A full understanding of the budget planning and preparation system is essential, not just to derive expenditure projections but to be able to advise policymakers on the feasibility and desirability of specific budget proposals, from a macroeconomic or microeconomic perspective. It is much easier to control government expenditures at the "upstream" point of budget preparation than later during the execution of the budget.

Thus, fiscal economists and general budget advisors need to know:

• what is the framework in which budget decisions are made;

• who is responsible for planning and preparing the budget;

• what are the basic steps;

• what are the typical weaknesses in procedures and how can these be overcome; and

• how can changes in budget plans be programmed and targeted?

Budget planning and preparation are (or should be) at the heart of good public expenditure management. To be fully effective, public expenditure management systems require four forms of fiscal and financial discipline:

1. control of aggregate expenditure to ensure affordability; that is, consistency with the macroeconomic constraints;

2. effective means for achieving a resource allocation that reflects expenditure policy priorities;

3. efficient delivery of public services (productive efficiency); and

4. minimization of the financial costs of budgetary management (i.e., efficient budget execution and cash and debt management practices).

Budget preparation is a process with designated organizations and individuals having defined responsibilities that must be carried out within a given timetable (see Figure 1 in Section 1 for a typical time line). This process is normally established and controlled by a legal and regulatory framework. While generally sharing broadly common procedures, budget preparation (and execution) systems do exhibit differences depending on their historic origin. Given the common heritage of many countries, it is possible to identify four main patterns--francophone, Latin American, (British) Commonwealth, and transition economies.

To understand the budget preparation process in a given country, it is important to:

• assess the basic soundness by judging the budget preparation system against certain internationally accepted standards or "budget principles";

• know where to find the rules governing the budget preparation process; and

• from those rules, identify who has the responsibility for what elements of the budget preparation process.

Based on the objective macroeconomic assessment of available revenues and financing, ideally, the expenditure budget should aim to be comprehensive, transparent, realistic, policy-oriented, and allow for clear accountability in budget execution. These concepts form a standard by which the soundness of budget systems can be judged (see Box 1).

Annuality. A budget is prepared every year, covering only one year; voted every year; and executed over one year. While maintaining the core concept of annual authorization, this principle has been modified at the preparation stage, such that most OECD countries now develop the annual budget within a multiyear perspective, through the preparation of medium-term revenue and expenditure frameworks. A very few are moving toward determining budget appropriations for more than one year at a time.

Unity. Revenue and expenditure (as well as borrowing constraints) should be considered together to determine annual budget targets. The budget should cover all government agencies and other institutions undertaking government operations, so that the budget presents a consolidated picture of these operations and is voted on, as a whole, in the parliament.

Universality. All resources should be directed to a common pool or fund, to be allocated and used for expenditures according to the current priorities of the government. In general, earmarking of resources for specific purposes is thus to be discouraged; but the case of extrabudgetary funds is considered in more detail below.

These three characteristics are essential to ensure that, in budget preparation, all policy proposals for undertaking government expenditure will be forced to compete for resources, and that priorities will be established across the whole range of government operations.

They are usually considered a prerequisite to meeting the first two of the four main goals of effective public expenditure management noted at the beginning of this Section: exercising the macroeconomic constraint of affordability on the total, and ensuring efficiency in the allocation of resources. These characteristics are typically enshrined in a legal and administrative framework regulating the budget process.

Knowing the rules

Although the precise legal framework for central government budgeting varies from country to country, it is usually set out at several levels.

The constitution is the highest in the legal hierarchy. Although it deals only with broad principles, the constitution may clarify three important aspects: (1) the relative powers of the executive and legislative branches with respect to public finances; (2) the definition of the financial relations between national and subnational levels of government; and (3) the requirement, for example, in Commonwealth systems, that all public funds be paid into designated accounts, and that these funds be spent only under the authority of a law.

The organic law is usually the main vehicle for establishing principles of public financial management. These laws may take the form of a single law that guides budget preparation, approval, execution, control, and auditing (loi organique relative au budget in the francophone system; ley de administración financiera in the Latin American system), or there may be several general laws covering specific areas of public finance management (e.g., under Commonwealth systems) that may also relate to subnational levels of government. They are called "organic" because they relate to organizational matters and systems, and do not therefore require annual reenactment. Moreover, they can often be modified only under certain conditions, such as qualified parliamentary majority.

Financial regulations. The organic budget law also gives to the government, or the minister responsible for public finance, the authority to issue detailed regulations and instructions (for instance décret portant réglement de la Comptabilité Publique in the francophone system, and decreto para la contabilidad pública in the Latin American system). These are often quite detailed.

The constitution, the budget organic law, and financial regulations are permanent and form the legal framework within which the annual budget law, which includes the revenue and expenditure estimates for a given year, is prepared, approved, executed, and audited. The annual budget law can take different shapes depending on the system.

In the francophone and Latin American systems, the coverage of the annual budget law (called budget or loi de finances in francophone countries and ley anual de presupuestos in Latin America) is rather wide, since it contains the amount and details of revenue and expenditure, the balance, and also any new tax legislation measures and some changes to spending. Under the Commonwealth system, both revenue and expenditure estimates are presented. Often the latter are further divided into recurrent and development estimates, sometimes presented as separate volumes. Typically, the presentation is detailed by institution and line item. By contrast, the annual budget in many transition economies has often been rather summary in format: prior to any recent reforms, budget estimates were presented by budgetary institution--typically only the major supervisory institutions and not their subordinate units--and broken down only by broad "functions," more or less the sectors used in the previous central planning framework.

Identifying the responsibilities within the budget system

The powers assigned to the legislative and executive branches, and, within the executive branch, who does what, essentially define the responsibilities for preparing the budget (Box 2).

The responsibility for preparing the budget usually lies with the ministry of finance with input from the line ministries and some smaller spending agencies. This exercise is normally controlled by a central budget department located in the ministry of finance, or sometimes in a separate budget ministry.

The character of central budget departments differs widely between countries, however. Some are only responsible for preparing the current budget, excluding debt. In such cases, the capital budget may be prepared by a planning or development ministry (or even at a higher level in the prime minister's or president's office), while the debt service costs are assessed (and paid) by another entity. Some budget departments are in charge of preparing the entire budget, although not involved in implementation of the budget. Others have a say on expenditure commitments, and some are also in charge of monitoring budget execution. It is therefore important to know the precise responsibilities of the budget department. It is particularly useful to know if the budget department is responsible for supplying partial or complete data on budget preparation, expenditure commitments, and full budget execution data.

In many developing countries, only partial data on budget preparation may be available in the budget department. It is important that all data on the current budget, the capital budget, and the debt service (including data on secondary and tertiary tiers of government) are consolidated to ensure that, in total, they are consistent with macro objectives. In some countries, research departments of the central bank may carry out this task.

In principle, the basic steps in a standard budget preparation system comprise the following:

1. The first step in budget preparation should be the determination of a macroeconomic framework for the budget year (and ideally at least the next two years). The macroeconomic projections, prepared by a macroeconomic unit in the ministry of finance or elsewhere, should be agreed with the minister of finance. This allows the budget department within the ministry of finance to determine the global level of expenditure that can be afforded without adverse macroeconomic implications, given expected revenues and the level of deficit that can be safely financed. In a few countries, there are fiscal rules in place that may limit total spending or recurrent spending (e.g., the "golden rule").¹⁶

2. The second step should be the allocation of this global total among line ministries, leaving room for reserves (a separate planning and a contingency reserve as explained below) to be managed by the ministry of finance.

3. The next step should be for the budget department to prepare a budget circular to give instructions to line ministries, with the indicative aggregate spending ceiling for each ministry, on how to prepare their estimates in a way that will be consistent with macro objectives. This circular will include information on the economic assumptions to be adopted on wage levels, the exchange rate and price levels (and preferably differentiated price levels for different economic categories of goods and services).

4. Step four is the submission of bids by line ministries to the budget department. Once received there needs to be an effective "challenge" capacity within the budget department to test the costing of existing and any new policy proposals.

5. The next step comprises the negotiations, usually at official and then bilateral or collective ministerial level, leading finally to agreement.

6. Finally, step six is Cabinet endorsement of the proposals for inclusion in the budget that will go to parliament.

Early in the preparation stage, that is before the budget circular is issued, those advising on the preparation of the budget should ask:

• Is the budget based on an aggregate level of general or central government expenditure, in cash terms, that is consistent with the macro framework, and any fiscal rules in place?

• Does the budget circular to the line ministries provide adequate guidance on preparing budget estimates? Does it include a guideline or limit for each line ministry on this total spending?

• Are there suitable reserves? Ideally, within the aggregate total there should be a planning reserve (not allocated in guidelines given to each line ministry), so the ministry of finance can assign extra resources later during budget negotiations for the most urgent priorities, without breaching the macroeconomic constraint. Moreover, after all final line ministry allocations have been made, there should still be a contingency reserve within the aggregate that will be held and administered by the ministry of finance to meet genuine contingency spending during the budget year.

There are often weaknesses in budget preparation systems: their nature, scale, and significance need to be understood, both to assess the value of the data produced and, where there are separate projections to be made by an IMF team or other external advisers, to accommodate such weaknesses. Eight common problem areas can be identified:

1. The central government budget is not really unified. It is a dual-budget system with separate recurrent and capital or "development" budgets that may be based on inconsistent macroeconomic assumptions, budget classifications, or accounting rules. Each budget may be compiled by a different ministry--for example, the ministry of finance for recurrent expenditures and a planning ministry for capital or "development" expenditures.¹⁷

2. The macroeconomic constraint is not explicitly taken into account in the budget process, or the economic assumptions underlying the estimated costs of expenditure programs are weak or erroneous.

3. Projections for the outturn of the previous and current years' budgets are not prepared, or the experience to date is not analyzed, so that budget preparation becomes a simple incremental exercise based on the previous year's (often erroneous) budget estimates.

4. Satisfactory procedures do not exist for review of expenditure policies and program prioritization.

5. There is no multiyear planning.

6. Extrabudgetary funds are used to divert spending to one or more "off-budget" accounts.

7. Quasi-fiscal expenditures, contingent liabilities, etc., are not taken into account.

8. Appropriations-in-aid are used inappropriately.

Table 1 provides a summary of certain weaknesses and some of their implications. The next subsection deals with the individual issues in more detail.

Is the central government's budget really unified?

While the budget document presented to the legislature may appear to be a unified one, in reality the current budget and the capital budget are often prepared following different procedures. In such cases, difficulties can be encountered in meeting macro objectives where the two budgets are prepared without full coordination, or on different economic assumptions. For example, in many developing countries the development budget or Public Investment Plan/Program (PIP) may include a combination of capital and current programs. Such a system can also lead to an inefficient use of funds because, for example, the same item of expenditure may be included in the two budgets, or, more typically, investment projects may be included in the budget, without providing for the necessary corresponding current expenditure. The supposed superior status of items included in the development budget may also tend to squeeze out current expenditures within the affordable total.

Information on planned capital expenditures may be partial, where donor-financed expenditure is significant and coordination with the donors is inadequate. It is important to check the extent to which the budget is unified in the above sense of ensuring the internal consistency of different components. Quite apart from checking whether the economic assumptions are common and consistent (see below) however, it is also essential to ascertain whether there has been policy agreement (e.g., on start dates for new policies, on levels of staffing for new development projects when completed, or whether the ministry of finance has ensured that the recurrent cost implications of capital spending in future years have been taken into account). If there is inconsistency, the coordination between the two budgets should be strengthened by whatever means available. A meeting with key donors may also be necessary.

Is the macroeconomic constraint explicitly taken into account?
Are the economic assumptions underlying the budget accurate
and consistent?

In some countries the budget is prepared with surprisingly little reference to the macroeconomic prognosis. Often, there is little macroeconomic analytical capacity in the government, or the budget department has no contact with those undertaking such analysis (e.g., a research department at the central bank). The absence of proper macroeconomic analysis is particularly common in countries that have a "dual-budget" system, that is, separate development and recurrent budgets as described above.

With inadequate macroeconomic analysis, there can be insufficient discipline to limit the size of the sustainable budget deficit at the beginning of the budget process. As a consequence, the budget preparation procedure can be principally driven by the requests from the ministries for increased spending (i.e., the bottom-up approach). Without a firm top-down limit, the ministry of finance can only challenge proposals on technical or policy grounds, rather than in terms of affordability constraints and priorities within a fixed total. There will be a higher probability that the deficit obtained through this procedure will not be sustainable. Fiscal adjustment will be easier if the macroeconomic constraint and the acceptable deficit is defined first (i.e., a top-down approach). From this, spending departments can be given some guidelines to limit their requests.

However, even if a macro constraint on aggregate expenditure is set, the fiscal economist needs to probe their validity. Since many countries have proven to be perennially optimistic in revenue forecasting, realistic revenue projections and the financeable fiscal deficit must be decided before the budget preparation procedure begins, not at some late stage just before or, worst of all, after, its completion. (In the worst examples, the revenue forecast can become a residual derived from line ministries' aggregated spending plans less external financing and "acceptable" domestic borrowing.) Those preparing the budget need to ensure that the budget preparation timetable is sufficiently long, and the process transparent and comprehensive, so that there is no need for arbitrary expenditure cuts late in the process, when revenue or borrowing constraints become clear.

Another source of weakness is that the economic assumptions to be used in estimating the cost of present and new policies may not be accurate, consistent across line ministries, or sufficiently discriminatory between different economic categories of expenditure. For example, a sharp fall in the exchange rate will have a much different impact on the cost of health programs (because of the import of medicines) than on the costs of servicing domestic debt. Poor unit cost estimates are one of the most common weaknesses in budget preparation. Fiscal economists need to urge the budget department to specify by category different price factors before budget estimates are prepared. The higher and more volatile the inflation rate, the greater the need to differentiate by category of expenditure.

Are recent budget execution figures known and analyzed?

The budget department--and others involved in budget preparation, such as the planning ministry--are often unaware of the provisional outturn for the last completed financial year, or the projected outturn for the current financial year, because the budget is executed by a separate treasury department, rather than by the budget department. Budget preparation for year t + 1 begins early in the current fiscal year (t) before the provisional outturn for the previous year (t ­ 1) is known, and usually before any projected outturn for the current year has been made available, with the consequence that the budget department/planning ministry prepares the budget by reference to the previous and current years' initial budgets, and not to the provisional or projected budget outturn for the current and preceding years.

If there is economic instability--for example, in times of high inflation--the budget preparation exercise can become seriously unrealistic. Uncertainty about likely price levels can also "excuse" and thereby perpetuate a lax attitude to budget preparation: when the budget is subsequently executed, the results may include wasted administrative efforts spent switching resources from one budget line to another (virement); excessive use of supplementary appropriations; loss of macroeconomic control over the total; poor allocation of resources among programs; and expenditure arrears.

At the preparation stage of the budget, when discussing the budget figures, in addition to the budget department and any planning ministry, the treasury (or budget execution department) should be fully involved. In particular, the treasury department should provide estimates of spending in the previous year and the spending to date in the current year (both in general and on specific programs or economic categories), as well as its forecast of the likely outturn for the current year. The best basis for forecasting expenditure on a given policy is usually the estimated cost of that policy for the most recent year available.

Do procedures exist for resource prioritization?

An efficient budget preparation procedure should aim at making the government's priorities clear and at selecting, from the many budget requests by spending ministries, those which are really important to the government. In principle that requires two elements. First, a budget strategy needs to be determined at a political (typically cabinet) level, which determines (1) the affordable total, (2) new policies to be accommodated, and (3) any changes (often reductions) in existing policy provision. Second, each spending ministry and the budget department/planning ministry should meet to discuss each ministry's estimates. To accommodate new policies, the budget department/planning ministry must require each spending ministry to prioritize its requests.

But this ideal is rarely matched by the practices in many countries. Quite apart from weaknesses in the institutional arrangements, decisions on priorities at the budget preparation stage can be wholly artificial because (1) subsequent cash allocations or supplementaries will render them redundant; (2) amounts given by line item are deliberately loose or unclear, in anticipation of a real allocation during budget execution; and/or (3) in practice, the priorities are set outside the formal budget framework, for example, by the president's office.

Ultimately, the allocation of resources across spending programs is a political decision, although those preparing the budget will need to advise on what is realistically achievable. For this, economic analysis should play an important role. For example, ministries need to have as much information as possible on expenditure policies and programs, on costs, and, ideally, on their outputs and outcomes.¹⁸

Whenever possible, however, the cost of all new policies that a line ministry wishes to pursue should be estimated separately from the estimates of the costs of ongoing policies. As supporting information, the spending ministry should provide data on expected results/performance from such new policies and incremental spending (ideally, outputs and outcomes) and preferably in a format that enables the requests across ministries to be compared. The ministry of finance should have a role in reviewing, and commenting on, such cost estimates. The data should be presented with enough detail to allow the budget department to judge the reasonableness of the budget request, the activities the request is intended to support, and the corresponding staffing levels.

Such systems are most advanced in a small number of industrial countries: even there, the practices (and results) are not wholly in line with the above principles. Real political agendas are sometimes nontransparent or inadequately articulated; the economic value of marginal expenditures across functions cannot be properly compared; and the measurement of policy outcomes, and their links to individual programs, has proved quite difficult in practice. Yet, considerable progress has been made, particularly on measuring output and on requiring better assessments of the results of new policies or programs proposed before they can be incorporated in the budget. Developments in this direction are to be encouraged, and there are some useful short cuts.

As noted earlier, to facilitate discussion on resource allocation, it is helpful for the budget department to set, within the macroeconomic total, guidelines/targets for each spending ministry on their total spending, when the budget circular is issued. In addition to targets by line ministry, an allowance should be made within the affordable total for suitable planning and contingency reserves (see below). This allows budget negotiations to coalesce around a realistic target for each ministry, consistent with the affordable macroeconomic total.

Such guidelines or targets can be normative (e.g., when they are derived from a medium-term expenditure planning framework; see below) or purely indicative (e.g., based on shares in the latest year's outturn figures).

Each line ministry/spending agency can be asked to put forward its estimates for its existing or baseline policies within that guideline. (This should automatically be the basis of the data when the figures are derived from a medium-term framework.) Separately, each ministry should be asked to identify what policies and programs would be enhanced/introduced or cut back, if their allocation were 5 or 10 percent above/below the guidelines. While such an approach can be abused (by line ministries offering only politically unacceptable items for reductions), with experience, and with a well-informed challenge capacity within the ministry of finance that identifies lower-priority items in advance, it can help to concentrate discussion on priorities at the margin, within an affordable total.

A planning reserve is a sum (usually one or two percent of total expenditure) not allocated in the guidelines, which the ministry of finance later plans to allocate to new programs, if necessary above the guidelines during budget negotiations. A contingency reserve is a reserve for in-year expenditures above appropriations for handling genuine contingencies; it should be modest in size (if too large, a bidding process from ministries may quickly set in) and thus it is unlikely it should exceed 2 or 3 percent of total expenditures.¹⁹ It should be under the control of the ministry of finance, and access should be granted by the ministry of finance only under stringent conditions.

Where priorities are not being clearly established during the budget preparation process, the budget department/planning ministry can establish benchmarks using these mechanisms and thus set the basis for a discussion by policymakers of the priorities among the requests.

Is there any multiyear planning?

Focusing on the current or next fiscal year's expenditures alone can be misleading. Expenditure planning should be extended beyond one year, not least to gain a full appreciation of the future spending implications of present policy decisions. Nowhere is this more important than on the recurrent costs of capital spending. For countries with multiyear PIPs, such plans need to be reintegrated with recurrent expenditures and into a multiyear expenditure plan that provides the basis for establishing a realistic global budget. Although the introduction of a regular procedure of medium-term planning frameworks by function, by ministry, and (ideally) by program takes time to develop, those analyzing and preparing the budget should begin this process by preparing medium-term fiscal scenarios.

There are several variants of such a planning framework. The simplest has only aggregate projections for public spending for the two or three succeeding years beyond the budget year. A second has "illustrative" figures by line ministries--sometimes on a mechanistic basis (e.g., shares of a global total are assumed to be held constant to the proportions in the budget year). A third is normative in that it projects costs of existing and any new policies agreed for introduction over the medium term, but these medium-term figures play no role in subsequent-year budget negotiations. The best approach uses these figures for the past budget year as the starting guideline for the next year's budget negotiations.

Is there a legitimate need for extrabudgetary funds?

Extrabudgetary funds (as defined in the GFS manual) generally refer to accounts of government transactions that are not included in budget totals or documents and typically do not operate through normal budgetary execution procedures. Such transactions may, for example, be financed through foreign aid or earmarked revenues not included in the budget.

Unfortunately, extrabudgetary funds are often set up for inappropriate reasons, not consistent with principles of good governance. For instance, they may be designed to allow the president or some parts of the executive branch to bypass the normal budget procedures (for example, the comptes spéciaux in the francophone system). In this case, the fiscal economist should aim to identify all such funds and then ensure that they are consolidated on a gross basis in fiscal tables. This may be difficult where expenditures from these accounts cover security or presidential spending, which can be considered highly sensitive issues. When consolidated, however, and when the political authorities can be persuaded to consider them as a legitimate component of the published budget, at some point those preparing the budget may be able to close these accounts or at least to reduce their number. The affected expenditures should then follow regular budgetary procedures and appear in the relevant heading in the consolidated budget.

Another reason to create this kind of account may be to earmark revenue for a particular purpose. In this case, a specific kind of revenue is transferred to this account when collected, and whatever funds are available must be spent on a given item. While there are advantages and disadvantages in operating such funds in many countries, in many cases the disadvantages far outweigh the advantages (see Box 3 on the pros and cons of extrabudgetary funds). In the worst instances, new extrabudgetary funds may be established specifically to divert expenditures out of the budget, sometimes with the aim of publishing a lower fiscal deficit. The practice of opening such accounts is often an indication that the budget process is not functioning properly, and that resources for priority tasks must be allocated through other mechanisms. Unfortunately, this practice gives rise to rigidities in the short and long term. In the short term, financial management will be impaired because resources transferred to a special account are typically not available to the treasury for cash management purposes--for example, to relieve short-term cash shortages (see Section 5). In the medium term, a shift in government priorities may be impeded by the fact that a part of the available resources is set aside for a special task.

Some operations of a fiscal nature are not conducted through the budget. Examples of such quasi-fiscal expenditures include interest subsidies paid by the central bank on loans to public enterprises, and special support operations for banks and public or private sector enterprises administered through the banking system. Quasi-fiscal expenditures also include spending by nonfinancial public enterprises that represents the provision (or subsidization) of public goods (e.g., schools or hospitals). By definition, such expenditures do not pass through the budget and cannot be easily consolidated with the statement of general government operations.

In general, it is difficult to extract information on, let alone estimate the cost of, quasi-fiscal activities so as to consolidate such data in the general government tables. But, to gain an overall assessment of the fiscal stance, it may be necessary to assess the size of such operations and to notionally add the figures to the information on general government operations. In addition, those preparing the budget should take every opportunity to persuade policymakers to transform such nontransparent activities into explicit subsidies, transfers, etc., to the extent they should continue at all, within the budget.

Governments also have, at any point in time, certain contingent liabilities. The most common is the existence of explicit government guarantees, usually on bank lending to industry or lower tiers of government, which can fall due. But there are other forms of implicit contingent liabilities: for example, there may be a challenge in the courts to the government interpretation of a law that, if the judicial decision goes against the government, will have expenditure implications.

In general, countries should be urged to ensure that a careful record of all such explicit contingent liabilities is maintained (while recognizing that there will always be some uncertainty on aspects like judicial decisions as well as moral suasion pressures on "implicit" government guarantees) and to make prudent allowance for such guarantees being "called" (i.e., payments being due) or for adverse judicial decisions, by ensuring that there are sufficient resources in the contingency reserve to meet such expenditures. Of course this will always be a difficult judgment; in some years the reserve may be more than adequate--in which case the unused balance can be used to improve the fiscal position relative to the budget. In other years, some excess, even after the contingency reserve, may arise and should be met transparently through supplementary estimates--see Section 4. Those preparing the budget should ensure that some estimate of expenditures from both explicit and implicit contingent liabilities is allowed for in budget preparation.

How should appropriations-in-aid be handled?

Many countries have spending agencies that are able to finance a large part of their activities from their own sources of revenue--normally fees and charges. An example might be a dedicated passport office that charges for the issue of passports but receives budgetary resources for its capital expenditures. These budgetary resources are often termed appropriations-in-aid, or sometimes net appropriations--that is, the amount sufficient to meet the gross service costs, after an assumed contribution from the fees and charges they raise.

There are three issues in this regard. First, irrespective of how far the costs of the service--for example, the issue of passports--are financed from earmarked charges rather than from general budgetary resources, the activity is essentially within the government sector. Thus, in terms of measuring the size of government, the appropriations-in-aid data are insufficient. The gross expenditures or gross costs of the service need to be identified, as well as how much is financed from own fees and charges, and how much from general budgetary resources. Second, though it is essentially a budget execution issue, there are often cases where the fees are paid into a separate bank account held by the relevant spending agency in a commercial bank. As explained in the next Section, this is generally poor budgetary practice, which can lead to abuse with the monies being diverted into other areas of expenditure. Third, in budget preparation, it is often necessary to be aware of deliberate underestimation of the likely revenues from fees and charges, so as to maximize the contribution from general budgetary resources. In particular the ministry of finance needs to insist on the annual updating of fees and charges to allow for inflation--quite apart from any separate expenditure policy issues about how much of the service cost should be met by users and how much by the general taxpayer.

Whatever the weakness of the budget preparation system itself, the fiscal economist or general policy advisor may be called upon to advise on options for changing expenditure plans (typically, but not always, for reductions in spending). In the past, fiscal adjustment through reductions in planned expenditures has often proved problematic. Changes in expenditure plans, relative to the authorities' original intent, have been implemented in ways that were disruptive to budget execution or were unsustainable in the long run. Where expenditure reductions have been undertaken, they have sometimes produced short-run savings at long-run cost--for example, by cutting needed capital expenditure or by so severely contracting maintenance expenditure that the capital stock was partially consumed. Where planned expenditure reductions have failed (in the sense that outturn expenditure was above the revised budget), they have typically led to payment arrears, and/or to excess spending above appropriations. This has damaged both the private sector economy (its bills are unpaid) and the credibility of the government in financial markets.

A fundamental problem is that changes in the budget are often proposed at too late a stage in budget preparation. Yet, whatever the time constraints, proper evaluation of expenditure policy options is vital. Those preparing the budget may be tempted to grasp quick solutions. However, budgets must represent an objective estimate of the costs of stated and agreed (within government) expenditure policies. Correspondingly, the only sustained (and sustainable) changes in expenditure plans are those rooted in changed expenditure policies.

Thus, expenditure reductions planned under a revised annual budget are not likely to be successful where:

1. they are made in appropriations without accompanying changes in the underlying expenditure policies. Just changing the estimates makes the budget provision less than objective; the likely consequence is overspending against appropriation and/or the emergence of payment arrears;

2. estimates for open-ended, demand-led programs are revised downward; again this is typically the triumph of hope over past experience;

3. inconsistent agreements are made between the ministry of finance and several line ministries to reduce the budget provision for certain line items, but with a "nod and wink" that access will be granted in-year to the contingency reserve; this reserve tends to become overcommitted when real contingencies arise;

4. "revised" economic assumptions on the exchange rate or inflation rate are invoked as justifying lower provision;

5. overoptimistic assumptions are made on "efficiency savings" through reductions in the number of civil servants and cuts in equipment purchases, utility charges, or fuel bills; and

6. reductions are made in transfers to lower-tier governments--this just passes on the problem.

1. they are imposed by the ministry of finance through cutbacks in planned appropriations (often at a late stage in the budget preparation process) without the concurrence, or over the heads of, line ministries;

2. the appropriations are not themselves changed but rather the ministry of finance undertakes to control total spending within the appropriated sum--for example, through controls in-year on monthly cash allocations to line ministries; and

3. they are to be accomplished by creative accounting measures--greater use of suspense accounts, the establishment of new or additional extrabudgetary funds, etc. (such transactions should, in any case, be consolidated within fiscal tables).

In terms of expenditure policies, the important questions include:

• Are the proposed expenditure policies soundly based? (Guidance here is contained in the IMF's Expenditure Policy Handbook).

• Do the proposals fit in with existing established policy priorities as laid out in any published medium term strategies of the government? Will they be rejected by parliament?

• Are the cost estimates for the new proposals accurate? Will the proposals achieve the projected adjustment? Is there an important quantitative difference between their immediate and longer-term costs (e.g., do they just have a short-term benefit rather than representing some fundamental fiscal adjustment)?

• How would they be enforced (through revised appropriations or cash allocations)?

While there are no hard rules about how planned public expenditure can best be adjusted, experience suggests some guidelines. Three broad approaches can be reviewed: (1) changes by program and policy; (2) changes by individual ministry; and (3) changes by economic category.

1. Changes to budget plans by policy or program are the optimal (though not always achievable) approach. Governments should use--or develop--mechanisms for identifying the most and least efficient and effective expenditure policies and programs, and target expenditure changes accordingly. (In this context, a number of more advanced countries are moving toward output-oriented budgeting.) In practice, country programs agreed with the IMF and the World Bank may include commitments for increases in expenditure in, say, health and education, together with reductions in unproductive expenditures. Outside such agreed priority (or nonpriority) areas, the ministry of finance should, in principle, assess the costs and benefits of alternative policy packages. In many cases, however, it will not be possible to review individual functions or policies, even in cases where good expenditure classification exists. Time pressures will often force consideration of other approaches.

2. Changes in expenditure plans by an individual ministry may be considered, for example, where there is a lack of information by economic category (see next item). Such an approach can be helpful in supporting or expanding initiatives in areas like health and education (albeit on a ministry rather than sectoral basis). Reductions, where needed, can be targeted elsewhere; for example, where one or more line ministries or spending agencies has a record of poor expenditure control or in support of a policy decision that affects only a few ministries.

   A common variant of this approach is "across-the-board" reductions by ministries, in response to a call for lower than planned expenditures. By allowing each ministry to decide how to cut a fixed percentage off its expenditure plans, it often seems attractive and broadly equitable. But there are many drawbacks to such an approach. Despite the apparent fairness, in reality across-the-board reductions avoid consideration of priorities and leave individual ministries to allocate among line items, with not only an uncertain economic and social impact, but also potential damage to the efficient delivery of services. Such reductions also may all too often be seen as temporary, so line ministries apply them in areas that allow payment arrears to build up (e.g., payments to utility companies). Across-the-board reductions should be avoided, therefore, with preference given to adjustments by economic category (if changes by specific policy or program are not achievable).

3. Changes in expenditure plans by economic category may have to be made where budgetary pressures emerge at a late stage in budget preparation. Again, they have the appeal of representing rough justice (e.g., if all ministries are asked to reduce their wage bill by a fixed percentage)--even though they do not imply proportionally equal aggregate changes by ministry. Adjustments based on this economic classification enable some economic analyses of expenditure patterns and prescription. Moreover, they can be targeted at wider expenditure policy objectives, such as reducing the wage bill or the number of civil servants, reining in travel costs, or cutting back generalized price subsidies to consumers or subsidies to industry. But they also have a downside. Often, they do not encompass any judgment about priorities between programs. Also, some of the measures applied tend to be simplistic (because they are "last-minute"), such as wage standstills or freezes, or percentage cuts in purchases of supplies. They are thus necessarily a blunt instrument, best seen as interim in nature, pending a deeper review of policy options. They may have short-term benefits but long-term costs--for example, increasing the financial cost of completing a capital project and postponing the benefits. Again, they may be seen as temporary, rather than representing a structural fiscal adjustment.

• make any revisions to emerging budget plans as soon as possible (last-minute changes tend to be ineffective);

• seek, as a starting point, expenditure changes that are in line with previously agreed decisions or views on expenditure policy priorities--this is especially important where there is room for additional spending;

• be sure that cost estimates for new expenditure proposals are realistic and accurate, not just for the year ahead but over the medium term, and that the proposals can be implemented at the political level;

• be wary of the individual ministry or agency approach, except where this is consistent with pre-agreed policy priorities or to address glaring past failures to exercise proper control;

• avoid "across-the-board" cuts;

• where expenditure plans need to be scaled back, use reductions by economic category if fundamental policy changes cannot be achieved. The first target should be any reductions consistent with the pursuit of outstanding policy goals, and ideally within the context of ongoing wider reforms--for example, measures to reduce civil service numbers or changes in wage policies to improve the alignment of public and private sector wages; and

• be cautious in reaching for the obvious but overly simplistic targets, like freezes in new or ongoing public sector capital projects or in public sector wages; or percentage reductions in the purchase of goods and services (unless and until the longer-term damage to the economy or to overall government operations is assessed as bearable).