This Technical Assistance Report focuses on Botswana’s Medium-term Expenditure Framework (MTEF). The Government of Botswana has committed to introduce the MTEF by 2016. The MTEF will provide a more explicit linkage between National Development Plan priorities and budget allocations by adopting a medium-term budgeting horizon. An MTEF model based on a binding nominal expenditure ceiling covering 100 percent of government expenditure is appropriate. To support the commitment to the resource allocations approved under the MTEF, a number of prioritization, control, and accountability arrangements need to be put in place.


This Technical Assistance Report focuses on Botswana’s Medium-term Expenditure Framework (MTEF). The Government of Botswana has committed to introduce the MTEF by 2016. The MTEF will provide a more explicit linkage between National Development Plan priorities and budget allocations by adopting a medium-term budgeting horizon. An MTEF model based on a binding nominal expenditure ceiling covering 100 percent of government expenditure is appropriate. To support the commitment to the resource allocations approved under the MTEF, a number of prioritization, control, and accountability arrangements need to be put in place.

I. Introduction

1. The Government of Botswana is committed to moving to a Medium-Term Expenditure Framework (MTEF) by 2016.2 The intention is to facilitate a smooth and sustained implementation of the National Development Plan. The introduction of an MTEF has the potential to: (i) facilitate aggregate fiscal discipline; (ii) foster better expenditure prioritization by lengthening the time horizon for budget decision-making and increasing the scope for shifting expenditures towards high priority areas; and (iii) increase certainty in public spending.

2. The NDPs have been developed based on principles of a medium-term planning framework. Within the plans, the government undertakes medium-term fiscal forecasts, identifies priority policies, and provides indicative expenditure profiles for the development budget for the plan period. The NDP 10 is currently driven by key strategic planning and management arrangements that include ministerial strategic planning, led by Office of the President to support the move towards an Integrated Results Based Management (IRBM) system,3 as well as high level policy guidance and monitoring of NDP implementation which was recently introduced and led by Thematic Working Groups (TWGs)4 and Mid-Term Review (MTR) of NDP10.

3. The existing planning frameworks have not been mutually reinforcing, leading to slow progress in realization of intended benefits from the NDP approach. The spending profiles for the NDP10 are restricted to development projects that only represent 20 percent of total public expenditure, leaving the majority of policy priorities uncosted. The recurrent budget remains on an annual incremental basis. In addition, there are major concerns regarding: (i) the disconnect between expenditure allocations and policy outcome; (ii) the little consideration of the longer term impacts of policies that have been endorsed during the annual budget; and (iii) the sustainability of the course of spending in light of anticipated medium to long-term falls in diamond revenues.

4. Accordingly, the authorities with FAD technical assistance have since 2010 embarked on a gradual introduction of a policy-driven budgeting within a medium-term framework. The 2010 FAD missions provided guidance on reform priorities for enhancing public expenditure management and specifically made suggestions to improve the annual budgeting process as well as strengthen macro-fiscal foresting and policy analysis as key prerequisites for a successful introduction of an MTEF. Since September 2012, the authorities, with support of an EU-funded FAD resident macro-fiscal advisor, have made significant advances in the development of a Medium-Term Fiscal Framework (MTFF), providing a sound fiscal framework for budget preparation. In addition, a recent November 2012 mission, working closely with the authorities, prepared a reform strategy and road map for the introduction of the MTEF.

A. Benefits of a Medium-Term Expenditure Framework

5. The uncertain and fragile external environment continues to pose risks to mineral export demand and calls for prudent macroeconomic and fiscal management. The authorities have reaffirmed their commitment to deliver a medium-term fiscal consolidation and rebuild fiscal buffers. The main thrust is to enhance fiscal management over the medium term and ensure prudent management of potential threats to macro-economic and fiscal stability. The Ministry of Finance and Development Planning (MFDP) has institutionalized a MTFF to forecast government revenue and expenditure over a three-year period, and serve as a foundation for the development of a Medium-Term Expenditure Framework by the end of 2016. The concepts are explained in Box 1.

What is the Difference between an MTFF and a MTEF?

A Medium-Term Fiscal Framework (MTFF) is a consistent set of fiscal projections underpinned by macroeconomic projections that describe the medium-term path of fiscal aggregates, such as total revenue, expenditure, balance, and debt levels. These are generally projected at an aggregate level, according to the macroeconomic projections.

A Medium-Term Expenditure Framework (MTEF) goes a step further, including all the elements of an MTFF, as well as detailing the policies, programs, or ministerial allocations underlying the aggregates, both on the revenue and expenditure side. These can be indicative allocations, based on forward estimates, or hard ministerial ceilings.

Note: The concepts are clearly illustrated in the Report of the November 2012 Mission executed by FAD expert Mr. Steve Gurr. The Report explains the concepts, principles and benefits of medium-term frameworks, including the prerequisites within the local context of Botswana (See Appendix I).

6. An MTEF provides a set of interrelated processes and institutional arrangements that further strengthens prioritizing within national development priorities and management of expenditure demands in a multi-year perspective. The framework, therefore, comprises all the systems, rules, and procedures that help government manage the tension between demands for expenditure (“needs”) and ultimately ensure effective allocation and use of expenditure between and within sectors and priorities, as well as enhancing predictability of funding for the accounting officers over several years. The ultimate outcome—also normally referred to as the MTEF—is the actual multi-year numerical revenue and expenditure projections and allocations presented alongside a given annual budget over a multi-year period.

7. Moving to an MTEF will bring the following benefits to the budget decision making process:

  • Strengthening of macroeconomic and fiscal forecasting to provide robust projections, enhance credibility of the budget and medium-term forecasts, and improve predictability in flow of resources. This would involve strengthening the existing institutions responsible for managing macroeconomic and fiscal policy framework.

  • Reinforcing fiscal discipline by raising the focus beyond the immediate year, ensure longer term consequences become a more important part of the fiscal decision making as well as promoting strategic reviews to focus on issues of efficiency and realization of potential savings.

  • Creating greater discretionary space for governments to allocate resources to higher NDP priority areas. This would require the government to address the existing legal constraints that foster dual budgeting process and undermine strategic allocation and efficiency in use of resources.

  • Providing guidance on the path of priority expenditures over the medium term by bringing a greater degree of rigor to those priorities through necessitating their better costing and prioritization.

  • Providing greater transparency and predictability to accounting officers about their likely future resources at the expenditure level that matters to them. This brings increased certainty over budget funding, encouraging better engagement in longer-term planning, and efficiency in execution of multi-year programs (such as the building of a road or the building of a new school).

8. Given the current capabilities within the planning and budgeting process, caution should be exercised in trying to move to an MTEF too quickly. There are a number of areas that should be strengthened to increase the effectiveness, credibility, and usefulness of an MTEF when it is ultimately adopted. Key among these are:

  • Building the macroeconomic, revenue, and expenditure forecasting capabilities in order to increase the accuracy of the forecasts underlying the annual budget and develop the rolling three-year forecasting capability.

  • Developing a medium-term fiscal framework which focuses on the key fiscal aggregates and uses a fiscal rule to convert the forecasts into a binding aggregate expenditure ceiling.

  • Unifying the budget process, particularly by fully integrating capital investment decision making with the ongoing recurrent budget process.

B. Building a Wider Political Commitment for the Reform

9. While the MFDP is committed to introduce the MTEF reform, there is an urgent need for endorsement and support from ministers and other policy makers. An MTEF is a complex reform that requires sustained political support and leadership, dedicated resources, and an extensive sensitization program for policy makers. The policy makers, in particular Cabinet, needs to understand the objectives, benefits, and implementation of the reform as well as links to other on-going public sector reforms, such as enhanced performance management and capacity enhancement. MFDP is encouraged to demonstrate to Cabinet the critical role of the MTEF in fiscal consolidation, and discuss their role in pursuing a smooth and sustained implementation of the reform agenda.

10. TWGs provide an appropriate political forum for pursuing the MTEF reform. While this major endeavor is outside the scope of this report, the mission recommends that during the on-going development of the TWGs role, the authorities should focus on mandates, deliverables, and institutional coordination arrangements that would provide strategic direction and influence budgetary decision-making at all levels of government, namely: sector, line ministry, MFDP, and Cabinet.

11. The report is organized as follows: Section II addresses current issues related to the macroeconomic and fiscal foresting and suggests areas of improvement to enhance the credibility of the MTFF; Section III discusses the design features of a suitable MTEF in the context of Botswana; Section IV describes the setting of medium-term budget and expenditure ceilings, including forward estimates (FE); and finally, Section V makes recommendations to strengthen the annual budgeting process to support multi-year planning and budgeting.

II. Improvements to Medium-Term Macro-Fiscal Forecasting

12. The government is committed to strengthening medium-term macro-fiscal forecasting as a prerequisite to the successful implementation of an MTEF. The government, with support of EU-funded technical assistance, has made satisfactory progress in developing and using an MTFF with evidence of increasing prominence being given to macro-fiscal issues within the MFDP. A review of progress against the recommendations of a previous 2010 FAD TA mission5 is presented as Appendix II. Notwithstanding, the current practices pose several challenges to a desired efficient MTFF process. This section examines the major issues as well as makes recommendations to enhance the credibility of the MTFF and refine macro-fiscal forecasting, including construction of aggregate expenditure limits.

A. Major Issues


13. Botswana faces a relatively volatile economic environment. As illustrated in Figure 1, GDP growth is particularly unstable in comparison to a peer group of middle income countries and Southern African Customs Union (SACU) neighbors. This reflects the unstable diamond sector and its impacts on GDP growth year to year. This impact was particularly sharp during the mineral sector downturn between 2008 and 2010 predicated on the 2008 global slowdown.

Figure 1.
Figure 1.

Volatility of Economic Environment

Citation: IMF Staff Country Reports 2014, 162; 10.5089/9781498302166.002.A001

14. Despite volatile GDP growth, revenues and inflation are both moderately stable. Whilst trends in the sub-components of revenue are volatile, in particular mineral revenues and transfers from the SACU, these have tended to counterbalance each other leading to relatively stable revenues in comparison to peer group countries. Similarly inflation, whilst remaining above the medium-term objective range, has been stable relative to peer group neighbors. This suggests that forecasting domestic revenues with some degree of accuracy should be possible.

15. The authorities are aware of the two significant sources of uncertainty that present challenges for both GDP and revenue forecasting.

  • The diamond sub-sector, which traditionally accounts for 25 percent of GDP and 33 percent of government revenues, has proved exceptionally volatile since 2007.

  • Transfers from the SACU, which have been growing sharply, almost doubling between 2010/11 and 2011/12, and accounted for a third of revenue in 2012/13.

16. In light of the above, the authorities need to ensure that the forecasting framework accords special attention to the developments in both mineral revenues and SACU transfers, including the major drivers highlighted below:

  • Mineral revenues are driven by volume of diamond production and international diamond prices. The former is growing increasingly volatile, as Botswana becoming a “swing producer” of diamonds and production rising and falling to balance market demand.

  • SACU transfers, determined by a revenue sharing formula, have two main sources of volatility:

    • 1) Revenue payments that are dependent on both intra-SACU and extra-SACU trade levels, which tend to be pro-cyclical and can swing by up to 30 percent from year to year.

    • 2) The two-year lagged technical adjustment, which adds or subtracts any over or underpayment from the estimated payment made two years prior.


17. The long-term forecasts of GDP growth published with the NDPs have typically proved to be optimistic. This is illustated in Figure 2. It is not possible to assess the credibility of the GDP forecasts6 that inform the annual budget as sufficient data on past forecasts are not available, which in itself is a concern. The current method of forecasting output is based on a 10 sector model and produces understandable and transparant forecasts. Other more technical tools based on econometric estimation are being developed with assistance from the South African Treasury. A Computable General Equilibrium (CGE) model is also being updated to allow modeling of policy simulations, and considers diamond stockpiling and economic diversification issues. It is important that legacy forecasting methods are retained as new ones are developed. This will allow each models’ results to be compared to the other, and act as a quality assurance tool.

Figure 2.
Figure 2.

Budget Forecasts and Outturns

Citation: IMF Staff Country Reports 2014, 162; 10.5089/9781498302166.002.A001

Source: Annual budget documents, National Development Plans

18. Forecasts of revenue and expenditure have traditionally varied significantly from the budget, with the Government usually returning surpluses larger than originally forecast in the budget. As Figure 2 shows, revenues are persistently underforecasted, with an average error of 3.6 percent of GDP, whilst expenditures usually underperform, with an average error of 2.8 percent of GDP. These variances necessitate significant within year adjustments to the budget whilst excess revenues can create pressure for new spending requests to be financed.

19. Credibility of the budget is further weakened by persistent underspending on the development budget. The development budget comprises “projects” which are incorporated into the NDP, and accounts for approximately one quarter of spending. Other than 2008/9 and 2009/10, when stimulus spending saw large upward revisions in development spending, the development budget underspent by an average 15 percent, or 1.5 percent of GDP, since 2005/6. The authorities are cognizant of the underspending, which reflects capacity constraints in implementing agencies to plan and execute projects.

20. Together, the systematic underforecasting of revenues and overforecasting of expenditures has had the effect of limiting government spending, producing higher than forecast surpluses. This approach is not compatible with an MTEF and presents a considerable challenge to establishing a credible expenditure framework.

21. The MTFF should present “best estimate” forecasts. An MTFF should aim at providing a realistic picture of future revenues and expenditures, and hence should not systematically under-forecast revenues nor over-forecast expenditures so as to limit ministerial spending demands. Rather a policy of expenditure restraint should be adopted through a set of fiscal rules or policy objectives, agreed by both Cabinet and Parliament, which are subsequently incorporated in to a fiscal framework.

22. Given that development expenditure is persistently underspent, the MTFF should include an underspending allowance. The underspending allowance would reduce development spending forecasts, and hence provide accurate forecasts of the actual expenditure outcome. The MFDP should formulate an under spending allowance based on average under spends over previous years, not including 2008/09 and 2009/10 when large revisions were made to the development budget as part of a fiscal stimulus package. This is a short-term measure. In the longer term, this issue should be addressed through reforming the development budget preparation process to make investment budgets more realistic.

23. The MFDP has no set process for testing the accuracy of past forecasts. Currently, there is no systematic record of part forecasts, other than sporadic publication in annual budget documents. Analysis of budget variances is hampered by the way the Government Accounting and Budgeting System (GABS) reports budget numbers. By default GABS reports only revised budget estimates, making it difficult to produce reports that detail the original budget, as approved by Parliament at the beginning of the financial year to make investment budgets more realistic.

24. The MPS should establish a tool to track and test the accuracy of past forecasts. This will reduce the risk of repeating forecasting errors in the future. The tool should simply take the form of a spreadsheet that records forecasts of GDP, inflation, and major fiscal aggregates, updated every quarter when new forecasts are produced. It is important that only final, agreed forecasts are recorded, otherwise the tool risks becoming overwhelmed with records of draft and provisional forecasts. The results from this exercise should be presented annually to the high level Botswana Modeling and Forecasting Group (see Figure 3), accompanied by a paper that details where the sources of errors emerged and recommendations on how processes can change to reduce the risk of repeating such errors.

Figure 3.
Figure 3.

Modeling and Forecasting Groups in Botswana

Citation: IMF Staff Country Reports 2014, 162; 10.5089/9781498302166.002.A001


25. The coverage of the MTFF is limited to Budgetary Central Government. The Fiscal Framework therefore does not adequately capture “Special Funds” nor the activities of various levels of local government or government parastatals. The MTFF only includes a single line item for “Grants & Subventions” under which transfers to parastatals and local government are captured. This risks creating an incentive for budget entities to channel spending through these areas increasing potential fiscal risks.

26. The coverage of the Fiscal Framework should be expanded to include Special Funds and local government activities. In Botswana, Special Funds are defined as revenue sources collected by government but retained by the collecting agency rather than being remitted to the consolidated fund. The Authorities report that these Special Funds are small (less than 1percent of GDP) but are growing in size. Local governments (comprising District Councils and Land Boards) are taking on even greater responsibilities for service delivery and collect various revenues that fund their ongoing operations. Capturing these revenues within the Fiscal Framework would assist in overseeing their fiscal operations. The authorities need to define key undertakings incorporated in a reform action plan to guide the expansion of the coverage.

27. There should be closer monitoring of parastatals to better anticipate emerging fiscal risks. The MFDP, working closely with relevant line ministries, should formulate the required processes and guidelines to enhance monitoring of the performance of parastatals helping in quantifying these risks and building mitigation policy measures into the MTFF. This would require an enabling legal and policy framework to support effective reporting mechanisms. The mission suggested a technical assistance mission to support the authorities in examining the existing legal, regulatory, and institutional framework and make recommendations that would enhance fiscal oversight over the agencies.

B. Refining Medium-Term Fiscal Forecasting

28. With the establishment of the Macro-Fiscal Working Group (MWG) in May 2012, Botswana has made satisfactory progress in establishing the MTFF. It provides a firm basis for formulation of credible medium-term macro-fiscal policy forecasts. The MWG comprises members from the MFDP and the Botswana Unified Revenue Service, and its principle mandate is to produce and update every quarter an agreed MTFF. The MWG joins two other groups7 which also deal with macroeconomic forecasting but are not yet fully functional. All these groups are coordinated within the Modeling, Forecasting, and Research Unit of the MPS, and headed by a Chief Economist. This is illustrated in Figure 3.

29. The MTFF—in its early stages of development—is still lacking some important elements. In particular: (i) lack of consistency between assumptions produced by the members of the MWG; (ii) the financing side of the Fiscal Framework does not have a feedback mechanism whereby government borrowing or saving affects future year interest payments; (iii) there is no link between the government balance and the stock of government debt or net financial assets; (iv) coverage of the Fiscal Framework is limited to Budgetary Central Government, and hence misses some local authority spending and revenues collected in commercial bank accounts, general funds, or Central Government special funds; and (v) future privatization proceeds (e.g., from the National Development Bank) are presently absent from the Fiscal Framework.

30. Strengthening the quality and reliability of the data used in the framework is an urgent priority. The 2013 IMF Article IV mission highlighted a number of shortcomings in macroeconomic statistics including limitations in the expenditure side of GDP and related GDP deflators as well as lack of timely quarterly data on revenues. It is therefore important to pursue a collective effort that would bring together the producers and users of the data to address the current weaknesses. A fully functional MWG should provide an appropriate forum for all responsible agencies to design and ensure sustained implementation of measures that would improve the quality and timeliness of information and data.

31. It is also vital that there is consistency between the assumptions underlying the macroeconomic and fiscal forecasts being produced by the Bank of Botswana and the MFDP. Currently the MFDP produces forecasts of GDP each quarter, whilst the Bank of Botswana updates its forecast of inflation every two months. It is planned that forecasts of the government balance will feed in to the Bank of Botswana’s inflation forecasts, and that projected government spending will feed in to the MFDP’s GDP forecasts. This sharing of information will assist in ensuring consistency between the modeling work and policy formulation undertaken by the fiscal and monetary authorities. The authorities have embarked on the development of a financial programming framework to promote consistency in the underlying assumptions for the MTFF. This is led by the Financial Programming and Policy Group.

32. Specifically, the mission encouraged the authorities to ensure that nominal GDP forecasts are consistent with forecasts of inflation. This would mean that forecasted GDP deflators are in line with forecasts of the Consumer Price Index. If these two forecasts are out of line this risks inconsistencies in the assumptions underlying the MTFF that will undermine the credibility of the Fiscal Framework, particularly given the role of inflation as a parameter driving the formulation of medium-term forward estimates (discussed in Section IV).

33. Forecasting of net financial assets is of particular importance given its decline since 2008. This decline is illustrated in Figure 4. The exhaustion of government financial assets limits the extent to which government can smooth out volatile revenue flows.

Figure 4.
Figure 4.

Government of Botswana Net Financial Assets

Citation: IMF Staff Country Reports 2014, 162; 10.5089/9781498302166.002.A001

34. The Fiscal Framework should be expanded to include stocks of government financial assets and government debt. This can be incorporated into the MTFF using the MFDP’s own records on external and domestic debt, netting off lending to parastatals, and the Bank of Botswana’s monthly Botswana Financial Statistics. An agreed definition of net financial assets will be needed to undertake this task.

35. The rolling MTFF process has still to firmly establish itself in the political decision making process. Whilst forecasts of revenues and expenditures are an established part of the NDP process, rolling MTFF forecasts have only been presented to Cabinet since 2012, as part of the Budget Strategy paper.

36. Efforts should be made to increase awareness of the macro-fiscal forecasts across government and engage Cabinet early enough in the formulation of the budget strategy. This can be achieved through ensuring the macro-fiscal policy objectives and forecasts are explained in Budget Options Paper which should be discussed and agreed by Cabinet early enough in the budget process, for example in August or September. The reform is elaborated in Section V. This would have to be complemented through traditional communication methods such as presentations and meetings with stakeholders, including budget officials and senior line ministry staff, and articles in the Daily News, a free newspaper that is distributed to civil servants every day.

37. Documentation8 explaining the assumptions on which the MTFF is based should be improved. This would include adding additional detail to the “Medium-Term Fiscal Framework Projections” report, providing sufficient information so that an outsider could replicate the forecasts. A set of short technical papers should also be drafted detailing the revenue elasticity calculations and efficiency factor assumptions.

38. Recommendations

  • The MWG should design and implement both measures to improve the quality and reliability of data generated from various sources and processes to track and test the accuracy of past forecasts so as to elimate systematic forecasting errors from the MTFF.

  • Formulation of a reform plan to support the increasing coverage of the MTFF over a medium term.

  • Review and strengthen the existing legal, policy, and institutional arrangements to facilitate both better strong fiscal oversight over the parastatals and identification of emerging fiscal risks.

  • Efforts should be made to broaden dissemination and understanding of the MTFF and its implication for the MTEF across government.

C. Construction of an Aggregate Expenditure Ceiling

39. Two linked but separate processes currently guide the construction of the recurrent and development budget ceilings. The recurrent ceiling is determined

  • Either as a residual after the development ceiling and targeted surplus or deficit is subtracted from the forecasted revenue envelope, or

  • As equaling the recurrent expenditure forecast in the MTFF, which is based upon a forecast of the wage bill and forecasted CPI inflation.

40. The development budget ceiling is usually set as a percentage of government revenues or expenditures. NDP9 set the development ceiling as equaling 30 percent of forecasted revenue. NDP10, drafted at the height of the financial crisis, called for a 4.6 percent year-on-year reduction in development spending from 2008/09 levels, whilst the June 2013 Mid-Term Review of NDP10 set the Development ceilings to equal 20 percent of forecast government spending, which in turn grew at half the rate of forecasted revenues. Botswana has historically adopted non-statutory fiscal rules, limiting the size of government expenditure to a given percent of GDP9 but have not been meaningfully binding. Box 2 discusses the setbacks of the practice. In particular, the current fiscal rule stipulated in NDP10 has two important drawbacks articulated in the 2010 technical mission report. First, a level of government spending of 40 percent of GDP is not consistent with long-term fiscal sustainability which would require a substantially lower level of spending. Second, it does not decouple spending decisions from changes in the international price of diamonds or other cyclical factors. In fact, when the price of diamonds increases, GDP also increases allowing an increase in government spending. The authorities are aware of the issues and are committed to revise the limit.

41. International best practice suggests that expenditure ceilings be set in line with an appropriately binding fiscal regime rather than systematically underforecasting government surpluses. The following steps illustrate how expenditure ceilings can be calculated using a rule that aims to balance the budget over the life of NDP10: (i) forecast total revenues for Budget Year (BY) and BY+1; (ii) calculate what deficit or surplus is required in BY and BY+1 to have a zero cumulative budget deficit over the years of NDP10; and (iii) subtract the deficit or surplus number from the revenue forecast in each respective year, giving an expenditure ceiling for BY and BY+1. These calculations are illustrated in Table 1.

Table 1.

Deriving Expenditure Ceilings from the NDP10 Balanced Budget Rule

(Pula millions)

article image
Source: MFDP.

Need to Strengthen Botswana’s Fiscal Regime

Fiscal rules are scarce in small middle-income countries such as Botswana.

In Botswana, there is one existing formal fiscal rule, complemented by two guiding principles embedded in the NDPs.

Implicitly these rules aim to keep government spending on a sustainable path. Nevertheless, the 40 percent expenditure limit was breached in financial years 2008/09 and 2009/10, and whilst the debt/GDP ratio is within its statutory limits, it places no restriction on the drawdown of government reserves and therefore provides no actual guidance for setting annual and medium-term budget ceilings. The government of Botswana’s net financial assets has fallen markedly from over 60 percent of GDP in 2008 to less than 20 percent of GDP in 2013, as illustrated in Figure 4.

In 1994 the Sustainable Budget Index (SBI) was introduced. The SBI is viewed as a framework for managing diamond wealth rather than a formal fiscal rule per se, and even then, there have been significant deviations from the SBI. By the early 2000s, the SBI had ceased to play a role in budget policy decisions.

The recently published Mid-Term Review of NDP 10, approved by Parliament, forecasts budget surpluses for 2012/13 through 2015/16. This supports a policy of running surpluses to build up Government reserves although the Mid-Term Review forecasts a cumulative deficit over the life of the Plan equivalent to 10 percent of 2013/14 GDP.

Hence, the existing fiscal regime (rule and guiding principles) is not meaningfully binding, and cannot be practically used in setting aggregate expenditure ceilings as part of an MTEF. The 40 percent expenditure limit rule, if used to set expenditure ceilings, would allow a large jump in spending. The debt limit rule is even less binding and would allow large deficits to be run, financed by running down reserves.

In conclusion, it may be necessary in Botswana to review the existing fiscal regime to establish alternative policy measures that would enforce overall fiscal discipline.

42. To improve budget discipline, the aggregate expenditure ceiling should be set for the coming two financial years, with an indicative ceiling for the third year. This approach would bring greater certainty and stability to government expenditures given the sometimes volatile revenue streams. This certainty would assist the Government to meet its ambitions to reduce the size of expenditure as a proportion of GDP, and fixed two year ceilings would benefit line ministries who can plan accordingly. This is discussed in greater detail in Section III. An appropriate time in the budget cycle for this ceiling to be set would be in August once the first quarter GDP data is released and GDP and MTFF forecasts are updated.

43. In addition, the mission encouraged the authorities to expedite the on-going revision of the fiscal rules, taking into account the major consideration raised by the previous 2010 technical assistance mission report. The report cautioned that movements in total GDP related to developments in the mining sector, usually prices, can cause the GDP ratios of fiscal variables to vary substantially. This could lead to an erroneous conclusion that there has been a change in fiscal policy instruments, even though all that has happened is that the international price of diamonds has changed. It recommended that formulation of the fiscal rules should target the non-mining balance (in percent of non-mining GDP) to decouple government spending from fluctuations in the international price of diamonds. To this end, technical assistance should be sought to examine the effectiveness of the existing fiscal regime and to advise on the scope of fiscal rules and objectives.

44. Recommendations

  • The government should adopt an appropriate and binding fiscal regime that would guide setting aggregate expenditure ceiling.

  • The aggregate expenditure ceiling should be more comprehensive and cover all public spending, not only those of central government but also those of EBFs, local finances, and parastatals.

  • Technical assistance should be sought to examine the effectiveness of the existing fiscal regime and to advise on the scope of fiscal rules and objectives.

III. Designing the Medium-Term Expenditure Framework

45. Before adopting an MTEF, the government should consider what form the framework will take in order to meet its objectives. An ideal MTEF model does not exist, but there are a set of choices to be made in order to instill discipline, create legitimacy, ensure enforcement, build credibility, and optimize resources allocation. Consideration will need to be given to:

  • How the multi-year spending limits will be designed and on what basis;

  • How the multi-year prioritization process will be organized;

  • What control mechanisms to employ to enforce the commitments; and

  • What accountability mechanisms will be used establish credibility.

46. The mission conducted a number of working sessions with the authorities—represented by BWG—to discuss the objectives and design characteristics of the MTEF. The group examined the design considerations as well as international examples in both the advanced and Sub-Saharan Africa contexts and suggested design aspects that relate to Botswana. This chapter discusses the major design considerations, raises the key issues discussed with Budget Reform Working Group, and presents recommendations for the design of the MTEF.

A. Design of Multi-Year Spending Limits

47. One of the main challenges with adopting an MTEF is ensuring that governments remain committed to expenditure levels for years beyond the legislated budget. As time goes by, government will be tempted to move away from what it had said it would do in the future. This can result from changes in the external environment (such as higher than expected revenues or inflation, or unanticipated drought), new policies, or simply due to oversights in the formulation of the original medium-term plans. In all of these instances, a case can be made for allowing for some upward revision to the original multi-year estimate.

48. However, accommodating these pressures reduces the credibility and value of the MTEF. If the government sets expenditure limits for the future, but fails to meet them, there is little credibility in the funding decisions and commitment path. This creates a risk that ministries of finance will cease to invest time and effort in the production of accurate and updated medium-term fiscal projections, and line ministries will not focus on medium-term implications of policies, thereby negating all of the benefits outlined above. This has been the experience of a large number of countries who have sought to introduce MTEFs, and resulted in the MTEF becoming simply a set of numbers that have little bearing on actual spending outturns over the medium term.

49. The MTEF should aim to resolve the tension between a firm commitment to medium-term expenditure ceilings and ensuring that previous plans are still relevant at the time of implementation. Four main questions arise in defining the multiyear expenditure limits that define that framework:

  • What should the nature of the commitment be? To a nominal or real figure, or to a specific level of policy deliverables?

  • What should the level of detail be? Aggregate expenditure ceilings or ministerial ceilings?

  • What should the coverage of the ceilings be? Should some items be excluded due to their unpredictable or unavoidable nature?

  • How often should multi-year ceilings be revised?

50. In initial discussions with the Budget Reform Working Group (BWG), they were inclined towards setting fixed ministerial ceilings for three years into the future, in much the same way as the annual budget is allocated. However, they recognized the difficulties inherent in adopting this approach, in light of international experience. Specifically, setting such rigid ceilings at such low levels means reducing the credibility of the ceilings, in recognition that some areas of expenditure are too unpredictable to set three years in advance, and that many of these expenditure areas, such as pensions and interest, would need to be funded regardless of the ceilings. Agreement coalesced around the approach laid out below.


51. Nature and level of commitment: The design for Botswana’s MTEF should be based on an aggregate nominal expenditure ceiling extending over three years. The government has a strong record of delivering aggregate expenditure within its annual budget appropriation (see Figure 5), and the current budget execution system is focused on ensuring this occurs. The relative stability of inflation in Botswana relative to peer group neighbors in SACU means that such an approach is workable and would avoid adding to the complexity of the ceilings by defining them in real terms. In addition, the MTEF should include indicative ministerial expenditure allocations over the medium term in order to improve the focus of allocative decision making as is the current practice.

Figure 5.
Figure 5.

Botswana Aggregate Expenditure: Original Budget vs. Outturn

(Percent of GDP)

Citation: IMF Staff Country Reports 2014, 162; 10.5089/9781498302166.002.A001

52. Coverage of expenditure: For simplicity and completeness, the expenditure ceiling should cover 100 percent of total expenditure. While there is an argument for excluding some items (i.e., such as interest expenditures because they are highly predictable, or extra-budgetary funds, social security, and local government grants, which are difficult to track10) authorities are encouraged to formulate two categories of expenditure ceilings.

  • First, the overall aggregate level of expenditure that maintains an overall fiscal position for any particular composition of spending in the medium term. In essence, it would be necessary to include a non-ministerial or “other” expenditure line within the framework to incorporate any non-ministerial expenditure coming from extra-budgetary funds or parastatals in order to align the MTEF with the aggregate ceilings in the MTFF.

  • Second, a medium-term expenditure ceiling that is available for allocation to line ministries.

53. Revision of multi-year ceilings: The recommended approach is for the MTEF to be set on a rolling basis, with fixed ceilings for years one and two, and an indicative ceiling for year three, as illustrated in Figure 6. This reflects the relatively volatile and uncertain economic environment facing Botswana. Thus the BY and the (BY+1) aggregate expenditure would be set and fixed, while the (BY+2) ceiling could be revised in the future if economic or revenue conditions change. As the new budget year rolls around, the annual budget ceiling remains as set the previous year (BY+1), and the previous budget year plus two (BY+2) ceiling can be adjusted if necessary to produce a new (BY+1), and a new indicative ceiling for BY+2 is added to the MTEF.

Figure 6.
Figure 6.

Illustration of Proposed Rolling Expenditure Ceiling Process

Citation: IMF Staff Country Reports 2014, 162; 10.5089/9781498302166.002.A001

54. Setting a fixed total expenditure ceiling should go some way to reducing the volatility in overall expenditure, by looking through year-to-year revenue fluctuations. As demonstrated in Section II, Botswana’s aggregate expenditure volatility is exceptionally high amongst comparator countries. This volatility has the potential to have a number of detrimental impacts. First, it can contribute to macroeconomic volatility, though it does appear that some of the volatility in Botswana’s expenditure reflects countercyclical fiscal policy. Second, it creates uncertainty for accounting officers, who must deal with uncertain overall resource envelopes and projects that take place over multi-year horizons.

55. This design also meets the government’s objective of maintaining expenditure discipline in an environment of slowing revenue growth. Over the past decade, average annual revenue growth of 11.2 percent allowed for large increases in budget resource envelope, meaning that the focus of budget processes have been on allocating spare fiscal space. This is not anticipated to continue, with revenue growth forecast to increase by a more pedestrian 7 percent over the next five years. Combined with the desire to reduce expenditure to 30 percent of GDP, return the budget to surplus over the medium term, and rebuild the government’s net asset position which has fallen from 80 to 20 percent of GDP since 2006, this will require considerable aggregate expenditure restraint into the future.

56. Nature of ministerial expenditure ceilings (limits): The option of set binding ministerial ceilings—designed to meet the government’s aim of shifting expenditure to higher priority areas—was considered but ultimately rejected. As indicated in PEFA 2013 report,11 the existing large variations between ministerial actual expenditures (out-turn) and budget appropriation meant that binding ministerial ceiling would not be a viable option in the immediate future. The degree of confidence that could be placed in the ceilings over the medium term would be limited, thus reducing their effectiveness as a commitment mechanism. Further, any ministerial ceilings set without the information provided by well-established bottom-up expenditure forecasts—forward estimates that take into account the expenditure drivers that alter the cost of delivering existing level of services over the medium-term—risk being unrealistic and unable to be met in following years.

57. Authorities agreed to set indicative ministerial expenditure allocations over the medium term to improve the focus of allocative decision making. These ministerial spending allocations would have be determined in the manner explained in Section III, based on the newly produced forward estimate process, with an initial allocation of fiscal space in line with priorities. While they would not be binding in nature, but in order to lend credibility and focus decision making on medium-term implications, they should represent the starting point in the annual budget process, against which any parameter changes and policy decisions would be reconciled against, as demonstrated in Figure 7 below.

Figure 7.
Figure 7.

Illustration of Proposed MTEF Model

(pula millions)

Citation: IMF Staff Country Reports 2014, 162; 10.5089/9781498302166.002.A001

B. Prioritizing Expenditure over the Medium Term

58. The credibility of and respect for MTEF allocations requires that they reflect both the government’s policy priorities and are consistent with the multiyear spending limits. Achieving these dual objectives requires institutional mechanisms that allow competing policies to be prioritized in a manner that takes into account their medium-term budgetary impact. These institutional mechanisms are:

  • an integrated medium-term expenditure planning and budgeting process;

  • a clear separation between the cost of maintaining existing policies and the cost of new policy initiatives in budget documents, based on an unambiguous and widely accepted methodology;

  • a forum for discussing and deciding on expenditure priorities that is perceived to be comprehensive, politically legitimate, evidence based, and binding; and

  • the provision of sufficient information to guide budgetary decisions discussed in Section V.

Integration between the Annual Budget and MTEF

59. There are different ways to integrate the annual budget and the MTEF

  • The first sees the medium-term aspect fully integrated into the annual budget process, with all budget decisions made over the medium term, and annual budgets and MTEFs merged into a single set of budget estimates. Thus, all decisions are taken over the three year period, and the MTEF is presented on the same basis as the budget estimates.

  • The second, less ambitious approach maintains an annual perspective at the most detailed budget level, with medium-term decisions and projections produced at a more aggregated level than for the annual budget.

60. Currently, Botswana’s planning and budget processes are not integrated, an approach that has found to have limited impact on actual policy prioritization. Medium-term planning occurs only once every three years, and is undertaken separately from the annual budget (Section III elaborates).


61. Initially, the MTEF decisions and presentation should be at a more aggregate level than the annual budget estimates, with the out years estimates being provided at line ministry level but with more accuracy as they evolve into the budget year. The MTEF and its allocation decisions should remain at the aggregate ministerial level, while the annual budget estimates continue to be provided at the detailed input line-item level. This is because the current level of specification of the budget estimates is too detailed, and trying to set the MTEF up on that basis would be too specified, too complex and too burdensome to maintain.

62. Eventually Botswana should aim to have the MTEF and Annual Budget prepared at the same level of detail and classification. However, this convergence would require the annual budget to shift towards a program-based budgeting, which will provide a more amenable platform to align both the MTEF and annual budget decisions and presentation. The introduction of program based budgeting should be seen as a long-term initiative that requires significant undertakings, not only improving an existing annual budgeting process but also designing a feasible program structure: restructuring of the line ministries to align the organizational set-up with the programs and introduction of performance based management culture.

Separation of existing and new policies

63. Currently, the practices do not present a clear separation between the cost of providing existing policies and the proposed cost of new policies in the budget. Thus, it is not possible to identify the added cost or impact in the budget of the government’s new policy announcements.


64. By using the forward estimates as a key tool within the MTEF, it will be possible almost immediately to separate the baseline from the impact of new policies over the medium term. Section IV explains the approach to calculating forward estimates, which includes a clear difference between maintaining expenditure on a no-policy-change basis and the impact of new policies.

65. This presentation should be included in both the Budget Strategy Paper and the Annual Budget immediately once the MTEF comes into effect. An example of such a presentation is provided in Figure 9.

Figure 8.
Figure 8.

Major Steps in Producing the Forward Estimates

Citation: IMF Staff Country Reports 2014, 162; 10.5089/9781498302166.002.A001

Figure 9.
Figure 9.

Reconciliation Table for Ministerial Expenditure Ceilings

Citation: IMF Staff Country Reports 2014, 162; 10.5089/9781498302166.002.A001

Deciding between different priorities

66. The final element of medium-term policy prioritization is the existence of a forum for deciding between competing expenditure pressures and proposals. Although the nature of these decision making mechanisms depend on the institutional and political context, the most effective of these forums tend to: (i) cover all government activities; (ii) engage high-level political decision makers; (iii) make use of evidence about expenditure performance; and (iv) represent the sole and final decision making authority within the executive.

67. Botswana’s recurrent budget formulation process is led by a committee of Permanent Secretaries, with Cabinet approving and acting as an appeal mechanism. This approach reportedly operates effectively. However, there are some improvements that could be made to the information that feeds into decision making, including multi-year costings and realistic costing of projects, to reduce cost overruns and delays in implementation.

68. The current approach appears to be working well, but may need increased political commitment to ensure that medium-term allocations have political buy-in. As detailed above, medium-term allocations free up considerably more space for discretionary policy decisions than annual budgets, and allocating that fiscal space towards priorities areas with any legitimacy requires political buy-in, albeit dependent on advice from within the bureaucracy.

69. Four thematic working groups have recently been formed to prioritize spending with broad sectors of the NDP10. These groups are only in their early stages of formation, and thus not yet fully operational. The responsible coordination agency, the National Strategy Office (NSO), has to deal with issues around accountability, as the TWGs cut across a range of different ministries and their role in setting budget priorities remains unclear. The recommendations from TWGs could provide policy guidance around the allocation of additional fiscal space undertaken by the Cabinet sub-committee, or to direct line ministries in how to fill in any unallocated fiscal space as determined by the Cabinet sub-committee.


70. The recommended option is to raise the political level of decision making early on in the budget process. This would see a sub-committee of senior Cabinet ministers making decisions on the allocation of any additional fiscal space as part of approving the MTFF early in the budget process. This would leave the Permanent Secretaries’ committee to determine the detailed budgetary allocations. The sub-committee would be informed by inputs from the TWGs and Permanent Secretaries in the form of the Budget Option Paper, consolidated by the MFDP. The reform is detailed in Section IV.

C. Maintaining Expenditure Control over the Medium Term

71. Once the expenditure limits and indicative spending allocations have been set, there needs to be a range of controls in place to ensure they will be delivered over a multi-year period. These controls should include:

  • regular updates of medium-term expenditure forecasts to ensure the government knows where it stands;

  • allowing sufficient margins between expenditure allocations and expenditure plans to absorb unexpected events without requiring reprioritization of policies;

  • firm controls on ministries’ and agencies’ ability to enter into multiyear expenditure commitments; and

  • controls over the accumulation, stock, or drawdown of carryovers.

Regular updates of expenditure projections

72. Frequent overshooting of expenditure limits set in the MTEF will quickly reduce its credibility. Safeguarding the integrity of the framework requires that initial plans be accurate and that decision makers be quickly informed of any emerging pressures so that they can take action if necessary. To ensure this level of accuracy and timeliness of information, high-quality expenditure projections, updated on a regular basis, are needed.

73. Currently the MFDP updates the revenue forecasts once a year and expenditure estimates three times a year. However, these are only focused on the budget year and not based on a coherent forecasting system that is based on a full and consistent macroeconomic forecast, though the recent introduction of the MTFF is going some way to filling that gap.


74. The mission recommended that the proposed forward estimates be updated at least twice a year. This process will provide the government with an indication of emerging spending pressures. This update will need to be embedded within a full forecasting round, where the macroeconomic forecasts are updated, medium-term fiscal forecasts are undertaken, and spending parameters are provided to feed into the forward estimates. This will allow a reconciliation of the top-down ceilings and bottom up estimates that encourages determination of policy and expenditure pressures, as well as formulation of spending decisions that are consistent within macro-fiscal policy objectives and targets as further illustrated in Section IV.

Margins and reserves

75. A credible MTEF requires a buffer between a binding expenditure ceiling and the allocation of resources. Even the best forecasters face uncertainty about future expenditures, and experience across MTEF’s indicate that almost all countries, developed and developing, tend to underestimate future expenditure pressures. Therefore, MTEFs should set aside an unallocated reserve between sum of all projected expenditure and the medium-term expenditure limit. In Botswana, where forward estimates have not been produced and will take some time to mature, this will be particularly important.

76. The function of this margin is to absorb the unexpected expenditure pressures that inevitably emerge. These pressures can come from a number of sources, including:

  • simple forecasting errors that emerge due to unforeseen shocks, such as higher than expected inflation or wage pressures;

  • biased underestimates of the cost of existing policies, which is a relatively common feature of forward estimates, particularly during the early stages of the their development;

  • unavoidable spending pressures that arise due to exogenous shocks, such as the impact of an unforeseen drought that requires support to farmers, or a disease outbreak that needs to be addressed; and

  • policy pressures that arise after the initial allocation of expenditures ceilings.

77. Without a margin, these shocks would have to be absorbed through disruptive expenditure cuts elsewhere, or by increasing the overall expenditure ceiling. The former is a difficult exercise and reduces the certainty provided to line ministries, potentially disrupting spending plans and projects, while the latter would have detrimental impacts on fiscal discipline and the credibility of the binding expenditure limit.

78. Such budget margins come in two main forms:

  • a contingency reserve to absorb any unforeseen forecast variations (the first two types of expenditure pressures) and that lead to forward expenditure estimates being revised up due to parameter changes; and

  • an unallocated planning margin to allow for future policy decisions (the latter two pressures) that aren’t currently built into the MTEF. The use of this margin allows the government to make future spending decision without breaching the aggregate expenditure limit or needing to identify offsetting savings.

79. The size of margins varies across countries but tends to be small in the budget year, increasing in size over the out-years. The budget year margin varies from -0.5 to 1 percent of spending, increasing to 1.5-3 percent of spending in the out years. This profile reflects the fact that budgets expenditures within the budget year are fully allocated, usually held to, and often under spent (thus leading to a negative margin in some countries). The increase in size over the out years reflects greater uncertainty around the range of expenditure forecasts further out into the future, as well as the fact that there are likely to be demands for new spending in the future.

80. Currently there is a P10M (0.02 percent of expenditure) contingency reserve within the annual budget. The existence of this margin is required under the PFM Act; however, it is very small relative to other countries. Nevertheless, the Budget Reform Working Group expressed concern about increasing it any further fearing the potential for ministers to treat the margin as an additional pot of money to be allocated, thereby weakening the primacy of the budget process. This is a valid concern.

81. However, budget margins can introduce an element of gaming into the medium-term prioritization process. Building expectations that greater resources will be available as time progresses can undermine the budget process.

82. This expectation can be prevented by introducing explicit rules and transparency provisions governing how the reserve will be accessed. Some examples were discussed:

  • In the UK, claims on the contingency reserve are permitted only for spending that is deemed to be unforeseeable, unavoidable, and unabsorbable within ministry budgets.

  • In Australia, the contingency reserve can only be used to deal with unexpected variations in forecast parameters in the forward estimates, and not to fund new expenditures.

83. Transparency provisions are also a reputational constraint on accessing the margin. Publishing details on what the margins are used for, particularly when used to absorb overspending and cost blowouts, creates a reputational and political cost to using the margin. This is particularly effective when Public Accounts Committees focus on this, as they tend to do, because cost blowouts are often politically embarrassing


84. The recommended approach in Botswana is to include a budget margin of 2–4 percent of expenditures beyond the budget year. Given the strong track record of delivering budgets within the appropriations, the size of the existing budget reserve can remain at the existing very small size in the budget year. Reflecting that there is no history of producing forward estimates to base any analysis of predictability of spending over the medium term, the proposed margins should provide a relatively large buffer to absorb forecasting errors in the two outer years. The proposed margin of 2 percent of expenditure in BY+1 and 4 percent in BY+2 sits at the top end of provisions amongst MTEFs (Figure 7). The margin should be equally split between a contingency reserve (for expenditure variations) and a planning margin (for new policies).

85. The use of the margins should be governed by clear access rules. The contingency reserve should be reserved solely for unforeseeable parameter variations, identified during later updates of the forward estimates process, and not be available for overspending or cost blowouts. The planning reserve should be reserved only for spending priorities identified later within the budget process, as time progresses, and BY+1 becomes the budget year.

Controls around multi-year contracts

86. MTEF’s often require controls to prevent ministries or agencies from entering into multiannual expenditure commitments that are inconsistent with the agreed on multiyear expenditure plans for their sectors. To be effective, multiyear commitment controls must apply to all manner of expenditure commitments ranging from political promises and new establishments, to proposed bills, and to contractual undertakings. Such a mechanism would require assessment of multi-year expenditure of any new contract or obligation and guidance on the extent to which the expenditures could be accommodated within the aggregate expenditure ceilings before incorporation in the annual budget. A number of countries12—for example, South Africa and Uganda—have restrictive regimes that require all multiyear expenditure commitments to be approved by Cabinet before inclusion in the annual budget and MTEF. Cabinet would seek advice on the spending decisions from the Minister of Finance.


87. The control mechanism has to be strengthened to provide clear guidelines on the approval of new policy decisions with multi-annual expenditure commitments. Authorities indicated that any multi-year contract, especially project investments, are approved by the MFDP before they are included in the annual budget. This process appears sufficient, but should be supplemented by tracking each ministry’s overall contractual commitments and comparing them to expenditure allocations within the MTEF before any contracts or new policy decisions are endorsed. In the case of project undertakings, this would be supported by maintaining an up-to-date contract database.


88. The introduction of an MTEF provides agencies with the possibility of improving planning and certainty of allocations by allowing carry-overs to take place. This allows unspent appropriations to be rolled over from the budget year into BY+1, thus preventing end-of year spending sprees, as unused appropriations are often wasted on low-priority expenditures such as overseas trips and new cars, and allows for capital projects that have uncertain and lumpy spending profiles to be adjusted with minimum disruption.

89. However, unfettered accumulation of carryovers by ministries can create a risk to the sustainability and transparency of the budget. To counter these problems, countries tend to adopt rules or numerical limits governing the amount of unspent appropriations that can be carried over. These limits can take the following forms:

  • limiting the type of appropriations that can be carried over. For example, Australia allows full carryover of administrative costs within departments but requires Cabinet approval and reappropriation for any carryover of program expenditure.

  • limiting the accumulation of carryovers from one year to the next. For instance, as a default position, France and Sweden allow only 3 percent of expenditure to be carried over from one year to the next.

  • limiting the drawdown of carryovers in a given budget year: For example, until 2010, the United Kingdom allowed unlimited accumulation of carryover entitlements by ministries but required treasury approval before those carryovers could be spent in a given budget year.

90. In Botswana, there are currently no carryovers allowed within the recurrent budget. Given the magnitude of under spends against the annual budget, allowing unrestricted carryovers would create a risk to overall fiscal management. However, there is room to allow some carryovers within the MTEF, particularly in regard to capital expenditures.


91. Consideration should be given to allowing some limited carryovers to occur,13 as allowed in the development budget. The rules could allow carryovers for capital expenditure and a limited amount of administrative expenditure (e.g., 2 percent of the appropriated amount) for goods and services expenditure. Salaries and wages should not be allowed to be carried over, nor administered expenditures such as transfers or pensions, which are usually under spent due to incorrect forecasting of their needs rather than due to reprofiling of spending across years. Any carryovers would need to be recorded as an adjustment in the overall budget ceiling, and the global size of carryovers taken into account while setting the macro-fiscal stance of the budget.

D. Accountability and Credibility Mechanisms

92. Ultimately, the credibility of a government’s MTEF depends on its ability to demonstrate that the government is delivering on what it committed to previously. Demonstrating this consistency between previous multiyear budget plans and current budgetary outturns and forecasts requires a set of accountability mechanisms that ensure that:

  • for any given financial year, the MTEF allocations in the BSP, the annual budget, and final accounts are presented on a comparable basis;

  • any variations between MTEFs of different vintages (i.e., from the 2014-15 Budget compared to 2015-16 Budget) are comprehensively and transparently reconciled and explained; and

  • governments and line ministries are held to account for any unjustified deviations from multiyear plans.


93. Presenting the MTEF, budget, and accounts on a comparable and consistent basis allows the original MTEF allocation for any year to be compared to the appropriated budget and eventually the final outturn. For example, this would allow the initial MTEF allocation for FY2016/17, as laid out in the BSP, to be compared to what was appropriated in the 2016/17 Budget, and how that budget was executed in the 2016/17 Financial Accounts. This may require some strengthening of the GABS system.

94. This entails classifying and presenting these three documents in the same way, even if not at the same level of detail. In Botswana, there are a number of ways that expenditures are presented that open up room for confusion when comparing MTEF, budget estimates, and outturn. These include:

  • spending from the consolidated fund and spending from the development fund, included in the financial statements, which require consolidation to put to the complete picture together;

  • the detailed spending presentation on an administrative basis in the budget estimates that are not aggregated and presented clearly; and

  • the MTFF presentation that is comprehensive but presented on an economic basis, rather than an administrative basis.


95. In preparing an MTEF, Botswana should develop a single comprehensive MTEF presentation that is provided consistently across all documents in the budget cycle. This should be presented on a high level administrative basis, with recurrent and development expenditures for each ministry, including statutory expenditures. Ideally, these would be presented on a gross basis, including expenditure funded by own-source revenues, even if these revenues are not covered by the expenditure ceilings (see Figure 9 for a stylized example).

Comparing the evolution of the MTEF

96. Successive vintages of MTEF allocations should be presented side by side with each update to help build credibility. At a minimum, this should compare the previously published total expenditure ceiling over the forward estimates with the new expenditure ceiling. Under the proposed approach, the BY and BY+1 should remain unchanged, with the only variation taking place in the third year as it can be adjusted (Figure 9 illustrates).

97. To focus ministerial decision making within the budget cycle, this reconciliation should also take place for each ministerial allocation. Thus, the previous expenditure allocation (BY+1) would represent the starting point of the new annual budget process, and any variations to that should be explained as either a parameter variation to the forward estimates or the result of a policy decision that changes the level of expenditure for that ministry. This process will make it clear that the original, albeit indicative allocation, is the starting point for the new budget process, and requires an explanation of variations that might otherwise go unnoticed or unexplained (Figure 9 illustrates).

98. These reconciliations will make clear the government’s commitment to previously stated targets and allocations. This will provide line ministries, external observers, and market financial markets with confidence that the government will deliver on the commitments to the future that it is making today.

Sanctions for overspending

99. If there are any unjustified deviations from the expenditure limits, some mechanism for holding those responsible to account should be considered. For the MTEF process to succeed and bring benefits to the budget process, the MFDP and Parliament must treat seriously any unjustified deviations from multiyear expenditure plans that threaten compliance with government commitments.


100. The mission recommended the following sanctions:

  • Administrative sanctions for agencies that forecast an unjustified breach of their indicative allocation, such as some freezing of proposed budgets, to be held as a reserve against overspending, until agencies can demonstrate that they can work within their agreed resource envelope.

  • Greater reputational costs, with public accounts committees focusing on unjustified spending variations, due to cost overruns.

IV. Setting Medium-Term Expenditure Ceilings

101. A sustained implementation of a credible MTEF hinges on reliable multi-year costing of ongoing and new policies of programs over a medium-term framework which is articulated in the previous section. In that context, this section provides practical guidance, including processes and techniques that should be adopted in construction of medium term expenditure ceilings by the lines ministries within a top-down macro-fiscal constrained framework.

A. Producing Forward Estimates

102. The MTEF should be based on a well defined, bottom-up expenditure forecast for each ministry on a no-policy change basis, commonly described as forward estimates. This should be the basis of the expenditure projections prepared by ministries, and should be the best estimate of the cost of continuing all existing policies and levels of service delivery.

103. The forward estimates inform the setting of ministerial expenditure allocations over the medium term. Rather than holding baseline spending allocations unchanged from the last budget year’s budget, forward estimates account for any off-one factors that won’t be continued into the future, as well as identifying spending pressures outside of the ministries control, which if not recognized would make an unchanged allocation unrealistic. For instance, if there was a growing number of school-age children, maintaining the spending allocation of the Ministry of Education fixed (even if in real terms) at the level of the last budget would ignore the growing spending pressure from more students in school, and entail a reduction in spending per student. Similarly, if the number of students were decreasing, such an allocation would result in greater spending per student, an unconscious spending increase that uses finite resources that might alternatively be better used in higher priority areas.

104. Spending allocations will not necessarily be set in line with the forward estimates. For instance, a forward estimate may indicate that there are sharply growing spending pressures in one area of the budget (for instance in the transport sector, due to sharply rising fuel prices). The government may well consider such an increase unacceptable, and set the spending allocation below the forward estimate forecast, requiring the ministry to identify savings. But the use of forward estimates will define just how large a savings decision will be necessary to bring the spending into line with the spending allocation, rather than simply ignoring the pressure to the budget.

Current process

105. Forward estimates are not currently prepared for the recurrent budget. Budget ceilings are determined from top down information only (i.e., the MTFF aggregates), without specific knowledge of the cost of existing programs and projects. The forecast aggregate expenditure ceiling is firstly adjusted to account for estimates of non-ministry expenditure, with the remainder pro-rated between the ministries based on budget share from recent years. The Development Fund adopts a multi-year approach to estimates as a result of multi-year costings being included in the NDP (Volume 2). However, these estimates are often not credible as they are not based on fully specified designs. Despite this, projects are taken into the budget without more rigorous costing. The budget documents currently present only the single year estimates for both the recurrent and development budgets.

Methodology for producing forward estimates

106. Producing expenditure forecasts or forward estimates at the ministry level involves a number of steps. These are focused on understanding the existing budget, understanding and applying the medium-term cost drivers, and aggregating the forward estimates and summarizing the overall sources of variations (Figure 8).

107. Each of these steps need to be identified, analyzed, and have the methodology agreed on between the ministry and MFDP, as the forward estimates form the basis of future budget allocations. Taking the time to agree on the methodology and inputs at the beginning will simplify future budget negotiations, as the agreed baseline can be approved quickly, leaving more time to focus on the higher value activity of assessing and costing new policy initiatives and proposals, rather than negotiating over existing budget allocations.

108. An important factor to remember while first producing forward estimates is that there is the trade-off between complexity and tractability in setting up expenditure forecast models. The forward estimates are models of how expenditure will evolve over the medium term. Thus they are an approximation of reality, and should not necessarily be thought of as extending the appropriations into the medium term. They do not need to be prepared at the same level as appropriations, particularly in Botswana, where appropriations are at a relatively detailed input line-item level. To do so would make the models far too complex, and difficult to both develop and operate. Similarly, the drivers of cost pressures will represent approximations of what drives spending. Taking the time early on to get the right balance between complexity and workability will make the forward estimates far more robust.

Understand the existing budget

109. The first step is to understand the current spending base and allocating it against four or five key spending units within the Ministry. This requires knowing where and what the ministry is spending its money on, and how that is linked to key program outputs. In some sense this mirrors program budgeting; however most ministries appear to be administratively set up broadly along such lines already. For instance, the Ministry of Education has 5 subsectors: Primary Education, Secondary Education, Tertiary Education, Headquarters, and Teacher Services. This map well into key spending areas, with the only major adjustment required in allocating teachers’ salaries away from the teachers’ services area, and allocating them to the primary and education spending areas.

110. The second step is to identify the current level of service delivery, or the no-policy change baseline. One of the important features of forward estimates is to separate the cost of providing existing levels of service, and the cost of introducing new policies. Thus, for education the current level of service delivery can be defined as the cost of maintaining enrollment rates (89 percent for primary education), no matter what happens to the population of school age children. Thus, if the population of children were to increase, no policy change would see the number of students enrolled increase, but the enrollment rate is to remain fixed. This is in contrast to an increase in student numbers due to an increase in the enrolment rate, which would be considered a policy change.

111. The third step is identifying the major spending areas within the spending units. These are often set up on an economic basis, such as salaries and wages, goods, and services, transfers, etc. However, at this point, it can be useful to break them up into key elements, so in the case of education, goods and services will be separated into textbooks, food, and utilities. Usually there will be an “other” category. As a general rule, this should make up no more than 10 percent of total spending within the unit. Each of these spending areas within the spending unit will be modeled separately.

112. The fourth step is to identify any one-off expenditure that need to be taken out to adjust the base. For instance in education, the National Teachers Day celebrations may not be expected to occur next year, so the amount allocated for it should be excluded from the expenditure base. Similarly, there may be some one-offs that need to be accounted for specific years over the medium term. Classic examples include elections and census collections.

Understand and apply the medium-term cost drivers

113. The fifth step is to identify the price parameters that will affect spending. A number of factors can affect the price of providing services. These factors, such as wages, inflation, and specific input factor prices (such as fuel and utilities) need to be identified and analyzed for their relationship with expenditure growth, as changes will not always be one for one. Thus, for education, the price parameters would be for the following:

  • salaries and wages, the annual incremental increase plus any negotiated wage increase;

  • textbooks, which are largely imported from South Africa, South African inflation multiplied by the BWP/SAR exchange rate;

  • food, the food component of the CPI; and

  • utilities, the utilities component of the CPI.

114. These parameters need to be linked to the relevant macroeconomic parameter forecasts. Often the price parameters will simply be linked to the overall inflation forecast. However, in some cases, there will be difference between the forecast of overall inflation and specific price parameters, such as fuel, which will be driven by world oil prices, and will have little relation to overall inflation. Again, the degree of detail here represents a trade-off between complexity and tractability. In general, the rule in setting price parameters should be to keep it as simple as possible, and only use unique parameters if large differences with overall inflation occur.

115. The sixth step is to identify changes in the volume parameters that drive the cost of providing the services. A number of factors will affect the amount of services being provided over the medium term. These factors can be due to: (i) demographic change (such as the population of school age people in the example above); (ii) past policy changes, where previously increased capital expenditures require a higher level of maintenance once those projects come on line; and (iii) macroeconomic factors, such as a higher unemployment rate that requires a higher level of unemployment benefits. Like the price parameters, these should be defined and linked to macroeconomic and demographic forecasts and projections that can be prepared by the macro-fiscal unit/Bank of Botswana or Central Statistics Office in the early part of the forecasting round, and communicated by the MFDP in the budget call circular.

116. The range of macroeconomic and fiscal forecasts provided in the budget call circular will need to be increased in order to provide the required parameter inputs. As parameters are identified, this will need to be communicated to the units who prepare the macroeconomic and fiscal forecasts, to check that they can be identified, and reach agreement on what basis the forecasts will be made on. For instance, the CPI forecasts may only be prepared for the overall CPI. However, the forward estimates may require parameters for fuel price CPI, so the macro-forecasters may need to develop a methodology for forecasting the lower level CPI series.

117. The seventh step is to apply the overall price and volume parameters to the base expenditure to yield an expenditure forecast for the spending area. This is done by multiplying the adjusted base (i.e., accounting for one-off expenditures) by both the overall price and volume parameters. Note that there will be some interaction in the process, as price variations are applied to volume variations.14

Aggregate for the ministry and summarize

118. The final step is to aggregate across spending units and spending areas to yield a forward estimate for the ministry. The expenditure forecasts for the individual spending areas within a spending unit should be summed, and then the spending areas summed in order to give the overall forward estimates for the Ministry. The same is true for the price and volume parameter variations, which can be used to provide a full reconciliation of the changes in the forward estimates (Figure 9).

119. A “dummy” forward estimates model, set up for secondary education for sector of the Ministry of Education, and the overall MFDP has been prepared for the authorities. These are based on the 2013-14 budget allocations, and include assumptions and stylized parameters to demonstrate how the forward estimates are prepared. The two ministries should become familiar on how these models operate, and agree on the methodology for preparing future forward estimates.

120. Recommendations

  • Begin the process of developing forward estimates three years ahead for each Ministry using the approach and model provided above.

  • Involve both the MFDP and line ministry budget and planning units in formulating the forward estimates models for each ministry.

  • Responsibility for producing the forward estimates should lie with each ministry, with agreement from the MFDP on the methodology, parameters, and one-offs before they are finalized.

B. Methodology for Preparing Medium-Term Costing for New Policy Proposals

121. The preparation of medium-term expenditure costings for all measures and policy proposals is a key element of MTEF process. They require analysis of the longer-term impacts of policies, and their use acts as a discipline for decision makers from agreeing to policies without acknowledging costs that may grow significantly over time, leading to unsustainable fiscal policies. They also demonstrate the impact of policy changes against the forward estimates, in order to meet the fiscal rules. Costing should be prepared for all new policy proposals, changes to existing policies, and alterations to eligibility criteria or assistance rates of existing transfer or social security programs.

122. The policy costing should provide financial costings of new policies over the three-year period as well as a range of other information. They should identify different components of expenditure (i.e., wages and salaries and capital expenditure) as well as revenue impacts (i.e., due to tax expenditures or increased own revenues). If the cost of the policy is likely to be significantly different in the years beyond the medium-term, for instance if the measure is not likely to commence within the medium-term period, there should also be a statement about the financial impact of the measure over the longer term. Appendix III provides an illustration on the determination of multi-year expenditures for a new policy or project.

123. The preparation of costings should be undertaken on a similar basis as the forward estimates outlined above. This serves to ensure consistency between costings, but also aids the inclusion of new measures into the forward estimates for the next forecasting round. In most cases, the methodology should be based on a basic price multiplied by quantity approach, with an allowance built in for both factors to vary over time according to well identified parameters. In cases where fixed cash costings are provided, it will still be necessary to ensure that the amount provided is sufficient for the designated purpose.

124. The forward estimates by program and medium-term costings should be formalized within the budget process through the use of budget submissions. Ministerial budget submissions should provide the forward estimates and any new initiatives to fill (or reduce) the remaining fiscal space (or gap). Submissions should distinguish between new initiatives within the ceilings, which should be accompanied by offsetting savings initiatives, and new initiatives that bid to use the unallocated planning reserve part of the budget margin. All new initiatives should be fully costed, with agreement of the MFDP.

125. Recommendations

  • The MFDP should develop guidelines for construction of medium-term ministerial allocations.

C. Setting Ministerial Expenditure Ceilings

126. Preparing the initial set of ministerial expenditure ceilings will involve both a top-down and bottom-up process. The following steps should be undertaken.

  • The first step is the consolidation of the forward estimates for each ministry into a credible expenditure baseline over the three-year period. Following the process outlined in Section IV, the MFDP—working closely with line ministries—should ensure that forward estimates for each ministry: (i) are consistently aligned with NDP priorities; (ii) reflect realistic costs of on-going policies and services being delivered; and (iii) enhance efficiency and effectiveness in use of resources.

  • The second step is to subtract the estimated non-ministry expenditure commitments from the total aggregate expenditure ceilings for each of the three years to give the amount available for ministry indicative expenditure ceilings.

  • The third step involves the construction of the expenditure ceilings by ministry. This requires a comparison of the consolidated sum of ministry forward estimates (from step 1) with the aggregate expenditure ceilings (estimate) for each of the three years of the medium-term (from step 2) and undertake appropriate measures as indicated the scenarios below:

    • Scenario 1: If the aggregate expenditure ceiling exceeds the sum of the ministry FEs for each year, the MFDP will allocate funds equal to the FEs to each ministry for each MTEF year, and reserve the remaining funds.

    • Scenario 2: If the aggregate expenditure ceiling is less than the sum of the ministry FE’s, the amount will be pro-rated between the ministries in relative shares based on their previous budget allocations.15

127. Finally, the MFDP will review the ceilings and make any necessary policy adjustments to the ministry budgets, based on the policy costing process in Section 3.C. Where the aggregate expenditure ceiling exceeds the sum of the ministry FEs, MFDP may give consideration to policy priorities for new spending where the particular initiatives have been strongly supported by Government. MFDP will be restricted in this by the amount of any unallocated funds within the aggregate estimate and/or the ability to identify offsetting savings elsewhere in the budget.

Rolling-over the forward estimates from one budget year to the next

128. The recommended approach for the MTEF is for that ministry budget ceilings be fixed for the budget year and indicative the first and second out-years. The indicative ceiling for the out-years will be revised in the next budget. While this approach provides considerable flexibility, it is important that the indicative ceilings are held to closely, with adjustments only where absolutely necessary to reflect significant changes in the fiscal forecasts, price and volume variations, and/or significant policy changes. This is necessary to ensure that the indicative ceilings remain a reliable guide for multi-year planning by ministries and government.

129. Preparing the forward estimates for the next year’s budget formulation involves a number of steps: rolling the estimates forward one year, revising the estimate for the budget and first out-year, and adding a new estimate for the second out-year. The following sequence of events/decisions should be used for each ministry:

  • First, transfer the existing estimate for the first and second out-years to be the new budget year and first out-year estimate. These become the starting point for determining the ministry ceilings for the new medium term estimates.

  • Next, update the estimates for budget and first out-year by applying any parameter changes to the forward estimate calculations. This will be calculated using the methodology for forward estimates described in Section IV.

  • Then, calculate a new estimate for the second out-year. This will be calculated using the methodology for forward estimates described in Section IV of this report, and will be agreed between the MFDP and each ministry. The MFDP will compare the sum of ministry FEs against the aggregate expenditure ceiling to determine if the full FE for each ministry can be accommodated.

130. The process for this comparison will be:

  • Take the fiscal forecast of aggregate expenditure. From this amount should be subtracted the estimated non-ministry expenditure commitments to give the amount available for ministry indicative Budget Ceilings.

  • If the aggregate expenditure ceiling exceeds the sum of the ministry FEs for the year, the MFDP will allocate funds equal to the FEs to each ministry’s second out-year ceiling, and reserve the remaining funds.

  • If the aggregate expenditure ceiling is less than the sum of the ministries FE’s, the amount will be pro-rated between the ministries in relative shares based on their previous budget allocations;16

  • Finally, the MFDP will review the new ceiling for the second out-year and make any necessary policy adjustments to the ministry budgets. Where the aggregate expenditure ceiling exceeds the sum of the ministry FEs, the MFDP may give consideration to policy priorities for new spending (raised by ministry in their Annual Budget Plan), where the particular initiatives have been strongly supported by Government and increase the ministry ceiling. The MFDP will be restricted in this by the amount of any unallocated funds within the aggregate estimate and/or the ability to identify offsetting savings elsewhere in the budget.

131. When presented in the budget, explanation/reconciliation between the old and forward estimates should be prepared at both the ministry and aggregate levels. The reconciliation is important to maintain budget discipline by ensuring that all changes to FEs are fully justified and agreed between the ministry and the MFDP. They should be separated between price and volume variations, as detailed above, and the impact of new policy decisions (Figure 9).

D. Development Projects and Budgeting

132. The Government is cognizant of the consistent record of under-spending of the Development Fund and has demonstrated interest to address the underlying causes. During the recent 2012 FAD TA mission on MTEF, the authorities agreed to the following areas of improvement:

  • projects need to be rigorously prepared and appraised, planned, and costed before they are considered for inclusion in the annual budget;

  • a more systematic mechanism that would allow effective management of the development budget cycle is needed: from project conception through to execution, and in particular promote high quality appraisal of projects to guide realistic costing and budgeting, and ;

  • an appropriate administrative actions need to be undertaken to support the unification of the recurrent and development budgets in the short-run and pursue legislative reforms in the future.

Improved planning and costing

133. The current practice encourages undertaking the detailed design, specification and costing of the project after inclusion in the budget. This has undermined the prioritization and realistic budgeting for the projects resulting in both cost over-runs and unmanageable delays in the execution of the Development Fund.

134. Re-sequencing the process of formulating the development budget has to be undertaken as a matter of urgent priority to establish an appropriate ordering of the development project cycle allowing only fully developed and costed project proposals to be submitted for consideration and inclusion on the FEs.

Management of the development projects “pipeline”

135. The MFDP should strengthen the Development Budget Division to provide appropriate capacity to oversee the process of identification, appraisal, costing, and financing of the development projects within the Development Fund. The major tasks of the unit are provided in the 2012 FAD TA mission report and summarized below:

  • Maintaining a data-base of: (i) all existing projects to ensure that on-going financial/contractual commitments are included in the MTEF budget; and (ii) pipeline projects to ensure they are subject to rigorous appraisal.

  • Provide policy advice on cross-government prioritization of projects, affordability, and medium-term financing of the development projects that should feed into budget strategy through the Budget Options Paper in the strategic phase of the MTEF.

  • Provide quality assurance to all processes in the development budget cycle to ensure a cross-government prioritization of projects, as well as proper design and costing of each individual projects.

136. The role could be managed by staff from the existing Development Program Division and may need to engage additional expertise to ensure it is adequately capacitated to fulfill its role. The expertise required for this role will include economics/finance, engineering, quantity surveying, architectural, and related disciplines. The unit should be structured to include staff from these disciplines or budget to allow access to contractors with these skills at key points in the process.

Towards unifying the recurrent and development budgets

137. The authorities have undertaken administrative actions to enhance coordination between recurrent and development budgets. The move is in recognition of the legal constraint enshrined in the PFM Act 2011 that requires the establishment of a Consolidated Fund (CF) to fund the ordinary operations of government and a Development Fund to finance projects for economic and social development. The intention of government is to reduce the dual budgeting process and the steps include:

  • New processes introduced in the annual budget calendar that allow joined-up decision making for both recurrent and development expenditure priorities.

  • The Thematic Working Groups, which have so far focused almost exclusively on the development budget, are now tasked with considering both recurrent and development budgets in order to ensure cross-ministry and cross-fund coordination and prioritization.

138. The mission encourages the authorities to address further areas of improvement to lay the foundation for effective integration of both budgets and enforcement of the desired legislatives reforms in the future. These include:

  • further refinement to the annual budget process and calendar to reinforce a unified budgeting process.

  • insertion of summary of both recurrent and development budget estimates of each ministry in the current Estimates Book..

  • review and update the GABS system to reflect process and reporting changes.

  • harmonization of the budgeting classification and Chart of Accounts for both Funds and ensure fully alignment with GFSM 2001.


  • The MFDP should re-sequence the processes for the development project cycle so that development projects are designed, rigorously appraised, and better costed before being included in the Development Fund.

  • The MFDP should strengthen its capacity to support and oversee effective coordination of the development project cycle including provision of quality assurance of the key processes and policy guidance on the medium-term prioritization and funding.

  • The MFDP should provide detailed instructions that would support implementation of the administrative reforms towards unifying the recurrent and development budgets.

V. Strengthening the Annual Budget Process

139. The Government of Botswana recognizes that a successful introduction of a MTEF is underpinned by major budgeting reforms. In pursuit of better budget preparation, the authorities, with support of several development partners, have embarked on several reforms, namely: (i) restructuring of the MFDP to allow better execution of its mandate; (ii) recent enactment of the Public Financial Management Act; and (iii) migration to 2001 Government Financial Statistics (GFSM2001). Notwithstanding, the authorities underscore the need to strengthen the annual budgeting process to provide multi-year planning and prioritization. This chapter reviews the existing budget preparation practices and suggests areas of improvements to foster policy-driven budgeting constrained within a medium-term fiscal framework.

A. Constraints to a Predicable Annual Budget

140. The authorities underscore the need to move towards a predictable budget for a successful introduction of medium-term perspective in budgeting. As noted in Section II, the recent performance of the annual budget has been mixed. The authorities have cited the major issues that constrain delivery of the annual budget to include frequent reallocation (virements) and to some extent, supplementary expenditure in an attempt to fund the persistently underfunded expenditure areas or priority pressures during budget execution. The mission discussed the issue with the Budget Reform Task Force and noted the underlying causes to include:

  • Inadequacies in planning and prioritization: Budgeting is still an incremental process with limited focus on the need to prioritize and rationalize allocation of ceilings in line with the NDP priorities. The practice does not allow early costing of policies and programs denying informed decisions on expenditure prioritization, affordability, and efficiency in allocation of resources.

  • Inadequate costing for expenditure proposals producing appropriations that do not represent an optimum allocation of resources. Examples of underfunded on-going policies include infrastructure maintenance and enhancing the relevance and quality of the education system, with insufficient funds to meet recurrent costs of improved implications of better academic facilities and recruitment of teaching staff. The officials consulted in the Ministry of Education and Skills Development explained that the poor costing is attributed to two issues: (i) lack of comprehensive information on budget parameters/drivers; and (ii) inadequate capacity to undertake rigorous costing of on-going and new policies.

141. To address the above issues and ensure better budgeting within a medium-term framework, the mission underscored the need to strengthen the existing planning and budgeting function in both the MFDP and line ministries. Special attention should be paid to key functions, including the preparation of no-policy change forward estimates and medium-term costing of new policies which have been articulated in previous sections. The processes should be integrated into the annual budget process and the requisite reforms are discussed in the subsequent subsections.

B. Enhance Annual Planning as an Instrument for Budgeting

142. The NDP and MTEF are two different but mutually reinforcing concepts. The NDP takes a longer view (five or ten years) and provides a broad and strategic picture of the national priorities which should inform the allocation of resources so that development challenges can be progressively addressed, while the MTEF guides the implementation of the NDP within a credible macro-fiscal policy framework. The MTEF focuses on the short and medium term perspective (the next financial year and the forward estimates), and ensures that annual strategic plans and budget submissions are aligned with the NDP but developed within the context of reliable macro-economic and fiscal framework. Figure 10 demonstrates the MTEF as a framework that reinforces the implementation of the NDP.

Figure 10.
Figure 10.

Relationship between NDP and MTEF

Citation: IMF Staff Country Reports 2014, 162; 10.5089/9781498302166.002.A001

143. In that context, the MFDP should institutionalize an annual budget planning process to provide a direct link between national planning and budgeting. Each ministry would undertake the process to produce and submit an Annual Budget Plan to the MFDP. The mission recommends that Annual Budget Planning be established as part of the strategic phase of MTEF as articulated in the 2012 FAD TA Mission. Annual Budget Planning should be undertaken in the May/June period and should involve the following tasks:

  • a review of the NDP, sector strategy, and ministry strategic plan, to distil the key priorities, both recurrent and development, for the Ministry in the budget year and out-years;

  • costing any new policy proposal required to address key priorities that are not addressed by existing services;

  • seeking to identify savings/re-allocations from within the existing budget to meet any additional costs identified above;

  • presentation of existing forward estimates for the budget and out-years with a reconciliation back to the FE’s last prepared; and

  • documenting the outcome of the planning exercise—this document should be made available to NSO (to review for consistency with the NDP) and MFDP (to review for accuracy of costing and to inform its preparation of budget planning ceilings for both recurrent and development budgets), and should provide an input to the TWG process.

144. In order to introduce the annual budget planning process for fiscal year 2015/16, MFDP will have to prepare and circulate a detailed schedule of guidelines/instructions spelling out the objective of the process, a step-by step approach including tools on how the process would be coordinated within the ministry, and an outline of the Annual Budget Plan. In addition, the MFDP will have to prepare training materials to train the staff in DBD who would have to provide technical support to line ministries. Technical assistance would be required to produce the guidelines, tools, and training manuals and conduct training. It is envisaged that the process could be initiated early next year for a few ministries to facilitate a policy and expenditure prioritization providing a base for formulation of FEs. Sustained technical assistance would be required to support the policy driven multi-year planning and budgeting within the budgetary central government agencies.

145. Recommendations

  • The MFDP to prepare guidelines, including tools to support the annual budget planning process, by end of this calendar year.

  • The MFDP prepare training materials and conduct training of staff of DBD and line ministries to support the preparation of multi-year expenditure projections by line ministries early next year.

C. Annual Budget Calendar

146. The existing budget calendar would need to be revised to include formulation of medium-term aggregate expenditure ceilings and updating of forward estimates, as well as construction of ministerial ceilings. In addition, the authorities would have to undertake key decisions on macro-fiscal policy objectives, including aggregate expenditure ceilings at the outset of the budget process. The top-down budget process, anchored in a medium-term framework, requires access to information that is currently not readily available. This underscores the importance of the strategic phase of the annual budget process.17

147. The strategic phase of the annual budget process includes a number of processes and outcomes which feed into the preparation of a Budget Outlook Paper (BOP). The BOP is submitted to Cabinet for consideration and serves to obtain political direction to the formulation of the detailed budget. The main elements of the phase are articulated in Figure 11 showing the proposed Budget Calendar, and include the following:

  • Activity 1(a): An early update on the medium-term macroeconomic projections plus fiscal policy objectives and targets undertaken by the MWG in June/July.

  • Activity (1b): An initial update of the policy and expenditure priorities together with medium term budget /expenditure ceilings and spending pressures. MFDP meets with the Department of Public Service Management (DPSM) on manpower planning, Ministry of Transport and Construction on vehicle fleet issues and line ministries and Project Review Committee to review the current and new policies and broad expenditure priorities.

  • Activity (3): Ministry Annual Budget Planning should be institutionalized in July/August to ensure that ministries update forward estimates and identify priorities for new policy proposals.

  • Activity (4): Policy guidance on expenditure prioritization for the medium term would be sought from the TWGs in August, and guided by the ministerial ABP prepared under the leadership of the permanent secretaries.

  • Activity (5): Development of the BOP that captures the medium-term economic and fiscal outlook and medium-term budget strategy including agreed policy and expenditure priorities. The paper benefits from consultations on the ABP and TWGs.

  • Activity (7): Cabinet consideration of the BOP introduced in August/September in which a sub-committee will consider the MTFF plus key budget policy priorities and indicative line ministry expenditure allocations.

Figure 11.
Figure 11.

Proposed Budget Calendar

Citation: IMF Staff Country Reports 2014, 162; 10.5089/9781498302166.002.A001

D. Budget Documentation

148. Today, the budget documentation prepared for the annual budget is limited. The budget documentation submitted to National Assembly consists of the budget strategy paper, budget speech, estimates of expenditure from the consolidated and development funds, and estimates of the consolidated and development funds’ revenues. The line ministries’ submission to the MFDP includes estimates for revenue budget estimates as well as recurrent and development expenditure estimates.

149. The existing documentation does not provide satisfactory information on the strategic orientation of the budget, government policy orientation, and the funding of NDP priorities. The content and analysis in the budget speech, as is expected, is limited to a very general statement of economic, social, and fiscal conditions under which the annual budget has been formulated. The Budget Strategy Paper (BSP) is published between September and November each year and outlines macro-economic developments, economic outlook, and risks, as well as the fiscal strategy and budget priorities for the upcoming budget. However, it lacks substantive analysis of macroeconomic development and future strategy and overall budget strategy and spending policy options. Line ministries’ submissions lack detailed explanation regarding the expenditure prioritization and allocation of resources.

150. Moving to the multi-year perspective in planning and budgeting would require improving the quality, coverage, and strategic direction of the document. The areas of improvement should aim at: (i) enhancing a comprehensive coverage of the fiscal operations; (ii) better analysis of macro-fiscal strategic decisions; and (iii) improving oversight and accountability of fiscal risks. In particular, the MFDP is encouraged to provide guidelines to line ministries to produce ABPs that would explain the strategic orientation of policy and funding proposals contained therein to better inform budget decisions and to facilitate accountability.

151. The mission therefore recommends the following:

  • Introduction of a Budget Options Paper (BOP) to be submitted to Cabinet early in September to foster early engagement of Cabinet in the budget process so as to consider and approve medium-term macro-fiscal and spending policy options. The Cabinet decisions on the BOP would assist in setting the aggregate expenditure priorities and ceilings. The proposed outline of the paper is attached as Appendix V and should later be updated and evolve into a medium-term BSP to be submitted to the Parliament with the Budget.

  • The proposed Annual Budget Plans (updates) should provide an outline of the sector and ministry’s policies/service priorities and alignment with NDP 10 over the medium term, identification of policies/services to be continued and discontinued or scaled down, provide medium-term cost implications of new or policies/programs, and an updated medium-term expenditure estimates (forward estimates) for on-going services. The plan could be updated at the end of the budget preparation process and submitted to the Parliament to provide explanations of budget proposals required to facilitate the budget debate.

E. Institutional and Capacity Arrangements

152. The mission examined the institutional and capacity implications for the introduction of the MTEF and underscored the urgent need to design and implement a capacity building program that would facilitate improvements in the planning, prioritization, and costing of the sector policy priorities and strategies, and strategic direction of the annual budget. Drawing from experiences of countries in the region, the mission advised that such competencies would include: (i) economic and fiscal forecasting skills, including sector policy analysis and costing skills to be able to advice on major spending proposals from ministries; and (ii) reform management skills to facilitate a well prioritized and sequenced implementation of the reform agenda.

153. Capability enhancement for policy analysis and costing of policies at the line ministry and in the MFDP will be a critical part of the reform. The mission stressed the importance of an on the job training approach which would not only allow fast application of skills, but also expedite the development of databases for required information and data; identification of relationships with macro parameters and instituting procedures to ensure quality control and regularly updated forecasts. For instance, focusing skills development at applying FEs tools will intrinsically enhance policy analysis skills and facilitate identification of cost and volume drivers for spending programs. In addition, further formal training would be offered to enhance the conceptual framework for policy analysis including impact analysis for significant expenditures aimed at producing a robust annual and medium-term budget strategy including forward estimates.

154. The Development and Budget Division in the MFDP needs to be strengthened to foster unified budgeting within a medium-term framework. The mission recommends creating a team unit within the Division to spearhead the MTEF reform as part of the annual budgeting process. Substantial and long-term technical assistance would be needed to build capacity to: (i) improve the annual process so as to formulate quality of medium-term expenditure policy and budget strategy; (ii) produce required guidelines and manuals, techniques, templates, and tools; (iii) coordinate the design and implementation of a capacity building program initiative; and (iv) provide the technical support and mentoring to line ministries. The mission noted that the authorities have embarked on recruiting an EU-funded resident advisor.

155. Likewise, the planning and budgeting function in line ministries and other spending units will need to be strengthened to support the development of the MTEF. The responsible staff in both the planning and budgeting units will have to work closely with the technical heads of departments to take greater responsibility for the medium-term expenditure framework. To ensure effective execution of the reform and related new tasks, the mission recommends merging these two units to facilitate a smooth execution of new tasks.

156. The MFDP is encouraged to conduct a capacity building needs assessment for both staff in the MFDP and line ministries. Special attention should be given to needs to spearhead the functions such as: (i) sector and expenditure policy analysis; (ii) costing of policies and programs aimed at producing more realistic and affordable annual budget forward estimates; (iii) process management skills including coordinating the strategic reviews and decision making processes within various department or sectors; (iv) sector-specific policy analysis and formulation, project design, appraisal, and costing of programs; and (v) planning techniques to ensure that the policy and budget decisions are translated into realistic and affordable annual plans that would contribute to realization of NDP objectives.

157. To address the above issues, the mission recommends the following measures:

  • The authorities should strengthen the existing planning and budget manual for both the MFDP and planning/budgeting units in line ministries.

  • The MFDP should strengthen the DBD in the MFDP and merge the planning and budget units in line ministries.

  • The MFDP should undertake a comprehensive capacity building needs assessment to ascertain the generic and specific competencies required to pursue the MTEF.

  • The MFDP should develop and implement an extensive capacity building program targeting the key concerned technical staff and policy makers.

Appendix I. Glossary of Terms

Medium-Term Expenditure Framework (MTEF)—institutional arrangements in the budget process governing the requirement to present certain medium-term financial information at specific times, procedures for making multiyear forecasts and plans for revenue and expenditure, and obligations to set numerical expenditure limits beyond the annual budget horizon.

Medium-Term Fiscal Framework (MTFF)—standing requirements to commit to, report against, and be held accountable for medium-term aggregate fiscal objectives, such as debt limits, surplus targets or deficit ceilings, or broad expenditure limits. For the purpose of this chapter, expenditure ceilings are considered part of the MTEF if they cover all or a subset of central government.

Binding framework—a framework that holds government accountable for the multiyear expenditure parameters (estimates or ceilings) set in year t - 1 or earlier, when, in year t, the budget for t + 1 and new medium-term estimates for t + 2 and t + 3 are set. Accountability means that some active measure or action is required if there is evidence that the previously set expenditure parameter is going to be exceeded. A fixed framework is a subset of binding frameworks in which medium-term expenditure limits are set, but are not subsequently revised.

Indicative framework—a framework in which updates of medium-term estimates can be made without reference to the same estimate set in the previous year, and in which the appropriations in government’s annual budget proposal for t + 1 are not reconciled against the medium-term estimates for t + 2 made in the previous year (or previous budget update, e.g., pre-budget report).

Appropriation—maximum limits for individual expenditure items as defined in the budget. In this chapter, appropriations refer both to the expenditure limits set in the budget as adopted by parliament and to any sub-limits imposed by government on ministries and agencies.

Forward Estimate—an assessment of the expected outturn of a revenue or expenditure item. In the chapter, “estimate” is used as a collective term for forecasts and no-policy-change assessments, for which the distinction between these concepts is unimportant. (In some countries, the term “estimate” is used to refer to the legislated appropriations in the budget.)

No-policy-change assessment—extrapolation into the future of a revenue or expenditure item under the assumption that today’s policies are kept unchanged. The definition of current policy differs between countries, with some countries emphasizing an extrapolation of existing legislation whereas others also incorporate policies that have been proposed to parliament but that have not yet been formally adopted.

Expenditure ceiling—a maximum limit on an aggregate of expenditure that is broader than an individual appropriation. A fixed ceiling refers to a limit that is not revised upward once it has been set, and applies to the ex post expenditure outturn. A flexible ceiling is a limit that can be revised upward after it has been set.

Thematic Working Groups—The composition and functions were revised by a recent Cabinet directive 33/2011. The functions include: determination of short, medium, and long term national priorities for Cabinet consideration and approval; alignment of sector development strategies, policies, and programs with national priorities as will be determined by government from time to time; monitoring and evaluation of achievement of high level national results; determine proposals for more effective use of financial and human resources; and ensuring reduced transaction costs for both Government and counterparts in thematic areas and in working together to achieve the goals set in the national development plans and BEAC Action Items.

Appendix II. Follow-up Implementation of Previous FAD TA Missions in 2010 and 2011

article image

Appendix III. Considerations of a Medium-Term Costing Process: Airport Security19

An illustrative example of a costing is to consider the cost of introducing a new process for passenger and baggage screening at Botswana’s airports.

The first step is to determine the number of airports affected that do not have the requisite screening process in place. Then, the price and installation cost of each piece of screening equipment required for each step of the process (such as explosive detection system capable X-ray machines for screened baggage, explosive trace detectors and walkthrough metal detectors for passengers, X-ray machines for passenger carry-on baggage, and other items, hand wands, divest tables, and barriers).

The second step is assessing the cost of operating the equipment. This includes the cost of employees with the technical expertise to operate and implement the screening process (based on the number and type of employees, their salary rates, and the length of time required), consultative forums between the administering agency and the airports (including travel costs, meeting expenses, and venue hire), as well as the cost of ongoing audit and compliance activities (the number of officers, salary rates, the number of visits, and the length of time required to undertake these activities).

Finally, the time profile of the expenses would need to be defined. Such a program would have a well defined upfront capital expenditure based on the first step, as well as depreciation/replacement/main-tenance costs over the medium term. The ongoing operational costs which will change over time, due to parameter variations in input costs (salary rates and utility expenses will vary over time). The volume of screening activities can be expected to increase over time as passenger numbers increase, requiring increased numbers of operational staff.

The total cost profile over the medium term should take account of all of these costs, with each major cost center being separately identified, understood, and verified by MFDP officials. The key parameters for each element of the costing should be identified and incorporated in the costing, so that changes in parameters can flow through to the policy costing once it is incorporated into the forward estimates in the future.

Appendix IV. MTEF Cycle: Strategic and Budget Formulation Phases

Strategic Phase of MTEF

The Strategic Phase of the MTEF budget process provides institutional and procedural arrangements that strengthen fiscal discipline and improves policy and expenditure prioritizations aligned with agreed macro-economic and fiscal policy objectives. In essence, this means that budget decisions are taken in a cascading manner, a total expenditure level is determined before the allocation between main policies or sectors is made, and sectoral or line ministry ceilings are set before the detailed division of expenditure within each sector is discussed and decided. In each step of the budget process, the allocation of expenditure is subject to the constraints that have been set at the previous stage.

Key elements to this phase include: macro-fiscal analysis and forecasting; policy and expenditure prioritization at ministry and sector level; public investment planning and prioritization; and external assistance planning. The outcome of these linked processes is a Budget Options Paper.

Strategic Stage of Budget Development

Citation: IMF Staff Country Reports 2014, 162; 10.5089/9781498302166.002.A001

Budget Formulation Phase of the MTEF

The phase is focused on developing the detailed budgets for each ministry within the context established through the earlier Strategy Phase. The elements that comprise this phase include: the preparation of the Budget Call Circular; preparation of the detailed estimates by the Ministries; analysis of these submissions by the MFDP and preparation of briefing notes for Budget discussions; and the consolidation of the Budget for the Minister to take to the Parliament. Much of the information required to complete these processes will have been produced during the Strategy Phase. The outcome of these linked processes is the documented budget for the next year and estimates for each of the outer years.

Resource Allocation Stage of Budget Development (October–January)

Citation: IMF Staff Country Reports 2014, 162; 10.5089/9781498302166.002.A001

Source: Concept Paper: Budgeting within a Medium-Term Expenditure Framework for the Government of Botswana, November 2012 FAD Mission Report.

Appendix V. Outline Structure for a Budget Policy Paper

article image
Source: Fiscal Affairs Department in the International Monetary Fund.

In the 2013 Budget Speech, the Minister reaffirmed government commitment to deliver a medium-term fiscal consolidation, rebuild fiscal buffers, and introduce the MTEF by 2016.


The IRBM was initiated in 2008 to foster improvements in strategic planning, coordination, and implementation of the National Development Plan so as to enhance public service delivery and achieve desired goals.


Thematic Working Groups constituted at both political and technical level to review and advise on short, medium, and long term national priorities for Cabinet consideration and approval.


Khan, A et al. “Botswana Public Financial Management Reform: Selected Issues and Action Plan” IMF, 2010.


The NDPs do present 6 year forecasts of GDP. However, these forecasts are soon out-of-date and are not used in the annual budget preparation process.


The Financial Programming and Policy Group, comprising members from the MFDP, Bank of Botswana, and Statistics Botswana was established in 2012 and is making good progress in developing a four sector financial programming framework that incorporates both the GDP and MTFF forecasts.


“Medium-Term Fiscal Framework Projections” drafted by the Macro-Fiscal Working Group.


The NDP8 planned to run a sizeable surplus, while the NDP9 and NDP10 stated an ambition to balance the budget over the life of the Plans.


Interest expenditures make up a relatively small (0.5 - 1.5 percent of total expenditure over the past 4 years) and highly predictable part of overall expenditure. Other items are difficult to define within Botswana’s current budgeting system.


PEFA 2013 reveals a significantly sized variance in the composition of expenditure that may indicate that: (i) the approved budget did not represent an optimum allocation of resources in the first place; and/or (ii) priorities changed during the year, the result being that those MDAs for which priority increased during the year were allocated a greater share of the available resources. The score for PI-2 is C.


In the United Kingdom, these controls take the form of a nominal delegated limit for each ministry, above which the ministry must seek treasury approval before entering into a multiyear commitment. Finland and Sweden have even more restrictive regimes that require all multiyear expenditure commitments to be approved by parliament as part of the budget.


The Technical Notes and Manual on Carry Over by Budget Authority recommends setting up general restrictions for granting end-year flexibility to broad categories of expenditure. It also encourages a case-by-case evaluation which could be appropriate for a limited number of large carry-overs but not cost effective for a large number of small carry-overs, such as for smaller operational expenditures. The TNM is an IMF publication prepared by Ian Lienert and Gösta Ljungman in 2009.


For instance, for a spending base of 100, a 10 percent increase in both price and volume will lead to an expenditure forecast of 121, greater than if price and volumes were applied individually, which would lead to a forecast of 120. This difference is described as the interaction term, and needs to be accounted for when reconciling the change in expenditure between the base and BY+1.


Dividing each MDAs budget by the sum of ministry budgets for the previous year and then multiplying this by the total amount of the aggregate expenditure ceiling available to ministries would calculate this: (ministry budget / total ministry budgets * aggregate expenditure ceiling available for all ministries).


Dividing each MDAs budget by the sum of MDA budgets for the previous year, and then multiplying this by the total amount of the aggregate expenditure ceiling available to MDAs would calculate this: (MDA budget / total MDA budgets aggregate expenditure ceiling available for MDAs).


The two phases of the MTEF—(i) strategic phase; and (ii) budget formulation phase—have been articulated in the November 2012 Mission Report and an illustration is attached as Appendix IV.


Fichani, K & Freeman, P. “Minerals and Energy Exports and Revenue Projections” Botswana Institute for Development Policy Analysis, June 2012.


Adapted from an example in Charter of Budget Honesty: Policy Costing Guidelines, issued by the Secretaries to the Treasury and Department of Finance and Regulation, Canberra, Australia 2012. (

Botswana: Technical Assistance Report-Introducing a Medium-Term Expenditure Framework
Author: International Monetary Fund. Fiscal Affairs Dept.