications... 27

5 Abbreviations and Acronyms ECA ESSPIN FCT FE GIFMIS IGR J-SEEDS K-SEEDS KADSEEDS LASEEDS LSG MDAs MDGs MTBF MTEF MTFF MTSS NEEDS O&M OAGF PEFA PEMFAR PFM SEEDS SPARC Excess Crude Account Education Sector Support Programme in Nigeria Federal Capital Territory Forward estimate Government integrated financial management information system Internally Generated Revenue Jigawa State Economic Empowerment and Development Strategy Kano State Economic Empowerment and Development Strategy Kaduna State Economic Empowerment and Development Strategy Lagos State Economic Empowerment and Development Strategy Lagos State Government Ministries, Departments and Agencies Millennium Development Goals Medium Term Budget Framework Medium Term Expenditure Framework Medium Term Fiscal Framework Medium Term Sector Strategy National Economic Empowerment and Development Strategy Operation and maintenance Office of Accountant General of Federation Public Expenditure and Financial Accountability Public Expenditure Management and Financial Accountability Review Public Financial Management State Economic Empowerment and Development Strategy State Partnership for Accountability, Responsiveness and Capability

6 SSG TA VAT Secretary of the State Government Technical Assistance Value Added Tax

7 Executive Summary Should an MTEF be introduced in a weak budget environment? The Nigerian Fiscal Responsibility Act recommends the use of a medium term expenditure framework (MTEF) to assist budgeting by the states. However, in the broader international environment there is currently discussion of whether the relatively sophisticated processes required to integrate MTEFs with annual budget preparation make unrealistic demands on the capacity of developing countries, particularly at sub-national levels. This report identifies the capacity for, and possible impediments to, the introduction and implementation of a sustainable MTEF within Nigerian State Governments. 1 There is abundant evidence in Public Expenditure and Financial Accountability (PEFA) surveys and a major World Bank review that the financial management of Nigerian states is very weak. The level of borrowing and grants assumed for budget financing is frequently driven by unrealistic spending aspirations, resulting in the failure of budget outturns to meet approved plans by a wide margin. However, an MTEF improves budget outcomes through the improvement it brings to budget plans, and if budget outturns differ markedly from the plan the benefits of introducing an MTEF may be lost. At worst, attempting to introduce an MTEF may be seen as a diversion from getting the budget basics right and at best may lead to a parallel process with little positive impact on budget outcomes. This report takes a more positive view, arguing that the design of the MTEF process itself can contribute to the increased realism of annual budgets. By providing a multi-year projection of budget financing needs, an MTEF may enable a longer lead time for organising finance for projects contained in the forward estimates (FEs), whether through release of funds from the Federation Account, improved state tax administration or a borrowing program. Where state governments are tempted to announce unrealistically large capital programs in the annual budget, the introduction of, for example, a two year budget frame allows project commencements to be distributed between the new budget year and the forward estimates for the subsequent year(s). The opportunity for the state government to announce in the budget speech projects for which disbursement begins in (say) eighteen months ahead in the forward estimates may contribute to a more credible budget plan. 1 See the terms of reference, which note that The assignment should draw on comparable international experiences in terms of the practicality, implementability and sustainability of such an approach.

8 Improved consistency between the annual budget plan and outturn may in turn open the way for realising the more conventionally recognised benefit of a MTEF in strengthening the link between budget composition and medium term planning strategies. Types of MTEFs Based on international experience this report suggests two alternative designs for a state level MTEF. Under the first (used at the sub-national level in Indonesia) local service delivery agencies (Ministries, Departments and Agencies (MDAs)) submit budget requests for funds not only for the next budget but the subsequent out-year(s). However, Indonesian experience suggests that local service delivery agencies may respond with inflated and unusable out-year bids. To be effective these out-year requests need tighter quality control. In Nigeria s case this could be achieved by basing out-year requests on programs and proposed initiatives contained in the relevant state Medium Term Sector Strategy (MTSS). Currently spending proposals in the MTSS are not constrained by multi-year resource envelopes, and indicative sector ceilings may be needed to filter MTSS proposals into the MTEF. 2 The Lagos State Government has prepared guidelines for preparing MTSSs which may meet this need, but involve large amounts of information. Given the tight timelines for budget preparation, coping with the amount of information and ensuring quality control would be a challenge. An alternative, and more data efficient approach to introducing an MTEF could be one based on the recent Philippines experience. Under this alternative approach rolling forward estimates are held in a central budget office database (using spreadsheets) rather than MDAs preparing a new set of FEs each year (as under the Indonesian approach). The first out year estimates are used as the starting point for the next annual budget preparation, (following revision/renegotiation for technical factors) and the detailed budget requests from MDAs focus on proposed new initiatives to vary their forward estimates rather than reapplying for the funding already contained in the FEs. Central management of the FEs ensures greater quality control and allows a greater focus in the budget preparation cycle on policy choices rather than funding of existing service levels. Next steps The Indonesian and Philippines approaches are compared in the body of this report. It would be possible in principle for The State Partnership for Accountability, Responsiveness and Capability (SPARC) to support both approaches in different states, responding to local preferences and capacities. In Lagos State (already apparently embarked on developing FEs along Indonesian lines) this could be combined with very tight MTSS links and assistance to develop MTSSs. This will help avoid the shortcomings in the Indonesian experience. For another state technical assistance could be provided to the ministry of planning and budget(or equivalent) to prepare a set of rolling FEs of existing commitments along Philippines lines, with MDAs bidding for variations on the FEs in each budget round. 3 This could be a preferred approach in states prone to unrealistically large capital budgets, since it emphasizes the future resource envelope and the lock-in implied by existing policy commitments, rather than (as in the Indonesian approach) a multi-year new spending proposal which is less clearly linked to a resource envelope. As experience is gained, the two approaches could be blended if so desired. The adoption of different approaches could also trial different numbers of out-years for which FEs are prepared (one, two or three out-years beyond the next budget year). 2 A key reason why most countries choose twelve months for their budget period rather than preparing two year or multi-year budgets is the difficulty of forecasting the resource envelope beyond the eighteen months following commencement of budget preparation (budget preparation plus the budget year itself comprises a minimum of eighteen months). 3 There would probably be interest outside Nigeria in a side by side trial of the two approaches.

9 In implementing such approaches, three issues would require early attention. The Education Sector Support Programme in Nigeria (ESSPIN) is currently providing Technical Assistance (TA) to the Lagos State Government on re-design (simplification) of the MTSS in the education context. This re-design should be coordinated with the strategy for a future Lagos State MTEF to ensure that each MTSS meets the needs not only of a stand-alone sector plan but a state MTEF based on hard sector data. An assessment will need to be made of the challenges in extending the resulting (hopefully best practice) MTSS formats to MDAs in other states. Attention will need to be focused on how the ministries of finance and planning in the states handle the required (big) increase in data associated with multi year expenditure estimates, and proposed new policy initiatives identified in the MTSSs. This may be assisted by the use of modern off the shelf budget preparation software which has the functionality required to integrate forward estimates with budget preparation (particularly through the use of change packages derived from MTSSs and budget version tracking). However, adopting an MTEF approach is far from a panacea for unrealistic budget preparation. A key requirement for states will be a demonstrated political commitment by the government, as well as key MDAs in the state, to use the MTEF to implement a more disciplined approach to preparing the state budget estimates.

10

11 Section One Background PFM in the Nigerian States This report forms part of a scoping exercise to identify the capacity and possible impediments to the introduction and implementation of a sustainable medium term budget framework within Nigerian State Governments, and the identification of possible risks and constraints to such an approach. The objective of the report is to identify a possible strategic approach to developing and implementing a comprehensive and cohesive MTEF with State Government, recognising the implications for such and approach and possible limitations, in particular capacity. While there is considerable experience with the introduction of MTEFs at the national level of government, experience at the sub-national level is more limited, and generally restricted to state level governments in wealthier economies. This reflects several factors: At the national level the success of MTEFs has been mixed, due partly to the relatively sophisticated budget planning and preparation processes which they require, including for management of additional data. This may have led to caution in introducing these processes in smaller government units with less depth in technical budgeting capacity; The benefits of an MTEF are fully realised only by the parallel preparation of medium term fiscal projections and a medium term budget framework (MTBF). 4 While there is wide diversity in the financing arrangements for sub-national governments, the majority (generally those without exportable natural resources) are heavily reliant on transfers from the national government (vertical fiscal imbalance), for which the specific purpose (discretionary) component may be difficult to predict. The Nigerian states, however, present some special issues. The 36 states and local government authorities receive approximately half of net federation account proceeds and 85% of the value- added tax (VAT) pool and have constitutional responsibility for delivering core public services. Achievement of national poverty and infrastructure goals therefore depends heavily on the quality of state spending. The combined budget activities of the states also influence macroeconomic and fiscal stability goals. Moreover, some Nigerian state level governments are themselves larger than many sovereign states. 4 A medium term fiscal framework (MTFF) refers to projections of the fiscal resource envelope for future budgets, based on projections of revenues, grants and deficit funding. A medium term expenditure framework (MTEF) refers to projections of future budget spending by sector/line item (either based on existing policy settings or including an allowance for new policy). A medium term budget framework (MTBF) refers to forward estimates of sector ceilings consistent with the medium term fiscal framework and government planning documents. Consistent with widespread usage, the acronym MTEF is used in this report to cover all three, unless otherwise specified. 1

12 Reflecting this, the Nigerian Fiscal Responsibility Act recommends minimum fiscal performance standards for all levels of government. 5 To achieve this the Act proposes use of an MTEF to assist annual budgeting to achieve medium term fiscal and planning objectives. 6 However, as with sub-national governments in larger countries, Nigerian states vary greatly in wealth, urban versus rural character and size and density of population, and there is a virtual certainty that progress toward budgeting in a medium term framework will be uneven across states. The case for State level MTEFs It is difficult to reject the logic of shifting the preparation of an annual budget into a medium term expenditure framework. This is for at least three reasons. First, annual budgets are often driven by historical allocations rather than responding flexibly to the government s planning priorities. 7 Many governments have trouble connecting their planning and budgeting processes. Planning occurs outside of the necessarily realistic resource envelope used to prepare an annual budget, and budgeting suffers from the rigidities of a short time horizon for reallocating resources in line with multi-year plans. 8 Nigerian states need an MTEF to ensure that successive annual budgets program achievement of policy goals in a way which is consistent with annual fiscal constraints, and to integrate donor spending with domestic resources. 9 Box 1: The case for an MTEF Facilitates a more realistic prioritization of expenditure by comparing multi-year spending plans with the medium term resource envelope (Planning Ministries). Encourages more efficient inter-temporal planning of budget requests by providing greater transparency to spending agencies about likely future resources (Sectoral Ministries). Increases control of advance commitments of public money by constraining budget appropriation and execution in future years to levels consistent with the Government s medium-term fiscal and sectoral objectives (Finance Ministries). Second, a further purpose of MTEFs is to take account of the medium term expenditure implications of decisions made in preparing annual budgets, and any necessary fiscal adjustments to be programmed over a number of years. Reflecting this, the MTEF is ideally suited to implementing major changes in budget composition, such as the Lagos State Great Leap budget in 2008, which boosted capital relative to recurrent spending. In the case of multi-year projects an MTEF helps ensure that 5 The Act provides incentives to encourage States and Local Government to pass similar fiscal responsibility legislation. 6 However, the states enjoy considerable autonomy in managing their spending, and have independently developed public financial management (PFM) systems. There is no common PFM reform framework and responsibility for PFM reform rests with individual states, although reforms undertaken at the Federal level may provide a template of convenience. There is limited constitutional authority for Federal Government leadership in establishing a common reform model for introducing an MTEF, or standards of multi-year fiscal management. 7 State planning priorities are contained in such documents as Nigeria: Millennium Development Goals (MDGs) Report 2005, the Lagos State Economic Empowerment and Development Strategy (LASEEDS) and the Lagos Economic Advancement Programme Ten Point Agenda. 8 Most states have not followed the Federal precedent of combining their Ministries of Finance and Economic Planning and Budget, increasing the benefits from introducing an MTEF but also the difficulty. 9 National Planning Commission, State Economic Empowerment and Development Strategy (SEEDS) Manual A Framework Guide for Development Planning, From Strategy to Action, See Main Report, p

13 provision is fully made for project completion. 10 State budgets generally have a high ratio of capital to recurrent spending and an MTEF will also help ensure that the future operation and maintenance spending required by the capital program is in fact fundable. Conversely, where the resource envelope available to the government is static, spending cuts needed to finance new capital priorities can be phased in over several budgets, based on the adjustment of forward estimates to reflect fiscal realities. This also applies to recurrent spending. Should a state government move to rationalize staff structures and adopt a medium term pay reform strategy, an MTEF will help determine an affordable multi-year program of severance costs and improved salaries. Third, recent state PEFA reports (discussed below), and a key World Bank study, suggest that at present state governments develop their budget spending plans with more concern for the announcement effects of new projects than the ability to finance them through the budget. 11 Subsequent unavailability of cash causes disbursement approvals to queue in state treasuries, even though MDAs are ready to proceed, and cash rationing typically undermines execution of the budget. Linked to this, the World Bank suggests that MDAs request funding for new projects in order to increase their power and influence rather than to implement a clear planning agenda. States routinely and repeatedly admit some projects into the budget yearly without executing them. They also do not subject new projects to a proper appraisal and feasibility study before admitting them into the budget. 12 Introduction of state level MTEFs may help resolve this problem of over-bidding by state MDAs and over-budgeting by state governments. Forward estimates of expenditures which are consistent with state resource envelopes (updated at the commencement of each budget preparation cycle) can be used as indicative ceilings for each MDA s budget request, with any bid in excess of the forward estimates needing justification in terms of the specific initiatives proposed, their cost detail and their contribution to achieving planning objectives. Of the three benefits from introducing an MTEF discussed in this section, the third (reduction of over-budgeting ) is a pre-requisite to achieving the other two. If budget outturns chronically undershoot budget plans the benefits of introducing an MTEF to improve budget planning are likely to be small. This leads to a further point. If the specific MTEF architecture best suited to reducing over-budgeting differs from the best MTEF architecture for linking budget preparation to state plans, initial support for introducing an MTEF should focus on the first architecture, with support for introducing the second only when budget outturns successfully correspond to budget plans. This issue of alternative MTEF architectures is discussed later in the report. 10 This assumes projects are correctly costed in the planning stage. 11 Public Expenditure Management and Financial Accountability Review (PEMFAR), Main Report p A footnote on p. 107 states Evidence entrenched planning routines is not hard to find in states. They range from the weak link between policy and plans/budgets, lack of sectoral strategies to underline the budget, improper definition of the capital budget to include items that have no investment or developmental values, poor record of implementation of the budget, to ineffective monitoring, evaluation and reporting mechanisms to guide budget implementation. 12 PEMFAR, p

14 4

15 Section Two How far does current State budgetting involve a multi-year dimension? The 2007 Fiscal Responsibility Act provides a model at the national level for introducing state level MTEFs. Part II of the Act requires an MTEF for the next three financial years to be considered by the National Assembly. The MTEF is required to contain: A macroeconomic framework; A fiscal strategy paper containing financing social and development strategies; A quantitative framework for aggregate revenue and expenditure, including minimum capital expenditure. Annual budgets prepared during the MTEF period are to be consistent with this medium term framework. Clause 17 of the Act states States and Local Governments which so desire shall be assisted by the Federal Government to manage their fiscal affairs within the medium term framework, implying that this is also seen as a model for the state level. Although not featured in the Fiscal Responsibility Act, a key part of the multi-year approach at Federal level is the preparation of eighteen Medium Term Sector Strategies (MTSSs) by Federal Ministries, intended to flesh out strategic sector policy priorities and link them to budget resourcing. These provide sectoral detail for the multiyear program of priority spending in the Federal Government s National Economic Empowerment and Development Strategy (NEEDS) and its Seven Point Agenda. Progress by State governments Nigerian states are gradually following the Federal example in developing high level planning frameworks for annual budget preparation. The Jigawa State Government has developed the Jigawa State Economic Empowerment and Development Strategy (J-SEEDS) which contained costed sectoral priorities for the period In Kano State medium term revenue and expenditure projections are provided in the Kano State Economic Empowerment and Development Strategy (K-SEEDS), a costed medium term development plan for three years from 2004 to 2007, aimed at the integrated economic and social development of the state This is being succeeded by the Kano State Roadmap for Development. 5

16 Moreover, expenditure ceilings in the call circular are now based on three year revenue forecasts. 14. For Kaduna the State Economic Empowerment and Development Strategy (KADSEEDS) contained year forecasts from 2005 to 2007 which assisted the preparation of annual budgets. 15 In early 2008 the Kaduna state government prepared multi-year estimates of total resources available to each sub-sector in the years 2009 to 2011, to facilitate the updating of KADSEEDS based on strategic prioritisation and costing. However, ceilings are not issued to state MDAs to guide the preparation of their budget requests. The March 2008 Kaduna PEFA concludes As regards the multi-year perspective, the consensus view is that (i) there are no reliable forward estimates of fiscal aggregates, (ii) there has been no debt sustainability analysis; (iii) such sector strategies as exist have not been reliably costed; and (iv) budgeting for investment and recurrent expenditure are separate processes with no integration of estimates. 16 Lagos State Government (LSG) has also prepared a high level planning framework (LASEEDS) and seeks spending information from its MDAs on two outer years when they submit their budget requests. Also individual MDAs receive spending ceilings in the budget process. However most MDAs still prepare separate capital and recurrent budget requests. LSG has yet to fully develop an aggregate fiscal strategy. In addition to statewide development strategies, individual MDAs at the state level are developing their own MTSSs, which flesh out the sector detail of the overall state development strategy, although it is understood that these tend to focus on prioritising capital projects and generally do not include recurrent spending. In the case of Lagos State Government, MTSSs were initially based on an elaborate, consultant prepared, template, with a simpler approach being adopted for the next round of LSG MTSSs. 17 Good plans may not mean good budgets Preparation of a high level planning framework is an important step in introducing policy based budgets. However, international experience suggests that annual budgets often respond sluggishly to formal planning priorities, if at all. Budget spending patterns become entrenched and ceilings for individual MDAs are based on last year s spending rather than medium term planning documents. Multi-year documents frequently are not updated and lose relevance as time passes. Planning and budgeting do not connect effectively. State level PEFAs point to poor planning-budgeting links in Nigerian states. The Kaduna PEFA notes that Apart from a broad instruction to take account of KADSEEDS in their submissions, MDAs are given no guidelines on how this is to be done and there is, as yet, no effective system of multi-year planning that integrates with the State s budget estimates. 18 LASEEDS is seen as a roughly costed set of additional spending initiatives which is not integrated with budget commitments and resource constraints. Reflecting these factors, the results of a succession of annual budgets often deviates massively from the longer term planning projections. The more effective MTEFs aim at 14 Kano State Government, Public Financial Management Performance Report and Performance Indicators, draft June 2007, p KADSEEDS set some difficult targets for 2005 to 2007, including increasing internally generated revenue (IGR), increasing the proportion of capital expenditure and controlling debt, as well as recognising the need for capacity building and investment in the state s Public Financial Management (PFM) systems. Kaduna State Government Public Financial Management Performance Report and Performance Indicators, March 2008, p Kaduna State Government, Public Financial Management Performance Report and Performance Indicators, March 2008, p See Lagos State Government, Ministry of Economic Planning and Budget, Guidelines for the preparation of Medium Term Sector Strategies for the Period Page 47. 6

17 preventing parallelism of planning and budgeting by tightly integrating annual budget preparation with MTEF management. 7

18 8

19 Section Three Challenges in introducing State level MTEFs While there is a strong case for introducing MTEFs at the state level in Nigeria, their successful operation requires certain conditions to be met. A view has begun to emerge that an MTEF is too sophisticated a budgeting tool for many developing countries, and can divert attention from more basic budget reform (see Box 2). The challenge in considering state level MTEFs is to find a business process for an MTEF which is a) readily grafted onto existing state budget processes b) in a migration path suited to a less than supportive political environment c) in which there are simultaneous improvements in annual budget processes (required for the benefits of an MTEF to be reflected in budget outturns). In short, the case for introducing an MTEF depends heavily on it being part of a reform strategy for the annual budget process rather than a stand alone change. This section assesses the scope for state level MTEFs in the light of: Predictability of state resource envelopes; A sound budget classification; Political constraints. Predictability of State resource envelopes Many Nigerian states do not currently prepare revenue projections 19 and for those that do the World Bank notes that for a sample of states there has been an inability to make accurate projections. 20 This mainly reflects variations in revenues from the Federation Account due to oil price and production fluctuations. Volatility in states internally generated revenues (IGRs) can also be significant but these make only minor contributions to the revenues of most states. In principle, preparation of a medium term fiscal framework (MTFF) at the state level requires forward estimates of three variables: payments to the state from the Federation Account, internally generated state revenues and reliance on net borrowing/lending by the state. 19 SPARC is supporting the preparation of revenue projections in Kano and Kaduna. 20 PEMFAR Main Report, p

20 Box 2:- An influential sceptic s view of MTEFs at the sub-national level Especially troublesome is the notion of extending the MTEF to general government, requiring sub-national entities to go through the exercise--as currently envisaged in some countries, e.g., Tanzania. Although that country has effectively introduced a number of sound improvements in public financial management, including a multi-year perspective for the annual budget process at central government level, it certainly does not have excess supply of planners, economists, accountants, and budgeting experts at the provincial level. Most provincial governments in Tanzania are unable to plan one year ahead, much less implement a medium-term expenditure framework. Alongside the improvement of the MTEF for central government, the priority is to help the provinces assemble an effective annual budget rather than pushing on them multi-year expenditure programming which will not only fail, but compromise the chance of improving annual budgeting into the bargain. Schiavo-Campo, Salvatore, Of Mountains and Molehills: The Medium-Term Expenditure Framework ; Paper presented at the Conference on Sustainability and Efficiency in Managing Public Expenditures, Honolulu, July Payments from the Federation Account Most states derive the bulk of their budget financing from the Federation account. 21 For example, between 2004 and 2006 Kaduna received 84% of its revenue from the Federation Account, while the figure is 95% for Jigawa State and 65-70% for Kano. 22 The key challenge is forecasting future payments from that Account, requiring projections of future oil prices and production, future threshold prices for payments to the Excess Crude Account (ECA) and future distributions from that Account (see Annex 1 for background). In principle transfers from the Federation Account should be predictable due to the formula base, combined with damping of the effect on oil price fluctuations by formula driven payments into the Excess Crude Account. 23 However, the Kano State PEFA notes Revenue forecasting is generally difficult in Nigerian states because revenue received from the Federation Account has sometimes been subject to large uncontrollable positive or negative variations and controllable local revenue is only a small proportion of the total. 24 The PEFA notes that Kano revenue forecasts have generally been met or exceeded. The World Bank notes the systemic poor projection of the revenues that drive the budget, arising from the states inability to forecast revenue flows from the center. Much of the states revenues derive from unconditional formulaic transfers from the federation account, which is largely composed of oil revenues. Granted that world oil prices are volatile and not easy to predict, however, states have not taken advantage of recent efforts by the Federal Government to reduce the negative impact of this situation by introducing an oil-based fiscal rule that pegs the oil price for budget purposes. Traditionally, at the state level there is the tendency to exaggerate these revenues at the time of budgeting, leading to larger but unrealistic budgets The oil price based fiscal rule introduced from 2004 uses a predetermined reference price of oil to determine revenues available for federal and state budgets. Earnings from oil above this reference price ( excess crude ) are held in an account with the Central Bank. Releases from the account need the agreement of all governments and do not occur automatically if the actual oil price falls below the benchmark (see Annex 1). 22 Kaduna PEFA p. 23. Kaduna IGR averaged 16% of total revenue between 2003 and However, withdrawal of funds from the Excess Crude Account is made on a discretionary basis by the National Executive Council and the National Council of States (for example, they recently approved a draw down for power projects). 24 Kano State Government, Public Financial Management Performance Report and Performance Indicators, June 2007, p PEMFAR Main Report, p

21 The Federal Government has recently improved its projections of both its revenues and transfers from the Federation Account. Forward information is intended to be available on the Office of the Accountant General of the Federation (OAGF) website. 26 However, due to the limited focus by state governments on financing when they develop their spending programs (described below in the section on Political Constraints) states do not necessarily capitalise on the availability of this forward information. Internally generated State revenues The World Bank notes that a sample of states it examined had difficulty in projecting their IGRs. In the case of the Bank sample, deviations in annual projections of IGRs were greater than deviations in annual projections of total state revenues. This could reflect the aforementioned tendency for financing projections to be a residual after desired budget spending has been decided, with shortfalls in financing being assigned to Bureaux of Internal Revenue and MDAs as balancing items rather than projections. This is consistent with the view that some state governors may be more focused on the announcement effects of budget initiatives than on whether the initiatives can be executed. It reinforces the point that the case for introducing an MTEF depends heavily on it being part of a reform strategy for the annual budget preparation. Net State borrowing/lending A further constraint on multi-year projections of the state resource envelope is the absence for most states of a publicly announced fiscal policy strategy, including deficit financing (if any). Lagos State is an exception, having recently adopted a three year borrowing strategy. 27 In other states deficit financing can be a residual. The PEFA review of Kaduna states there was a consistent weakness in the budgeting system: total cash income was over-budgeted, always by assuming that loans and grants receivable would be higher than they actually turned out to be. On this basis the score assigned to PI-3 (total revenue compared with budget) is a D. 28 In 2006 only 13% of the budget estimate for grants and loans was actually realised. This was reflected in a 50% underspend on the capital budget. 29 The World Bank notes... state governments also overestimate borrowing, which in their budget presentation, they wrongly treat as part of revenue. The tradition in many states is to beef up budgeted resources with loan proceeds even while negotiations are not completed and/or they have not yet fulfilled conditions for the drawdown of the proceeds of negotiated loans. As often happens, delays may affect the completion of negotiations, or states may not be able to fulfill all conditions precedent to drawdown, such as counterpart funding. These affect the ability of states to draw on the loan proceeds and, consequently, their ability to implement the budget as planned. 30 Reflecting the above, annual budget documentation on macroeconomic and fiscal information is typically brief This OAGF web site was down at the time of preparing this report. However information on actual distributions to state governments is available on the Federal Ministry of Finance web site. A SPARC study has used the Federal allocation model for developing revenue projection methodology for Kano and Jigawa. 27 A debt management unit has been established in Lagos. 28 p. 7. The reference is to the period The PEFA continues Because much of the capital budget was assumed to be financed by loans and grants, actual capital expenditure was necessarily much lower than budget. While recurrent expenditure was consistently in line with budget, in 2005 and 2006 capital expenditure was only 50% of its budgeted figure. 29 Op cit p PEMFAR Main Report, p See Kaduna State government, Public Financial Management Performance Report and Performance Indicators, March 2008, p

22 The conclusion of this section is that predictability of the state resource envelope is not a major impediment to introducing state level MTEFs. The effect of uncertain oil revenues is damped by the excess crude account, and the immediate problem is the unwillingness of many state governments to project their resource envelope rather than the difficulty of doing so. A sound budget classification An MTEF is essentially a set of forward estimates of budget spending, and is based (in varying degrees of disaggregation) on the classification used in the annual budget. The effectiveness with which an MTEF links state policy objectives to annual budget allocations depends in part on the suitability of that classification, ie. on whether the budget classification includes (as well as the usual administrative and economic classification) a program (or functional or sectoral) classification capable of linking detailed funding allocations to specific policy objectives. In the case of the Nigerian states this is more of a problem for recurrent spending than for capital, since individual projects are usually linked to policy goals. In contrast, recurrent spending is classified on an administrative basis (ministry, department, agency), and sometimes an economic basis (salaries, operating costs etc.), but is not attributable to the achievement of particular policy objectives. Hence it is difficult to link the components of annual recurrent spending to the achievement of specific poverty or development goals, or to projects in the capital budget. Capital and recurrent budgets therefore lack coordination. The Jigawa PEFA states Of greater concern is the fact that the lack of consistency between the recurrent and capital budget classifications has hindered the production of a uniform MTEF and budget which can transparently relate capital and recurrent expenditures on projects. 32 In Kano a new program classification has a program category, but this is used for capital and not recurrent expenditure, and the capital and recurrent budgets remain separate. Effective policy based planning through an MTEF is difficult without reform of state budget classifications and integration of the capital and recurrent budgets. To introduce performance budgeting in Nigeria the IMF has recommended the introduction of programme budgeting by 2013 in place of the current sector focus. Other budget experts point to mixed success of programme budgeting in other countries, where it has added a layer of complexity rather than displacing an over-focus on inputs. They emphasise instead (as identified in the PEFA Performance Indicator 12 for policy based budget preparation in Box 7 on page 27) the need to separately identify new policy initiatives embedded in ministry budget requests from adjustments of baseline funding. This is to enable choices about which reform agenda priorities to support and which to defer to be made on the basis of actual spending proposals rather than broad sector priorities. It is not the purpose of this paper to explore options for budget classification reform in Nigeria. Moreover, a new budget classification has been designed for Kaduna, Kano and Jigawa, and effective implementation will assist the introduction of an MTEF in those states. Importantly, improvements in state budget classifications will support more effective annual budgeting as well as the introduction of an MTEF. Political constraints on introducing an MTEF The World Bank PEMFAR states... the public financial management performance of Nigerian states is very weak and in need of urgent reform. There are scattered cases of individual success stories; however, taken together they amount to little and do not have a meaningful effect on the national economic development and poverty reduction agenda. 33 This conclusion 32 Jigawa State Government, PFM Steering Committee, Public Financial Management Performance Report July 2007, p World bank, Public Expenditure Management and Financial Accountability Review Main Report, p

23 is borne out by the poor Performance Indicator ratings in the PEFA studies of Kaduna, Kano and Jigawa. PEFA reviews of state budgets give poor scores for budget credibility due to major underspending compared with the approved budget plan. The Kaduna PEFA notes Actual expenditure differed from the original budget by more than 15% in all three years [ ]. The difference was explained by actual capital expenditure being lower than budget, which in turn was caused by below-budget total receipts (and in particular grants and loans). 34 In contrast to Obadan s conclusions about Federal level budget underspends (referred to below), state underspends appear to be due to cash rationing as a result of implausibly large capital spending plans rather than inability of MDAs to disburse funds within a twelve months budget time frame. The Kaduna PEFA also notes There needs to be a clear political will by both the executive and the legislative arms of Government to approve realistic budgets and to execute them as planned and identifies a need for achievable change strategies to underpin capital spending proposals. The Kano State PEFA suggests that divergence of budget outturn from plan reflects the inyear budget changes that have been introduced and regularised by supplementary budgets. It is clear that in-year budget changes of this magnitude undermine the longer term planning approach and further work is needed to maintain the medium term changes that have been introduced and regularised by supplementary budgets. 35 A further cause of divergence between budget plan and outturn in Kano is the use of Authority to Incur Expenditures to fund unbudgeted items. The Jigawa PEFA also points to the negative influence of in-year budget changes that have been introduced and regularised by supplementary budgets. It concludes. The key problem causing the poor scores for budget credibility is that execution has usually proceeded without much reference to the approved budget. 36 Lack of budget credibility is linked to the role of capital projects in the budget. The World Bank notes On average, budget execution for five states and the Federal Capital Territory (FCT) amounted only to 64 percent between 2001 and The largest shortfalls are associated with the capital budgets, with Bauchi and Enugu states showing capital budget outturn of only 40 per cent of plan in the period State Governments tend to see public spending primarily in project terms and the Bank notes a chronic tendency for states to over-budget for capital spending relative to budget outturn. It has been common for states to approve budgets with the exceptionally high shares of capital spending, in excess of 60 percent, while actual budget execution usually shows capital spending below 40 percent of the total. 39 The Bank attributes this to an ideological bias towards the capital budget believing that the larger capital budget reflects their greater 34 Kaduna State Government, Public Financial Management Performance Report and Performance Indicators, March 2008, p Kano State Government, Public Financial Management Performance Report and Performance Indicators, draft June 2007, p. 43. The Kano PEFA notes that The present funds release mechanisms for capital expenditure and for recurrent expenditure special releases are unnecessarily centralised, and cause bureaucratic delays which can undermine the process of budget management. p Jigawa State Government, Public Financial Management Performance Report and Performance Indicators, July 2007, p. 8. The approved budget is never translated into a month-by month plan for matching expenditure against cash inflows. In the absence of proper cash planning, approved requests for expenditure queue up in Treasury, which performs the role of expenditure prioritisation. 37 PEMFAR Main Report, p PEMFAR Main Report, p PEMFAR Main Report p

24 commitment to social and economic development. 40 This is linked to the tendency for state governments to be given credit for the projects they start rather than those that they finish. Lack of budget credibility undermines an MTEF It is difficult to justify support for introducing a medium term expenditure framework when budgets lack credibility even in an annual framework. An MTEF is normally seen as strengthening budget outcomes through the improvement it brings to the budget plan. However, if budget outturns differ markedly from the plan the benefits of an MTEF are likely to be lost. In the last resort it will be difficult to improve state level budget planning if state governments ignore resource constraints and budget in an ad hoc and opportunistic manner. 41 At worst, attempting to introduce an MTEF could be seen as a diversion from getting the basics right and at best as a parallel process with little impact on annual budget preparation. Reflecting this, there is general acceptance of the wisdom of ensuring effective annual budgeting before attempting to introduce a multi-year framework. On the other hand, as alluded to earlier, caution needs to be exercised in recommending deferral of support for an MTEF until annual budgets are more credible, since introduction of an MTEF and associated resource envelope may itself strengthen the link between annual budget plan and outturn. However, the case for introducing an MTEF in a state for which the annual budget outturn deviates markedly from the budget plan would be contingent on the design of the MTEF process itself improving the realism of the budget plan. In short, the case for introducing an MTEF depends heavily on it being part of a reform strategy for the annual budget rather than a stand alone change. The only rationale for investing in an MTEF to improve the budget plan where the budget itself lacks credibility would be if the MTEF itself could be designed in a way which increases that credibility. 40 PEMFAR Main Report p This conclusion reflects a situation, where some states (for example, Enugu) carry particular capital items in the budgets from year to year without even beginning to fund them at all. 41 The World Bank s PEMFAR attributes this to the failure of the planning function to be re-established following the end of the military period (during which military governments did not use plans to formulate their spending programs). 14

25 Section Four Using an MTEF to improve credibility of the annual budget An MTEF is usually seen as improving the quality of the budget plan rather than the effectiveness of budget execution. In this consultant s experience little attention has been paid to whether an MTEF will help budget disbursement conform to the approved plan. However, it is likely that there could be links, depending on the reasons for underachievement of budget plans. The most systematic work on under-disbursement of planned budgets in Nigeria has been undertaken at the Federal level, by Professor Obadan, who focused particularly on capital spending by federal MDAs (see Box 3). He concludes that in most instances under-spending of capital budgets reflected failure of the MDA to organise to spend their project votes before the budget year ended, rather than denial of access to the vote due to within-budget-year cash rationing. The problem was poor coordination within MDAs rather than cash rationing by the Treasury (see Box 3). However, as previously indicated, state level PEFAs tend to suggest a different conclusion at the sub-national level that state governments develop their budget spending plans with more concern for the announcement effects than the availability of finance. 42 Subsequent unavailability of cash causes disbursement approvals to queue in state treasuries, even though MDAs are ready to proceed, and cash rationing undermines execution of the budget. 43 It is possible that a suitably designed MTEF may reduce the likelihood of cash rationing. The approval of FEs in conjunction with the budget estimates provides a multi-year projection of budget financing needs, and a longer lead time for organising capital raisings for those projects approved for inclusion in the FEs. Central agencies can develop a state financing plan not only for the next budget year but also the forward estimate year(s), providing a longer 42 Budget preparation should begin with an estimate of available resources, with spending options then prioritised to fit this resource estimate. State PEFAs suggest that a spending program is adopted without reference to financing constraints with unachievable financing plans being included in the budget. 43 See for example the Kano PEFA, page 47, which indicates Requests for the release of funds are assessed by the Governor with approvals communicated to the Treasury Department for release of funds to the MDA. Approvals are determined by the Governor s (and the Secretary to the State Government (SSG s)) priorities, and release of funds depends on availability. There is some evidence that persistent releases for recurrent items force larger capital items to be deferred for long periods. Also, in years when unbudgeted loans to parastatals increase (see PI-7), funds are further restricted. These factors have resulted in unpredictable availability of capital funds to some MDAs. 15

26 lead time for organising financing, whether through release of funds from the Federation Account, improved state tax administration or a borrowing program. Moreover, where state governments are tempted to announce unrealistically large capital programs for the annual budget, the introduction of a multi-year budget frame could allow these to be distributed between the next budget year and the forward Box 3 Lack of Credibility of the Federal Budget Following massive under-realisation of the approved 2008 Federal budget plan a review of the causes was commissioned. The consultant, Professor Obadan, concluded Further analysis of available data shows that in the last four years, capital budget implementation, in terms of the proportion of funds released that was actually utilised, exhibited declining rates of performance. High rates of implementation were achieved in 2005 and 2006: 88.1 and 96.5 percent, respectively. A relatively lower rate of performance was recorded in 2007 (67.8%) while that of 2008 was much lower at 43.9 percent. (Italics added) A committee formed by the Federal Executive Council and headed by the Head of the Civil Service of the Federation concluded that The bureaucratic causes of delays in the budget implementation process identified are as follows: Lack of familiarity and understanding of extant laws, regulations and guidelines that govern the process of project execution by MDAs; Delays in obtaining approvals of Ministries and Permanent Secretaries for various stages of the procurement process; Challenges in documentation; and Delays in processing memos to Council. Source: Mike I. Obadan, 2008 Federal Capital Budget Implementation: Factors Affecting Performance, p. 16. estimate year(s). The opportunity for the state government to announce in the budget projects for which disbursement begins in (say) eighteen months and which are included in the forward estimates rather than the budget may contribute to a more credible budget plan and also offer a politically attractive double announcement effect (project first announced as included in the FEs, and a year later re-announced as included in the subsequent budget based on those FEs). This may help reduce the temptation for unrealistic shoe-horning of too much project spending into each budget period. Linked to this, the MTEF presents a political opportunity to demonstrate commitment to social and economic development over several years, reducing the pressure to demonstrate the commitment in successive single year budgets. Finally, where a twelve month budget year provides insufficient time for MDAs to organise procurement and disburse project funds (the Obadan view), the FEs may encourage them to implement their capital spending program more confidently by providing them with a multiyear (rather than one year) financing approval. 44 Where a project involves lengthy procurement processes procurement planning can commence in the next budget, with the first drawdown of funds only appearing in the FE Under this approach the FEs are used as the baseline for preparing the following budget, so that MDAs always know their funding framework for a two year (plus) period. While not being guaranteed the amount in their FE they are guaranteed that this is the baseline for preparing the subsequent budget (the second year is a quasi budget ). 45 There are similarities here with the use of a multi-year commitment control system for authorising capital expenditures. When Australia adopted multi-year FE based budgeting in the late 1980s, along the lines now adopted by the Philippines, the commitment control system was dropped, because inclusion of a project in the FEs instead provided the necessary certainty to the spending agency that funds required beyond the next budget 16

27 It should be emphasised, however, that the FEs do not provide the same level of certainty as estimates enacted in the budget. 46 When eventually the FEs do become the baseline for preparing the next new budget the fiscal projection originally used to prepare them has to be updated for unforeseen events since the FEs were prepared. The ceilings for MDAs in the new budget circular may therefore be higher or lower than the FEs on which they are originally based. The point is that while complete certainty of financing requirements and availability cannot be guaranteed for the first out year (a key reason why budgets are only prepared on an annual basis) a much higher level of certainty is provided by a rolling two year (plus) budget framework containing forward estimates than serial preparation of single year budgets. Multi-year forward estimates? It is suggested above that adoption of a two year (plus) budget frame may help improve the credibility of state budgets, by reducing the incentive for state governments to squeeze unrealistically large spending plans into single annual budgets, and providing a greater (ie. two year plus) time frame in which: MDAs can plan for project execution; central agencies can plan for necessary future budget financing; state governments can announce budget funding for new projects. The resulting improvement in the credibility of the annual budget would be an important achievement in its own right. It would also open the way for realising some of the more conventionally recognised benefits of an MTEF involving a stronger link between budgets and medium term plans. However, there may be a trade-off in simultaneously realising the short term budget credibility benefits and the longer term budget /planning benefits of introducing an MTEF. An MTEF is usually assumed to include forward estimates for two or three years beyond the next budget year, and the discussion in the previous section has emphasised a single out year only (ie. the next budget is prepared in a two year framework). While this may be a disadvantage when programming multi year planning priorities into a series of annual budgets, a two year budget frame may be more appropriate for reducing the discrepancy between annual budget plan and outturn. This is because to achieve the budget credibility benefits listed above it is important that the FEs are themselves credible, ie. have a quasi budget status arising from their consistency with future resource availability. Not unsurprisingly, this is less true of FEs for a second and third out-year, since the accuracy of projections of the resource envelope diminishes as the time horizon is extended. If the FEs have little credibility they will provide little additional certainty for MDAs or central agencies in programming their spending and financing activities. Possible transition strategy In this regard, the greater predictability of financing following the introduction of an MTEF should be attractive to MDAs responsible for capital programs, but it would be possible to commence with a two year budget frame for all MDAs. Subject to satisfactory improvement in budget credibility, the number of out-years could then be increased to two or three to achieve a better connection between the budget, MTEF and state planning documents. A phased approach to increasing the number of years in the MTEF has the potential to help progress year had been programmed and would be made available. Other countries have found other solutions, such as to appropriate capital funds on a two year basis, or to allow automatic carryover of unspent project funds. 46 Budgets are almost always limited to twelve months duration due to the difficulty of projecting the resource envelope for longer periods than this. Even twelve months proves too long for many governments (including Nigerian states), who are obliged to introduce supplementary budgets before the twelve months has ended. 17

28 MDG goals by social ministries by first improving budget credibility, then budget planning, particularly if combined with a quid pro quo by the MDA to improve performance reporting. The initial use of a two year budget frame will not of itself prevent unrealistically large spending programs being announced by state governments for a two year period, and introducing an MTEF is far from a panacea for unrealizable budgets. A key influence on the rate of progress in individual states would be a demonstrated commitment on the part of the state government, as well as MDAs in the state, to a more disciplined and responsible approach to budget preparation through the use of a multi-year framework. States capacity to forward plan An MTEF will make little contribution to annual budget outcomes if the state government has limited capacity to plan or there is an absence of sectoral strategies for implementing overall state policy goals. As discussed above, this does not appear to be a problem in the Nigerian states. However, linked to this is the need to translate desired changes in key policy areas into tightly defined expenditure proposals for which costs can be accurately estimated for budgeting purposes, and benefits tracked. This requires design and costing capacity at the level of MDAs, as well as effective organisation given the short time available to them between the issue of the budget circular and the lodgment date for their budget request. An early step in laying the foundations for state level MTEFs will be to refine MTSS documents to a form which can directly input to the budget/mtef process. A key reason why MTSS type documents are ineffective in some countries is that the MTSS is seen as a parallel or subsidiary process to the normal input-based budget request and therefore there is little incentive to spend time on detailing and costing each proposal in the MTSS. Line ministries are able to produce vague and poor quality MTSSs but nonetheless win increased funding through their routine input based budget request process. This can be avoided if the budget office instructs MDAs that the only route for winning increases on the previous year budget allocation (other than for technical parameter variations) is to link the requested increases to their MTSS. This will help to reinforce the status of sector planning in guiding budget allocations, and provide an element of synergy between getting the budget basics right and laying the foundations for an MTEF. This is discussed further in regard to the MTEF design options in the following sections. Concluding comment The TOR for this report require identification of a possible strategic approach to developing and implementing a comprehensive and cohesive MTEF with State Government, recognising the implications for such an approach and possible limitations, in particular capacity. The conclusion of this section is that there are at least some synergies between improving annual budget basics and laying the groundwork for an MTEF. This would suggest that a platform approach, in which fixing the basics is a precondition to beginning to work on an MTEF, may be an oversimplification. The two can and should proceed as complementary processes provided the initial MTEF is a basic (perhaps single out-year) design, which helps equate annual budget outturns to annual budget plans. The following section discusses two possible design options, based on Indonesian and Philippines experience. 18

29 Section Five Approaches to MTEF design There is widespread acceptance of the case for preparing the annual budget within an MTEF. However there is limited focus on appropriate MTEF architectures (see Box 4). Drawing on experience in Indonesia and the Philippines, two alternative MTEF options for Nigerian states are described in the following sections. Indonesian Option: adding an out year to existing budget preparation processes Following Indonesia s far reaching 2001 decentralisation the Government legislated for the introduction of MTEFs at both the national and district level. 47 Under the resulting Indonesian model district service units (the equivalent of Nigeria s state MDAs) prepare their annual budget request not only for the new budget year but also one out year (at the national level it is for two out years). As indicated above, the Lagos State Government has also adopted this approach. The list of Features of Multi-year Budgeting in the Kaduna State Government budget manual also appears to be consistent with the Indonesian approach. 48 Box 4 Should an allowance for new policy be included in an MTEF? There is a debate over whether an MTEF should be a forward budget, containing combined projections of existing commitments and an allowance for (unspecified) new initiatives within a projected resource envelope (a ceiling based approach). Alternatively, as in South Africa, Australia, Canada and the Philippines, the MTEF is a baseline budget for existing commitments, defined independently of resource envelope projections. In this approach the future resource envelope and baseline commitments in the MTEF together define uncommitted resources available for financing new initiatives. This fits more comfortably with annual budget preparation, in which proposed initiatives are programmed into the annual budget based on the latest projection of available revenues. In Nigeria s case, at the Federal level there is no clear commitment to which alternative MTEF architecture to adopt, although the decision will need to be made before the specifications for the new Government Financial Management Information System (GIFMIS) are finalised (the draft functional requirements for the GIFMIS allow for the baseline approach as well as the forward budget approach). 47 In contrast to Nigeria, Indonesia is a unitary state, and the Ministry of Home Affairs can prescribe processes used by the districts for budget preparation. The Indonesian approach reflects that country s strong planning environment (including for sub-national administrations) combined with the previous failure of the planning and budgeting processes to connect meaningfully. 48 Kaduna State Government, Budget Manual, p

30 In the Indonesian case the out year request is for information purposes only, and the budget itself continues to be prepared on an annual basis. However, at least in principle, the out year data could form the beginning point for negotiations in each new budget preparation cycle. Since each MDA s request for funding in the next budget also includes an estimate for funding needed in the out-year beyond the next budget, this should in principle force the local service unit to justify its budget year spending proposals in terms of a two-year spending plan. Moreover, in principle the local budget committee can then review the budget request by each local service unit in a two year context and approve requests based on the local government s two year fiscal program. In order to distinguish the Indonesian forward budget request approach from the alternative ( Philippines ) approach described below it is important to note that 1) the forward estimates for each local service delivery unit are prepared by the unit itself rather than by the local budget office; 2) the forward estimates do not separate out year funding required to maintain currently approved service levels from funding required for new policy proposals (ie. proposals to improve service levels) 49 ; 3) the forward estimates proposed by individual service delivery units are not constrained by ceilings to be consistent in the aggregate with the projected out year fiscal envelope. The net effect is that there is a strong wish list element in the forward estimates submitted by each service unit. There is also no control of whether the FEs for a particular calendar year are serially consistent across successive years. 50 These considerations reduce the contribution of the forward estimates prepared by district service units to formulating the district budget. Reflecting capacity limitations, many local service delivery units do not take the out year seriously. Some do not in fact fill out the out year field in their budget request template. Others either paste their budget year request into the out year field or assume an arbitrary percentage increase. Not unconnected with this, the local budget committees make little or no use of the out-year information, so service units cannot be blamed for failing to take this field in their budget request template seriously. The bottom line is that the provision for forward estimates by local service units in conjunction with their annual budget request has thus far made no real difference to the quality of district annual budgeting. Making the Indonesian approach work In order to increase the effectiveness of the Indonesian approach to an MTEF it would be necessary to ensure that individual MDAs submit better quality forward estimates than occurs in the Indonesian case. These should be based on clearly defined and well costed spending proposals for the two year time frame. The groundwork for achieving this is in fact already in place in Nigeria in the form of the spending proposals contained in the MTSSs. This is illustrated by the guidelines for the preparation of MTSSs prepared by the Lagos State Government. 51 While these appear to be over-ambitious in some respects, particularly in regard to the use of a scoring system for ranking competing initiatives and an implied zero based budgeting approach, the guidelines do conform to PEFA Performance Indicator 12 (see Box 7 on page 27) in requiring budget requests and FEs submitted by MDAs to include explicit costing of the implication of proposed new policy initiatives (and) clear, strategy linked, selection criteria for investments. 49 As currently implemented the Indonesian approach (in contrast to the Philippines approach described below) is not PEFA-consistent (see Box 7). 50 For example in 2009, when preparing the 2010 budget, a set of FEs is submitted by MDAs for 2011 and 2012 along with their budget request. A year later, ie. in 2010, when preparing the 2011 budget, FEs are submitted by MDAs for 2012 and However, the FEs for 2012 submitted in 2009 may differ arbitrarily from those for 2012 submitted in 2010 ( serial inconsistency ). 51 Lagos State Government, Ministry of Economic Planning and Budget, Guidelines for the Preparation of Medium- Term Sector Strategies for the period

31 A good example of the implementation of these LSG Guidelines is provided by the LSG Ministry of Education s MTSS. In this best practice MTSS the Ministry lists its existing measures and proposed new initiatives for achieving each of the six State education goals. Additionally, for each of the proposed new initiatives the MTSS provides detail on the outputs/outcome of the proposed initiative and the initiative s cost (both capital and recurrent) in 2009, 2010 and PEFA PI 12 requires this information to be integrated into the annual budget formulation process (see Box 7). Under the Indonesian approach this would require each MDAs budget request to include (in addition to funding requested for the new budget year), the agency s bid for the out-year, both referenced to the initiatives proposed in its MTSS. 53 One difficulty is that, as a review of the proposed initiatives in the Ministry of Education MTSS for suggests, the amount of data involved is very large. Experience also suggests the likelihood of version control issues for individual proposed initiatives where the MTSS is not updated as an integral part of annual budget preparation. Under an Indonesian approach the use of modern budget preparation software, with change package and version control functionality, would appear to be essential for integrating the MTEF with annual budget preparation (see Box 5). 52 See Lagos State Government, Ministry of Education, Medium Term Sector Strategy ( ). The new initiatives are proposed by the departments and agencies within the state education sector and compiled by the Ministry. 53 This could be achieved in those Nigerian states which might adopt the Indonesian approach (such as the Lagos State Government) by requiring that each MDAs forward requests (say for 2011 and 2012) in its (say 2010) budget request are derived from its MTSS. The difference between the 2009 appropriation to the MDA and its out-year requests would need to be transparently broken down into changes in the cost of existing service levels and the costs of proposed new initiatives. This information is lacking in Indonesia. Although Indonesia s local service delivery units do prepare multi-year plans these tend to be general in nature and do not link to the units out-year budget requests. This reflects lack of planning capability in local service delivery units. 21

32 Box 5 Why is real MTEF based budgeting so elusive? An MTEF is intended to substitute policy based budgeting for traditional incremental budgeting. However it will achieve this only if it recognises one of the main reasons why incremental budgeting is so doggedly persistent. MTEF basing of an entire public sector budget requires the processing of vast amounts of information for the out-years as well as for the next budget year often more than can be handled within a typical six month budget preparation period by a typically small group of budget office officials and cabinet ministers. Moreover, including performance information to guide budget decisions (part of the rationale of budgeting in an MTEF) further increases the data load. Indonesian experience suggests that increasing the volume of data by adding out-years will of itself achieve little. A key reason why incremental budgeting is so persistent is that it is the only way in which the vast amounts of data required to prepare a budget can be processed in the six months or so available for budget preparation. Linking budget to MTEF requires a parallel technical breakthrough in the way in which the additional multi-year data is managed for budget preparation. This requires a re-assessment of what data is important and a new for old approach introduction of new data in an MTEF should result in a reduced focus on incremental budgeting control data (which should ideally be managed at MDA level) and an increased focus on data relating to policy options, capable of being used to prepare a policy based budget. Currently a typical MDA budget request will involve enormous amounts of unsmart data. This comprises quasi mechanical requests for ongoing financing of existing service levels, adjusted for inflation, operational issues such as staff adjustments and ambit claims. While there are undoubted control issues in the budget office agreeing such numbers, they tend to swamp strategic issues in preparing the budget and there are much greater benefits for achievement of policy goals from focusing budget preparation on data relating to proposed new initiatives and how to use the budget to achieve policy goals. Modern off the shelf software, designed for public sector budget preparation, helps resolve the risk of data swamping in an MTEF. Typically this has the capacity to include performance information and financial information in a single database a functionality which is often built into modern GIFMIS budget preparation systems. The pre-existing but updated forward estimate for the next budget year would become the initial data entry screen when budget preparation for that year commences. The GIFMIS would store the programme and sub-programme performance information for annual comparison, and would add the costs of approved change packages to the budget for the coming budget year, as well as for the out-years. The approved change package for each budget can remain visible through the various iterations of the draft budget. The Government can then add to the budget speech, and published budget papers, a list of its approved new initiatives (including savings) embedded in the budget, for each of its policy goals. It will also be possible for the ministry of economic planning and budget to retrospectively check that the claimed performance impact of a previously funded new initiative has actually been achieved an important aspect in re-positioning states to a performance budgeting environment in which they manage budget allocations so as to achieve results. Philippines Option: managing changes in spending in a multiyear context A second and more stream-lined approach to introducing an MTEF is currently being introduced at the national level in the Philippines, after several abortive attempts along Indonesian lines. 54 It differs from the Indonesian approach in two key respects: 54 The Philippines approach had earlier been adopted in a number of developed countries including South Africa, Canada and Australia. 22

33 A distinction is made between the baseline budget required to meet existing service level commitments and new initiatives proposed by MDAs to improve achievement of their sector goals; The overall resource envelope is explicitly introduced to the MTEF process through the assignment of ceilings for each MDA. In the course of preparing traditional input based budgets line agencies usually submit ambit claims for more funds than they need. Their actual budget allocation is then determined through a negotiation process with the budget office, and sometimes cabinet. In some countries (but not in some Nigerian states) the budget office pre-empts ambit claims by assigning each line agency a binding ceiling for its budget request, usually conveyed in the budget circular. However this can have the disadvantage that ceilings are set without consideration of the quality of reform proposals by each agency. Agencies in high priority sectors may receive large ceilings but can organise only to submit poor proposals, or have poor absorptive capacity, while other better managed agencies are starved of funds which they could use relatively effectively. The following paragraphs outline the Philippines approach to an MTEF, simplified to involve a two year budget frame. 1) Budget office: As a one-off event to get the MTEF started, 55 the budget office negotiates with each line agency forward estimates of the financing needed by the agency to maintain its current service levels in the next budget and the following out year. 56 At the commencement of each subsequent budget preparation cycle this two year estimate is updated for parameter changes (eg. changed inflation assumption) and the out year is sent to each agency as its indicative ceiling for its up-coming budget request (ie. the updated FE becomes the indicative ceiling assigned to the agency for its next budget request). 2) Line agencies: Each line agency is able to request budget funds in excess of its indicative ceiling, but only if it justifies over the ceiling proposals in terms of the goal of each proposed initiative, expected impact on policy objectives and detailed cost build-up. This should be based the MTSS provided by the line agency, containing its sector strategy. Each over the ceiling policy proposal must include costs both for the new budget year and first out year (since FEs will then be updated simultaneously with legislative approval of the budget estimates). Cabinet deliberation: using its two year fiscal projection, the state government adopts as many of the new proposals as can be financed in the next budget and first out year. The FEs are then updated and rolled forward by one year, ready for the same process to be repeated in the next budget preparation cycle. 57 The Philippines Government has only recently adopted the approach described here, for the current budget preparation round, and it is not yet possible to assess its success. It may be challenging for service agencies to clearly justify their new policy proposals rather than burying them in an ambit claim to increase a total program allocation. However, the intention 55 After the initial creation of the set of FEs they are updated annually. Implementation in the Philippines involves more than one out-year. 56 In formal terms there are three out years. The out year estimate is the future cost of maintaining existing budge policies. It excludes terminating programs and projects and includes the effects of inflation and demographic changes. Key principle is to separate existing commitments from proposed new initiatives. Existing commitments include public debt interest, personnel costs, other operation and maintenance (O&M) costs, programme costs where service levels are already approved (including adjustments for demographic changes in demand driven programs) and costs of existing activities and projects which are not due to terminate. Existing commitments are reviewed for cost effectiveness when the forward estimates are negotiated annually. 57 If state governors were to introduce new measures during the course of the budget year these would also have to be added to the forward estimates. 23

34 is to force agencies to justify their requests for additional funding in terms of clearer impacts on policy objectives. Comparison of the two approaches Which is the more feasible approach to introducing an MTEF in a capacity constrained environment? The Indonesian approach has its conceptual basis in classic program budgeting, which has a mixed track record of success. The Philippines approach is grounded in the incremental approach to budgeting, which does not require information about the total outputs/outcomes of a program in order to decide its budget allocation, but only the effect on outputs/outcomes of changing the allocation. 58 The Philippines approach probably places lesser demands on MDAs, since they are not required to submit annual requests for their baseline funding. Baseline funding is already included in their forward estimates, held in a budget office database, although these forward estimates are re-negotiated each year. 59 This is a significant point since, in providing external technical assistance for introducing a state level MTEF, the assistance can be focused to a greater extent on a single state budget office rather than being distributed around a large number of state MDAs. It also ensures better quality control of the FEs, and avoids serial inconsistency in the FEs mentioned above. Box 6 The Philippines approach applied to Nigerian states 1. Use the MTSSs of each MDA as the core element of MTEF preparation 2. Require each MTSS to CLEARLY separate the projection of funding needed to maintain currently approved service levels over the next three years from proposed initiatives to increase service levels over the same period (ie. initiatives to achieve planning goals). 3. In the course of the budget preparation cycle the state ministry of economic planning does the following a) reviews and re-negotiates each MDA s proposed baseline spending for the new budget year and out-years (as presented in its MTSS) b) in parallel, estimates the state s resource envelope for the new budget year (no out year projections involved) c) sums the agreed baseline spending for the new budget year across the MDAs and compares this with the resource envelope for the new budget year; this gives an estimate of unallocated resources for the new budget year d) now turns its attention to the new initiatives proposed in each MTSS and prioritises as many of these into the next budget as is consistent with the unallocated resources e) when the state budget is passed the multi-year cost of the approved new initiatives is added to the relevant baseline spending in the MTSSs. Usually the functional requirements for a GIFMIS (such as those for the Federal level GIFMIS) do explicitly cater for this sort of budget preparation approach. There is a requirement that first out-year estimates in the data base can be used as the starting point for budget preparation, and (consistent with PEFA PI 12) that new initiatives can be separately identified in the database. 58 The main point here is that it may be easier for MDAs to link sector plans to initiatives to change a particular program allocation than to the program allocation in its entirety. The latter involves a more complex set of performance indicators and, taken to its logical conclusion, an activity based attribution of overhead costs to programs. 59 The first out year of the FEs (year t+1) is the initial ceiling for the agency budget request (after updating for changed inflation, efficiency gains etc.) 24

35 The Philippines approach also brings much more transparency to reasons for variations in funding requested by individual MDAs than does the Indonesian forward budget approach as currently implemented by the Indonesians. This forges a stronger link with the MTSSs drafted by each MDA. As indicated above, the Indonesian approach is unlikely to work if the forward requests by each MDA blur the distinction between the additional funds being requested to maintain existing service levels and those being requested for proposed new initiatives. In the absence of this transparency the forward requests risk being meaningless ambit claims which add little value to annual budget preparation. In short, for the Indonesian approach to work it would need to take on a key characteristic of the Philippines approach, ie. separation of MDA forward funding requests into funding requested to maintain existing service levels and funding for new proposed initiatives. This would in fact be supported by the Nigerian approach to preparing MTSSs (absent in Indonesia). 60 Separate identification of new initiatives in the Indonesian approach would also bring it much closer to the Philippines approach, with the main difference being whether the central budget office holds a rolling database of the FEs which it uses to set initial MDA ceilings in each annual budget preparation cycle, or whether MDAs themselves prepare their FEs each year. Next steps The Indonesian and Philippines options identified here require technical working through with donors, PFM experts and state government officials. It would be possible in principle for SPARC to support introduction of both approaches in different states, depending on state capacity and preferences. In Lagos State (already apparently embarked on an Indonesian approach) this could be the more devolved Indonesian approach combined with strengthened links between the FEs and the MTSSs. There may also be a need for assistance to develop all Lagos State ministry MTSSs along the lines of the Lagos Education Ministry model. For another state the ministry of planning and finance could itself create a set of FEs for existing commitments, with MDAs subsequently proposing variations in their annual budget requests. 61 Different states could also trial different numbers of years for which FEs are prepared (one, two or three out-years beyond the next budget year). However, care will need to be taken that drafting of state fiscal responsibility acts will not pre-empt technical options before their respective merits have been discussed. Three related issues would require early attention: TA is currently being provided to the Lagos State Ministry of Education on re-design (simplification) of the Lagos State education sector MTSS. This re-design should be coordinated with the strategic design of the Lagos State MTEF, to ensure that the state education MTSS meets the need not only of a stand-alone sector plan but the preparation of the education component of the annual budget in a multi-year frame; An assessment will need to be made of the challenges in extending the Ministry of Education best practice MTSS model to other state MDAs; Attention will need to be focused on how state ministries of finance and planning will handle the additional data associated with multi year expenditure estimates, and proposed new policy initiatives. This is likely to be facilitated by the use of modern off the shelf budget preparation software with the functionality required to integrate forward estimates with budget preparation (such as functionality for managing change packages and budget version control). Lastly, and perhaps most important, a key criterion in designing the MTEF for different states would be the extent of commitment by the state government, as well as key MDAs in the state, to use the MTEF to implement a more disciplined approach to budget planning. Where 60 The guidelines for preparing MTSSs adopted by the LSG are similar to those at the Federal level. 61 There would probably be interest outside Nigeria in a side by side trial of the two approaches. 25

36 there is a tendency for annual budgets to contain unrealistically large capital programs the MTEF could initially be introduced with a strong emphasis on estimation of the future resource envelope and the lock-in implied by existing policy commitments, rather than being a launching pad for a menu of new spending proposals. This would imply a preference for the Philippines approach, which compares the future cost of existing policies with the available resource envelope and hence provides greater transparency of the room for new initiatives than the Indonesian approach. Such an approach would help strengthen annual budget discipline which is a pre-condition for realising the advantages of preparing budgets in a multiyear framework. Box 7 Consistency of MTEF Approaches with PEFA Principles The Indonesian approach as currently implemented is not PEFA consistent. However the Indonesian approach when combined with an effective MTSS link could be so consistent, as is the current Philippines approach. The relevant extract from the PEFA Sub-national Guidelines is as follows. Expenditure policy decisions or options should be described in sector strategy documents, which are fully costed in terms of estimates of forward expenditures (including expenditures both of a recurring nature as well as those involving investment commitments and their recurrent cost implications) to determine whether current and new policies are affordable within aggregate fiscal targets. On this basis, policy choices should be made and indicative, medium term sector allocations be established. The extent to which forward estimates include explicit costing of the implication of new policy initiatives, involve clear strategy-linked selection criteria for investments and are integrated into the annual budget formulation process will then complete the budget policy link. Source: PEFA, Guidelines for application of the PEFA Performance Measurement Framework at Sub National Government Level, Volume 2 Annex, Performance Indicator (PI) 12. Multi-year perspective in fiscal planning, expenditure policy and budgeting, para

37 Appendix One Implications of the Excess Crude Account (ECA) for State MTEFs Payments into the ECA are made monthly and determined by the formula: 1. actual production*actual price LESS 2. planned production*benchmark price where 1. = actual income 2. = budgeted income (as estimated at budget preparation time) However, an asymmetry arises if actual income is less than budgeted income (ie. because of an under-run on planned production or actual price falling below benchmark or both). In the under-run case the resulting hole in budget financing is not filled by an automatic release from the ECA. The reason for this asymmetry is that under the Constitution all revenues go the Federation Account. The ECA is a sub-account of the Federation Account and does not belong to the Federal Government. In contrast to the formula driven payments into the ECA, withdrawal of funds is made on a discretionary basis by the National Executive Council and the National Council of States (for example, they recently approved a draw down for power projects). To introduce further uncertainty, the existence of the ECA is not mandated in the Constitution and the use of its funds, the formula for automatic payments into the fund and even its continued existence depends on decisions by the two levels of government. The revenue sharing formula between the states is implemented by the monthly meeting of the Federation Accounts Allocation Committee, which has a technical committee to determine mechanistically the payments credited to each state based on [actual production*actual price]. MTEF implications What does this mean for a multi-year projection of a state s resource envelope? 27

38 The above arrangements complicate the projection of state resource envelopes or out-year ceilings for state MDAs. This is a common problem for commodity economies subject to boom and bust cycles. On the one hand the smoothing effect of the use of the rolling benchmark oil price imparts greater stability to projections of future oil prices, by evening out the highs (but not the lows ). This reduces the variance in oil price driven payments to the states (since the impact of temporary high prices is diverted into the ECA). However, uncertainty re-emerges in the unpredictability of releases from the ECA, for which there are no policy guidelines. 28

39 29

40 30