5 Program Classification

Marc Robinson
Published Date:
October 2007
  • ShareShare
Show Summary Details
Marc Robinson and Holger van Eden1 

The programmatic classification of expenditure—that is, the grouping of expenditure by common objective for budgeting purposes—is a basic information tool used by most contemporary performance budgeting systems. The program classification of expenditure is intended to introduce clarity about the objectives being pursued through public expenditure, the government activities aimed at achieving these, and about amounts being spent on them. Expenditure objectives would usually, at the program level, be defined by the intended results expected for society, or outcome. To borrow the words of the French legislation of 2001 which established the framework for the new performance budgeting system in that country, a program “groups together expenditures which finance actions or coherent groups of actions…[and] which have precise objectives defined in terms of outcomes of relevance to the public. “2

As noted in Chapter 1, the continued prevalence of the idea of classifying expenditure by objective is obscured by wide differences of contemporary program terminology: while some countries still use the word “programs,” others have adopted a confusing array of other terms (for example, output classes, requests for resources, business lines, key result areas,3beleidsartikelen4) to mean essentially the same thing. Whatever the terminology, the basic requirement for the programmatic classification of expenditure is always that programs are clearly defined by reference to their policy objective or, more precisely, that they are coherent groups of outputs (together with their supporting activities) which are aimed at a certain outcome. Program classifications involve the development of a program “hierarchy” classifying programs into constituent components (sub-programs, activities, and so on). In some countries, programs are the highest level of the program hierarchy, but in quite a few others programs are the second level of the hierarchy. The highest level of the hierarchy then encompasses groups of programs aimed at achieving outcomes in a certain policy area. This highest level of the program hierarchy is then referred to as key result area, main policy area, or high-level program. In this chapter, the term “program” is used to mean the highest level of the program hierarchy.

The classification of expenditure by objectives is not intended only for budgetary planning and reporting purposes. The aim is, rather, that it should serve to promote budgetary decision-making which more closely aligns government activities with political priorities, and contributes to better public sector performance. Budgeting in terms of programs is intended to facilitate good allocative decision-making, and is (as discussed in Chapter 1) intrinsically more suitable for this purpose than traditional budgeting in terms of so-called “line-items” (“object classes” in the terminology of the US national government) such as employment costs, supplies, and travel expenses.

It is when the programmatic classification of expenditure is combined with good information about the results delivered by programs (performance measures, evaluations, and so on) that it is of greatest use in assisting decision-makers at both central and line-agency levels to make more informed judgments about whether to increase or reduce the amounts spent in the pursuit of competing public policy objectives. However, not many countries have progressed to the stage of actually using performance information in the budget process in a systematic way. Defining a program structure of expenditure already provides a major step in reaping the benefits of program budgeting.

In addition to its core role in facilitating improved expenditure prioritization, the program classification of expenditure is also widely seen as contributing to improved technical efficiency. There is, in particular, a strong contemporary consensus that the relaxation of line-item controls is the natural concomitant of the increased used of programmatic controls on expenditure. The relaxation of line-item controls gives managers greater flexibility in the choice of the mix of inputs used to produce outputs, which can potentially result in substantially improved productivity. This increased freedom to function without externally imposed line-item controls, however, carries with it an obligation by these managers to track accurately the costs associated with pursuing various outcomes and outputs.

The use of program classification for budgeting purposes commonly involves all or most of the following:

  • The legislative body appropriating expenditure by programs, that is, programs being used in the annual budget law as appropriation items.

  • The use of the program format in the supporting budget information papers presented to the legislature (and made available to the public) as the basis for the exposition of expenditure plans. Typically, this involves for each program a statement of the program objective, a description of the key program activities and associated outputs by which the program objective is to be achieved, a short narrative discussion of past and planned program strategies, major investments, and selected measures of program performance. In more sophisticated systems, such a description can be augmented by an overview of past and expected future performance (including possibly performance targets, if the country concerned has decided to engage in target-setting).

  • A requirement that line agencies present their budget “bids” by program categories, and, in more sophisticated systems, that these bids be accompanied by supporting program performance information and sometimes also by proposed performance targets.

  • The formulation of forward expenditure estimates according to the program classification—in other words, the preparation of bottom-up costed multi-year estimates of ongoing/approved expenditure agreed between Ministry of Finance and line ministry in terms of program categories.

In essence, the program classification of expenditure becomes the main budget classification under program budgeting.

The objective of this chapter is to articulate key principles of a good program classification—that is, to identify the principles and practical considerations which should determine how programs and their component elements are defined. The chapter will argue that a solid program classification facilitates the budget process immensely. It is not the objective of this chapter to prescribe the manner in which program expenditure information may be used to best effect in budgeting and management—this is part of the subject matter of Part Two of this volume. However, the manner in which program information is to be used has a very important bearing on how programs should be defined, and for this reason certain key issues concerning the use of program information are covered in this chapter.

An important practical consideration affecting a number of program classification choices is the need to attribute costs between programs in a reasonably accurate manner. Cost attribution between programs was treated in detail in Chapter 4 on cost information for performance budgeting. This chapter will draw on the information in Chapter 4 where it is relevant to issues such as the role of administration programs and the expenditure coverage of programs.

Key program classification issues

There are a number of recurring issues which arise when governments embark on the program classification of expenditure, and which this chapter explores. These are:

  • How should program objectives be defined, and what relation does this have to the way in which programs themselves are defined?

  • Is it appropriate that there be “administration” programs or other programs based on support services, or should the costs of these support programs be allocated to programs with a common intended outcome?

  • What expenditures (or transactions more generally) should be included in the program structure?

  • What should be the size, scope, and number of programs? What would constitute too many programs, or too few?

  • How do the various layers of the program’s structure connect? Is there a logical relation between programs and the lowers levels of the program structure?

  • How should the center of government, that is, usually the Ministry of Finance and the chief executive’s (Prime Minister or President) office, guide the decision-making process on the allocation of expenditure between competing purposes, using the programmatic classification of expenditure? What form should central guidance take, that is, how directive should it be?

  • To what extent should the budget law lock in the allocation of expenditure between programs during budget execution? Put differently, what scope should the executive have for reallocating expenditure over programs or over lower levels of the program classification?

  • In the context of program-based expenditure control, how far can and should the relaxation of input (“line-item”) controls go?

  • How should the program classification relate to the organizational structure of government? Should programs be permitted to cross ministry/agency boundaries? And what should be the relationship of program structure to the internal organizational structure within agencies?

Outcome-focused programs

It is fundamental to good program classification that the objective of each program should be clearly articulated. In practice, program objectives are often specified poorly, and this is frequently the result of misunderstandings about the principles which should guide the definition of the programs themselves. It is therefore important to emphasize and elaborate on the point that program objectives should be aligned with the strategic policy objectives of government. Strategic objectives are, mostly, best formulated in outcome terms. As discussed in detail in Chapter 3, outcomes are the intended impacts of the services which government delivers. More precisely, they are changes brought about by public interventions upon individuals, social structures or the physical environment. It is only possible to state the objectives of a program in outcome terms if the constituting parts of the programs, its concrete outputs and activities, are all focused on a common intended outcome. The following are examples of good practice in the specification of program objectives:

  • The “Nature” program (“business line”) of the Canadian Department of Environment (2004-05 fiscal year), the stated objective of which was “the conservation of biological diversity in healthy ecosystems.”

  • “Outcome 1” (effectively the school education program)5 of the Australian Department of Education, Science and Training (2006-07 fiscal year), the objective of which was that “individuals achieve high quality foundation skills and learning outcomes from schools and other providers.”

It is as a rule inappropriate to specify program objectives in a way which does not indicate the intended outcome of the program or, more fundamentally, to create programs which are not outcome-based. It is, in particular, important to avoid describing program objectives in terms of activities (that is, in terms of the type of work task6). This occurred, for example, when one African country relatively new to program budgeting recently specified the objective of a newly-established “general education” program as being to “manage the development, implementation, evaluation and maintenance of national policy, programs and systems for general education and quality assurance.” While programs can and should be looked at regularly from an activity perspective, an organizational perspective, and even from an input perspective, when defining programs it is essential to look at them from the perspective of government strategic objectives, which in most cases would be specified by a outcome.7 Defining programs by the nature of activities performed or, worse yet, by the organizational entity implementing these activities, runs counter to the essential theme of performance budgeting (and “managing-for-results” more generally), which is that budgeting and management should be oriented towards results delivered to the community.

To say that programs should be outcome-focused does not mean that programs can or should always be defined only by reference to the common intended outcome of the outputs which they group together. In many cases, program outputs share not only a common intended outcome, but also other defining characteristics. They may, for example, have a common mode of intervention. Thus, for example, a “Preventative Health Care” program is defined by reference not only to its intended outcome (lower morbidity and mortality), but also by the fact that all program services aim to achieve that outcome by preventing, rather than treating, disease and injury, that is, a type of activity.

Program objectives are sometimes specified by reference to the intended proximate outcome of the program, rather than its intended higher-level outcome(s).8 This will also depend on where programs are positioned in the program hierarchy, that is, at the first or the second level of the program structure. For example, the objectives of school education programs are (as in the Australian example above) best specified in terms of student educational outcomes, rather than by reference to higher-level outcomes such as a better-educated labor force or a more competitive economy. As this example indicates, stipulating the proximate outcome tends to make clear the program’s client or target group,9 in a way which the high-level outcome often will not, thus defining the program more clearly. High-level outcomes are often served by a wide range of very different types of government services and organizations,10 so that programs based on high-level outcomes would tend to be large and diffuse.

Precisely because program objectives tend to be specified in terms of proximate outcomes, a number of countries take the additional step of explicitly identifying the high-level outcomes to which each program is intended to contribute. For this purpose, they define, either across government or per ministry, a list of key result areas (alternatively known as main policy areas, or high-level programs). These key result areas would encompass all line ministry programs and would be defined by high-level outcomes. In this way the program level becomes the second-level in the program hierarchy.

Arguably, one contributing factor in the bad practice which may be observed in many countries in respect to the specification of program objectives (and in the underlying definition of programs) is a specific conceptual framework which is sometimes used in public sector strategic planning, which demands that strategic plans be structured around the concepts of missions, values, goals, objectives, and so on. The problem with the use of this framework in program and performance budgeting is that the concepts upon which it is based are much less well-defined than the standard international “logical framework” based on the concepts of inputs, activities, outputs, and outcomes. For example, it is often not at all clear what exactly constitutes a “goal” and an “objective,” how (if at all) goals and objectives differ, and, most importantly, how they relate to outcomes.

Administration programs

Having stipulated the principle that programs should be defined by their objectives, and that these objectives should be specified in outcome terms, it is necessary immediately to acknowledge that there is one widely—although not universally—accepted exception to this principle: “administration” (“corporate services”) programs. In many countries, program structures have been designed which include an administration program within each ministry. Such programs typically cover the costs of central directorates of the ministry such as personnel, legal services, IT, and security, as well as top management (the expenditures of the minister’s office and top ministry-wide civil service management). Administration programs thus bring together a range of activities which do not deliver services directly to external parties, but which support those areas of the ministry which do deliver such services. These are, in other words, programs not focused on own outcomes, but on supporting those of other programs.

The rationale for administration programs is a pragmatic one, and reflects program costing considerations. The issue is discussed in detail in Chapter 4, and the key points are only briefly summarized here: it is quite possible to avoid administration programs in order to maintain uncompromised the basic principle that all programs should be outcome-based. A small number of countries (Australia, for example) do precisely this. However, if one is to adopt this approach, it is necessary to allocate with an acceptable degree of accuracy all of the “indirect” costs of support services to a set of outcome-based programs. To do this requires accounting systems and personnel capable of carrying out the task of indirect cost allocation. Good cost allocation is a quite demanding and resource-intensive task and, even in the most sophisticated systems, always requires a number of “rules of thumb,” that is, some imperfect decisions on cost attribution. In practice, proper cost allocation is often beyond the capacity and/or affordability of many countries. There is therefore a good pragmatic case for the use of administration programs in many countries, particularly in the early years of the implementation of a program budgeting system in countries where cost accounting is not yet well developed. If administration programs are used, it is necessary to specify their objectives in terms not of intended outcomes, but of their expected outputs supporting the implementation of other programs. There remains, however, a strong case for moving progressively over time, subject to the availability of resources, to reduce the size of administration programs by allocating more and more support services to outcome-based programs (starting, of course, with those support services which are least difficult to attribute, such as training). Ministries of Finance should be vigilant in pushing line ministries to improve cost attribution. Care should also be taken not to allow line ministries to differ too much in the extent to which they make use of “administration” programs.

Similar considerations apply to programs of the handful of central agencies (such as Ministries of Finance, public service commissions, and departments/offices of the President or Prime Minister) which do not deliver services to the public, but which rather regulate and support the activities of service delivery ministries. The objectives of such programs cannot usually be clearly defined in terms of specific intended outcomes. Rather it is more useful to define them in terms of the specific contribution they aim to make to good governance. For example, a Ministry of Finance might legitimately define the objectives of a “budget management” program as “the maintenance of fiscal discipline and the efficient allocation of public expenditure,” notwithstanding that this is not, properly speaking, an outcome.

In addition to administrative programs some countries have also chosen to identify “policy” programs to cover staff activities within line ministries aimed at policy preparation. This approach often reflects internal organizational structures where there are separate policy and operational divisions, with the policy division responsible for policy preparation for multiple program areas. Under such circumstances, it might be argued that cost attribution considerations justify a separate policy program in the same way that they may be used to justify administration programs. The case for this approach is again dependent on the practicality of cost attribution. Often, however, it is less persuasive than for administration programs. The reason for this is that much policy development work is a specialist activity performed by individuals who are specialists in certain subject areas, and who would therefore spend all of their time on policy development related to particular service delivery areas. To the extent that this is not the case, and there are individuals who are involved in policy-making across multiple program areas, a first-best solution is to use a time-recording system to allocate the costs of their work to the relevant program. A further alternative to this is to include the “overhead” component of policy-making in an administration program. Certainly, with time recording of staff becoming more common in bureaucracies in developed countries, the case for policy programs will become less in the coming years.

Which expenditures should be covered

Countries with mature program structures generally try to allocate all or most government expenditure to programs. That implies that the central government’s budget can be seen as a collection of programs encompassing all of the central government’s expenditure. These programs in turn reflect all the government’s policy objectives, all the activities geared to these objectives, and their associated cost, that is, the total of budgetary expenditure.

Developing and transition countries introducing program budgeting often are unable to include all expenditures under the program format—even those that are in principle attributable. For example, personnel costs of programs are sometimes omitted, even for staff working exclusively for specific programs. The reasons for such a stance can be that the accounting department has not made the necessary steps to allocate personnel or other administrative expenditure to programs. As accountancy units are usually not organized according to program expenditure, the easiest category to attribute initially to programs is often so-called external expenditure, that is, transfers and other payments to third parties. By determining the objective of these expenditures, these expenditures can usually be easily attributed to programs. In such cases the program structure omits all internal expenditure within the line ministry. This was and is the practice in a number of transition countries. Internal expenditure is often in such cases still budgeted on the traditional line-item basis.

Another way that developing and transition countries sometimes limit the comprehensiveness of program expenditure is by only identifying expenditure on explicit policy priorities under a program structure. The ongoing activities of government then remain classified under normal input-oriented budget lines. Sometimes, only new expenditure initiatives put in place by the present government are identified as programs. Alternatively, programs are sometimes defined as all government activities of finite duration, with all permanent expenditure classified under the traditional input budget.

Any partial coverage of expenditure in the program structure has a detrimental effect on allocative decision-making. By not fully attributing all expenditures to programs the full costs of programs cannot be compared. This means government cannot compare the cost of pursuing different societal outcomes. Not including personnel costs of the civil service staff would obviously bias expenditure towards programs executed by the civil service, rather than by, for example, lower levels of government or through outsourcing to third parties, even though the latter two options may be more cost-effective for society as a whole.

The non-inclusion of ongoing, or permanent, expenditure in the program structure is detrimental for a similar reason. Expenditure that maintains the basic functions of government will serve particular government objectives. Just because certain functions have been part of government for a long time does not mean they should not be considered as candidates for increased expenditure, or that they should never be cut back. By leaving permanent expenditure outside the allocation mechanism that program budgeting allows, permanent expenditure is, on the one hand, protected against cutbacks, but also, on the other hand, not eligible for increase of expenditure.

A third issue that needs to be tackled in defining program coverage is whether programs should encompass gross expenditure or net expenditure, the latter being defined as gross-expenditure minus own-source revenues attributable to the program in question. A reason for the latter position might be that a program that finances itself through user charges can be seen as cheap, and thus in allocative decision-making government could have a preference for such programs. There are two reasons for not endorsing this position. First of all, user charges can be unrelated to the actual costs incurred by public sector providers and may also be compulsory, so that it has more the character of a quasi-tax than a market price for a service in a competitive market. Second, for transparency reasons it is important to see the full extent of the resources being channeled through government. Without encompassing gross expenditures, a program may look to be very cost-effective, while in reality a lot of public resources are channeled to it. For this reason most countries favor encompassing gross expenditures in their program definitions.

A final issue to note on the coverage of programs, discussed in more detail in Chapter 4, is that mature program budgeting systems integrate current and capital expenditure related to the same policy objective in one program. The reason for this is that if, for example, only the current costs of programs are compared during allocative decision-making, then the capital investment needed to keep the program effective would be seen as a free resource and there would be a bias for more capital intensive programs. In many developing countries, programs have been defined as only covering recurrent expenditure, with capital projects, which are mainly funded by donors, not being linked to recurrent programs. This has led to recurrent expenditures not being aligned with capital expenditure, and has cast in doubt the long-term sustainability of certain public investments. Integration of current and capital expenditure helps present more accurately the true costs of the policy objective a program is pursuing, and provides an opportunity for balancing recurrent costs with capital expenditure, thereby improving the overall effectiveness of government expenditure.11

Design, size, and scope of programs

The development of any good system of program classification requires that the center of government (generally the Ministry of Finance (MoF)) provides guidance to line ministries about the appropriate size and scope of programs. This guidance should be consistent with an appropriate balance between centralized and decentralized use of program budgeting in allocative decision-making, reflecting the fact that the program classification should serve as a tool for budget management both for the center of government and for line ministries and agencies.

The development of line ministry program structures should be a collaborative endeavor between line ministries and the MoF. In general, it is advisable that each line ministry start the process by developing a proposed program structure based on the strategic objectives of the ministry. Thus developing a program structure should be intimately linked to a ministry’s strategic planning. The main difference is often that strategic planning tends to focus insufficiently on ongoing objectives and activities. In developing a program structure line ministries should make sure that all ongoing policies and activities are adequately encapsulated in the program classification. Also, future strategic objectives and polices which have not yet been approved by cabinet would usually only be incorporated in the program structure once budget resources have been allocated to them.

It is very important for the MoF to provide guidelines to line ministries which specify a clear methodology and uniform approach to developing a program classification. These guidelines should, first, provide a clear statement of the fundamental principles of the definition of programs (for example, that programs should be outcome- and output-based) and, second, indicate a clear position on matters where international practice in program classification may otherwise vary (for example, the level of detail in which outcomes are specified, whether there are to be administrative cost programs, the size and numbers of programs). In some countries with a strong culture of decentralized budget management, such as the Netherlands, line ministry program structures have to some extent developed without sufficient uniformity across ministries, making the program budget less effective for allocative decision-making between ministries than it might otherwise have been.

The importance of ensuring appropriate uniformity of program structures between ministries should not be taken as advice that the MoF should dictate the program structure of ministries. The experience is that, in most countries in which the MoF has unilaterally determined the program structure without substantive input and concurrence from line ministries, those ministries do not really “own” the new program structure. This may lead to a situation where line ministries largely retain their input or organizationally driven budget for internal budget management purposes. To assure use of the common methodology, and to balance usefulness of the program classification between the needs of line ministries themselves and the center of government, it can be a good development strategy to have joint line ministry-MoF working groups develop the program structure at the technical level.12 Once the line ministries have developed concrete proposals, the MoF should review these on uniformity across ministries and appropriate application of the agreed methodology. Final negotiations with ministries to resolve remaining issues may be necessary. It should be a prerogative of the MoF to give final approval to the program structure for each line ministry, just as it approves economic and functional classifications of the budget as part of its responsibility for coordinating budget management.

One of the seemingly simple, but crucial practical issues that development of a program structure must address is how many programs each ministry should have. Often line ministries will initially take quite extreme positions. For example, a Ministry of Agriculture might take the position that “our activities are all about improving the agriculture sector,” and that there should therefore be only one program called “Improving agriculture.” At the other extreme, the same ministry may take the position that each of a myriad of services, such as advice to farmers, drought relief, an agricultural research institute, and the like, are all so important that they should be recognized as programs in their own right, rather than being relegated to lower levels of the program hierarchy—with the result that the program structure may comprise 50 or more programs.

Obviously, for allocative decision-making such extreme positions are not helpful. MoFs have often sought to avoid a profusion of small programs, by indicating minimum sizes for programs, or suggesting the approximate numbers of programs that line ministries or agencies should develop. This reflects the desire to keep allocative decision-making of cabinet and parliament—which, as we will see, is often focused on the program level—at a fairly strategic level, and avoid getting these two institutions too involved in details which can, for the most part, be better left to the line ministry itself. There are other reasons for avoiding too many small programs. First, as discussed below, it reduces the flexibility available (given the prevailing virement rules13) for line ministries and their agencies to deal with unexpected changes in the demand for services by switching expenditure allocations on their own authority. Second, the magnitude of the task of cost attribution between programs increases more than proportionately as the size of programs falls (see Chapter 4). Third, a structure with too many small programs can make intra-ministry allocations more problematic. As the program level is usually the level at which expenditure is appropriated by parliament, programs and/or agencies described as programs can be insulated from intra-line ministry reallocation through the action of sympathetic parliamentarians.14

A further consideration is that a certain stability in the program structure at the higher levels is advisable. This reflects the fact that strategic priorities of government do not usually change much from year to year. It also allows for better tracking and evaluation of expenditure programs over the medium term. A program structure with too much detail at the higher level will not show this stability. Stability is, however, not something which can usually be expected to be achieved in the early years after introducing a program classification. Rather, it is often sensible to let line ministries have the opportunity in the first couple of years to make adjustments to their program classification to let it better reflect their strategic objectives. However, after the early years of implementation, the program structure at the higher level should solidify and only then be changed to reflect important changes in strategic priorities.

Although it is important that the MoF provides guidance on the desirable number of programs per line ministry, and therefore implicitly the minimum program size, there is a case to be made against any excessively rigid minimum size requirements. To insist that no program should ever fall below some arbitrary expenditure level could lead to the creation of incoherent programs, grouping together services with different objectives, simply because those services individually account for relatively small portions of the line agency’s expenditure. Programs which lack coherence are not of great value for decision-making about expenditure priorities. There is much to be said for the view expressed by the French parliamentary budget commission that

The coherence of programs is fundamental to the success of (performance budgeting) reform…The need to attain a critical size is often put forward to justify programs which mix activities with little in common. Such an argument is difficult to accept: the coherence of programs is more important than their size. (Commission des Finances, 2004)

It is equally important from the point of view of central decision-makers that programs do not become too big relative to the size of the line ministry budget, because if they do so, they tend to cease to be useful to the government for prioritizing expenditures. Excessive size also tends to produce incoherent programs. In the extreme, for example, if each ministry were to have only one program, many allocative decisions of great interest to the cabinet and parliament would be beyond its reach. This has at various stages been an issue in a number of countries.

For example, under the “accrual output budgeting” system developed in Australia during the late 1990s, ministries greatly reduced the number of programs (now called “outcomes”) they had relative to the previous “program budgeting and management” systems which had operated from the 1980s. The concentration was such that the Department of Defense had only a single outcome, which grouped all of the department’s services under the common intended objective of “the prevention or defeat of armed force against Australia or its interests.” After a few years’ experience with this approach, a strong feeling developed among politicians that such highly aggregated programs gave them too little capacity to influence the allocation of resources to more specific objectives when they felt the need to do so. The result was a deliberate move to increase the number of programs so that, for example, in 2005-06 the defense department had seven programs (including separate outcomes for navy capability, army capability and air force capability).15

The same concerns arose in France after ministries made their initial attempt to define the program structures required by the new reforms flowing from the new organic budget law (LOLF) of 2001. The parliamentary budget commission complained that too many ministries where suggesting structures which gave them either only one mission, or only one programme within each mission (Commission des Finance, 2004). Noting that fully two-thirds of ministries proposed single-mission program structures, the Commission posed the rhetorical question of whether one could really reduce the wide-ranging activities of, say, the mega-Ministry of Economy, Finance, and Industry to one single mission.16

As a very rough rule of thumb, taking into account the need for coherence in programs, and depending on the number of line ministries and agencies which receive direct allocations in the budget, the number of programs (that is, expenditure classes at the highest levels of the program classification) should generally be between 5 and 15 per line ministry. This indication reflects the importance of line ministries focusing on a limited number of related strategic objectives.17

In addition to the overall number of programs, line ministries also often need guidance on the differences in size of the various programs. It has proven not very useful to have one or two huge programs and ten very minuscule ones in the first level of the program structure. While here rules of thumb are even harder to give than for the number of programs, it should be clear that program structures with vastly different program sizes are not very efficient for allocative decision-making, as reallocations in the smaller programs usually have almost no impact on the amount of fiscal space. Decisions about reallocation between minor programs would require the cabinet to spend inordinate time and resources on relatively unimportant government objectives and activities.

Logic of the program structure

We have seen that the program structure consists of various layers. While the names of items in these layers will differ from country to country, the most common program structure consists of three layers that in this chapter will be referred to as: programs, sub-programs, activities. (There are, however, variations on this—for example, in some countries, high-level programs are defined above the program level, while in others a “measure” level is inserted between sub-programs and activities, and in yet others the activity level is further broken down into sub-activities.) Sub-programs typically pursue the same overall policy goal as the program to which they belong but do this by either producing slightly different outputs (resulting in the same societal outcome) or producing similar outputs in different ways, that is, through different public policy instruments. Like the program, the sub-program should be defined by objective, related activities, and amount of resources spent. In contrast to the program, sub-program objectives would be defined by outputs rather than outcomes. Outputs of sub-programs would contribute to the realization of the desired outcome of the program to which they belong.

Beneath the sub-program level, activities will usually represent discrete measures, initiatives, or operations, or projects (geographically or with respect to time, or in some other way distinct) that together form the sub-program. Again the items at the activity level should be connected to well-definable objectives, as all parts of the program structure have the some basic configuration, that is, are based on an objective, encompass certain activities of government, and require a certain amount of resources. The objectives at the activity level would probably be defined as sub-outputs of the output of the encompassing sub-program. Thus, objectives in the program hierarchy generally become more operational the further down one goes; the targeted results tend to become more concrete. At the program level the objectives generally point to outcomes, while at the lower levels of the hierarchy the objectives relate more and more to concrete outputs for which government can be held accountable. Box 5.1 presents the program structure of the Dutch Ministry of Agriculture as an illustration of the relationship between program levels.

Most countries with mature program structures use the first level of the program hierarchy as the basis of allocative decision-making by cabinet. Cabinet and other central executive government decision-makers do not in general concern themselves with the lower levels of the program hierarchy. This is largely left to the line ministries and MoF to discuss in the agreement of the more detailed planning of the budget. In addition to looking at activity and sub-activity levels, MoFs generally would want information on the use of inputs at all levels of the program structure, to verify costing assumptions. The input information would not necessarily need to be detailed, and certainly would be less detailed than under input budgeting systems. It could be limited, for example, to the first layer or layers of the economic classification.

At the line ministry level, more detailed planning of expenditure at the lower levels of the program structure is usually required by the finance department of the line ministry itself. Finance departments would also require input costing of program items. This planning could take place either before or after the approval of the line ministry budget and would be binding to a certain extent for, for example, lower levels of managers in the line ministry. However, this detailed planning under program budgeting should provide more managerial freedom than the very detailed input control that many countries employed under input budgeting. Any mistaken perception that program budgeting does not involve detailed planning and costing of expenditure should, however, be dispelled.

Box 5.1.Budget structure of the Netherlands Ministry of Agriculture & Nature in 2004

Below we present part of the program structure of the Ministry of Agriculture in the Netherlands as it was used for the 2004 State Budget. The first level consists of 11 programs, or beleidsartikelen (policy articles) as they are called in Dutch. We then go deeper into the program structure by looking at the program “Promoting Sustainable Production” (program 5, emboldened below), and look at the sub-programs of that program. We then repeat that process by looking at the sub-program “Eco-friendly fishing” (sub-program 5.5, emboldened below) to see its components at the activity level, which is the final level of the ministry’s budget.

The budget structure below clearly shows the logical connection between the three levels. The sub-programs and activities all fall within the overall policy objective of promoting sustainable (agricultural) production. However, the objectives of the lower levels are more focused and concrete, and at the activity level more time-limited.

Level 1:
ProgramsBudget (€ millions)
1.Strengthening Rural Areas256.4
2.Nature Realization.a176.5
3.Nature Maintenance164.3
4.Economically Viable Agricultural71.0
5.Promoting Sustainable Production121.8
6.Food Safety, Quality, and Animal Health90.3
7.Research & Development294.8
9.Informational Activities62.1
10.Nominal and Contingency-8.4
Total expenditure1,915.1

The presented program structure is by no means perfect. It is obvious that the programs at the first level of the budget structure relate only partially to policy objectives. The ministry seems to have chosen a sizeable Administration program with an unspecified objective, as well as R&D, Training, and Informational Activities programs that seem to reflect support activities of the ministry rather than specific objectives. The Netherlands only introduced program budgeting in 2002 (after a trial period of three years) and to some extent the Ministry of Agriculture’s budget shows that a substantial part of the ministry’s expenditure is not yet being attributed to policy objectives. In the future the three activity-based programs may be further attributed to specific policy objectives.

Level 2 of program 5:
Program 5Budget (€ millions)
5. Promoting Sustainable Production121.8
5.1 Promoting organic farming8.8
5.2 Decreasing use chemical fertilizers7.3
5.3 Protection of crop safety7.6
5.4 Improved animal welfare1.1
5.5 Eco-friendly fishing15.4
5.6 Administration81.7

The second level of the program structure shows that the program “Promoting Sustainable Production” has within it six sub-programs, including again a very substantial Administration sub-program. Obviously, it would be better for allocation purposes if this Administration sub-program within program 5 could be allocated to the five more output-oriented sub-programs. Given that the ministry has administration budgets both at the program and at the sub-program level, a substantial part of the overall budget is not really covered in the program structure, making it less usable for allocative decision-making.

Level 3 of sub-program 5.5:

Sub-program 5.5
5.5 Eco-friendly fishing15.4
5.5.1 International coordination2.8
5.5.2 Improvement of domestic fishing1.0
5.5.3 Innovation projects5.0
5.5.4 Technical measures & research4.7
5.5.5 Various1.9

Level 3 further divides the sub-program “Eco-friendly fishing” into five activities. It could be noted at this level that the identified classification items are often described in activity terms. In a well-developed program budgeting structure items at all levels should, however, remain linked to desired results at varying levels of concreteness. Without the link to performance targets, program budgeting reverts to activity-based budgeting. An activity item called “various,” while understandable from a practical point of view, is not acceptable from a program budgeting point of view. Activities, like any level of the program structure, should be linked to an encompassing objective. This objective should be the designation of the activity in question.

a In a country such as the Netherlands the government has a policy to increase the total land area that could be designated as “nature.” Land is purchased from private landowners and then actively turned into nature reserves. As it is a densely populated country it then needs to be maintained.

Role of the legislature: programs and appropriations

In many countries with program budgets, expenditure is appropriated by, and at the level of, programs. In other words, the annual budget law passed by the legislative body specifies the allocation of expenditure between programs, but is silent on the allocation of expenditure within programs. There are quite a few exceptions to this generalization, however. In Australia, for example, allocation of program expenditure is left to the executive under a system of so-called “global” appropriations under which parliament appropriates an aggregate amount to each ministry and is only informed of the planned program allocation of that aggregate. Conversely, in a country like the Netherlands, parliament has taken the view that the program level of expenditure does not always provide it with sufficient scope for political direction. As a consequence, the government conceded during the introduction of program budgeting that although programs (beleidsartikelen) would remain at the level of appropriation, parliament would retain the right to demand changes at lower levels of the program hierarchy when approving the budget.

A key issue which arises in systems where the legislature appropriates expenditure by program is whether the legislature has the power to vary the program appropriations proposed in the budget presented by the executive. There is considerable variation in international practice on this point. In the United Kingdom, at one extreme, the legislature has no power to alter the allocation of expenditure between “Requests for Resources” contained in the budget proposed by the Chancellor of the Exchequer. At the opposite end of the spectrum, the United States Congress has essentially unconstrained power to alter the program allocations suggested in the President’s budget. In France, the position under the new performance budgeting system lies somewhere between these two poles. In that country, parliament appropriates expenditure by both the higher-level missions and by the second-layer programs. However, the parliament may only alter the allocation of funds between programs—and not between missions.

There is no correct answer to the question of how power over program allocations should be shared between the legislature and the executive. The approach taken depends very much on the political system and on social preferences concerning the more general relationship between legislature and executive. The US approach, for example, is a natural expression of a system based upon the deliberate separation of executive and legislative powers as a safeguard against the undue concentration of powers. The greater concentration of powers in the hands of the executive under Westminster-style systems, by contrast, is often justified by a view that efficient government demands strong executive leadership. As noted in Chapter 20, the provisions permitting the French parliament to reallocate between programs are part of a quite deliberate broader effort on the part of the parliament to claim a greater role in budgeting, in a context where traditionally the parliament had very little capacity to influence the executive budget.

Within-year flexibility in program allocations

In countries where expenditure is appropriated by program, the question arises whether changes in legislated program allocations are permitted, and, if so, by what mechanism. It is not uncommon, both in developed and developing countries, that there are unanticipated within-year changes in the demand for certain types of public services. Also, it is not uncommon for program costs to be incorrectly calculated, or for foreseeable circumstances not to have been taken into account, in budget planning. In many countries, a degree of flexibility is given to the executive to change the program allocations determined by the parliament. The main way of providing such flexibility is through provisions allowing line agencies to transfer (vire) funds between programs and within programs with the approval of a suitable authority (usually depending on the scope of the reallocation: the line minister, the finance minister, or cabinet). Any such virement provisions need to be specified in law—usually in the organic budget law. For virements within programs, which we will also describe below, the organic budget law usually authorizes the Minister of Finance to regulate these in financial regulations.

Because virement between programs clearly reduces the degree of legislative control of expenditure priorities, the extent of such virement authority is usually limited (for example, in both France and New Zealand to a maximum of 5 percent of the amount of any program appropriation (Lienart, 2004, p. 334)). The argument usually made for virement provisions is that it would be too slow and cumbersome to require that any and all changes to program allocations be authorized by parliament through supplementary budget legislation. By contrast, however, no virement of funds between Requests for Resources is permitted the UK, and all reallocations of money between programs must be taken back to the parliament. In defense of such an approach, it may be argued that other means do exist with which to deal with sudden and unexpected changes in expenditure requirements—in particular, the Minister of Finance’s contingency reserve. Many countries have judged, however, that on balance some recourse to virement is expedient. Virements do, however, in general need to be reported on in budget execution reports and/or supplementary budgets.

Aside from the legal possibility of virement between programs, the question arises to what extent the center—the Ministry of Finance—should use the program structure as the basis for budget execution control. That is, should it control the reallocation of funds between and within programs by line agencies during the budget year, so as to make it conform to planned allocations? Or should the lower levels of the program allocation of expenditure be seen merely as an expenditure planning tool, with agencies free to vary the actual allocation of expenditure during the year as they see fit? Universally, the answer given to this question is that the program structure should be used for expenditure control. For the first layer of the program structure, that is, the programs themselves, there is the classic principle of budgeting that any legislative appropriation should only be amended by another legislative action as otherwise the authorization function—or legislative “power of the purse”—would be negated. As indicated, the organic budget law of some countries has softened this principle to some extent by allowing the MoF to approve requested changes in the budget appropriations up to a certain limit. Such changes are usually only allowed within a line ministry budget, that is, the resources cannot be vired to another ministry’s program, and the virement can only take place on request of the line minister in question. For control of fiscal policy, the Minister of Finance also usually has the right, most often only with approval of cabinet, to cut expenditure to below appropriation levels by program if macro-fiscal circumstances require this. Such budget cutbacks should, it is often specified, not invalidate the relative allocative choices authorized by the legislature.18

Below the appropriation level—the level in the program hierarchy at which the parliament appropriates money in the annual budget—the degree of allocative flexibility accorded to line ministries differs considerably internationally. In some of the more advanced countries, there is no central control over the manner in which line ministries allocate funds between sub-programs and activities. In other countries that have just introduced program budgeting, line ministries will in general have some, but not full, flexibility to make changes within lower levels of the program classification. Compared to the line-item control imposed by MoFs under input budgeting the amount of flexibility for line ministries has generally increased greatly. A certain level of control by the MoF is deemed necessary in countries with weak decentralized budget management in order to maintain the integrity of the budget, since the budget is the product of allocative decision-making by cabinet and costing review by the MoF, based on detailed justification of program expenditure. If line ministries were able to make unlimited changes to the constituting parts of programs, this would invalidate the screening and approval process of both the MoF and the cabinet.

Virements frameworks will vary from country to country, and, as a general rule, the stronger the planning, financial management, and accountability mechanisms, the more room line ministries can be given to make changes to their budget within the appropriation limits set by parliament.

With respect to the breakdown of program expenditure between inputs, performance budgeting in general calls for the relaxation of central input controls so as to accord line agencies greater managerial freedom to produce services efficiently (see Box 5.2). This does not mean, however, that all input controls should be abolished—that is, that countries should in general not allow for unlimited substitution between elements of the economic classification. Virement rules usually place some restrictions on the reallocation by the line ministry from one main economic category to the other. For example, there are almost always limitations on the extent that line ministries can vire into personnel expenditure and away from capital expenditure. The rationale for these rules, especially in developing countries is, on the one hand, to avoid turning government into an employment vehicle, and, on the other hand, to protect capital expenditure against the ever-present danger that it will be cut particularly severely as a short-sighted response to fiscal pressures.

Above the level of autonomy for line ministries, the Ministry of Finance will have to approve within-year budget amendments. One of the desirable effects of program budgeting is that the number of such approvals has been reduced substantially compared to a systems of input based budgeting. A framework for line ministry autonomy as described above would lead perhaps to one-tenth or even one-hundredth of the approvals by the MoF compared to a system of input budgeting where the MoF has to give approval for any budget change at the lowest level of the line-item classification. (See Box 5.2 for further information on the trend of reducing input controls and replacing them with control on outputs.)

Under program budgeting an interesting development has been that virement powers of line ministries have in general been increased to allow more scope for line ministries to allocate resources so as to maximize desired outcomes. The thinking behind this increase in powers and flexibility was that accountability of ministers should be based on the achievement of stated objectives, not on the mix of resources used for attaining those objectives. However, to avoid invalidating the screening and costing of proposals and to attain some measure of quality in the planning of budget execution, governments have, in general, even under program budgeting, limited the extent to which in-year changes to budget planning can be made.

Box 5.2.Loosening input controls without losing control

When introducing program budgeting, governments are faced with the option of removing or loosening input controls. In many countries, input controls are extensive. The Ministry of Finance in Greece, for example, controls expenditures up to the lowest level of the four-level economic/line-item classification. This means that office supplies need to be planned by type (paper, pens, furniture, and so on), and that any reallocation of funds of a line ministry or agency during the year has to be approved by the MoF. In a country like the Dominican Republic many reallocations even need to be approved by Congress. In practice, countries with input controls can have thousands of virement approvals by the MoF and hundreds of supplementary budget approvals per year.

Not only is such a system of input controls costly, time consuming, and disruptive for budget execution, it is also often ineffective in what it wants to achieve. Input controls have generally been used to maintain appropriate expenditure allocations, to combat fraud, and to achieve fiscal discipline. In effect, very tight input control has often made budget execution inefficient, as it does not allow for government managers to allocate resource to the best of their abilities. Especially in countries where the government budget is prepared by budget analysts on an incremental basis without much input from government managers, input control can lead to large inefficiencies, and/or a process of systematic alteration of the budget during the year. In Greece, for example, line ministries have indicated that they tend to start preparing virement proposals as soon as the budget is published.

Input controls have often proven to be unreliable or insufficient in stopping fraud as they are based largely on a process dominated by paperwork, and not on the physical, financial, or system controls used in audit processes. In developed countries internal and external audit processes have largely replaced input controls as mechanisms much better attuned to ward against fraudulent use of funds.

While input controls may play a role in limiting aggregate expenditure, there seems to be not much reason for the very detailed controls seen in many countries. Controlling overall expenditure can also be done by controlling expenditure at somewhat higher levels of the economic/line-item classification.

In loosening input controls in the context of introducing program budgeting governments generally choose to substitute input controls with outputs controls based on agreed outputs at the subprogram and activity level. Thus while line ministries will be allowed to vire expenditure over different items of the economic classification, outputs achieved by the line ministry will need to be reported on, and ministries are held to those outputs. Thus while the input accountability under program budgeting decreases, the accountability for that which really matters, namely government outputs, increases.

It should also be noted that program budgeting does not exclude some input controls. As indicated, spending on items in the lower levels of the program classification is not fully free. MoFs and finance departments usually impose virement limits for the sub-program and activity levels on line ministry managers. Many countries have also chosen to impose control on aggregate input categories parallel to the program or subprogram level. In such a system the input reallocation is then limited to within the identified main economic categories. A more simple combination of input controls and program budgeting is a system where personnel expenditure and capital expenditure are controlled within the program structure.

One aspect of the increase in budget flexibility under program budgeting is the increased scope for budget planning and for in-year virement by program managers. Line ministries have not only seen their powers increase, they have to some extent also decentralized their powers to program managers and managers responsible for items in the lower levels of the program classification. The reason for decentralizing these powers is the same as for the shift of these powers from MoF to line ministries: individual managers of programs or sub-programs are better aware of which resource mix will optimize their outputs. Given enough accountability for results, such freedoms should be warranted.

However, the experience with providing more budget flexibility to program managers in developing and transition countries has not always been unanimously positive. Loosening of expenditure control has been associated with larger wage increases, especially at the managerial level, and more expenditure on luxury overheads such as cars and office space. In this sense a consensus has grown that control of program expenditure needs accountability for both outputs and program inputs at an aggregated level.

Relationship of program structures to organizational structure

As noted above, it is widely agreed that it is desirable that programs should to the greatest possible degree be defined by a government’s strategic objectives defined as outcomes. The sub-components of a program can then be seen providing outputs and sub-output contributing to this societal outcome. This approach poses a number of problems, one of which is that a program structure based on desired outcomes would tend to differ in important ways from the existing organizational structures in government. The question therefore arises whether it is appropriate to seek to align program and organizational structures, and, if so, whether this should be done either by designing program structures in such a way as to respect organizational boundaries or by modifying organizational structure so as to reflect more closely the outcomes being pursued.

Programs and ministry/agency boundaries

Many countries confine programs to ministry boundaries. In other words, programs are not permitted to group together expenditures from different ministries. The Netherlands, Australia, and the United Kingdom are cases in point—although, as discussed further below, some of these countries permit programs to cross agency boundaries within the same ministry.19

There are, by contrast, some countries which make selective use of interministerial programs. Sweden and France are examples of this approach, and concrete examples of such programs are given in Chapter 20. The argument for permitting programs which cross ministerial (or agency) boundaries is that this facilitates better central allocative choice, by recognizing that in important cases high-priority government objectives will be the shared responsibility of a number of different agencies. As a 2001 report to the New Zealand government put it:

Outputs produced by different agencies can constitute related services and having them funding from separate votés [that is, separate appropriations] makes reallocating resources across agencies but within related interventions more difficult…Reducing the number of votés by allowing more than one department and portfolio to draw money from a single vote should provide greater scope to enhance coordination between agencies, make resource allocation decisions across agencies, focus on intended outcomes, and reduce compliance costs. (Advisory Group on the Review of the Centre, 2001, p. 41)

The proponents of such interministerial or interagency programs see this as an important means of counteracting the tendency of agencies to pursue their own objectives with insufficient regard to the broader objectives of the government and the related activities of other agencies—what has sometimes been referred to as “silo” behavior. It has been something of a trend in public administration over the past decade to emphasize the need for mechanisms to build a strong “whole-of-government” or “cross-portfolio” perspective to public administration, by encouraging and compelling ministries and agencies to work collaboratively. (See Box 5.3 for a Swedish cross-ministry example.)

Box 5.3.Cross-ministry budget planning in Sweden

Sweden introduced its present form of program budgeting in the mid-1990s. The highest level of its program structure encompasses 27 expenditure areas which can overlap one or more ministry in the total of 13 ministries. If appropriations are planned for an expenditure area, relevant ministers will discuss the interministerial allocation jointly with the Ministry of Finance. Also during budget execution and evaluation of performance, line ministries need to coordinate their activities. At the ministry level, 48 programs (policy areas) are identified and below that sub-programs and activities (activity areas and activity branches). Parliament allocates at the expenditure area level, and then at the level of individual agencies, and for special investments, grants ands transfers. The lower levels of the program structure are not used for expenditure planning and control, but for managing outputs. In total Sweden has around 500 financial appropriation items spread out over 13 ministries. While most performance budgeting systems would have an explicit link between financial allocations and performance, that is not fully the case in Sweden. The Swedish authorities have indicated that they do not find the coordination requirements of this system overly burdensome.

It might be argued that the case for creating programs crossing ministry (or perhaps even agency) boundaries would disappear if the responsibilities of ministries were properly defined in the first place, so as not to overlap. A moment’s thought will, however, make it clear that this would be a simplistic position, and that there are often good reasons why services aimed at the same problem are spread over multiple agencies or ministries. For example, the Ministry of the Interior and the Ministry of Justice in many countries may have parallel aims to ensure the rule of law in society. Most countries have found it sensible, however, to separate law enforcement from the criminal justice system. Organizational structures need to take into account not only the intended outcomes of public services, but also a range of other factors including economies of scale and scope in the production of those services. In the example of the two ministries above, balance of powers and protection of civil rights would be seen as more important considerations than efficient allocation of resources. Also, advocates of cross-ministry programs argue that ignoring institutional boundaries has the advantage of putting the spotlight on existing societal problems and the organizational structures available to address them. Thus such an approach would encourage organizational restructuring on a more results-oriented basis where appropriate.

The main reason why cross-ministry programs exist in only a handful of countries is probably a view that the practical problems which arise are too great to make the exercise worthwhile. Agencies, and ministries as a whole, need clearly defined organizational budgets. The existence of pools of “shared” funds without clear ownership rights can be expected to be a source of interagency conflict and also to reduce accountability for the results achieved with those funds. This line of thinking leads to the view that it is wise to temper the principle of outcome-based program classification to reflect the realities of ministerial (and possibly also agency) organizational boundaries.

The need for agencies to have clear budgets suggests that if there are to be cross-ministry programs, it is necessary that the ministry shares of these program expenditures be clearly specified in advance. The most straightforward way of doing this is to specify this allocation at the sub-program level—such as is done in France where the budgets of both the missions (effectively programs) and programs (effectively sub-programs, which are confined to ministry boundaries) are centrally specified.

The same issues arise in respect to cross-agency programs within ministries. However, one means of facilitating the construction of programs across agency boundaries within the same ministry is for the center to focus primarily on funding ministries, rather than agencies. Under this approach, the center delegates to ministries a large measure of authority over the allocation of the ministry budget between agencies within the ministry. Such an option requires the development of significant ministry-wide budgeting capacity within the core finance department of the ministry.

Programs and the internal organizational structures of agencies

The same type of tension between the program classification principle and organizational structures arises within individual agencies. Here again, it is a mistake to believe that the problem can be resolved by basing organizational structure solely upon the program principle. This would involve defining the “divisional”20 structure within the agency solely upon product lines—or, to use language more appropriate to the public sector, upon groupings of services delivered to external parties with common outcomes. If the divisional structure were to be based solely upon programs, there could be no regional offices delivering more than one program but managed as a single unit for locational reasons. In addition, there could not be organizational units delivering support or corporate overhead services such as finance and human resource management—because support services are not products/outputs (see Chapter 3). It would be necessary instead for each internal product-based division to provide all of its own support services from separate units within the division. Although it is often the case that support/overhead services tend to be unduly centralized in public sector organizations (and indeed to become ends in themselves), there are good reasons why it does not make sense to totally decentralize all such support services.

It is therefore a mistake, inconsistent with the basic results-oriented principle of budgeting by programs, to assert that the program structure of each ministry should be made to fit its organizational structure.

Managing programs coherently across organizational boundaries

Managerial systems are not within the scope of this chapter. However, it is useful to make a few comments about management issues which arise as a result of inevitable differences between program structure and organizational structure, whether within or between agencies. One common solution to this tension is to operate two cross-cutting layers of management—one for management of the organizational unit, and one for management of the program. This is sometimes referred to as the principle of matrix management. A contemporary example of this is the creation under the new French reforms of the position of program manager (responsable du programme), which may in many cases require the person concerned to oversee the work of several agencies within the same ministry (see Chapter 20). Such an approach can clearly raise difficult questions about the power relationship of the program manager and the organizational manager—in particular, that of how to endow the program manager with enough real power to make the organizational manager take his or her authority seriously. One response to this question taken by some jurisdictions has been to endow the program manager with significant power over the use of the program budget—possibly even to the extent of creating something analogous to a “purchaser-provider” division. A very similar approach has sometimes been applied to the supply of support services. Thus, for example, in the US Department of Defense (and within certain other departments), many support services are provided on an internal “purchase” basis under what are known as “Working Capital Funds” (see Chapter 1). Whatever approach is taken, it is an important part of managing-for-results that public sector organizations reorient themselves toward external service delivery, rather than internal processes.


The program classification of expenditure is the most important financial management tool that countries have to get right when introducing program budgeting. A program classification should be based on a hierarchy of outcome- and output-oriented programs, sub-programs, and activities. A program classification will clarify the policy debate in cabinet and the legislature and should result in much improved expenditure allocation. Line ministries will also use the lower levels of the program classification to steer internal operations to more effective and efficient service delivery. The MoF should ensure that while line ministry strategic priorities are fully reflected in line ministry program structures, there is an adequate level of homogeneity in defining the program structure across line ministries. Line ministries should not only adhere to the specific guidelines on how programs and lower levels should be defined, but also take into account more practical guidelines on number of programs, relative size, and number of layers in the classification.

Countries should be encouraged to use the program classification as the main presentational format for expenditures in the budget, while the legislature should appropriate spending by programs, or larger parts of the program structure. In introducing a program structure, at least initially compromises should be accepted with regard to the use of administration programs, and keeping programs within line ministry boundaries. While a program structure provides much more flexibility to line ministries and their managers in changing inputs to best achieve desired outcomes and output, program structures should be used to control expenditures. A virement framework should define to what extent line ministries and the MoF are allowed to change allocation, and to what extent parliament has that prerogative. While cabinet and the legislative body would control expenditure allocation at a fairly high level, the lower levels of the program structure would be the instrument with which MoF and finance departments in line ministries monitor and control expenditure.


The authors would like to thank Philip Joyce for his valuable comments on a draft of this chapter.

Article 7(1), Loi Organique relative aux Lois de Finances (LOLF) no. 2001-692.

KRAs are part of the terminology often used in forms of strategic planning, but they have also been used as labels for groups of expenditure programs.

Terminology used in, respectively, the Australian, the British, the Canadian, the Swedish, and the Dutch budget systems.

Note, however, that it is not good practice to leave programs without a summary title in this manner. For most programs, it is both possible and desirable that the short title of the program should give an indication of its objectives.

See Chapter 3 for more detail on the concept of “activities” within the standard “logical framework.”

In the Netherlands program budgeting has been communicated to line ministries as the three “W’s” system: “What do we want to achieve? What are we going to do to achieve this? What is it going to cost?” The second, instrumental question on activities being carried out is one line ministries are familiar with even under input budgeting. This logical structure of questions allows program budgeting to provide an opportunity for line ministries to better align activities (and organizational structures) to objectives.

See Chapter 3 for the distinction between proximate and high-level outcomes.

More generally, it tends to make clear who or what the program directly targets in order to achieve its intended higher-level outcomes.

The high-level outcome of a strong economy is, for example, served not only by education, but also by macroeconomic policy, physical infrastructure provision, and a wide range of other government services.

This does not necessarily mean that this allocation can be changed without limit during the budget year, as this would undermine the stimulus to plan government expenditure in the pre-budget year conscientiously.

Consultants, or an independent third party, can play a useful role in arbitrating in these technical discussions.

Virement refers to the authority to change allocations of expenditure authorized by parliament or the executive during the budget year.

For this reason the development of the program classification in line ministries often starts with the proposal by line ministry policy departments and agencies for a multitude of programs, to safeguard their own interests.

Defining organizational entities as programs is usually not appropriate in the program budgeting framework, as we will discuss further on in this chapter.

From parliament’s perspective, the development of mono-program missions was also of particular concern. As discussed in Chapter 20, the LOLF represented an attempt by parliament to reassert a role in expenditure prioritization which it had for decades been denied. To this end, the LOLF asserted the power of parliament to reallocate funds between programs in the same mission (but not to change the allocations to missions, which would remain the prerogative of executive government). If, however, missions were to have only one program, parliament would remain powerless in respect to expenditure prioritization.

As indicated, in some countries the highest level of the program structure is equated with the program level, in others this level is that of key result areas or high-level programs. Further distinctions in sub-objectives and activities of line ministry expenditure should be made at lower levels of the program classification so as not to overload the strategic allocation process at the center of government.

In some countries, notably the United States and China, the budget appropriation is both a maximum and a minimum. This implies that appropriations cannot be cut by the executive without supplementary budget approval of the legislature. Exceptions to this can, in principle, only be made if program execution has been delayed for reasons outside the control of the executive.

An “agency” is defined here as an organizational unit which has day-to-day management autonomy (so that, for example, its chief executive is not subject to direction and supervision by the chief executive of another agency). A “ministry” is defined here as a group of one or more agencies under the responsibility of a single minister. In many systems, each ministry will have a core department, and a number of other non-core agencies. The core department will tend to have principal policy responsibility for the issues covered by the ministry, and may have some ministry-wide coordination or control function (including usually over the approval of the budgets of the non-core agencies, that is, the core ministry often plays the role of finance ministry of its sector).

Using the term to refer to the highest level of organizational unit within the agency.


    Advisory Group on the Review of the Centre2001Report of the Advisory Group on the Review of the Centre Presented to the Ministers of State Services and FinanceNew ZealandNovember.

    Commission des Finance2004Note D’etape sur La Mise En (Euvre De La Re forme OrganiqueJanuary152004<> (accessedDecember102004).

    LienartI.2004The Legal Framework for Budget Systems: An International Comparison (Paris: OECD).

    Other Resources Citing This Publication