Many emerging economies are trying to improve their budget processes and move to performance-based budgeting modeled on OECD-type reforms. This section first presents a brief historical perspective of these reforms by reviewing the evolution of the “new” performance budgeting model being applied in industrial countries. It identifies the main components of this budget management model as a means to determine the challenges facing emerging economies when converting their existing budget systems to this model. This review highlights the critical importance of a good program basis for budgeting, which is not easy and typically requires four important reform elements to be in place. First, any existing program structure must be set in the wider context of strategic budget planning. Second, such strategic planning should be anchored on a medium-term budget framework. Third, this usually implies redesigning and refining existing program structures. Fourth, and perhaps most difficult, a new system of accountability and budget incentives needs to be introduced. For emerging economies, these four elements should be viewed as the prerequisites for a successful introduction of performance budgeting.
The Road from Old to New Performance Budgeting
Budget reforms that were introduced in industrial countries following World War II, under the rubric of performance budgeting, have been introduced in many guises and generally have endured in some form in most countries. Unfortunately, owing to these many variants, the term itself has been interpreted differently at different times and in different countries. At the broadest definitional level, the term is associated first with a budget presentation that emphasizes the outputs rather than the inputs associated with government operations16 and, second, with a restructuring of government operations on the basis of programs and activities producing these outputs. As a consequence, the term is often used synonymously with program budgeting.
There is a vast body of literature on program budgeting,17 which unfortunately also tends to be somewhat confusing. There is no agreed form of program budgeting, and the term program budgeting is often used interchangeably with related terms, such as performance budgeting, planning-programming-budgeting systems (PPBSs), and output-based budgeting. Accordingly, to impose some order in this terminological confusion, this section attempts a stylized characterization of each of these budgeting concepts and their specific features. As indicated in Box 8, the progression toward the new performance budgeting follows distinct steps.
Origins of Performance Budgeting
Elements of program budgeting were in evidence in the United States prior to World War II,18 but the term performance budgeting is more clearly associated with the 1950s reforms undertaken in the United States. These focused on developing performance information for budgeting and reorienting the U.S. federal budget process from its focus on inputs to one that also included the outputs derived from using those inputs. The approach really took off when the Hoover Commission on the organization of the executive branch of the government promoted this approach and encouraged its widespread implementation in 1949.19 The Budget and Accounting Procedures Act of 1950 required agency heads, in coordination with the Bureau of the Budget, to support their budget requests with information on performance and program costs by organizational unit. The U.S. budget for the 1950–51 fiscal year was the first to show the effects of this changed perspective. Even from these beginnings, the Hoover Commission used the terms program budgeting and performance budgeting almost synonymously.20
The Steps toward the “New” Performance Budgeting
Step 1. Planning, programming, and budgeting systems (PPBSs)
Program, output, and “new” performance budgeting occur only after the first two elements of PPBS have been achieved.
Elements:
Identify and examine goals and objectives in each major area of government activity.
Analyze the output of a given program in terms of objectives.
Measure total program costs, not just for current year but also for several years ahead.
Formulate multiyear expenditure programs.
Analyze alternatives to find the most efficient and effective means of attaining program objectives.
Establish these procedures as a systematic part of the budget review process.
Step 2. Program budgeting
Program budgeting leaves the PPBS’s higher-level strategic planning functions out of the budget process. It entails an interactive process for refining cost assignments and output definitions.
Elements:
Group organizational units within common functions and subfunctions.
Identify costs of a function and subfunction.
Given these costs, decide what the unit’s output should be.
Step 3. Output budgeting
Elements:
Group together all costs of achieving a particular output, regardless of the number of agencies involved.
Emphasize full costing, including overhead assignment.
Define outputs in terms of measurable indicators and assess the quality of goods and services delivered, analogous to what is done in the private sector.
Compare with actual output to gauge efficiency.
Step 4. New performance budgeting
New performance budgeting contains all elements of output budgeting. It also includes higher-level accountability with associated rewards and sanctions, i.e., moves to a performance budget management system.
Elements:
Define outcomes/impacts in terms of measurable indicators, and compare these with actual outcomes to gauge effectiveness.
Incorporate explicit performance measures and systems of performance assessment.
In these early days, performance budgeting was not viewed as a fundamental change in budget decision making, but rather as a useful adjunct that was perhaps relevant for only some service-based operations. These reforms were seen as being severely handicapped by inadequate accounting systems that could not identify the full costs of government operations, the equally constraining problem of developing measures of performance, and, most particularly, the lack of interest of decision makers in using performance information (Burkhead, 1956). In practice, program and performance budgeting proved time-consuming and required much administrative support, namely, improved accounting, costing, activity, and performance-measurement systems.21 Nor was it very popular with appropriation committees, who were more comfortable with detailed line-item controls than with providing lump-sum appropriations in return for specified levels of performance. Budgeters also often preferred that information on operating targets, workload, and performance standards not be explicit but remain largely hidden in detailed objects of expense.
The Planning-Programming-Budgeting System Period
Nevertheless, the application of program and performance budgeting persisted in the United States at the federal government level. The PPBS, as a complete system of budget making, was first adopted by the Department of Defense in 1961, and later applied to all agencies. It subsequently spread to states and local governments and also to some other countries.
The PPBS process—as its name implies—had three basic phases which linked planning with budgeting through programs. The planning phase sought to identify present and future objectives and to evaluate different possible ways of achieving these objectives. The programming phase then took the proposals of the planning phase and integrated them into programs arranged by a hierarchy of priorities that would be subject to decision making at different levels of the political hierarchy. For example, the setting of broad priorities (program categories) would be a government or cabinet-level responsibility, while the optimization within individual programs would be the task of sector agencies. The third phase of budgeting was the translation of each multiyear program into a set of specific annual actions, by determining who does what and then assigning required resources. This was a difficult stage of the process. Since the administrative structure of the budget could be different from that of the program, it was difficult to apportion resources among programs. Moreover, the U.S. federal budget process was strictly regulated by a complex set of laws that had evolved over time and which governed not only the budget structure but its presentation, approval, and execution.
By 1971, the system was largely abandoned. In part, the system was a victim of its own ambition. As Axelrod (1995) argues, “many of the problems of the PPBS in the federal government stemmed from its simultaneous introduction across the board in agencies that were ill-prepared to implement it” (p. 288). The diagnosis of what went wrong holds lessons for the future adoption of a full-scale budget programming approach.
Despite its name, the PPBS was not a single system; each agency had to develop its own version and agency directors often did not appreciate its value and were reluctant to undertake such work. Central leadership of the reform process was lacking. The reluctance to fully convert to the new system was fueled by the fact that the Bureau of the Budget continued to review cost and performance on the traditional basis in making its recommendations. Its adoption was further undermined by a lack of commitment by legislators: Congress insisted on the traditional budgetary presentation and on preserving the existing appropriation structure. Additionally, there were important practical problems that were never fully resolved related to program definition, how to develop a program or subprogram around each objective, and how to allocate costs. In many ways, as will be evident from subsequent sections in this study, many of these questions still beg answers. Further, there was a question of coverage. The system was probably not useful for dealing with large segments of government expenditures such as entitlements, transfer payments to state and local governments, debt servicing, tax expenditures, and off-budget expenditures.
In the end, the critics were damning. Wildavsky (1969) characterized the system as “a vast amount of inchoate information characterized by premature quantification of irrelevant items.” The system imposed what was considered an unbearable burden of calculation that choked the budgetary process and impeded budgetary decision making instead of expediting it. Schick (1971) was equally damning, focusing on what he felt was a chasm between planning and budgeting—the ideal of analytical policy analysis, as seen in the PPBS, was stifled by the routines of budgetary process reality.
Consolidation and Refinement of Program Budgeting
Despite the demise of the full-blown PPBS, program budgeting continued to be applied in the United States in a less ambitious form. The passage of the Congressional and Impoundment Act of 1974 advanced multiyear budgeting; budget classifications by mission, function, and program; the use of sophisticated analytical techniques by the Congressional Budget Office (CBO) and the General Accounting Office (GAO);22 the development of performance indicators; and the improvement of accounting and information systems. This approach persists to the present day, resulting in a situation where it is possible for federal budget decision makers to have data on program and performance as well as on objects of expense.23 Moreover, in the meantime, the program approach to budgeting spread to other countries.
With the 1965 publication by the United Nations of a Manual for Program and Performance Budgeting, the use of program and performance budgeting was reinforced as a tool of development planning. In the 1960s, nearly 50 countries introduced variants of program and performance budgeting, although never on the scale of the United States. This then spread to developing countries, so that by the end of the 1960s, nearly all Latin American countries, several Asian countries, and some African countries had introduced versions of program budgeting.24 On the whole, the impact on budget decision making has been disappointing25—perhaps indicating that for many developing countries, its adoption was premature. The UN manual, heavily based on the U.S. experience, did lay down preconditions for successful performance budgeting in developing countries: sound budgetary operating procedures, financial discipline, efficient methods of recording and reporting financial and physical data, and close coordination between the central budget agency and other government agencies. All, or most, were typically absent in developing countries.
Output and the New Performance Budgeting
In the United States, after the PPBS system, there were brief initiatives, such as Management by Objectives (1973) and Zero-Based Budgeting (1977), but these were limited in impact, being pushed by the executive branch with little or no involvement of Congress.26 However, between 1990 and 1996, there was a spate of new laws that emerged from Congress that created a government-wide platform for the new performance-based management. The term new performance budgeting was used by Mikesell (2003) to categorize the following reforms in the United States of the 1990s: the Chief Financial Officers Act of 1990, the Government Performance and Results Act of 1993, the Federal Acquisition Streamlining Act (FASA) of 1994, and the Clinger-Cohen Act of 1996.27 All represented an attempt to convert the federal budget process to a more results-based one with specific references to the use of performance information for budgeting.28 The 2003 Program Assessment Rating Tool (PART) is the most recent initiative (see Box 10).
However, in many ways this new legislation was only the means for the United States to catch up with international developments in PEM reform. Following the early lead of New Zealand, whose public sector reforms culminated in the passing of the Fiscal Responsibility Act in 1994, other countries, such as Australia and the United Kingdom, were quick to follow, and in turn their attempts were mimicked in many northern European countries as well as in the United States. The penetration of the new ideas to lower levels of government has been almost universal in these countries.29 The themes of these new reforms—often termed the New Public Management—were the primary emphasis on outputs or outcomes, performance in budgeting, and particularly the use of performance information in the budget process. This was coupled with other major PEM reforms, including the introduction of greater flexibility for government bodies (agencies), the introduction of contractual arrangements, and the move away from cash accounting to the use of accrual accounting and budgeting along the lines of the private sector.30 The first aspect, the new performance approach, has been the most universally accepted internationally. A 2001 OECD survey found that 70 percent of OECD members included performance information in their budgets, whereas the remainder included such information for a limited set of programs. Moreover, 40 percent of the countries surveyed claimed their performance measures distinguished between outputs and outcomes, with 20 countries using systematic annual reporting on performance against outputs and 15 countries on performance against outcomes targets.31
However, as noted, performance budgeting is more than simply introducing performance information into the budget process. A key feature of the new performance budgeting approach is the recognition that, if performance is to matter, the objectives of the budget management system must be integrated with overall accountability, so that good budgetary performance is rewarded, and poor performance is penalized. Although countries have taken different approaches to this, there are some common elements that can be identified from OECD countries’ experience.
While using a program structure to introduce a performance orientation, it is clear that, to be effective, a program budget format must also be integrated into a wider model of budget management. Experience suggests that the benefits of a budget program format will soon be lost unless departments (and most central agencies) continue the momentum and purpose of the reform by moving on to develop standards of service delivery and to search for ways by which these standards can be continuously improved and services delivered more effectively. That is, government service delivery needs to be redefined in a more results-oriented manner. How to do this? The most effective way, at least from the experience of a few pioneering industrial countries, involves a fundamental revision of accountability relationships within the public sector. The implications for budget management are profound: first, performance needs to be specified and reported in a way that is operational for budget managers; second, government agencies need greater managerial autonomy and freedom from tight input controls, so that they can determine their most efficient delivery of results; and, third, changes are required to the range of incentives and sanctions facing departmental managers.
Accordingly, while a good program structure is the basis for the new performance budgeting approach, experience indicates that it will not be sufficient to ensure the success of budget reform. If the program structure is to yield benefits, it must be linked to other, much wider reforms, which are taken up in subsequent sections of this study.
Issues of Design in a Program Structure
From the previous, necessarily brief, review of the evolution of performance budgeting, some key features emerge. First, an essential element in all systems is the concept of a program and the restructuring of the budget on this basis. At first glance, the principles determining programs are fairly clear: a program can be viewed as any suitable and meaningfully integrated group of activities and projects, under a single manager, which consumes resources to contribute to a specified policy objective. The operational objectives of each program and activity can then be identified. Budgeting and accounting can be carried out on this basis, so that the separate costs and revenues of each program/activity are made clear to decision makers. It is apparent from this description that later stages in the evolution of this program approach maintain the fundamentals but add some elements. Thus, output budgeting involves measuring the outputs and performance of activities so that they can be related to their costs and operational objectives. In turn, the new performance budgeting involves adding the use of the resultant data to establish standards and norms, so that costs and performance can be evaluated and government resources used more efficiently/effectively through an incentive system that enforces accountability. In all approaches, however, the basis of performance budgeting is a meaningful program structure.
The second essential element is that the program approach to budgeting is an enduring one which, despite its critics, is viewed by budget practitioners as having value added. Specifically, program budgeting is seen to counter the alleged deficiencies of traditional line-item budgeting, including its short-term focus, its incremental nature, its emphasis on inputs rather than program outputs, and its excessive detail which obscures alternative means of reaching the same objectives (Babunakis, 1976). At the same time, there is little agreement on the guiding principles of program classification, and as a consequence, the term is used in a variety of ways—in some cases a program corresponds more closely to a broader concept of a function, and in others to a line item of expenditure. Because it is policy oriented, a program structure is inevitably country specific.
The third essential element is that its successful implementation has proved to be universally difficult regardless of the country setting and the form attempted. Some of these problems have already been indicated: difficulty in providing meaningful information, particularly relating costs to performance data; suspicion of information overload; legislators’ reluctance to yield detailed line-item controls associated with the concern that somehow budget compliance will be compromised; and failure to implant this in wider budget management reforms, particularly increased delegation—indeed, perhaps some view it as a way of circumventing centralized controls. Despite the difficulties, however, program budgeting is undoubtedly a central element of OECD budget practice (Box 9).
If one views the progression from old to new performance budgeting as stages in the development of budget management, it is possible to characterize many emerging market economies as having reached the stage of program budgeting, with some already experimenting with the next stage of output budgeting. To successfully move on to the next stages, emerging market economies must introduce a program structure into their budgets, and where this already exists, must fundamentally redesign their program structures to accommodate the new performance budgeting approach. Specifically, it requires reviewing programs to ensure that there is a clear policy statement or list of objectives that adequately defines the purpose of the programs, so that the results expected from the programs can be assessed, measured in some way, and reported. In turn, this allows introduction of the more rigorous accountability mechanisms that characterize the new performance budgeting approach. Following is a further exploration of these considerations, which stresses four main features:
OECD Practices: Use of a Program Structure
| Percentage of OECD Countries | ||
|---|---|---|
| • | Legislature approves budget appropriations: | |
| – Disaggregated appropriations at program level | 25 | |
| – At program level | 17 | |
| – Split between personnel and program spending at program level | 17 | |
| • | Expenditure classification includes program classification, with government policy objectives and individual program budgets | 44 |
| • | Expenditures linked to strategic goals: | |
| – For all goals | 19 | |
| – For some goals | 31 | |
| – For a few goals | 23 |
| Percentage of OECD Countries | ||
|---|---|---|
| • | Legislature approves budget appropriations: | |
| – Disaggregated appropriations at program level | 25 | |
| – At program level | 17 | |
| – Split between personnel and program spending at program level | 17 | |
| • | Expenditure classification includes program classification, with government policy objectives and individual program budgets | 44 |
| • | Expenditures linked to strategic goals: | |
| – For all goals | 19 | |
| – For some goals | 31 | |
| – For a few goals | 23 |
1. The program structure should be set in a wider strategic framework. The program structure is basically a way of describing the expenditure plan of the budget in terms of objectives. To be relevant, the program structure must be anchored in a wider strategic view that allows it to describe the contribution of government operations to achieving nationwide objectives. To use the language of strategic planning, as pioneered in New Zealand, the national objectives or “strategic result areas” are translated into detailed and fully costed “key result areas” which are the responsibility of specific government units. The latter should then formulate the required detailed implementation plans required for preparing the agency’s budget bid. The wider strategic view has typically been addressed in many countries by adopting a medium-term budget framework, as discussed below, but also through a constant review of program design to ensure its alignment with government objectives. This is very evident in the U.S. government’s Program Assessment and Rating Tool (PART) (Box 10).
2. The program structure should be defined in a way to support political decision making and prioritization. Following from the need to relate programs to wider policy and planning priorities, it is necessary to make clear the relationship between the resources used by the program and the proposed outputs and/or policy results (outcomes). This implies that programs are provided with sufficient resources to meet the objectives assigned to them, and in turn, that there be adequate costing of the program’s outputs. The latter has often proved technically difficult.
3. The program structure should ensure accountability. Programs and subprograms should be disaggregated to activities and projects in such a way as to support clear managerial responsibility in attaining the proposed outputs or outcomes. This implies each program has an appropriate size for efficient management and clear managerial responsibility, usually within a single organizational unit. It also implies that programs should be designed and “owned” by the organizational unit and not by the MoF or budget office.
The U.S. Government’s Program Assessment Rating Tool (PART)
In 2002, the Office of Management and Budget (OMB) developed PART, a questionnaire to collect information to develop a consistent program rating. On this basis, the OMB and agencies rated 20 percent of programs for the 2004 budget, added 20 percent more for 2005, and plan to add another 20 percent for 2006.
Objectives of PART
Systematize performance information into the budget preparation process to better align objectives with budget structure.
Make agencies link information on program purpose and design with strategic planning and program management and results.
Work with agencies to improve the operating and financial performance of their programs.
Reach overall assessment of program effectiveness by better outcome measurements and more systematic evaluations.
Features of PART
Comprises 30 questions on program design, strategic planning, program management, and program results.
Represents a clear attempt to link the program’s purpose with its design, reveal the strategic planning behind its implementation, and generate performance-based program management and results information.
Questions differentiate types of program: grants, regulatory-based programs, capital assets, service acquisition, credit, and research and development programs.
Rates and systematically assesses programs as part of formulating the president’s budget.
Measures and diagnoses program performance and indicates areas requiring attention.
Assists agency and budget decision makers on management and regulatory improvements, legislative proposals, and budget requests.
Unanswered questions
How can PART be integrated with the GPRA (see Box 51)? PART is clearly designed to serve the president’s interests, but the GPRA process is designed to be consultative and to seek agreement on goals and measures.
How can support from Congress be sought? Congressional involvement is critical if PART is to be used beyond the executive branch. While some congressional management committees have shown interest, the appropriation committees have few incentives to change the status quo and use the performance information.
4. The program structure should be integrated into a wider performance-enforcing budget management model. Gaining maximum benefit from the program structure in terms of improved performance and allowing accountability to be enforced requires wider budget management reforms. It entails that responsibility for implementing each program/subprogram should be aligned to budget appropriations as much as possible, providing a clear link between budget and policy outputs/outcomes. This implies that programs are directly linked to one budget appropriation manager, as described further below.
The Wider Strategic Framework for the Program Structure
The core building block of performance budgeting is a meaningful program structure for government operations. However, underlying this reorientation are two wider objectives: to improve the quality of budgeting, by providing a link with policy, and to help ensure the relevance to wider sectoral policies of government operations supported by budget appropriations. However, designing an operational program structure for the budget should not be regarded simply a way of reclassifying the government’s expenditure plan. Rather it should involve some more fundamental rethinking of government activities and the way they are managed. That is, budget managers should be given the opportunity to—
reconsider their respective roles and missions.
review the way to best organize themselves to carry out their missions.
redesign cost allocation and financial reporting systems to improve the efficiency of management.
review the efficiency and effectiveness of operations and delivery of goods and services.
The starting point in such a review of government activities is to connect them to underlying government objectives. This can be achieved by linking the operations of each spending agency with the implementation of its sector’s policies. Broad sector functions can be divided into subfunctions and further into programs that comprise a number of activities, of both a recurrent and capital (project) nature. These activities and projects are then considered as cost centers for which an economic, or object of spending, classification can be prepared. In this way, the program structure is dependent on sectoral policies, and these in turn are dependent on the system of prioritization among sectors. The set of programs will then reflect the major priorities of the government. The program structure is thus the link between the budget and overall strategic planning, which reflects the policy framework for government operations. Only in this way is the program structure a tool for policy analysis and not simply another way of classifying expenditures. It is a way to consider what should be funded, the level of the funds that should be applied, and the policy impact of those funds. If effective, therefore, program budgeting should represent a substantial change in the way that budgets are prepared, approved, managed, reported, and evaluated.
This would imply a more top-down approach to budget planning, in contrast to the bottom-up approach of more traditional budgeting. The drawbacks of the bottom-up approach to budget planning are well appreciated (Box 11). Despite the attraction of developing a more proactive, top-down approach, this is not easy because it implies having a strategic view of the state’s role in society, with important implications for budget policy. First, it implies a sharper differentiation between private and public sector activities and a clear view of what government should and should not do. Second, it requires a determination of the maximal size of the resource envelop available to the government sector. Third, it demands a clearer view of where and for what purposes the government resources are optimally deployed within this resource envelop. The impetus for performance budgeting comes from this latter perspective.
For many countries, the implications of this reorientation have been profound. The more precise specification of activities to be carried out by government has, for many industrial countries, resulted in a considerable downsizing of government and other fundamental public sector reforms. It also has resulted in a restructuring of the relationship between government and the private sector, which typically involves a greatly reduced role for the government in influencing private sector activities. For the government, it has usually involved borrowing the techniques of the private sector for planning and delivering services, including the adoption of “strategic planning” in marked contrast to more traditional development planning methodology. This can be viewed, following the New Zealand model, as a top-down process in three stages: first, a determination of the overall sustainable level of government resource use in the economy; second, within this envelope, a definition by the government of its core activities (“strategic results areas,” in the terminology of strategic planning); and third, translation of each of these basic core activities into action plans for each department or agency involved in that core activity (its “key results areas”). This is schematically represented in Box 12.
Limitations of a “Bottom-Up” Planning Process
The spending plans may not be sustainable. The bottom-up aggregate demands may result in an unsustainable level of resource use because no one has an overall aggregate view. The absence of an effective resource constraint results in poor budget discipline.
There may be counterproductive resource allocations. People working in individual sectors do not fully understand and/or allow for complex development linkages. Projects and programs may not be fully compatible and mutually supportive.
There is a tendency for bottom-up demands to reflect powerful interest groups and not necessarily general welfare considerations. One might suspect the interests of the weak or disadvantaged may not be adequately heard.
There may be significant gaps in the coverage of such plans. Important areas may not be covered because no one acts as a “product champion” by accepting responsibility for formulating a needed project, or because ambiguities in the public’s understanding of the role of the state result in neither the government nor the private sector undertaking some types of activity (for example, some types of research or information gathering).
Are there any guidelines for formulating these strategic plans? First, it is easier if there are relatively few strategic results areas and if they represent those activities or achievements considered to have critical bearing on a country’s development.32 For this reason, they should perhaps center on a government’s core “public good” function: law and order, infrastructure, education, and health facilities as investments in human resources. They should, to the greatest extent possible, be presented in output-oriented terms.33 Within this framework, the next step is to translate each strategic results area into a set of key results areas for each department which are represented in the departments’ budget vote, incorporated in their business plans, and reflected in their level of budget funding. In this somewhat idealized system, the strategic plans are translated into budget allocations—set in an MTBF to ensure sustainability. The role of the budget then is to provide the link between higher-level planning and formulation of more detailed implementation plans.
Introducing Medium-Term Budget Frameworks
A realistic MTBF is often viewed as the cornerstone of a performance-oriented budget approach, linking resources to the policy objectives that define performance. As indicated, this approach involves a reorientation of the budget process to employ systematic use of medium-term forecasts of economic and fiscal aggregates to assist in developing a multiyear framework for spending policies. However, the MTBF does not imply multiyear budgeting in the sense of appropriations extending beyond a single budget year.34 Rather, the first-year outturn estimate becomes the starting point for preparing the budget for the following year. While most OECD countries have adopted some type of MTBF, the exact forms vary from country to country. Box 13 summarizes OECD experience in this area. However, most of existing MTBFs include all or several of the following key characteristics:
A clear fiscal policy statement which enables the MoF to set a medium-term path for a specific public expenditure aggregate;
A medium-term macroeconomic forecast within which the annual budget and multiyear budget estimates can be presented to the legislature each year;
The capacity to produce realistic revenue estimates for the plan period (a crucial element);
A formal requirement that ministries maintain estimates of expenditures covering several years beyond the proposed budget year;
A budget preparation process that assesses and gives formal status to the “forward” or “out-year” estimates as well as the estimates of outlays for the budget year;
A process by which the budget figures for the individual spending ministries and agencies become hard budget constraints expressed in cash terms.
In introducing this approach to budget preparation, middle-income countries have perhaps not fully exploited the MTBF’s role as part of a wider shift in budget management to simultaneously limit and take strategic choices at the macro level while introducing a performance perspective at the micro level. Also, it has not always been fully appreciated that a number of preconditions are required for success. For example, given that an important objective of an MTBF is to lengthen the time horizon of fiscal policy, a degree of macroeconomic stability is important, especially on the revenue side. Simply put, if planned resources are not forthcoming, even for priority activities, no great attention will be given to either the defining priorities or the resulting framework.
Furthermore, the required degree of transparent prioritization at the political level can create problems from different perspectives. First, introducing a transparent process like an MTBF that makes explicit policy choices presumes an adequate level of governance and transparency. Second, to introduce an MTBF is also to introduce a process of budget rationalization. Portfolio rationalization, especially where it involves reductions in ministries’ activities, is a difficult and often painful process with direct political implications. Third, there are other political considerations: the timetable for budget preparation must be changed, and this may need some legal backing. Most OECD countries use up to one year to prepare the next year’s budget and MTBF in order to allow for extensive interaction between the MoF and individual ministries, to facilitate a common agreement of the economic situation and the scope for government spending, and thus to enhance the legitimacy and creditability of the budget preparation process. Last, the MTBF must be supported by adequate capacity in the rest of the PEM system. In isolation, an MTBF cannot compensate for inadequate financial planning, project/program management, access to information, and forecasting skills. This theme is pursued further in subsequent sections.
Connecting Planning with Budgeting
A more detailed description of the key issues involved in the design of an MTBF can indicate the scale of the challenges of operationalizing this approach.
Clarity in the policy statement: At a minimum, there should be an explicit statement of fiscal policy with a target fiscal deficit for the budget year and some planned track for the deficit in the two to three following years. On the basis of revenue forecasts, that target can be translated into a medium-term path for an appropriate public expenditure aggregate. Such policy statements should be stable (not changed every year or so), realistic, and transparent. Accordingly, the successful implementation of an MTBF requires that at least two types of procedure be institutionalized: first, a system of regular reporting of actual fiscal developments compared to planned performance, with reasons for any deviations; and second, a requirement that decisions be made and acted upon for required policy adjustments in current and future years in light of the medium-term targets. Country experience shows that this requires a commitment by all units within the budget process to make the system work. Of particular importance is that the aggregate target be endorsed by the cabinet and the legislature. In subsequent policy proposals, the government and/or legislature should be obliged to show how these proposals can be undertaken within the norm established for the current year and the medium term.
OECD Practices: Use of MTBFs
| Percentage of OECD Countries | ||
|---|---|---|
| 1. | A medium-term framework exists with targets/ceilings for expenditures, deficits, and revenues: | 63 |
| a. Ceilings for expenditures, but targets for revenues and deficits | 20 | |
| b. Ceiling for deficits, targets for expenditures and revenues | 36 | |
| 2. | An MTBF is a legal requirement | 50 |
| 3. | An MTBF covers: | |
| a. 3 budget years | 40 | |
| b. 4 budget years | 24 | |
| c. 5 budget years | 24 | |
| 4. | Based on a growth assumption with a margin for “prudence”: | |
| a. Margin less than half a percent lower than forecast | 17 | |
| b. Margin more than half a percent lower than forecast | 4 | |
| 5. | Multiyear cost estimates are made for: | |
| a. All new spending | 56 | |
| b. Only for selected expenditures | 18 | |
| 6. | Made available to the public | 64 |
| 7. | Legally required to be made available to the public | 36 |
| 8. | Budget documents contain a statement of medium-term fiscal policies | 81 |
| 9. | Budget documents contain a forecast of fiscal aggregates for two future years | 58 |
| 10. | Budget documents contain formal rolling medium-term (3- to 5-year) estimates of revenues and expenditures | 50 |
| Percentage of OECD Countries | ||
|---|---|---|
| 1. | A medium-term framework exists with targets/ceilings for expenditures, deficits, and revenues: | 63 |
| a. Ceilings for expenditures, but targets for revenues and deficits | 20 | |
| b. Ceiling for deficits, targets for expenditures and revenues | 36 | |
| 2. | An MTBF is a legal requirement | 50 |
| 3. | An MTBF covers: | |
| a. 3 budget years | 40 | |
| b. 4 budget years | 24 | |
| c. 5 budget years | 24 | |
| 4. | Based on a growth assumption with a margin for “prudence”: | |
| a. Margin less than half a percent lower than forecast | 17 | |
| b. Margin more than half a percent lower than forecast | 4 | |
| 5. | Multiyear cost estimates are made for: | |
| a. All new spending | 56 | |
| b. Only for selected expenditures | 18 | |
| 6. | Made available to the public | 64 |
| 7. | Legally required to be made available to the public | 36 |
| 8. | Budget documents contain a statement of medium-term fiscal policies | 81 |
| 9. | Budget documents contain a forecast of fiscal aggregates for two future years | 58 |
| 10. | Budget documents contain formal rolling medium-term (3- to 5-year) estimates of revenues and expenditures | 50 |
The distinction between new and old policy: An MTBF generally requires that a clear distinction be drawn between expenditures associated with new and existing policies. As emphasized previously, in this way the MTBF supports strategic planning at the highest level and allows for transparent prioritization between sectors and within sectors. Accordingly, expenditures can be divided into existing policy and “agreed new policy,” with a reserve left for the admission of new policy proposals in later years. This requires all types of spending to be clearly identified and costed in a standardized way in advance, typically necessitating more detailed knowledge of the composition of government expenditures. This, in turn, requires improved budget classification, usually by development of a more functional economic classification that includes subfunctions and clearly relates programs in a hierarchical and unique way. Second, there is a need for improved reporting: ministries will have to maintain and report not just on past expenditure levels but also on planned future expenditure levels. The MTBF should also lead to more detailed budget proposals for strategic planning purposes, including why government intervention is required, specific outcomes sought, expected outputs required for those outcomes, and multiyear estimates of the costs. Third, there must be clear differentiation between those expenditures that are obligations arising from other laws (statutory spending) and those expenditures that are truly discretionary.
Reliability in the macroeconomic framework: The government must have a macroeconomic forecasting capacity and set realistic envelopes for the affordable total government expenditure within the medium term. Clearly, the more stable the main parameters, such as inflation and the exchange rate, the easier the task. However, even for economies experiencing some instability, the MTBF is a useful approach for investigating the areas of uncertainty and setting a realistic path, while allowing for uncertainty through a planning reserve (discussed below). To be effective, implementation of an MTBF requires enhanced planning and forecasting capability not only in the MoF but also in the ministries, as well as a stronger review capacity in the MoF. This may require capacity building in the MoF in macroeconomic forecasting techniques, particularly with regard to revenues, and capacity building in most ministries for supplying information regularly to the MoF on the prognosis for current and future expenditure levels. When projected expenditures seem likely to exceed budget estimates, ministries will have the responsibility of finding savings within their own areas and proposing new policies to the MoF to respect the estimates within their appropriations.
Allowance for cyclical factors: A number of countries also make some allowance for cyclical factors in their MTBFs by accommodating the variation in certain well-defined expenditure programs. Such an approach allows for the operation of automatic fiscal stabilizers on the expenditure side. It also imposes discipline during upswings when, say, unemployment expenditure programs are temporarily low and policymakers are tempted to use those resources for other expenditure programs. When first developing macro forecasts it is recommended that, at least initially, no specific allowance be made for cyclical factors.
Accommodating the effects of inflation: The question arises whether expenditures should be forecast in terms of cash rather than volume (or constant price) terms. Several countries that started with a real-expenditure-based norm have switched to a cash planning system. This is due to its transparency as well as to a reluctance to set the norm in real terms, which provides ministries with complete protection against inflation at a time when it may be desirable to cut spending. Using a cash-based system does not mean the impact of inflation is ignored, but if inflation is higher than expected there is no presumption of an accommodating increase in line ministry appropriations. By setting a clear cash limit, giving greater flexibility to managers, and establishing a degree of performance accountability, governments using this approach in the MTBF have provided stronger incentives to line managers to use their budget resources more efficiently.
The coverage of the MTBF: There is a tension between planning for those expenditures that are wholly under the government’s control (which may only cover parts of central government expenditure) and a wider expenditure aggregate, such as general government expenditure (which may be less subject to central government control but more meaningful for macroeconomic analysis and thus more relevant to establishing the acceptable size of the deficit and tax burden). While practice varies, most OECD countries now employ an aggregate that includes the activities of the secondary and tertiary levels of government—although generally excluding public sector enterprises, except for transfers and net lending to them.
Choice of time horizon: While practice has varied, there is a growing consensus for MTBFs of three to four years, with the majority of OECD countries now using three years. Of course, for some purposes, the government should look much further forward—for example, to answer questions about the longer-term consequences of existing pension or health policies as the population ages. For such questions, the MTBF needs to be supplemented by periodic longer-term analyses.35
The inclusion of planning and contingency reserves: A planning reserve can be defined as a reserve in the forward years available to be allocated for provision to priority expenditures in forthcoming annual budget negotiations. A contingency reserve is usually thought of as a reserve for dealing with unexpected, but unavoidable, expenditures during the budget year. Reserves accommodate uncertainties of several types. They accommodate changes in macroeconomic assumptions, and so countries with higher or more unpredictable inflation paths must have higher reserves. They accommodate exogenous shocks such as natural disasters and changes in the costs of programs, for example, when the projection of demand for government services underestimates volume requirements. They also accommodate new and changing policy priorities. The aim should be to set a reserve large enough to accommodate all of these uncertainties, without it becoming so large that it negates efficient budgeting. For example, a figure much above 5 percent of general government expenditure would be hard to justify except for a high debt/high inflation economy. In setting reserves, a simplified 1-2-3 rule of thumb may be a useful starting point: if the unallocated contingency would be set around X percent in the current fiscal year, it would rise to 2X percent in the second, and 3X percent in the third, reflecting the fact that uncertainty increases over time. Of course, use of the contingency reserve in the current year would require approval by the MoF, which should grant it only reluctantly. A record should be kept of the agreed claims against this reserve, which should be published at the end of the fiscal year. Above all, it should be made clear that the contingency need not be fully utilized.
Formulation of Program Structure in Budget Format
Emerging market economies typically must fundamentally redesign their program structure to allow a new performance budgeting approach. As argued in the previous section, this needs to occur in the context of a wider strategic plan for government operations to strengthen the links between policy and planning and budgeting. In this context, the review of program structures should involve a clear policy statement, or list of objectives, that adequately defines the purpose of the programs and the results expected from the programs so that they can be assessed and measured in some way. The process for this review is outlined in Box 14. In this way, the central characteristic of a program is that a given collection of government activities shares a common set of objectives—that is, the program structure is based on policy.
While the program structure should be seen as the means whereby the budget is linked to strategic objectives, the issue still arises how programs should be best designed to meet these objectives. There appear to have been two basic approaches to this task, largely reflecting the trade-off between viewing the program either as a tool for planning or as a tool for budgeting. The route pioneered by the United States is to make the program structure agency specific—take a spending agency (say, a ministry) and design the program only within that ministry’s activities, hence anchoring the program to a specific budget. Other countries have designated broad policy areas and have identified programs on this basis, which means that individual institutions can end up contributing to only part of a program. The agency-specific approach constrains the logic of the policy basis of programs by the prevailing organizational structure of government, and it is possible to end up with cost centers that are meaningless from a policy analysis viewpoint. This approach does, however, make accountability easier to specify, monitor, and enforce. The broad policy approach, while purer from a policy viewpoint, depends on having adequate classification and accounting capability to capture all inputs associated with programs, regardless of where they arise. This can be a problem in countries without this capacity or where budgets are still firmly rooted in approved inputs, and this makes accountability more difficult to specify, monitor, and enforce.
Process for Agreeing on Program Budget Format
Identify strategic results areas.
Define policy objectives in the agency’s area of operations (sector).
In light of this, identify the key changes required for the agency to deliver its strategic goals.
Realign strategic action plan.
Review each spending unit’s role.
Redefine the spending agency’s current mission (may change from year to year).
Define key results areas.
Given these key results areas, decide whether the agency is organizing itself in the best way to achieve these results.
Decide on the final distribution of accountabilities, i.e., who is responsible for doing what.
Create responsibility centers.
Identify accountable officers.
With their supporting staffs, accountable officers should prepare implementation plans that elaborate in detail how, when, and by whom key results will be achieved/delivered.
Prepare budget format and costing.
Finalize budget formats and cost-allocation procedures so that agencies are able to prepare mock-ups of financial and accounting reports to be produced by the agency or the MoF systems.
Agree with MoF on any new codes required for new activities and on nonfinancial performance measures.
In general, the agency-specific program design is preferable for most countries, but in exceptional cases, multi-agency programs may be unavoidable. A useful starting point in determining the policy framework for program design is to use the functional classification of government expenditures. In presenting categories of government expenditure useful for international comparisons, as well as a means of presenting budget data that is meaningful to the general public, functional classifications have been standardized internationally (for example, the UN’s COFOG system). Because the functional classification aggregates data from a range of government agencies that are engaged in similar activities, functional classifications provide a comprehensive view of government expenditures in such broad areas as defense, health, social welfare, and education and thereby provide a useful framework for constructing a program structure. However, within this framework, the way that functional categories can be broken down into programs and directly related to policy will vary from country to country. General guidelines on how this might be done are presented in Box 15.
When reviewing country experience in adopting ground rules for the design of program budgets, it is possible to identify three key issues: the agreed rules for aggregation to a program structure; the institutional coverage of programs; and the existence of adequate management information.
Agreed Rules for Aggregation to a Program Structure
This is an important precondition for making program budgeting a success. These rules should reflect the level of aggregation of basic activities and/or projects that is most useful for defining a program. If the programs are too disaggregated, then some programs may have only one subprogram, calling into question the usefulness of this distinction. However, if the programs are too aggregated so that their implementation cuts across institutions, it will be difficult to define specific program outputs and assign specific accountability. Care should also be taken to agree on the specificity in program definitions: Should programs be expressed in terms of a specific major policy area, rather than as a catch-all for a loosely related set of activities? Is it possible for each subprogram to capture activities that produce homogeneous outputs? Can there be a clear specification of the outputs sought in terms of timeliness and accuracy, and in a way that can be measured?
General Guidelines on the Design of Programs
Programs are “unifunctional,” so that each program is linked to only one function.
Programs are hierarchically constructed, so that each program has a number of subprograms, and each subprogram can be decomposed into a number of activities and projects. Each subprogram is related to only one program; likewise each activity and project is related to only one subprogram.
Each program has an appropriate size for efficient management. This varies from country to country, but often implies that for broadly defined and resource-intensive programs, the main unit for accountability and performance specification is at a lower level (say, at the subprogram rather than the program level).
Programs and subprograms should be defined in a way that supports political decision making and prioritization by making clear the relationship between the resources used and the expected outputs and policy results (outcomes).
Programs must consider all related activities (including regulatory ones) and projects that together assist in achieving their objectives. This means that capital and recurrent spending should both be considered in judging program performance relative to its objectives.
Greater levels of disaggregation to activities and projects will be used to support, and be designed to facilitate, management’s pursuit of the subprogram objectives and results.
Accountability for subprograms should be clearly assigned to particular managers, usually and preferably within a single organizational unit.
Responsibility for implementing each particular program should almost always align by administrative unit to one chapter (or “vote”) in the budget. Where this is not possible, it is important to assign lead roles to a particular chapter.
The Institutional Coverage of Programs
In redesigning programs, the following range of questions typically needs to be addressed: How closely should programs be aligned with the organizational structure? Can programs extend beyond single ministries? Can a work unit contribute to more than one program? A related accountability issue is the alignment of programs with appropriations: How should appropriations be allocated to programs?
When originally introduced, program budgeting described programs irrespective of organizational lines of responsibility. It was not unusual for responsibility for the design of the program structure to be given to the central budget agencies. Sometimes, the line ministries were not heavily involved, especially at the design stage, even though they had the most detailed and comprehensive view of the functions and objectives of their entities and were responsible for implementing the programs. Many countries have tried this approach since the 1960s, and few have truly succeeded in making it work. Increasingly, the emphasis is on a modified approach to program budgeting, which focuses on outputs and performance with improved accountability—the output budgeting and new performance budgeting approaches. However, this should not detract from the conclusion that introducing a good program structure should involve substantial changes in the organizational structure of government. The reality is, of course, that this is usually not a practical option.
A dominant reason for arranging for a given output to be managed within one organizational unit is to assign clear accountability for performance. In practice, the bulk of outputs are produced within a single organization, for example, medical treatment at hospitals, teaching services in schools. Problems arise, however, where a program seeks complex outcomes, such as crime prevention or tourism development, which necessarily require outputs from a range of agencies. When programs are conducted jointly by two or more administrative units, responsibility is also divided, so that specific activities and end products are difficult to assign to individual organizations. In such cases, costing becomes difficult and requires elaborate costing systems, discussed further in Section V. At the same time, it is unwise to reorganize government agencies in the interests of introducing a simplified program structure. Instead, in designing a program structure, it is necessary to recognize a variety of operating units: some are merely supervisory, others implement their own programs, and some do both; some fund projects implemented by other levels of government; and some have responsibility for public enterprises and quasi-autonomous bodies. Each type may require a different approach.
Existence of Adequate Management Information
Improved accountability is dependent on adequate information to judge performance. The design of programs therefore should take into account the practicality of constructing a comprehensive and regular information and reporting system to provide relevant data in a timely manner. A major user of a performance budget should be the program management. However, a government consists of vast hierarchical systems of many levels and organizational relationships. Designing information systems for such structures is complex: At what levels should costs be collected? Can performance indicators be provided for those cost collection centers? From what sources should originating data be collected? How can the reliability of this data be ascertained and improved? How should data be reported to higher levels? What information do different managers need, and what regular reporting system would supply that information?36 What are the implications for the redesign of government financial management information systems? The establishment of a performance budget management system requires finding answers to these questions, and how to do so is discussed in subsequent sections.
Or as Burkhead (1956) put it, “that emphasizes the things which government does, rather than the things which government buys. Performance budgeting shifts the emphasis from the means of accomplishment to the accomplishment itself.” (p. 133)
For example, Wildavsky (1969) and Henke (1992).
An early definition by the United States Bureau of the Budget quoted in Axelrod (1995): “A performance budget is one which presents the purposes and objectives for which funds are required, the costs of the programs proposed for achieving these objectives, and quantitative data measuring the accomplishments and work performance under each program” (p. 266). The Tennessee Valley Authority’s program budget system of the mid-1930s, while a “for profit” agency, is a classic example of the approach, and other federal agencies followed by developing activity schedules within program budgets during the war years.
The Hoover Commission also publicized the concept of a performance budget in recommending “that the whole budgetary concept of the federal government should be fashioned by the adoption of a budget based on functions, activities, and projects; this we designate a performance budget.”
“A program or performance budget should be substituted for the present budget, thus presenting in a document of much briefer compass the government’s expenditure requirements in terms of services, activities, and work projects, rather than in terms of the things bought” (A.E. Buck in the Hoover Commission’s Task Force Report).
As a consequence, Schick (1971) complained of hybridization, whereby various aspects of the performance system were assimilated into traditional budget procedures and thereby watered down. He writes: “once the initial allure of performance budgeting had worn off, it was viewed with indifference. Performance budgeting was not salient to the interest of budget participants, nor was it regarded as an important reform.” (p. 62)
Now the Government Accountability Office
However, to some critics, the data on programs and performance are just window dressing, with decisions being taken based on traditional line-item objects of expense (Schick, 1966).
Waterson (1965) identifies 16 countries, 10 in Latin America, that had experimented with performance budgeting; Premchand (1977) provides brief case studies of similar budget innovations in Asia.
These problems led Dean (1989) to conclude that “the history of performance budgeting is one of high hopes and disappointing achievement” (p. 123).
For a fuller description of these approaches, see Anderson (2004, pp. 5ff), which also discusses the impact of the 1981 President’s Private Sector Survey on Cost Control (The Grace Commission).
FASA included provisions for agencies to establish cost, performance, and schedule goals for major acquisitions and related pay to performance; Clinger-Cohen included requirements for agencies to develop performance measures for information technology used or acquired by the agency.
See Mikesell (2003).
See Bellamy and Kluvers (1995, p. 39).
For a review of the experience and difficulties in implementing this approach, see Schick (1996).
For full details, see Kristensen (2002).
These will be country specific because what constitutes a “public good” is debatable, and governments need not necessarily deliver all such public goods.
That is, the expected results should be explicitly stated and should be quantified where appropriate and relevant.
Although multiyear appropriations are used by some countries for some expenditure items, such as capital spending, others have the flexibility of carrying forward unspent budget appropriations to the next year.
One technique that has been employed for this purpose is inter-generational accounting, developed in the early 1990s by Auerbach, Gokhale, and Kotlikoff (1999). The United States was the first to present such accounts, in 1993, with other OECD members following, including Germany, Italy, New Zealand, Norway, and Sweden. For a fuller description see OECD (1997b).
The particular problems of performance measurement are addressed in Section IV.
