Before considering how the expenditure side of the government's budget is planned, prepared, and executed, it is necessary first to clarify the coverage and sources of data on public spending. This brief section discusses three critical, if basic, questions: What is the appropriate definition of government expenditure from a macroeconomic perspective? What are the data sources for public expenditure aggregates? How can expenditure projections for a short-term perspective best be prepared?

When considering the macroeconomic impact of government expenditures, fiscal transparency demands that the economist work with a very broad definition of the government sector.⁶ While several different aggregates may be referred to in IMF programs, "General Government Fiscal Operations" (GGFO), or general government for short, is usually identified as best suited to the macroeconomic perspective. This aggregate covers the activities of all levels of government (central and state level), and the quasi-fiscal operations of nongovernment entities.⁷ General government should thus reflect the overall magnitude of government operations, the aggregate burden of taxation, the allocation of public resources, and the size of the government borrowing requirement. This last item is, of course, crucial to IMF work since the financing of the government deficit plays a critical role in financial programming.

Tables on past and projected general government expenditures can normally be prepared by consolidating inputs from a series of spreadsheets, covering at a minimum the central government, aggregate state and local governments, and social security expenditures within any separate fund. In some cases, however, other important government transactions (besides social security) take place through special or "extrabudgetary" funds (e.g., road funds or health funds). In others there are quasi-fiscal operations undertaken through the banking system (e.g., subsidized loans to state-owned enterprises). Those preparing general government tables should, ideally, try to identify all such transactions, quantify their magnitude, and include them within the consolidated total.

Some countries--but by no means all--already provide a good consolidation of their data. But it is often desirable to check that the data conform to appropriate norms for coverage, as well as acceptable standards of accuracy and consistency. In practice, problems are often encountered in ensuring the data are sufficiently accurate and comprehensive, including the following:

1. Some extrabudgetary funds are set up by a special law and may follow different accounting rules, classification systems, or even different fiscal years. A good understanding of these differences is necessary before consolidation can be attempted.

2. It is also not unusual to encounter government transactions that are reported on a net rather than gross basis--for example, by spending agencies;⁸ this practice creates a transparency problem by hiding the scale of government operations and their impact on the economy. These transactions should be unraveled and, wherever possible, the data prepared on a gross basis.

3. Some central government expenditures take the form of transfers to local governments. When consolidating general government spending, it is necessary to avoid double-counting such transfers.

4. Some developing countries still maintain "dual budgeting," that is, separate recurrent and development budgets. While this is now generally viewed as undesirable because it risks both inadequate macroeconomic control and poor resource allocation, it is often a given in the short term. Those preparing a general government budget need to consolidate the two carefully, keeping a watch for inconsistencies (e.g., timing, economic assumptions, etc.).

5. In some instances, debt servicing is handled separately--for example, by the central or a commercial bank. Data have to be obtained from the bank(s) and consolidated with the information on other central government expenditures.

6. The misclassification and/or exclusion of expenditures is a frequent problem. There is often a confusion between expenditure and financing transactions. A common example is the incorrect inclusion of loan amortization as expenditure. Net lending operations by government are often difficult to track down and are sometimes "off-budget";⁹ but they should be consolidated in the government accounts. Similarly, within a general government table, only transfers and net lending to the nonfinancial public enterprise sector from general government (not any "commercial" expenditures such as the wage bill of the enterprise) should be included. But it can often be difficult to unravel these transactions. There may also be quasi-fiscal expenditures involved in lending to this sector directly from the central bank or from "commercial" banks, under government or central bank direction.

7. Data on quasi-fiscal activities may be very difficult to obtain, if possible at all. Quantification of such activities is, in itself, often very difficult (see Section 3).¹⁰

• In francophone Africa, the main tables prepared by the authorities usually cover the central government budget--which does not include social security operations (typically rather small). To this, social security operations and other extrabudgetary fund operations (like commodity stabilization funds) should be added whenever possible.

• In Latin America, much the same practice applies. However, social security operations can often be more readily included, since information on the social security budget has to be provided annually to the legislature.

• In the (British) Commonwealth system, local governments are usually highly dependent on central transfers for support and the extent of major extrabudgetary funds is minimal. Central government budgets are typically the main operational tables.

• In transition economies, extrabudgetary funds are extensively used, especially for social purposes; are usually important; and need to be consolidated. Moreover, very centralized systems are giving way to greater autonomy at the subnational government levels. The movement of budget authority to subnational levels, however, has often not (yet) been translated into significant local revenue assignments or important shares in total taxes.

The key point is that, in deciding which public expenditure aggregate will best allow a macroeconomic perspective, a balance needs to be struck between the desirable (as wide a coverage of fiscal transactions as possible) and the realistically achievable. Ideally, all relevant nonbudgetary transactions should be included or at least some adjustments made to the aggregates reported in the budget. But how far this will be practicable is inevitably constrained by the availability of information.

To sum up, the key questions to ask about the coverage of the data are:

• Is it possible to work with general government or with central government only?

• Is the coverage of government operations complete as defined in general government? If not starting with central government budget data, which other transactions should be included and how feasible is it to estimate and monitor their size? Have intersectoral transfers (e.g., central government grants to local government) been counted only once in the consolidation?

• Are estimates gross or does netting take place? If netting takes place, are these transactions sufficiently large or strategically important enough to necessitate conversion to a gross basis?

• Are the scale and nature of activities undertaken by nonfinancial public enterprises sufficient to justify working with the nonfinancial public sector rather than GGFO?

Those preparing expenditure aggregates need to be able to get timely, reliable, and accurate information on both outturn and planned expenditures (see glossary for definition of outturn). Data on outturn central government expenditure as a whole are generally available from some unit in the ministry of finance. More detailed information on month-to-month spending can sometimes be obtained from line ministries or spending agencies. But there are important differences, particularly when gathering information during the fiscal year ("in-year"), in terms of data availability on outturn expenditure among the francophone, Commonwealth, Latin American, and transition economy systems. Such in-year information is crucial for fiscal monitoring and fiscal adjustment.

• In francophone systems, accounting for government expenditures is more centralized, so that data are available, usually quickly, on cash outturn--and even on earlier stages in the spending process (see Section 4)--from the ministry of finance.

• Commonwealth systems are, by contrast, often more decentralized. Typically, consolidated reports of payment orders and checks issued by government are available; checks cashed may also be available, but often only with a reporting lag from the banking system and through the central bank. Data on earlier stages of spending, however, require special reports from the line ministries, except where a centralized general accounting office is in charge of issuing payment orders directly.

• In Latin American countries, the expenditure process is quite decentralized. Data on monthly limits for payment orders are available; however, there are typically no data on the earlier stages of the spending process.

• In transition economies, the position is in flux: now the information is normally available from the ministry of finance, often the treasury department, but there are still some reporting lags.

All these outturn data are termed "above the line;" that is, accounting records of government expenditures undertaken. The monetary survey can be an important source of "below-the-line" data on cash outturn expenditures; that is, data extracted from the financing of government expenditures from banking, nonbank, and external sources. While the banking system provides data on government bank accounts, rather than on a detailed line item budgetary basis, it can provide a useful check on the budgetary data from the ministry of finance (this is explored further in Section 4).

It is thus necessary to determine what data on outturn government accounts are available within the ministry of finance; and what ancillary sources--line ministries, planning commission, central bank--can be used to get the necessary information on the consolidated government sector.

Finally, data on planned expenditures for the setting of the next budget can usually be found from a combination of the ministry of finance, any planning ministry, and in some instances the central bank (for debt payments)--see Section 3.

Most country authorities, whether working with IMF programs or World Bank projects or not, prepare and publish at least annual projections of public spending.¹² Many also make projections, at least in aggregate and often by line ministry, of annual expenditure in each of the two subsequent years. Budget advisors can readily subject these to economic analysis, particularly on volume measures, cost factors, and underlying economic assumptions. In the context of macroeconomic forecasting, the techniques for projecting annual expenditures are well known. However, the fiscal economist may be less aware of how to make short-term projections on expenditures, on a monthly, quarterly, or semi-annual basis. Yet, such projections can be critically important in the context of taking in-year fiscal remedial action.

Whatever data are provided by the authorities, a fiscal economist may wish to make a separate assessment of short-term expenditure trends--for example, for the second half of the fiscal year. (For financial programming work in the IMF, for example, it is always important to estimate a time path for public expenditures in order to derive the quarterly--and sometimes monthly--financing requirements of the government.)

The starting point will usually be the latest annual expenditure projections, usually presented in the annual budget by the authorities and, ideally, with estimated figures for the last year and projections at least for the next two years. (Some such forward projections represent expected expenditures; but more often, they are normative. For example, in many low-income countries, the annual figures for years two and three are likely to be residuals from a targeted fiscal deficit and projected revenues. The positive or normative nature of projections needs to be properly understood.)

A breakdown of the annual projections into quarterly and monthly figures may be provided by the authorities, but this is not always the case. Short-run focus projections do not usually require sophisticated modeling approaches. Rather, to prepare expenditure projections with any precision requires a close understanding of (1) the different stages of the spending process; (2) different categories of expenditure, identified by economic type, and often individual projects; and (3) the patterns of spending in preceding years.

Ideally, all government expenditure transactions should be classified in four ways:

1. by administrative responsibility--the ministry, department, or spending agency that undertakes the expenditure;

2. by economic category--defined by Government Financial Statistics standards;

3. by function (e.g., health, education)--defined by the United Nations; and

4. by program (e.g., by policy goals and objectives).

But expenditure classification is important, particularly for short-term projections. It is usually easiest to adopt a disaggregated approach that projects government expenditure transactions, by month or by quarter, separately for broad economic categories. Apart from its relative simplicity, this classification facilitates broader economic analysis; for example, it allows easier simulation of the effects of varying economic assumptions or projections on the exchange rate, inflation, etc. Starting from the very simplest economic categorization,¹³ wages and salaries are nearly always paid on a well-defined timetable (weekly, biweekly, monthly, or bimonthly).¹⁴ Some grants and transfers are paid monthly but others quarterly. Debt interest payments can usually be projected from a schedule of due payments. So, once the budget is set, the likely pattern of such recurrent expenditure outlays can be determined in a fairly straightforward way. By contrast, the profile of the purchase of other current goods, services, or capital goods may be more variable from month to month and thus more difficult to assess.

Projections should thus be based on a combination of budget plans, known commitments (see Section 4), and past expenditure patterns. For analytical purposes, once a base projection has been established, the economist may also need to examine particular scenarios and thus to distinguish between those economic categories that are most sensitive to different economic assumptions, such as the exchange rate or inflation rates; or consider whether some expenditures are particularly sensitive to changes in output. Especially when the inflation rate is high or volatile, even three- or six-month projections need to be built up from a very short-time profile, typically monthly. Also, particularly during periods of economic crisis or fundamental macroeconomic structural problems, such that revenue receipts (and often external financing) are very uncertain, monthly expenditure projections are especially important and may need to be constrained by available financing.

Greater disaggregation than the broad economic categories identified above may also then be desirable, to build up projected spending patterns. Moreover, it may also be necessary to identify large or volatile items of expenditure from past years' outturns (e.g., on capital projects, or payments of transfers) and make special efforts to estimate their magnitude and timing in the period ahead. Any additional or exceptional expenditures that arise will have to be factored into the projections separately.

In developing short-term expenditure projections, therefore, the following are key questions to bear in mind:

• What information by economic category can be identified?

• What components are susceptible to changes in which economic assumptions?

• What guidance can be found from past expenditure patterns?

• What special expenditures may occur in the projection period?

• Does a high or volatile inflation rate require the development of a very short-time profile (e.g., monthly) projections?

• What constraints on expenditure projections may be implied by the projections or revenue or financing?