The IMF's Statistical Systems in Context of Revision of the United Nations' A System of National Accounts

16 A Discussion of Public Sector Accounts

Vicente Galbis
Published Date:
September 1991
  • ShareShare
Show Summary Details
Jonathan Levin, Jan van Tongeren, Brian Newson and Derek Blades 

This paper was prepared as a guide for discussion at the January 1988 Expert Group Meeting on Public Sector Accounts, with particular reference to the structure of the present United Nations' A System of National Accounts (SNA) and its relationship to the methodology of the IMF's Manual on Government Finance Statistics (GFSM). It is organized in response to questions with respect to the main issues, as follows.

Section I examines the objectives and roles of the two data systems, the resulting structural and data relationships between them, and the principles that should guide any changes designed to clarify and strengthen these relationships. Section II considers proposals for clarifications, revisions, or additional alternatives to coverage of transactors in the light of developments over the past two decades in institutions, economic practices, and analytical needs, with a brief discussion of the statistical units used in the government sector. Section III examines data compilation procedures in both systems that may require revision and their bearing on more explicit reconciliation of the two systems. Section IV considers possible modifications to some analytical concepts and whether the addition to the SNA of several cash-based alternative concepts applicable to government but not to other sectors would enhance analytical usefulness of the SNA or generate ambiguity. Section V examines whether various previous SNA classifications should be revised to reflect general usage in fiscal analysis, national institutions, and data availability.

This paper was prepared by Jonathan Levin, except for subsections that should be attributed as follows. Jan van Tongeren: in Section I, “Data Links”; in Section II, “Community Production of Services and Capital Goods,” “Nonprofit Institutions,” “Majority Ownership,” “Accounts and Tables for Nonfinancial Public Enterprises and Public Financial Institutions,” and “Statistical Units in the Government Sector”; in Section III, “Accrual Versus Cash Basis,” “Depreciation of Government Fixed Assets,” “Rent on Government-Owned Buildings,” and “Debt Cancellations as Transfers”; in Section IV, “Distinction Between Current Expenditure and Capital Formation” and “Government Operating Surplus”; in Section V, “Indirect Taxes” and “Withdrawals of Entrepreneurial Income.” Jan van Tongeren and Jonathan Levin: in Section II, “Government Employees' Pension Schemes” and “Government Employees' Welfare Schemes”; in Section III, “Gross and Net Treatment.” Brian Newson: in Section II, “Supranational Authorities”; in Section III, “Separate Identification”; in Section V, “Functions of Government: Revision of COFOG.” Derek Blades: in Section II, “Ownership, Control, or Both”; in Section III, “Valuation in Current and Constant Prices.”

I. Conceptual and Data Links Between A System of National Accounts and the IMF's Government Finance Statistics

The structural and data relationships between the SNA and the Fund's system of government finance statistics (GFS) are discussed in this section.

Structural Relationships

Differences between the two systems—in objectives and in concepts—are discussed, and areas for reconciliation identified, in the following subsections.


What types of data on government are needed by the Fund and member countries as a guide to fiscal and financial policies, what types of data on government are needed for the construction of national accounts, and to what extent could the objectives of either statistical system be incorporated and met by the other?

Basic differences between the SNA public sector accounts and the Fund's system of government finance statistics (GFS) arise from the development of the two systems along separate tracks, with two separate sets of objectives.

The GFS system codifies forty years of practice by Fund missions and Fund publications in attempting to measure and present, in a framework of financial analysis, the actions of a sector that cannot be expected to respond like other sectors to monetary and exchange rate stimuli and constraints. To meet the Fund's needs, and those of its member countries, for an immediately verifiable current record of government actions, GFS data are derivable directly from government accounts, with no reliance on estimates or imputations. Because they must be closely integrated with analyses focusing primarily on bank deposits, bank credit, and other financial targets, GFS data are based on government payments and receipts, with only secondary reference to the accrual-basis assets and liabilities generally unavailable in government bookkeeping records.

The primary aggregates of policy importance in the GFS are the overall deficit or surplus; financing from abroad, from the monetary authorities, and from deposit money banks; and aggregate revenue, grants, expenditure, and lending minus repayments. The GFS reflect the characteristic differences between the motivation and significance of government actions and those of market-oriented sectors by organizing these major aggregates differently from those in other sectors—grouping government lending with government expenditure, for example, rather than with government borrowing. Within these primary aggregates, the more detailed components also carry significance for the light they shed on the government's impact or dependence on particular groups and activities. Of limited policy significance in the framework of financial analysis within which the GFS operate are a number of aggregates of greater importance to other sectors and to the economy as a whole. These include concepts of physical output (such as the government's product or value added) as a component of gross domestic product (GDP); the concept of government consumption, as distinct from the delineation of current versus capital expenditure or of purchases of goods and services versus transfers; and the concept of change in the government's net financial position, a concept of little applicability to government units that take and give rather than purchase and sell.

The SNA public sector accounts, in contrast, focus primarily on government product and value added, government consumption, and changes in a government's financial position, as reflected in its net lending. Organization of these aggregates for government in the SNA public sector accounts is largely symmetrical with their organization for other sectors, so as to permit calculation for the entire economy as the consolidated sum of the aggregates for the sectors. To measure the physical output constituting product, and to parallel also other measurements for other sectors with which they must be aggregated, the public sector accounts present the data for government operations not on the basis of payments and receipts but on an accrual basis, requiring various estimates and imputations. In addition, to facilitate use of the more readily available data on government transactions to measure the operations of transactors in other sectors, classification and timing of transactions in the public sector accounts are kept fully symmetrical with those for other sectors. The perception and timing of transactions from the government point of view (as reflected in the GFS, for example) may, however, be asymmetrical with that of the other transactors.

Given the continuing needs for both SNA measurement of government production, value added, income, consumption and capital formation, and for GFS measurement of government financial performance, it is unlikely that either statistical system can be entirely replaced by the other. Considerable saving of statistical resources and avoidance of users' confusion may be achieved, however, by closer integration of the two systems in aspects that do not face basic differences of objective and by the clarification of the relationships between the two systems' major concepts.

Most issues under consideration in the review of GFS and SNA measurements of government involve bringing both systems closer to evolving government practices. Other issues involve modification of either or both systems to bring them closer to each other without jeopardizing their basic objectives. These may be viewed as efforts at harmonization to the extent consistent with differing objectives and reconciliation to the extent feasible at a relatively aggregate level of detail.

A description of progress over the past decade in delineating the relationships between concepts in the two data systems is contained in Chapter 17 of this volume.

Changes to Facilitate Reconciliation

What principles should govern decisions to change the treatment of particular account items to provide additional or alternative groupings of existing categories in order to identify aggregates common to both theSNAand GFS?

The sources of difference between the two systems should be identified and related to the basic nature and objectives of the two systems, and possible flexibility should be considered in this context.

•Reconciliation of major items, facilitating cross-use of data with few and identified adjustments, should be attempted.

•Changes should not destroy or sacrifice items needed to measure other sectors or parts of either system; treatment should not be out of keeping with the system's basic principles.

•Alternative treatment could supplement, but should not replace, items needed for a system or for the building of a component of a system.

•When treatment in SNA is the result of an asymmetry in the nature of an item with respect to the government sector and the other transactor, a presentation of the item from the government standpoint for both sectors, or with different treatment and an explicit adjustment to reconcile the SNA sectoral accounts with each other, should be considered.

•Reroutings that facilitate major concepts should be favorably considered.

Identification of the major components of both GFS and SNA and of the relationships between them is the focus of Chapter 18 of this volume.

Small Conceptual Differences

What should be done about conceptual differences between the GFS andSNAthat remain after conceptual reconciliation efforts, that are quantitatively trivial, and for which no data are available?

All data systems follow conventions on the treatment of items insignificant in magnitude and not measured by available data. A standard of reasonableness governs compilers' efforts to search out data or information permitting estimation of such items. Why, therefore, does this question arise in connection with the reconciliation or harmonization of the two data systems of SNA and GFS? Probably because in the detailed articulation of relationships between GFS and SNA tables and concepts a long list of minor, insignificant, and unmeasurable difference items emerge. This is to be expected when a general system designed to measure activities and concepts applicable to all sectors is contrasted with a particular system formulated to measure the characteristic activities of a single sector with different objectives, behavior patterns, and accounting standards. Items of significance in other sectors may have little if any bearing on government operations. For consistency in the measurement and aggregation of all sectors, however, they remain part of the concepts applied to government in the SNA.

The question of what to do with difference items between the universal, accrual-based, estimate-supplemented concepts of the SNA government accounts and the particular, cash-based, accounting-number-restricted concepts of the GFS data is posed in several forms. First, does inclusion of such insignificant or unmeasurable difference items in the “bridge table” spelling out the relationship between the two data systems overemphasize differences and underemphasize basic similarities? Second, for practical purposes, would bridge tables omitting insignificant difference items provide derivation procedures more likely to encourage compilation of national accounts for government and cooperation between GFS and SNA compilers? Third, should presentation of data for SNA concepts derived from data for approximately similar, but not identical, GFS concepts require explicit qualifications that such data differ from the strict SNA standard? Clearly to be avoided is the publication of differing numbers for the same SNA concept—one based on data derived from GFS sources without adjustment for insignificant or unmeasurable differences, and the other based on GFS data but adjusted by estimates of such items. The confusion to users resulting from such practices would undermine users' confidence in the integrity of the compilation procedures. The importance of relatively simple reconciliation procedures between the two data systems may call for a revaluation in some cases of the degree of accuracy needed.

Data Links

What should be done about discrepancies that arise in the compilation of data for GFS andSNAand that are not a result of conceptual differences?

Frequently, discrepancies between GFS and SNA data are simply the result of the different compilation procedures followed for each, rather than a reflection of conceptual differences. It has been suggested that compilation of SNA and GFS data from the same sources would be preferable in preventing such discrepancies, avoiding unnecessary duplication of effort, and making explicit the subsequent adjustments required to meet the standards of each system.

This would appear to create unnecessary difficulties if there are two data sources—one on an accrual basis meeting SNA standards, and one on a cash basis meeting GFS standards. Neither standard is likely to be met completely by government sources. More frequently, other elements, unconnected with differences between the two data systems, may be encountered and may be adjusted for differently in the two compilation procedures, introducing additional, unnecessary discrepancies between their results.

Few countries are likely to have two complete sources of data on government operations. They are more likely to have two sources for particular parts of their operations, such as tax assessments and tax collections or payment orders and checks issued for purchases of goods and services. In these circumstances, evaluation of alternate data sources and determination of how they may best be reconciled for overall consistency would be necessary for both data systems and could be carried out uniformly with provision for explicit adjustments to meet the different standards of each system. This would facilitate a sequential compilation procedure, starting with the basic government accounts, adjusting them to GFS, and then making the additional adjustments necessary for SNA.

Procedures for the adjustment of GFS data for use in the SNA are illustrated in Chapter 19 of this volume.

II. Transactor Coverage

This section provides a discussion of coverage in the SNA and GFS of transactors at the borderline of general government, borderline transactors within general government, public sector transactors, as well as a discussion of statistical units for the government sector.

Borderline of General Government

Four classes of transactors are examined: nonfinancial enterprises, financial institutions, nonprofit institutions, and international organizations.

Nonfinancial Enterprises or Establishments

Within this category of transactor, departmental enterprises, ancillary enterprises, and community production are considered.

Departmental Enterprises and Dual Sectoring. Should departmental enterprises that sell to the public and have identifiable current costs be excluded from government and classified as quasi-corporations in theSNAand as nonfinancial enterprises in GFS, thus ending this aspect of the dual sectoring of government?

Production accounts for producers of government services differ from production accounts for enterprises in that the former have no market price valuation of output, which is not sold, and instead value output by the sum of input costs and calculate no operating surplus. The two types of producers differ also in that enterprise sales proceeds are required to meet production costs, whereas government revenues (primarily taxes) approximate disposable government income available for whatever purposes governments choose. Similarly, government production costs are assignable to various government functions or purposes, whereas enterprise production costs, except to the extent that they exceed sales proceeds and are covered by subsidies, are not. To avoid the addition of government service production costs, equal to government services output, to government enterprise production costs, and to avoid the addition of enterprise sales proceeds to taxes, two choices are possible: classifying all government-owned or government-controlled enterprise activity with identifiable production costs in the enterprise sector, or leaving such activity in government but measuring its production activity in a separate production account and aggregating only the outcome of such activity—the operating surplus—with data for the producers of government services. The present SNA chooses the first alternative for government-owned or government-controlled enterprises that are incorporated or have large-scale sales to the public and the second alternative for unincorporated units with only small-scale sales to the public. Government sales to the public of industrial or commercial goods and services that have no separately identifiable costs remain in the production account for producers of government services. This sectorization is followed also in the GFS, which nets departmental enterprise sales to the public against their identifiable operating costs and aggregates with the producers of government services only the resulting cash operating surplus or deficit.

One reason for leaving in the government sector unincorporated enterprises with small-scale sales to the public may have been the difficulty in some instances of separating their income and outlay, and particularly their capital accounts, from those of the government. The proposal to classify such enterprises in the nonfinancial public enterprise sector, rather than in government, is associated with a desire to eliminate the “dual sectoring” in the production account for government. This would restrict the government sector production account to operations whose output is valued as the sum of their inputs, with no market price of output and no operating surplus derivable as the difference between their sales to the public and identifiable costs of inputs.

The effect of classification of departmental enterprises as nonfinancial public enterprises would appear to be as follows.

•Although GDP would be unaffected, departmental enterprises' production would be attributed to the enterprise sector rather than to government. This change would generally be small in magnitude, but it might be argued that, since departmental enterprises differ only by size and incorporation from nonfinancial public enterprises, their inclusion in enterprise production totals would be preferable to their grouping with nonmarket services provided to the public without a sales price. Classification by purpose (as in The Classification of Functions of Government, COFOG)1 of government outlays, for example, would more appropriately include only a government subsidy to cover a departmental enterprise that is operating a deficit rather than full departmental enterprise production costs.

•Classification of departmental enterprises with nonfinancial public enterprises should have little effect on government income and outlay accounts, which would continue to reflect the enterprises' operating surpluses or deficits financed by government. To the extent that interest or transfer transactions were carried out by departmental enterprises, they would be assigned to the enterprise, rather than to the government sector.

•Removal of departmental enterprises' capital accumulation account transactions from government, to the extent they are identifiable, would alter the totals for both government and enterprise sectors. A more unitary concept of capital formation would result because changes in enterprise stocks would not have to be added to the more limited concept of changes in government strategic stocks.

•Similarly, the effect of such removal on the government's capital finance account would separate any enterprise lending from the characteristically different lending undertaken by government for public policy purposes. Should such public policy lending—or any other clearly governmental activity—be undertaken by departmental enterprises, it could be attributed to government by considering it to have been carried out by the enterprise for government in an agency capacity. This would be the case with similar occurrences involving nonfinancial public enterprises or other institutions.

It must be noted also that the criterion of small-scale versus large-scale sales to the public used to distinguish between nonfinancial public enterprises and departmental enterprises raises difficult problems in practice, leading those seeking to apply the standard to question its validity.

On balance, a decision on whether to classify departmental enterprises with the nonfinancial public enterprise sector rather than with government would appear to depend on whether the advantages of (1) separating market-valued output from input-valued output, changes in operating stocks from changes in strategic stocks, and commercial lending from lending for public policy purposes outweigh (2) the difficulties of separating out departmental enterprises' capital transactions from the capital transactions of government. Whichever choice is made, however, the elimination of dual sectoring in the government production account without removing departmental enterprises from government—that is, leaving departmental enterprises' gross receipts and payments to be aggregated with those of producers of government services—would exacerbate present difficulties and hinder the measurement of government activity.

Ancillary Enterprises. Should ancillary enterprises selling to other parts of government, at sales prices that cannot be taken to represent market prices, and own-account production of capital goods be treated as part of governmentwith their output valued, like that of the rest of government, as the sum of their inputs rather than at their sales price—thus ending this aspect of the dual sectoring of government?

Ancillary enterprises may be defined as government-owned or government-controlled (or both) industries that are primarily engaged in the provision of goods and services to other parts of government and that, because they are unincorporated and do not have large-scale sales to the public, are classified in the government sector. Under present dual sectoring practices in the government production account, the output of ancillary enterprises is valued at the prices such enterprises charge other parts of government for the goods and services the enterprises provide them, giving rise to an operating surplus or deficit as the difference from the cost of the enterprises' inputs. In practice, however, an ancillary enterprise's charge to other parts of government may differ substantially from market prices, for political or internal administrative reasons. This would undervalue or overvalue GDP, government gross output, government consumption, and capital formation under present SNA practices and misstate the enterprise's operating surplus or deficit.2 To eliminate dependence on such unrealistic internal pricing, such ancillary enterprises could be assimilated to producers of government services and have their output valued at the cost of their inputs. Own-account capital formation within government also lacks a marketlike output sales price, since the capital formation may be carried out for the unit itself or may be carried out for another part of government at an unrealistic, administratively determined charge. More accurate valuation of such own-account capital formation could be based on the cost of inputs. This would appear to be preferable also to attempting estimates of own-account government capital formation on the basis of similar capital formation by enterprises.

Regardless of whether departmental enterprise production is measured at its market sale price in a separate (dual sectoring) production account within government or in the nonfinancial public enterprise production account outside government, measurement of ancillary enterprises' output sold to other parts of government at unrealistic prices and of own-account capital formation by government could more appropriately be carried out within the production account of producers of government services, with output valued at the cost of inputs. Although this may result in including within the production account for producers of government services some activities classified as industries by International Standard Industrial Classification (ISIC),3 the preferable criterion for inclusion in the different production accounts would appear to be not industrial character but whether output should be valued by market sale or by the cost of inputs. This need not detract, however, from the breakdown of production account information for the government sector into separate activities corresponding to ISIC categories.

Community Production of Services and Capital Goods. Should the services and capital goods resulting from informal community production be attributed to government or to the household sector?

Some services and capital goods resulting from informal community production are closer in character or organization to households, some to private nonprofit institutions, and some to government. In GFS, for lack of payment transactions, there would be no registration of the services or capital goods produced. In SNA, estimating procedures for valuation of output, inputs, and consumption or capital formation would be necessary regardless of the sector to which the activity is attributed.

Consideration may be given to the character of the activity by which the services or capital goods are produced, to the character of the service or capital goods themselves, to control of their distribution or use, to whether depreciation of the capital goods will have to be included in future accounts, and to whether maintenance may have to come from households, nonprofit institutions, or government.

Discussion at the SNA Expert Group Meeting on the Household Sector (Florence, August 31-September 4, 1987) concluded that capital assets constructed on a communal basis should be attributed to the sector responsible for their upkeep; that sector may be different from the sector that produced the assets.

Financial Institutions

The treatment of flows originating in monetary authorities' functions, of government employees' pension schemes, and of government employees' welfare schemes is discussed in this subsection.

Monetary Authority Functions and Acceptance of Deposit Liabilities. Should flows to or from government—from the performance of monetary authorities' functions or the acceptance of demand, time, or savings deposit liabilities—be rerouted and shown as government inflows from or outflows to the financial institutions sector?

Some reroutings of transaction flows are carried out in the SNA for analytical purposes. Thus, employer contributions to pension schemes or social security schemes are rerouted through households and give rise to a corresponding increase in the compensation of employees that constitutes household income. For purposes of money and banking statistics, similarly, it is useful to gather all transaction flows constituting the acceptance of money or near-money liabilities or the management of the money supply and international reserves—including transactions with the Fund—and attribute them to the financial institutions sector and its appropriate parts. This permits money and banking statistics to keep track of the liquidity in the economy as indicated by the monetary and near-monetary liabilities of the banks. Were such activities to be attributed to the government sector directly, the coverage of money and banking statistics would have to expand to include a part of government. The task of measuring financial intermediation, liquidity, and bank credit would be possible but more difficult.

To facilitate simplification of the task of money and banking statistics, government sector statistics reroute, through the financial institutions sector, all such banking function flows reaching government. Deposit flows reaching government, therefore, are classified as flows reaching government through the financial institutions sector. So long as money and banking statistics continue to group in the financial institutions sector the provision of liquidity to the economy and management of the money supply and international reserves, the reflection of this rerouting in the statistics for government is necessary for the sake of consistency. Depending on practices to be followed with other SNA reroutings, it may be found appropriate to identify such reroutings explicitly for the use of analysts who prefer the identification of all transaction flows by actual transactors. An alternative to such rerouting would be to redesign money and banking statistics to cover the banking activities of government.

Government Employees' Pension Schemes. Should government employee pension funds, invested with the employer government or elsewhere, form a part of government or of the insurance and pension fund subsector of the financial institutions sector?

For pension funds invested entirely with the employer, the present SNA provides an exception to the classification of pension funds in the financial institutions sector. Because the conduct of such pension funds is thought to follow more closely the conduct of the employer than of households or of an independent pension fund subsector, pension funds invested entirely with the employer are classified as part of the employers' sector rather than in the pension fund and insurance subsector of the financial institutions subsector. This classification is followed also by the GFS. In the case of government employee pension funds, however, the distinction drawn on the basis of investment with the employer is not so clear. Pension funds are frequently invested in government securities for reasons of prudence, whether by legal requirement or by choice, so that investment of government employee pension funds with the employer—the government—does not indicate the degree of employer control that investment with a nongovernment employer does. Classification of such government employee pension funds with the majority of other pension funds, in the financial institutions sector, may therefore be preferable. This would permit the type of analysis of government employee pension funds that can best be provided in the context of pension funds and insurance companies, including the calculation of actuarial liabilities and reserves. It would remove from government accounts several noncharacteristic actual and imputed flows, such as the imputation of household equity in pension fund reserves, and the distribution and reinvestment of interest earnings on such equity. Analysis of pension funds would thus be carried out entirely in the context of pension funds, independently of whether reserves are invested entirely or partially with the securities of the employer government.

With the classification of government employee pension funds in the financial institutions sector, the separate identification of contributions to such funds would be retained. All imputations of service charges, net equity of households with pension funds, and interest on this household equity would be omitted from the government accounts and registered instead in the accounts of the financial institutions sector.

Government Employees' Welfare Schemes. Should government employee welfare funds and unfunded welfare and pension schemes for government employees be considered social security schemes, with their contributions and benefits classified as social security contributions and benefits?

For the government sector, the SNA groups together government employee welfare schemes and unfunded government employee pension schemes and distinguishes them from pension funds, social security schemes, and social assistance grants. In practice, however, the distinction with social security schemes may be rather difficult to make. Although schemes provided by employers and those imposed by government are clearly different in the rest of the economy, they may not be readily distinguishable from each other when the government is the employer. Government employees are often not covered by the social security schemes that apply to other sectors, so that, as an examination of country practice reveals, many countries include as social security schemes the pension and welfare schemes that are exclusively for government or public sector employees.

Other countries that do cover government employees in their general social security schemes have supplementary pension and welfare schemes for government employees that may or may not be classified as social security schemes, depending on the interpretation of the SNA by the national accountant. As a result, SNA distinctions between different kinds of contributions and benefit payments are generally not applied in the prescribed detail in most countries, either because all schemes envisaged in the SNA are not relevant in all countries or because the distinctions are difficult to make. Many countries, in fact, do not distinguish between social assistance grants and unfunded employee pension and welfare benefits and make no imputations of contributions to unfunded employee welfare or pension schemes.

To simplify the prescribed breakdown in contributions and benefits, therefore, it has been suggested that in the SNA for the government sector the distinction be eliminated between social security schemes and government employees' welfare funds and unfunded welfare and pension schemes. This would bring no change in the total of employee and employer contributions and of benefits registered in the SNA for the government sector, but all such contributions and benefits would be shown as social security contributions and social security benefits, respectively.

The imputation of a service charge component of contributions would not continue, since no service charge is calculated for social security contributions. The service charge component of pension and casualty insurance premiums does provide a measure of output and value added when applied outside government, but its elimination from contributions to the government would have no effect on government output and value added because these are calculated from the sum of inputs rather than sales. Elimination of the imputed service charge on contributions to government employee welfare funds and unfunded pension and welfare schemes would merely shift, to services produced for own use, the service charge amount now attributed to government sales in the production account for government.

Imputed employer contributions to unfunded pension and welfare schemes for government employees would continue to appear as part of the compensation of employees and, at the same time, as a government sector receipt from households in the income and outlay account.

As in the past, noncontributory pension or welfare benefits provided to government employees as members of the general population rather than as employees would continue to be classified as social assistance grants.

Nonprofit Institutions

Should government include nonprofit institutions that serve households or enterprises but that have majority financing and control by government? If so, should the criteria of control and finance be maintained, or should other criteria be applied?

At present, the SNA includes in the government sector all nonprofit institutions that are mainly financed and controlled by public authorities. Private nonprofit institutions that serve enterprises but that are not wholly or mainly financed and controlled by the public authorities are included in enterprises. In the present SNA, however, private nonprofit institutions serving households form a separate sector on their own that is of an equivalent status with government, enterprises, and household sectors.

The SNA Expert Group Meeting on the Household Sector in September 1987 dealt with nonprofit institutions as one group, including those to be allocated to the enterprise sector, those to be included in the household sector, and those to be incorporated with the government sector. The distinction between nonprofit institutions included in the government sector and the rest was left open for recommendations by the Public Sector Expert Group Meeting. With regard to the remaining nonprofit institutions, the Household Sector Expert Group recommended that the allocation of these to the household and enterprise sectors should be based on whom the institutions serve, not on who pays for most of their outlays. It decided, furthermore, that nonprofit institutions serving households should be included as a subsector of the household sector, and that this subsector should include all such institutions serving households regardless of the number of their employees. Also included as nonprofit institutions should be religious missions, private aid agencies, and community activities including those production activities carried out on a communal basis that result in capital assets such as roads, schools, and the like.

The present SNA criterion for inclusion of nonprofit institutions in government requires both majority government financing and government control, leaving institutions that are not controlled by government outside the government sector, regardless of their degree of dependence on government funds. This has had advantages in leaving outside government institutions that maintain operating independence and for which operating statistics would, as a consequence, be less likely to be available through government accounts. Such institutions may be quite sensitive about their separate existence outside government, moreover, and may feel with some justification that their ability to raise funds from private sources could be jeopardized by their classification as part of government.

Adoption of a criterion of majority government finance or control as a condition for including nonprofit institutions in government would also bring in institutions controlled but not majority-financed by government. Institutions that are controlled by government but are not majority-financed by government are likely to form a part of government in any case, perhaps collecting fees or compulsory levies for the greater part of their receipts. Their classification would not be affected by adopting a criterion of majority government financing or control to determine inclusion of nonprofit institutions in government.

International Organizations

Issues pertaining to the treatment of two types of international organization—supranational authorities and other international organizations—are examined in this subsection.

Supranational Authorities. Should general government include, as a nonresident subsector, the nonheadquarters operations of supranational authorities within the country; that is, international organizations empowered to levy taxes within the country?

Supranational authorities are defined in the GFS as those international organizations that are empowered to levy taxes within countries. Seemingly the only case that exists at the moment is the institutions of the European Community (EC).

The SNA does not discuss supranational authorities but stipulates that “international bodies, such as political, administrative, economic, social or financial institutions, in which members are governments, are not considered residents of the country in which they are located or operate” (SNA, paragraph 5.113). The European System of Integrated Economic Accounts (ESA)4 follows the same line but specifies further by showing the EC institutions separately as a subsector of the rest of the world and in specifying that taxes collected for (or subsidies paid by) Community institutions are to be recorded as direct transactions between producer units in the country and the Community institutions, without passing through the accounts of the national general government sector. Hence, for instance, total taxes paid by the resident producers are the sum of those destined for the national government and those for the Community institutions.

In the GFS the “general government” is expanded to encompass both the national general government (the SNA sector) and a nonresident subsector for the operations of supranational authorities (excluding their headquarters). This is not done, however, for other international organizations. In the GFS as in the ESA, taxes collected for Community institutions (and subsidies paid by them) are recorded as direct transactions between the producer units and the supranational authorities.

The advantage of the GFS treatment is to bring together all taxes paid in a country into one “government” receiving sector instead of two. The disadvantages from the SNA point of view are as follows.

  • nonheadquarters operations of supranational authorities, at least in the EC case, are merely an amalgam of income and outlay account transactions, not a real institutional unit at all.
  • The enlarged government sector is both resident and nonresident at the same time.
  • It would seem that the normal consolidation rules applied to the enlarged sector would be inconsistent with both the balance of payments and the gross national product (GNP).

A rationale for the GFS treatment of supranational authorities is delineated in the following paragraphs.

Over the years since adoption of the SNA in 1968 there has been a growth of governmental activities on a supranational level not fully anticipated by the national framework of the current system. Some of these issues have been approached in the Fund's Manual on Government Finance Statistics (GFSM) and GFS Yearbook, and consideration may be given to whether review of the SNA may call for a similar approach.

The view adopted in the GFSM and GFS Yearbook is that, just because the countries of the EC have assigned several of their governmental functions to authorities that encompass several countries, there is no cause for a reduction in the measurement of overall governmental activities in these countries, as regards both taxes and governmental expenditures. This requires statistical recognition that, in these countries, government is no longer solely national and that there are transactions being carried out within the country by governmental authorities that are nonresident. The GFSM and GFS Yearbook recognize this fact by adding to the traditional resident, or national, government the activities within the country of the nonresident, supranational government authorities to reach the overall concept of general government for the country. This causes no contradiction with balance of payments concepts so long as it is recognized that, although national government is resident, general government need not be entirely resident.

The delineation of the transactions of supranational authorities from those of national government is accomplished in the GFS by defining supranational authorities as a separate level, or subsector, of general government and by using the same principles of attribution of receipts and payments—for example, between collecting and beneficiary governments—as are employed for other levels of government. Thus, particular taxes set by supranational authorities but collected for them by national governments are attributed directly to the supranational authorities and appear in neither the receipts nor expenditures of the national governments. Payments levied on the national governments by the supranational authorities without specifications as to how the funds are to be raised within each member country are registered as receipts of the supranational authorities from each national government. Both the raising of money domestically and its payment to the supranational authorities are registered in the data for the national government. Similarly, expenditures paid by supranational authorities to individual and enterprise residents of a member country are registered as transactions of the supranational authorities with these residents and not as transactions with the national government of the country. Data for both national governments and supranational authorities are shown separately for ten EC countries in the GFS Yearbook. Data are also shown for general government, consolidating the data for all levels of national government and the supranational authorities and eliminating any transactions between them.

To reflect the nature of its activities and relationships, data for the supranational authorities subsector of general government exclude headquarters or central office operations—such as those in Brussels and Luxembourg and the several nuclear research centers—and are restricted to operations within the country, with all transactions abroad viewed as proceeding through headquarters. Transactions of supranational authorities with their headquarters are not viewed as giving rise to claims or liabilities but are shown as transfers that serve as balancing items, averting the calculation of any overall deficit or surplus for the supranational authorities subsector of general government that would be parallel to the deficit or surplus concept of other levels of government. All transactions of supranational authorities with nonmember countries are classified as transactions of the supranational authorities' headquarters, and no supranational authorities subsector of general government is considered to exist in nonmember countries because no governmental functions have been taken on by supranational authorities there.

Other International Organizations. Should the operations of all international organizations excluded as nonresident from all countries' national accounts be aggregated as an additional unit to complete the universe of national accounts?

Participants in the Expert Group Meeting on External Sector Transactions concluded that, for completeness and symmetry of national accounts, an additional unit embracing international organizations that are classified as nonresident everywhere they operate should be defined and measured through the compilation and consolidation of data on their activities. Such international organizations were defined to include organizations that meet three criteria: (1) authority derived directly from the authority of the organization's members, which may be independent states or other international organizations; (2) sovereign status (that is, the laws and regulations of the country or countries in which the international organization is located do not apply to it); and (3) production of services that are primarily nonmarket services.

To facilitate the work of national compilers and international organizations working in the field, the expert group recommended that a list of such international organizations be drawn up. Accordingly, the Fund now collects balance of payments data from international organizations, consolidates them, and uses them to round out global totals; the results are presented in the Balance of Payments Statistics Yearbook (BOP Yearbook).

Such aggregate data for international organizations would not provide data on their activities in individual countries. This would form a part of national balance of payments data on transactions with nonresident official entities, which include both international organizations and public sector entities of other individual countries.

International organizations covered by the aggregate data would cover both organizations governmental in nature and those performing functions of financial intermediation. A strict parallelism with the distinction of government versus financial institution used in national sectorization, and any thought of aggregate governmental activity at both the national and international level, would require the preservation of subcategories for groupings of governmental and financial institutions within the overall aggregates for international organizations.

Borderline Within General Government

This subsection considers treatment of various government levels and of social security funds.

Central, State, Local, and Other Government Levels

Should state or provincial governments and local governments be shown as separate subsectors of general government when they operate as levels of government separate from central government and from each other?

Although the 1968 SNA distinguishes only between central government and local government, the significance of separate state or regional governments within the noncentral government group has led the Fund's GFS Yearbook, and more recently the revised UN-SNA questionnaire, to distinguish a separate intermediate level of government (where it exists) between central and local governments. Data for state, provincial, or regional governments are published for some 22 countries in the GFS Yearbook. State, provincial, or regional governments are defined in the GFSM as governmental units exercising a competence independently of central government in a part of a country's territory that encompasses a number of smaller localities—that is, that occupies an intermediate position between the central government and any independent local governments that may exist. Governments are considered to have an independent competence if they have the power to raise a substantial portion of their revenue from sources they control and if their officers are independent of external administrative control in the actual operation of the units' activities.

For some purposes, such as the usual definition of the money supply to exclude only holdings of the central government, state governments may be grouped with local governments. For other purposes, perhaps where closely coordinated central fiscal policy extends effectively through the state government level, state governments may be grouped with the central government. In other instances state governments may most fruitfully be measured alone, permitting analysis of the separate pattern of revenues and expenditures they may represent.

Although separate state or provincial governments exist in a minority of countries, they are not restricted to federal states and have been added to a number of countries in recent years to meet political conditions of regional or cultural identity. Distinguishing between state and local governments may be difficult where there are several layers of local government, such as counties and towns. Separation of all distinctive categories of government units may be warranted for analytical purposes, but identification of state or provincial governments will usually depend on their regional character and their exercise of jurisdiction over lower units of government. The benefits of separating state from local government data, where state governments exist, are most clearly evident in clarifying relations between levels of government and financial flows between them. The role of a separate level of regional government in such flows may be particularly significant, justifying separate measurement.

Social Security Funds

Should social security funds be shown as a separately identified portion (or subsubsector) of each level or subsector of government at which they operate or as a subsector separate from other levels or subsectors of government?

The SNA classifies social security funds as a separate subsector of government, whereas the GFS shows their operations separately but incorporates them as a part (subsubsector) of the level of government at which they operate. This difference reflects differences in the organization of social security, both between countries and with the passage of time, and different considerations of policy and analysis.

In countries with older social security systems, early employer-operated arrangements, mutual benefit insurance, or other mechanisms came under laws governing their operations beginning in the nineteenth century. “Finally, to consolidate earlier piecemeal efforts, a particular program was codified for the first time for an industry or for the country as a whole, almost always on a compulsory basis”5 As separate decision-making centers managing their own finances, social security funds required separate analysis both to portray their impact on the economy and to report to participants on the stewardship of their funds.

Over the past two decades, however, broadening social concerns have extended government budgetary operations into areas previously served exclusively by social security funds. The separateness of the functions, finances, and decision-making powers of social security funds has been substantially diminished. In practice, for example, unemployment benefits have been integrated with employment services and means-tested welfare benefits; old-age benefits have been tied to universal or means-tested pension plans or awarded early to ease unemployment in recessions or in particular industries; family allowances have been adjusted to replace tax credits or exemptions; and sickness plans have been coordinated with government hospitals, clinics, or health services. (For a further discussion of these issues, see Chapter 20 in this volume.)

The financial independence of social security funds was diminished by the indexing of benefits in the face of unemployment-caused reduction of contributions, the maturing of many systems, demographic changes, and the broadening of many benefits without corresponding increases in contributions. Increased portions of social security costs were met out of general budgetary revenues. The independent decision-making powers of social security funds diminished as central decisions were taken to integrate social security changes with other countercyclical, social, or political moves and as the financial base for independence eroded. Although the functional, financial, and decision-making independence of social security funds declined, however, recognition of their separate existence remained important to the confidence of their contributors in many countries and to evaluation of their financial soundness.

Against this background, although separate identification of the operations of social security funds has continued to have value, separate measurement of central government operations excluding social security funds has become increasingly inadequate. Social security contributions comprise 44 percent of combined central government budgetary and social security fund revenue in France, 55 percent in Germany, and 40 percent in the Netherlands. The impact of government on the economy and the potential weight of central countercyclical fiscal policy is not adequately measured by data for central government excluding social security funds. Separate presentations for social security funds or for central government excluding social security funds would fail also to give a complete picture of government expenditures for various functions, since some activities are carried out in combination by social security funds, by social security schemes that do not constitute funds, and by other government programs.

To meet the needs for both separate identification of social security fund operations and combined data for central government operations including social security funds, an appropriate characterization or status for social security funds is necessary, whatever words may be used. In practice, two issues need to be resolved. First is the classification of any state or local level social security funds, which are now classified as part of the social security fund sector by the SNA and as part of the state or local level of government by the GFS. Second is the working out of appropriate consolidation procedures for presentation of social security fund operations in a manner portraying both their own operations and their relations with budgetary central government. This would involve, more specifically, whether budgetary government transfers to social security funds or lending by social security funds to budgetary central government would be eliminated in showing a consolidated deficit or surplus for social security funds, with memorandum items for eliminated amounts, or would not be eliminated in showing an unconsolidated deficit or surplus for social security funds. Although alternative consolidated and unconsolidated presentations are possible, it would be useful to avoid the confusion that would result from users encountering unexplained data in the two formats.

Public Sector

Questions about treatment of various public sector transactors are addressed in this subsection: definition of the ownership of public enterprises, a public-private delineation in the national accounts and supporting tables, and issues relating to the nonfinancial-nonmonetary public sector.

Definition of Ownership

The distinctions between majority ownership or control by government are discussed in the following paragraphs.

Ownership, Control, or Both. Should public enterprises be defined as enterprises majority-owned or controlled by government, or both?

Because of their special relationship with government, public enterprises may be subject to different influences and motivations than private enterprises, so that their analysis in separate public enterprise groupings may serve to delineate separate patterns of behavior. Whether the relationship influencing public enterprises' behavior is the result of government ownership, government control, either, both, or a combination of factors is difficult to determine in any one country at a given time and probably cannot be determined for all countries in general. The need for general guidelines to promote comparability in the identification of public enterprises among countries and over time, however, has stimulated efforts to formulate generally applicable criteria.

The present SNA (see Table 5.1, for example) proposes that both nonfinancial enterprises and financial institutions should be divided into public and private groupings on the basis of ownership “and/or” control. This criterion for distinguishing between public and private has been criticized for lack of clarity: does “and/or” mean “and” or “or,” and how is “control” to be defined?

At a national accounts meeting in May 1984, the Secretariat of the Organization for Economic Cooperation and Development (OECD) proposed that the next SNA give four necessary conditions for classifying an enterprise as “public”: it is owned by government, it is controlled by government, it is large, and it is intended that the enterprise will be retained in public ownership on a more or less permanent basis. The Secretariat also offered suggestions about how “control” and “large” should be defined.

All participants in the May 1984 meeting agreed with the ownership criterion and recommended that, in general, all enterprises with 50 percent or more public ownership should be counted as public. Some participants noted, however, that guidance should be given for deciding borderline cases. Sometimes public ownership, although majority ownership, is split among different levels of government so that the largest single block of shares is actually held by private interests. There may also be difficulties in deciding on the ownership of subsidiaries of public enterprises.

As regards the control criterion, it was argued that this was usually superfluous because ownership almost always implies control. One participant noted, however, that governments could temporarily acquire enterprises in rescue situations without any intention of controlling their commercial operations, and in such cases it was not helpful to classify them as public enterprises. There was also support for the view that the degree of government control could be useful in borderline cases—for example, in deciding how to classify subsidiaries of public enterprises when the extent of public ownership could not be clearly established. Moreover, in cases where government ownership was less than 50 percent, the extent of government control—through the power to appoint board members or directors, for example—might justify classifying an enterprise as public. In summary, although most participants did not object to control being regarded as a condition for classifying an enterprise as public, they felt that it would usually be superfluous.

The OECD meeting in May 1984 had suggested a size criterion on the practical grounds that the elimination of small public enterprises—usually municipal undertakings—would simplify the work of compiling accounts for the public enterprise subsector and thus encourage more countries to publish such data. Most participants in the meeting rejected the size criterion, however, because no single rule for measuring enterprise size could possibly be appropriate for countries as diverse as the United States, Japan, Iceland, and Luxembourg. Some participants also noted that, with the increasing use of central enterprise registers, all enterprises regardless of size would necessarily be classified as either public or private, so that there would be no practical advantages in eliminating small enterprises.

The fourth proposed criterion—that there should be an intention to keep the enterprise in public ownership on a permanent basis—was criticized by one participant on the grounds that statistics should record current facts rather than future intentions. Policymakers may change their intentions, or a new set of policymakers with different intentions might come to power. It would be unreasonable to expect statisticians to change their public enterprise subsector to accommodate new statements of intent. The view was also expressed that, although “intended permanency” was not a necessary condition, it would often be an important practical consideration in deciding whether or not an enterprise is to be counted as public. Under that condition, government ownership and government control would also normally go together. But if the public authorities have temporarily taken possession of an enterprise with the expressed intention of returning it to the private sector at the earliest opportunity, control would normally not be intended, and it would be best to leave the enterprise in the private sector.

In light of this discussion, varying experiences and conditions in different countries, and discussions in other forums, it will be necessary to decide whether the revised SNA should retain the present ownership “and/or” control criteria or adopt a new formulation designed to encompass enterprises with behavior patterns reflecting their special relationship to government.

Majority Ownership. Should government majority ownership be defined to include units in which majority ownership is held by a parent unit in which the government holds majority ownership?

Regarding the definition of ownership, the Handbook of National Accounting: Public Sector Accounts contains the following statement:6

The owner of an enterprise is the holder of the majority of the equity or common stock. This owner should also be regarded as the owner of subsidiaries in which the enterprise holds the majority of the equity or common stock, and iteratively if subsidiaries have subsidiaries.

Accounts and Tables for Nonfinancial Public Enterprises and Public Financial Institutions

Should the nextSNAprovide for a public-private breakdown in the accounts or supporting tables?

The present SNA does not provide for a public-private split in the main accounts of the system although it does recommend a public-private breakdown in two supporting tables: “Domestic Factor Incomes According to Kind of Activity and Institutional Sector of Origin” (Table 17) and “Capital Transactions of the Private and Public Institutions” (Table 19). The UN-OECD annual questionnaire calls for the public-private split for capital formation, saving, consumption of fixed capital, and capital stocks.

Nonfinancial or Nonmonetary Public Sector

Should the revisedSNAinclude accounts for a public sector concept and should it recognize also alternative formulations of the public sector, such as the nonfinancial public sector (excluding public financial institutions) or the nonmonetary public sector (excluding monetary public financial institutions)?

To measure not only the performance of government functions but also the commercial and financial activities carried out under government majority ownership and/or control, the present SNA suggests that in countries where the public authorities play a particularly important role in the economy it may be useful to prepare accounts for the “Public Sector” consisting of general government plus nonfinancial public enterprises and public financial institutions.7

Thus, although the disparate nature of some government and enterprise activities poses obstacles to their aggregation, aggregates may be calculated for other aspects of their operations, such as saving, capital formation, and financing, recommended by the GFS. Because financing transactions between units included in such aggregates are eliminated in consolidation, however, analytical needs may be better served by restricting the units covered for some purposes. Consequently, the GFSM recommends consolidation of only the government and the nonfinancial public enterprises, leaving out the central bank and other public financial institutions, so as not to eliminate in consolidation their lending to the government and to nonfinancial public enterprises. This partial coverage of the public sector is referred to as the nonfinancial public sector.

In some countries, government policy is carried out extensively through nonbank lending institutions that rely heavily on borrowing from the central bank and other government-owned and/or government-controlled monetary financial institutions (that is, from institutions whose liabilities primarily constitute money). In such instances, one measure of the financial impact of government operations and government-controlled operations has focused on the nonmonetary public sector, so as not to eliminate in consolidation the borrowing from the central bank and other public monetary institutions.

Recognition of the validity of such limited-coverage variations of the public sector for particular analytical purposes could add a degree of flexibility to an SNA public sector concept.

Statistical Units in the Government Sector

Two questions will be raised—one concerning statistical units in COFOG, and the second concerning the definition of an institutional unit in the government sector.

Should the unit of classification for COFOG be in principle a transactor unit rather than a transaction unit, applicable to all government expenses? If so, should this transactor unit be the same as the establishment-type unit that is recommended in theSNAfor the classification of production accounts and capital formation by economic activity?

The present SNA (paragraphs 5.86–5.90) and COFOG (pages 2 and 3) opt in principle for the transaction as the unit of classification. However, only for practical reasons, both recommend using the establishment unit for the classification of production and capital formation expenses, leaving the transaction unit as the basis for the classification of other expenses such as transfers and loans.

It could be argued that an orientation toward transaction units would not be in line with the analytical aim of COFOG, which is to analyze by purpose the government policies carried out through government expenditure programs. The expenditures are coordinated through programs, offices, or bureaus that are the focal points for government budgeting. Each program integrates a series of alternative or supplementary expenditures that serve the same purpose or the same group of purposes. For instance, education can be promoted by running public schools (production of government services for own consumption), constructing school buildings (government gross fixed capital formation), “subsidizing” private schools (current transfers), or providing funds for private school construction (capital transfers, loans). Thus, it seems inconsistent with the aim of the functional analysis for the COFOG scheme to determine the function or purpose of each expenditure separately. Instead, the function of an outlay should be determined with regard to the function of the program as a whole.

Taking the above point of view does not mean that one has to accept that all outlays of one unit serve one function only. In the same manner that establishments produce characteristic and secondary products, a government unit may serve a primary purpose and a secondary one. If the secondary purpose is important and if separate records are available, there may be a need to try to split off those expenses that serve the secondary purpose and treat them together as the expenses of another COFOG unit.

It is not only convenient but also logical to use the same unit for COFOG as is used for the activity breakdown of government production and capital formation. The reason is that government production cannot be analytically separated from government consumption, since the major part of government services are assumed to be consumed by the same government unit. Furthermore, government consumption is only one of the expenditure types that should be looked at in a COFOG breakdown, in addition to other types of government expenses that are registered in the SNA income and outlay and capital accounts for the government sector. Production analysis in the government sector therefore cannot be separated from analysis of other government expenses. The connection of the two types of analysis is recognized in the SNA, which defines the establishment type unit in the government sector as a unit for which separate accounting records are available on production cost, sales, capital formation, and employment and which serve in principle one purpose (SNA, paragraphs 5.31–5.34).

Using the same unit of classification for classifying government “production expenses” by COFOG and ISIC categories does not imply that the two classifications are the same. For instance, if a ministry of education has one program for the construction of schools and another to run public schools, the two units should be allocated to different ISIC categories (construction and education) but included in one purpose category (education).

The modified orientation of the statistical unit in COFOG may also help to clarify some of the questions recently raised by the Statistical Office of the EC (EUROSTAT) with respect to revision of COFOG. The distinction suggested between administrative expenses of a specific nature (to be included in individual consumption) and those that are directed to more general purposes (to be included in collective consumption) is the same as distinguishing between expenses that serve the purpose of the transactor unit by which they are carried out and general administrative expenses that serve a different purpose and therefore should be integrated with another COFOG unit. Also, the question of classifying subsidies by COFOG category may be clarified; subsidies should not be classified by the purpose of the transaction itself or on the basis of the industry that receives them, but by the function served by the unit as a whole that makes the subsidy payments.

The second question concerns the institutional unit in the government sector.

Is there a need to clarify the definition of the institutional unit of government as a unit of decisions, control, and management, or should additional criteria be included, such as availability of separate accounting records on income and outlay and capital transactions or existence of a separate accounting or budget office, or should the unit be further restricted to be one that is funded for the majority of its expenses by its own revenues (taxes)?

The present institutional unit of government in the SNA is very broad (SNA, paragraph 5.68). The central government is taken as one unit, and various state, provincial, and local governments and individual social security schemes are each considered as separate institutional units. The units selected are those that are units of decision, control, and management.

For more refined analysis of government operations (by government subsectors, regional analysis, and the like) there is a need to have a further breakdown of the government sector, particularly of central government. Thus, it may be argued that the above definitions and coverage of government institutional units may be inadequate.

Government subsectors can only be distinguished if complete institutional sector accounts can be drawn up for the units included in each subsector. These sector accounts should cover not only expenses but also revenues. With regard to revenues, the government differs from other sectors. Its revenue generation is generally much more centralized, for example, than that of enterprises. Central government revenues usually are all channeled through a central treasury, making it impossible to consider ministries, for instance, as separate subsectors of government. There are government agencies, however, that keep separate accounts, and those are not restricted to the local government sector but also include units that generally carry out central government programs. In Latin America there are many decentralized government agencies for education, health, and so on that are separately administered and keep separate accounts.

But separate accounts do not mean that such agencies are necessarily financially independent. In fact, many at the central government level and some operating at the regional levels are funded to a major extent by central government transfers of tax revenues.

Development of these criteria for an institutional unit in the government sector might help to clarify some of the other questions raised here and assist in arriving at correct answers. For example, should the monetary function of the government be split off only if complete income and outlay and capital accounts can be set up for that function? Should social security funds, government pension funds, and welfare schemes be split off from other government units only if complete and separate accounts are available or are separately funded, or both?

III. Registration of Transactions

This section considers questions regarding registration of transactions in the national accounts: accrual versus cash accounting, consolidation practices, gross and net treatment, imputation and rerouting, and valuation in current versus constant prices.

Accrual Versus Cash Basis

How should accruals be defined for various government payments and receipts, and how should they be related to cash-basis statistics?

The SNA recommends that transactions should be recorded on an accrual basis, whereas the GFS adheres to cash-basis reporting. The accrual basis used in the SNA was defined before the IMF's GFSM defined the cash basis of related transactions in the GFS. Since then various queries have arisen with regard to the difference between the basis of recording the same transactions in the GFS and SNA. Within the SNA, questions have also arisen about what should be included in the capital finance accounts in the items “other accounts receivable and payable” and “trade credit and advances,” which reflect in the SNA the difference between the cash and accrual bases.

Particular attention should be paid to the difference between the recording in the SNA and GFS with regard to the following transactions.

Transactions in goods and services. These are to be recorded in the SNA as of the date when the legal title to the goods changes, or when the services are rendered. The only exception is capital formation in building and construction, where the transaction is deemed to occur as the work is put in place. According to the accrual principle, payments for capital goods may be made either before or after the date when the transactions should be recorded. The GFS recording of transactions in goods and services takes place when payment is made.

Compensation of employees. The SNA records it at the time it is earned, but the GFS records it at the time when payment is made.

Indirect taxes and subsidies. These are to be recorded in the SNA at the time of the production or sales transactions to which they relate; in the GFS, they are recorded when the taxes are received or the subsidies are paid by the government. A difficulty in interpreting the SNA rule occurs when tax authorities allow a period after the obligation arises during which the tax can be paid without penalty. Many enterprises take advantage of this grace period and delay their actual payments to the government as long as possible, although recording the tax in their accounts and including it in their prices at the time the obligation arises.

Property income, direct taxes, and other transfers to or from government. These transactions are to be recorded in the SNA when due for payment without penalty; in the GFS they are to be recorded at the time of government payment or receipt.

Transactions in financial claims. These are to be recorded in the SNA when the ownership of assets is transferred or when liabilities are incurred or liquidated. In the GFS, they are recorded at the time of government payment or receipt.

In considering the recording of the above transactions, guidance is necessary particularly with regard to whether the difference between the GFS and SNA criteria of recording should result in actual differences between what is included in each transaction item in the GFS and SNA, and if so how they should be calculated or estimated and perhaps explicitly recorded.

Additional questions arise when noncash transactions are registered on an accrual basis in the SNA but are not recorded at all in the GFS except in particular instances as memorandum items. Examples would be transactions in kind, assumption of debt, or debt cancellations. Explicit identification of such transactions is necessary for reconciliation of SNA and GFS concepts and data.


Should different consolidation practices be applied for different purposes in the SNA?

Consolidation consists of eliminating transactions between transactors within a group in the process of combining or aggregating the group's mutual transactions. In the SNA, transactions within the group are eliminated only if they appear in the same account for both transactors. Transactions within the group that appear in the production account of one transactor but in the income and outlay account of the other, such as indirect taxes, are not eliminated when operations of the two transactors are combined into a single set of accounts. This limit on consolidation is necessary because elimination of such intragroup transactions would alter input costs, value added, the output of government, and, hence, GDP. The same basic consolidation rules are followed in the central framework of the ESA.

In the GFS, however, and in the ESA supplementary accounting system for analyzing general government expenditure and receipts, the objective is not to measure government production, value added, or consumption but to measure the flow of revenues, expenditures, lending, and financing between the government sector and the rest of the economy. In the GFS and the ESA supplementary accounts for general government, therefore, all transactions between transactors within the group are eliminated in consolidation. This includes the additional elimination, beyond SNA rules, of intragroup payments and receipts for indirect taxes, subsidies, and government social contributions to itself that are paid as employer but are deemed to pass through the income and outlay account of households.

The United Nation's Handbook of National Accounting: Public Sector Accounts recognizes that such further consolidation may be warranted—for example, of payments and receipts of taxes and subsidies between government units—when the object of the analysis is not the measurement of gross output but the relationship between government and the rest of the economy. The particular flows to be measured as indicators of the relationship between government and the rest of the economy are not specified in the Handbook, however, and the extent of further consolidation warranted is therefore not detailed. Whether such key sectoral indicators of relations with the rest of the economy as tax revenues, total revenues, total expenditures, lending, and the overall deficit or surplus are to be included in the SNA or left to sectoral data systems, such as GFS and the ESA supplementary analysis, remains to be resolved. In either case, however, explicit identification of the additional flows—such as intra-governmental tax payments, social security contributions, and subsidies—eliminated in calculation of such key sectoral indicators, although not in calculation of the regular SNA accounts, would be useful as memorandum items to both the SNA accounts for government and the GFS and ESA supplementary analysis for general government.

Gross and Net Treatment

Should flows previously shown net in theSNAbe shown gross so as to increase information on actual flows?

When flows in opposite directions are different in nature or significance, there are advantages in showing each separately in addition to any net entry for the two entries offset against each other. In the GFS, as a general rule, all nonfinancing flows are shown gross with the exception of corrective transactions, such as refunds, and departmental enterprises' sales to the public, which are offset against corresponding operating costs.

Aside from the netting that occurs in consolidation, eliminating an intragroup transaction from the accounts of both transactors rather than offsetting against each other two transactions of the same transactor, few SNA entries are not shown gross. Entries previously shown net only for statistical convenience, such as transfers given and received, have been converted to separate gross listings in SNA questionnaires of recent years. Several concepts defined as net by nature, however, continue to be shown in the SNA without a separate listing of their gross components. These include net acquisition of fixed capital assets, which does not show separate entries for the gross acquisition or gross disposition of fixed capital assets, and the net acquisition of land and intangible assets. It includes also government final consumption, which is only shown net of any disposition of previously purchased goods registered in consumption at the time of purchase because government is deemed to carry no inventories other than strategic stocks.

The addition of separate gross entries could be useful in such cases. Thus, without prejudice to the net character of the concept of fixed capital formation as the increase in the government's stock of fixed capital assets during the period (before or after allowance for consumption of fixed capital), separate information on both the acquisition and disposition of fixed capital assets could be significant in portraying the nature of government operations. Separate information on gross flows of fixed capital assets or land may have little significance among households or enterprises, since most sales and purchases may represent both sides of the same transactions taking place within the sector. This would not be the case with government, however, for which acquisitions and dispositions of fixed capital assets or land represent significantly different activities with the rest of the economy.

A third category of flows that are not shown gross in the SNA is the capital finance account. Borrowing and amortization are not registered separately but only as net borrowing. There may be several reasons for this practice. Data may come from balance sheets, in which a combination of flows results only in changing a single entry for the stock. With liabilities of a short-term character, one year or less in maturity, the turnover can result in borrowing and amortization that exceed the total amount outstanding and have little independent significance. In other cases, the automatic rollover of obligations at maturity may cast doubt on the separate identity of borrowing and amortization.

Recognizing the validity of these reservations, there may still be merit in the separate identification of borrowing and amortization, excluding if feasible both short-term maturities and automatic rollovers. The burden of debt service—amortization plus interest payments—constitutes an important aspect of economic developments, both domestically and for the balance of payments. Projections of future amortization coming due and data for past amortization payments made reflect important elements of economic policy. Data on borrowing, moreover, indicate the extent to which the government finds it necessary to enter the capital market or approach potential lenders. Therefore, the addition of separate borrowing and amortization entries, perhaps for liabilities of more than a year, may be a useful means of providing such information.

Imputations and Reroutings

In this subsection questions with respect to four issues are addressed: separate identification of imputations and reroutings, depreciation of government fixed assets, rent on government-owned buildings, and debt cancellations as transfers.

Separate Identification

Separate identification of imputations and reroutings facilitates alternate presentations, but to what extent is it feasible?

For the SNA system as a whole, the consensus of previous expert group and working party meetings has been that, in general, the present imputations and reroutings should stay in the SNA, but that they should be identified separately in order to (1) facilitate the interpretation of the accounts, (2) permit a wider range of types of analysis, and (3) allow micro-macro links.

A document entitled “Imputations and Reroutings in SNA,” discussed at the SNA Expert Group Meeting on the Household Sector in September 1987, attempted a complete catalogue of all imputations and reroutings in the SNA and the repercussions in the accounts and tables of attempting to identify them all separately. The expert group concluded that it would overburden data publications if all cases of imputation or rerouting were shown, and therefore that only the most important should be identified. But the expert group also agreed that it is necessary to specify all the imputations and reroutings in the manuals for the benefit of both compilers and the most sophisticated users.

For the government sector, the SNA's imputations and reroutings are the origin of many of the differences between the SNA and the GFS and are therefore already itemized in Bridge Table II of the GFSM (pages 263–73). Apart from two general differences, one of coverage (net treatment of identifiable costs of departmental enterprises in the GFS) and one of recording (cash basis in the GFS),8 the imputations or reroutings that give rise to differences are as follows:

  • Employer contributions imputed with respect to unfunded employee pension and welfare benefits
  • Employer contributions to social security schemes at other levels of government (not consolidated in the SNA because routed through households)
  • Consumption of fixed capital
  • Cancellation of bad debts (capital transfer in the SNA and balance of payments)
  • Casualty insurance service charge separate from net premium
  • Maintenance and property tax components of rent received
  • Imputed interest with respect to household equity in employee pension funds' reserves with government
  • Stock valuation adjustment
  • Payments in kind (only memorandum items in the GFS)
    • —compensation of employees
    • —social security and social assistance benefits
    • —international cooperation (aid).

Few if any of these items would be considered important enough for the economy as a whole to warrant being identified separately in statistical publications; some are even trivial for the government sector. It would, however, be difficult to modify the SNA concepts to eliminate them.

One approach, which has been applied in the EC for the past ten years, is to specify some of these—the first three listed above as well as the subitems needed for consolidation—within the SNA (ESA) general government questionnaire, so that they can be eliminated to produce a budgetary presentation similar in intention, but not identical, to GFS. The other items listed above are treated in the EC's budgetary presentation in the same way as in the national accounts. EUROSTAT then publishes data in two forms: in accordance with the SNA (ESA) in the national accounts, and in the budgetary presentation in “General Government Accounts and Statistics.” It would be possible, although EUROSTAT does not currently do so, to add a table showing the transition from one form to the other.

Depreciation of Government Fixed Assets

For what government fixed assets should consumption of fixed capital be calculated?

Consumption of fixed capital (or depreciation) in the present SNA is calculated for all assets that have finite lives. These include buildings, plant, machinery, and vehicles but exclude roads, dams, and bridges, which are assumed to last forever with normal maintenance.

As experience suggests, however, no matter how well the roads, dams, and bridges may be maintained, they do become obsolete and so have finite useful lives. In these circumstances, the present SNA rule will give a false picture of the total value added generated in the government sector. This view was specifically stressed in the Economic and Social Commission for Asia and the Pacific (ESCAP) Seminar on the Review and Development of National Accounts (Bangkok, July 1–7, 1986), which recommended that capital consumption estimates should be calculated for public goods such as roads.

Rent on Government-Owned Buildings

Should rent on government-owned buildings be imputed?

The present SNA does not include imputations for rent on government-owned buildings or on buildings owned by nonprofit institutions. The ESCAP seminar noted this as a serious omission, which would result in underestimation of the contribution of the government sector to GDP. This is particularly so because the contribution of this sector to GDP is based on cost, which includes only intermediate consumption, depreciation, and compensation of employees, and does not include any operating surplus that would have resulted from valuation of gross output in the market. Inclusion of imputed rent on government-owned buildings, based on market value, would imply that GDP originating in the government sector would include an operating surplus in addition to depreciation and compensation of employees paid out to government personnel.

Debt Cancellations as Transfers

Should debt cancellations be treated as transfers and, if so, when should they be registered?

The GFS does not include debt cancellation because it involves a noncash transaction. The SNA now treats debt cancellation in the flow accounts; that is, as a reduction in the value of an asset or liability with a counterpart flow of current transfers.

It has been argued that the present SNA treatment is inconsistent with the rules regarding the reconciliation accounts, which should include all changes in the value of assets and liabilities that are the result of price changes or the creation or destruction of assets rather than the consequence of current production or income generation. Debt cancellation clearly falls in this category and should therefore be dealt with in the reconciliation accounts. This would also be in line with the recommendations of the Expert Group Meeting on the SNA Structure (Geneva, July 1986) that capital gains and losses and other changes in the value of assets should not be included in the flow accounts.

Valuation in Current and Constant Prices

The treatment of transactions in kind and of deflation to real terms are discussed in this subsection.

Transactions in Kind

How should transactions in kind be valued—at cost, or at prices for comparable marketed goods and services?

Governments commonly engage in two types of transactions in kind. First, as for any other employer, the government's wage and salary bill may include some payments in kind—free food and lodging, for example. Second, and quantitatively more important, governments typically provide a wide range of free education and health services. The SNA, in common with all national systems of accounts, values these transactions in kind “at cost.” Thus, free meals provided to government employees, or free schooling provided to the general population, are valued at direct production costs: as the sum of compensation of employees, consumption of fixed capital, and intermediate consumption.

It can be argued that valuation at cost understates the intrinsic value of these transactions in kind in the sense that, if they were provided on a market basis, their valuation would also include a profit margin. It might therefore be appropriate to revalue these in-kind transactions at prices that include an imputed profit—either by using the prices of comparable marketed goods and services or by directly imputing a profit margin calculated, for example, as some average rate of return on capital employed.

Deflation to Real Terms

How should final consumption expenditure and value added of government be calculated in real terms?

Most countries presently calculate all government value added and final consumption expenditure in real terms by some kind of “input method.” This involves revaluing cost components—mainly compensation of employees and intermediate consumption—at base-year prices, either by deflating current price values by wage and price indices, or by extrapolating base-year values by volume indices. The basic objection to input methods is that they do not reflect changes in productivity.

In 1975 EUROSTAT published a report on the measurement of non-market services at constant prices, which recommended that government services that are supplied to individuals, such as health and education services, should be calculated at constant prices by the use of output measures.9 Examples of such measures include pupil-hours of instruction, numbers of medical treatments supplied, and the number of social security transactions handled. The EUROSTAT report recommended that for “pure public services”—such as defense and most public administration—values at constant prices should be measured by reference to labor inputs, but that inputs should be measured in person-hours of duty and that different types of person-hours should be weighted by total cost per person-hour, including the cost of equipment and consumables used. In discussions of the EUROSTAT report in Luxembourg and at the OECD, virtually all national accounts experts agreed with its main recommendations, which were subsequently included in the United Nations' Manual on National Accounts at Constant Prices.10

In practice, very few countries have adopted output measures of this kind. Thus, whether these recommendations may need to be modified in the light of country experience may need further consideration, especially as regards the three output measures described above.

IV. Analytical Framework

Three analytical issues are considered in the questions raised below: the distinction between current expenditures and capital formation, saving versus the overall deficit or surplus, and the operating surplus of government.

Distinction Between Current Expenditure and Capital Formation

Should military expenditures classified as capital formation be restricted to dependents' housing or extended to include military hospitals, structures, or durable equipment?

According to the SNA “Blue Book,” government outlays with respect to construction works and other durable goods intended for defense purposes are all classified as current expenditures and not as capital formation.

However, outlays by producers of government services on the construction or alteration of family dwellings, but not military barracks for personnel of the armed forces, are classified as gross fixed capital formation. The distinction between family dwellings and military barracks is based on the type of housing provided; family dwellings are facilities that are similar to the dwellings normally used by civilians. In contrast, the construction of schools, hospitals, airfields, or roads for use by the armed forces is classified as intermediate consumption even though these facilities might be put to civilian use. This is also the case for motor vehicles used for military purposes.

The transfer of such facilities to civilian purposes at a later date constitutes an addition to the stock of fixed assets. The value of the assets when transferred to civilian uses plus other outlays made on the conversion of schools, hospitals, motor vehicles, and the like to civilian use are to be included in gross fixed capital formation. The counterpart entry is a reduction in the intermediate consumption of the relevant government service.

In many countries, assets are used for mixed civilian and military purposes; examples are airfields, roads, hospitals, and the like. Because the SNA does not address this issue, many have argued that the present guidelines should be clarified, possibly toward some extension of what is now included in gross fixed capital formation. This discussion is reflected in the United Nations' Handbook of National Accounting: Accounting for Production: Sources and Methods, which proposes to include certain items of a clearly nonmilitary character in government fixed capital formation, even if they are financed out of military budgets.11 These are family-type housing, schools, hospitals caring for civilians as well as military personnel, and highways, port facilities, and airports, if they are not limited to military use. Because this treatment is different from that recommended in the SNA Blue Book, guidance is needed.

Saving and Overall Deficit or Surplus

Should theSNAinclude two major balancing items that are important for analysis of the government sector, and are therefore included in the GFS, but are unnecessary for analysis of the economy as a whole or inapplicable in other sectors?

This question is broken into further questions pertinent to saving and the overall government deficit or surplus.


Should theSNAfor the government sector, although not for other sectors, recognize, in addition to the concept of government saving, representing the portion of all current income remaining after current outlays, the concept of the government's own saving, comprising the portion of current income other than grants from governments and international organizations remaining after current outlays?

The net saving of government, as for other sectors, is calculated as the residual in the income and outlay account. That is to say, government saving is calculated as the portion of current income remaining after current outlays, excluding capital and financial transactions from both income and outlay. As distinct from other sectors, however, in the case of government significance is found also in the measurement of saving excluding grants received from other governments or official international organizations. This concept is referred to in the GFSM as own saving. It is used to reflect the special position of grants (that is, official transfers) in government operations. Because such grants are sometimes provided as a form of budgetary support or to meet a deficit that would otherwise ensue, there is a need for measures of budgetary performance that exclude the effects of the grants themselves. It is for this reason that such official grants are identified in a separate category in the GFSM, permitting alternate calculations of both the overall deficit and of saving. Recognition of the alternate concept of government's own saving would facilitate such analysis in the SNA as well.

It must be recognized, however, that this concept is not applicable to other sectors. The distinction between the continuing, ordinary flow of tax and property income to government and the extraordinary, politically determined flow of official grants to government finds no easy parallel in the household or enterprise sectors. Nor is own saving a concept that preserves symmetry between grant-receiving and grant-making governments. While own saving is lower than saving for the receiving government, own saving is not higher than saving for the government making the grant, since current transfers paid, whether to governments or to others, would not be excluded in calculation of own saving. Grants between levels of government within a country would in any case be eliminated in consolidation of general government as a whole. It is in the measurement of the performance of individual governments, or of general government insofar as there are official grants from abroad, that the concept of own saving offers an additional measure useful in the analysis of government.

Overall Deficit or Surplus

Should the concept of government's overall deficit or surplus, met through the net incurrence of liabilities plus the net decrease in holdings of currency and transferable deposits and in holdings of any other financial assets acquired for liquidity management purposes, be introduced into theSNAfor the government sector, although not for other sectors?

The single measure of government performance most frequently cited in analysis and policy discussions is the overall deficit or surplus, which represents the result of current, capital, and lending operations and the consequent government financing requirement. Because it has no close parallel in other sectors, the overall deficit or surplus is not explicitly defined in the SNA and is not generally derived from SNA data on government. Whereas the SNA saving concept measures the government's current account deficit or surplus, the SNA net lending12 concept represents the balance drawn after government capital receipts and expenditures, in the capital accumulation account, but not after government lending, which is included in the SNA capital finance account.

Government lending is included in determination of the government's overall deficit or surplus, however, because the government, unlike other sectors, generally lends and acquires equity not for purposes of liquidity management or earning a return but to promote public policy objectives. In this respect, government lending is like government spending, in that whatever portion cannot be met out of government revenues, grants, or repayments of previous government lending affects the deficit or surplus and requires government borrowing or the drawing down of previously accumulated cash balances. From a policy and analytical point of view, therefore, government lending is not symmetrical with either government borrowing or the borrowing of those receiving government loans, since borrowing is undertaken for liquidity management purposes. Although this asymmetry requires care in the consolidation of two governments that have lent to or borrowed from each other, it is an essential element in the measurement of government operations in accordance with the nature of such operations. The overall government deficit or surplus concept makes it possible to measure the extent to which government spending and lending undertaken to carry out the government's policy objectives are covered by the government's tax and nontax revenues, grants, or repayments of previous government lending.

In general, the overall government deficit or surplus is met through the net incurrence of government liabilities plus the net decrease in government holdings of currency and deposits. That is, a government deficit would normally be met from government borrowing less amortization and any drawdown of government balances. In general, this would permit calculation of the overall government deficit or surplus in the SNA as the balance drawn in the capital finance account before the net incurrence of liabilities and changes in holdings of currency and deposits. In some instances, however, governments hold assets other than currency and deposits for liquidity purposes. Sinking funds may acquire securities to match the maturities and currencies of their liabilities; social security funds may acquire financial assets for liquidity management rather than public policy purposes; and local governments may invest their temporary excess cash in the securities of higher levels of government or of corporations. Such government acquisition of financial assets, other than cash and deposits, for liquidity management purposes is usually minor in magnitude but can be quite significant. U.S. state and local government holdings of Federal government securities, for example, total about US$100 billion. Were such acquisitions for liquidity management purposes counted as a part of lending, they could seriously distort the meaning of the deficit or surplus. Accommodation of this distinction is achieved in the GFS in accordance with guidelines set out in the GFSM. Provision of a parallel distinction when required in the determination of a deficit or surplus concept in the SNA would maintain the meaning of the concept and avoid the occasional occurrence of disparate concepts for the deficit or surplus in the SNA and GFS presentations.

To accommodate instances in which government net acquisition of financial assets other than currency and transferable deposits occurs for liquidity management purposes, it would be necessary to insert an adjustment item above and below an overall deficit or surplus line for government in the SNA capital finance account. It might read “Less: financial assets (other than currency and transferable deposits) acquired for liquidity management purposes” above the overall deficit or surplus line, and “Financial assets (other than currency and transferable deposits) acquired for liquidity management purposes” below the line.

As with the consumption of fixed capital item appearing with opposite signs in the SNA production account and capital accumulation account, insertion of this adjustment item would permit calculation of the appropriate balance within the accounts. Because procedures are already carried out for the derivation of such numbers for GFS presentations in some 130 countries, determination of the amounts involved for the SNA compilation should not pose significant difficulties.

The cash-based GFS concept of overall deficit or surplus has become a widely accepted standard for analysis of government operations. There may be both advantages and disadvantages, therefore, in adding an SNA-based concept of overall deficit or surplus, which may yield a result differing from the GFS-based concept and could instill ambiguity in the analysis of government operations.

Government Operating Surplus

Should the value of government output be calculated to include an element of interest payments in addition to other cost items of government inputs so as to avoid undervaluation of government output and approximate an operating surplus element in the value added of government services?

Gross output of government is defined in terms of costs, including wages, depreciation, and the purchase of goods and services, which is called intermediate consumption. Interest payments are not considered as costs in this sense because they reflect the price of borrowing money, and interest on the government debt is therefore excluded from the contribution to GDP of the government sector.

It has been suggested that at least part of the interest payment on government debt should be included as costs so as to avoid undervaluation of government output and thus approximate an operating surplus element in the value added of government services. This addition to the value added of the government sector is particularly important in order to avoid the distorting effects of the present SNA, which suggests that government output is to be valued at cost because of the lack of information on the market value of government services.

V. Classification

This section addresses possible revisions to various SNA classifications: taxes, property income, social assistance to households, and the functions of government.


Indirect taxes, estate and gift taxes, social security contributions, and fees paid to the government are considered in the following paragraphs.

Indirect Taxes

Should indirect taxes be restricted to commodity taxes only, thus eliminating the need to distinguish between tax payments by business and others?

Commodity taxes are indirect taxes on commodities (that is, goods and services) or activities of an industry that are proportional to the quantity or the value of commodities produced or sold by the industry. These commodity taxes are included in indirect taxes, which in addition cover noncommodity taxes. The latter are payments by business of regulatory and administrative fees, such as court fees, business license fees, or airport taxes, that are treated as current transfers if paid by households. Noncommodity taxes also include real estate taxes if these are not levied as a replacement of income taxes.

The GFS includes a detailed breakdown of tax revenue but does not identify indirect taxes. It also does not distinguish between payments made by households and those made by business. As a result, GFS includes in some of its tax categories payments by households that are treated in the SNA as transfers (for example, motor vehicle tax) and includes in some of the nontax categories payments by business that are treated in the SNA as indirect taxes (for example, court fees).

In practice, countries do not usually make the distinction of tax payments by business or nonbusiness units when responding to the SNA. In most instances, they allocate the total tax payment of one category to either indirect taxes or current transfers, depending on the major component of the category in question. In addition, analysis of national accounts data shows that the difference between total indirect taxes and commodity taxes is very small.

Because of practical considerations and in order to facilitate the link with GFS, it has been suggested that indirect taxes be restricted to commodity taxes only. If this restriction of indirect taxes were accepted, the present noncommodity indirect taxes might be treated in the SNA as direct taxes or as other current transfers, depending on their nature.

Estate and Gift Taxes

Should estate and gift taxes, inheritance taxes, and nonrecurrent taxes on property be classified as taxes in theSNAand, if so, should they be classified as current?

The present SNA classifies estate and gift taxes and nonrecurrent taxes on property not as taxes but as capital transfers to government. In contrast, the Fund's GFSM, the OECD's Revenue Statistics,13 and general usage in fiscal analysis classify such taxes as taxes and classify all tax revenues as current revenue of government. Although the magnitude of such taxes is relatively small—constituting less than 1 percent of central government tax revenues in almost all countries—this represents a significant conceptual difference between the SNA and GFS as regards tax revenues, current revenues, and saving.

Governments levy estate and gift taxes in order to raise revenue and, in some instances, to reduce inequality or help shift property to more productive uses. Although occasional to the taxpayer, such taxes are recurrent to the government, which receives a continuing flow of such payments and regards them, as it does other taxes, as current, regular, tax revenue. No connection is made in government policy between estate and gift tax receipts and government capital expenditure, and efforts are not made to earmark such receipts for particular purposes. There is no perception in government that estate and gift taxes are capital in nature or different from other tax receipts.

The perception in the household sector matches this government view of estate and gift taxes. They are viewed, as are other taxes, as a cost that must be met by a tax payment to government at specified events or occasions with no expectation that it will be used differently from other taxes. What distinguishes estate and gift taxes from other taxes is the irregular transmittal of wealth, at death or before, that occasions their payment, usually out of accumulated wealth rather than current income. When the cost of estate and gift taxes is met out of accumulated wealth rather than out of income, its payment constitutes dissaving—payment of a cost in an amount exceeding current income and resulting in either a decrease in assets or an increase in liabilities. A part of the value of a taxed estate, or of the proceeds of its sale, is viewed as unfortunately used up, lost, or “consumed” in payment of the tax to government. Payment of the tax represents a loss of previously accumulated household savings, and this is most appropriately represented as involuntary, unavoidable, household sector dissaving. If payment of estate and gift taxes brings no corresponding government capital formation, this dissaving of the household sector becomes the dissaving of the consolidated economy as a whole.

The present SNA, by classifying estate and gift taxes as capital transfers, shows the same dissaving for the consolidated economy as a whole but attributes this dissaving to government, rather than to the household sector. Given the prevailing government perception of estate and gift taxes as an indistinguishable part of total government tax revenues, with no particular tie to capital expenditures, an increase in government estate and gift tax receipts, under the present SNA calculation, is likely to lead to an increase in government dissaving. The policy implications of such a classification would tend to discourage government use of such taxes because of their negative impact on one measure of government operation—saving—although more careful analysis of the economic, social, and revenue effects of such taxes may point in the opposite direction. For household sector dissaving results not only from taxes on wealth, but also from any level of taxes that households do not meet out of current income and find it necessary to pay for by drawing down previously accumulated wealth or by borrowing.

Thus, the classification of estate and gift taxes as taxes, and their removal from consideration as capital transfers, would more closely approximate the concepts generally accepted for the measurement of government by those carrying out or analyzing government operations, as well as by the taxpayers involved.

Social Security Contributions

Should social security contributions be included in total taxes?

Social security contributions are levied on payrolls, or on a proxy for payroll in the case of the self-employed, and are earmarked for operation of a system of benefits to which the contributor, by virtue of the contribution, is admitted. Such contributions represent as much as 30 percent to 50 percent of total nationally collected government revenues in many developed countries, but less than 10 percent in most developing countries outside Europe and parts of Latin America. They are vital, therefore, to a complete portrayal of government finances.

Because they are associated with admission to a benefits system, though not vested and returned like provident fund or retirement fund contributions, social security contributions may be more readily acceptable than other tax payments. Their separate identification is often held to be important, moreover, to ensuring their dedication to social security purposes. Reflecting these characteristics, social security contributions are not classified as taxes in the SNA. Employer contributions are deemed to be routed through households as compensation of employees and simultaneously paid to government by households, whereas employee contributions are collected from households.

In the OECD Committee on Fiscal Affairs (Working Party 2), after several years of deliberation on the classification of government revenue, and in regional and country discussions preparatory to publication of the GFSM, it was determined that the balance of advantages lay with classifying social security contributions as a separate category within the overall total of taxes. Given the definition of taxes as compulsory, unrequited, nonrepayable payments to government, social security contributions, other than the exceptional voluntary contributions provided for in some systems, are classified as taxes. To preserve the distinctiveness of social security contributions, they are separated from payroll taxes that are not devoted to social security. They are also distinguished from levies collected on a personalized income tax base, rather than on payroll, although they may be identified with social security, and from taxes on other bases, such as sales taxes, earmarked for social security purposes in some countries. In both the OECD Revenue Statistics beginning in 1973 and the Fund's GFS Yearbook beginning in 1977, social security contributions are separately identified as a category of taxes levied on a payroll base and devoted to social security purposes. The resulting portrayal of total taxes explicitly includes social security contributions but identifies them separately for the analytical purposes noted above. There would appear to be some advantage in maintaining such a separate identification of social security contributions within an overall total of taxes in the SNA.

Fees Paid to Government

Should fees paid to government, whether by business or others, be classified as payments for government services rather than as taxes, except for fees out of all proportion to the cost or distribution of government service provided to the payer?

At the border between taxes and sales lies the difficult-to-classify realm of administrative fees and charges. The SNA (paragraphs 7.65 and 7.66) classifies fees paid to government by households as direct taxes

… if they are not directly connected with the provision of non-regulatory service by the public authorities, and are primarily designed to raise general revenue. . . . However, if the service furnished is regulatory in character, the fees which households pay for the service…are classed as compulsory fees since they are obligatory and unavoidable in the only circumstances in which they are useful. . . . The compulsory fees and duties are classed as indirect taxes when paid by producers.

In practice, these distinctions have been conceptually difficult to understand and difficult, if not impossible, for governments to apply. Practice has varied widely, and attempts to reach consensus among practitioners on the application of these criteria to specific administrative fees have encountered opinions that are sharply divided.

In these circumstances, extensive discussions were held over a period of years in the OECD Committee on Fiscal Affairs (Working Party 2) with a view to clarifying, at the borderline, the distinction between tax and nontax revenues of government. The conclusions reached are now reflected in both the OECD Guidelines and Revenue Statistics and in the Fund's GFSM. Taxes are defined as compulsory, unrequited, nonrepayable payments to government and as any collections of fees and charges out of all proportion to the cost or distribution of government service provided to the payer. Classified as taxes are business and professional licenses and, reflecting prevailing administrative practice among governments, taxes on permission to hunt, shoot, or fish, taxes on the ownership of dogs, and radio and television licenses, unless the public authorities provide general broadcasting services.

Nontax revenue, therefore, includes fees for the provision of a service, whether compulsory fees for provision of a regulatory service, such as driver's licenses, passports, or court fees, or voluntary charges for the provision of a nonregulatory service, such as museum admission fees, or school or hospital charges, unless such fees are out of all proportion to the cost or distribution of the government service provided to the payer.

In the application of this classification, a determination of whether a service is regulatory or nonregulatory or whether it benefits the payer or others is not necessary. Should it be found desirable to assign consumption of the services rendered in exchange for the fees to business or households, the distinction presently applied in the SNA to payment of fees could be continued for the more broadly defined category of fees, with treatment of the business and nonbusiness-paid components classified appropriately.

Property Income

In this subsection, questions are addressed regarding withdrawals of entrepreneurial income from quasi-corporate enterprises, operating deficits of departmental enterprises, and indexation payments on debt.

Withdrawals of Entrepreneurial Income

Should the category of dividends received by government be combined with withdrawals of entrepreneurial income from quasi-corporations and, in the case of GFS, with the cash operating surplus of departmental enterprises, which would thus no longer be treated separately from corporate nonfinancial public enterprises?

In the income and outlay account, the SNA includes as receipts the operating surpluses of departmental enterprises and withdrawals of entrepreneurial income from quasi-corporations; it includes interest, dividends and land rent, and royalties as both receipts and disbursements. Dividends are defined as those received only from public and private corporations (and also from cooperatives), whereas withdrawals from entrepreneurial income cover actual payments of property income made to government by quasi-corporations.

The GFS lists on the receipts side only two property income categories—cash operating surpluses of departmental enterprise sales to the public, and other property income—and on the expenditure side only cash operating deficits of departmental enterprise sales to the public and interest payments.

In practice, countries often do not make the distinction between dividends received by government from public corporations and withdrawals of entrepreneurial income from public quasi-corporations. The latter are frequently registered as dividends received.

If departmental enterprises were included as quasi-corporations in the SNA and as public enterprises in the GFS, and all own account production of capital goods were separated out as a public enterprise or quasi-corporate activity, this would eliminate the present distinction between the operating surplus of departmental enterprises and withdrawals from the entrepreneurial income of quasi-corporations. Both would then be treated as withdrawals and could be combined with dividends as suggested above.

Operating Deficits of Departmental Enterprises

Should current transfers to quasi-corporate enterprises and to cover the operating deficit of departmental enterprises be shown as subsidies in the SNA?

Current government transfers to enterprises are classified as subsidies, which by definition permit market price to vary from factor cost. Several elements have complicated the application of this principle to departmental enterprises, however: their close relationship with government, determination of price, and inadequate data.

Because departmental enterprises are owned by government and operate within government administration and accounts, no explicit government transfer permitting sale at lower prices is generally identifiable. Instead, prices set below operating costs are reflected in an operating deficit that is met by government as an explicit subsidy payment would be. Although an operating surplus reaching government would be registered as entrepreneurial or property income, the dual role of government as owner and provider of public policy subsidies would appear to offer the alternative of classifying operating deficits as either subsidies or negative withdrawals from the entrepreneurial income of quasi-corporate enterprises. Given the basic public policy orientation of government, including its responsibility for the operation of departmental enterprises, classification of the operating deficits of such enterprises as subsidies would appear to be preferable.

When departmental enterprises sell to other parts of government at prices set administratively without regard to market price or costs, calculation of an operating deficit or surplus does not yield a realistic value of either subsidy or property income or of the enterprise's output or value added. In these circumstances, integration of such departmental enterprise activity in the production account of producers of government services, with no calculation of an operating deficit or surplus, would yield a more valid measure of government and enterprise activity, as discussed elsewhere in this chapter.

The operation of departmental enterprises within government administration and accounts frequently makes it difficult to measure the consumption of fixed capital within operating costs and, hence, the operating deficit or surplus. Thus, in the GFS, a cash operating deficit or surplus for departmental enterprises' sales to the public is calculated that does not take into account consumption of fixed capital, changes in stocks, or changes in accrued trade credits outstanding. Any resultant cash operating deficit is classified in the GFS as a government subsidy, and any cash operating surplus is classified as entrepreneurial or property income received by government. Where data for consumption of fixed capital, changes in stock, and changes in accrued trade credits are difficult to obtain, they may have to be estimated so as to calculate the fully accrual-based measure of output, value added, and the operating deficit or surplus for the SNA. This would be true in any case, whether departmental enterprises selling to the public are classified within government or in the nonfinancial public enterprise sector.

Indexation Payments

Should indexation payments on debt be classified as interest paid for use of capital or as amortization paid for repayment of capital in real terms?

Although the issue of classifying indexation payments as interest or amortization goes well beyond the public sector accounts, it may be useful to note its major implications in the analysis of government.

Government debt instruments are usually of considerable magnitude in countries that find it necessary, because of price inflation, to index their liabilities. At present, indexation payments are in general classified as interest payments, both in GFS and in the SNA, where they would consequently appear in the income and outlay account rather than in the capital finance account. This classification results in lower totals for saving and net lending for the paying sector, although there is no change for the economy as a whole because debt payments abroad would usually be denominated directly in foreign exchange rather than being indexed to domestic prices. Correspondingly, classification of government indexation payments as interest rather than amortization increases a government's deficit.

Several arguments have been cited in discussions about the classification of indexation payments as interest or amortization.

•The inflation component is not separated out and treated differently for payments other than for debt.

•All adjustments for inflation should be reserved for the revaluation account.

•Because repayment of debt denominated in foreign exchange is classified as amortization, classification of indexation payments on debt denominated in local currency as interest could be viewed as inconsistent.

•Although indexation payments formally separate out the explicit inflation component, an implicit inflation component is included in interest in the absence of indexation and is treated as interest.

•Because the implicit inflation component in normal interest payments may be viewed as a return to capital, it should be classified as amortization, with a consequent reduction in the deficit of the paying government to reflect only real interest payments.

•Because a primary purpose of the compilation of both the GFS and SNA is to understand, relate, and forecast behavior and its impact on the economy, determination of the appropriate classification of indexation payments should be based on the behavior of those receiving indexation payments. If lenders in an inflationary environment understand their indexation payments to constitute a return to capital to be reinvested for maintenance of their capital's value, classification of such payments as interest in the income and outlay account would tend to overstate income and saving.

Whatever the merits of these arguments, classification of indexation payments remains a contentious issue for the few countries that index debt payments. The implications are substantial for the presentation of government accounts and for treatment of the implicit inflation component in normal interest payments. Consistency of treatment will require a determination of appropriate practice.

Social Assistance to Households

Classification of two forms of social assistance to households—food coupons and welfare work assignments—is considered in the following paragraphs.

Food Coupons

Should the issuance of coupons to households for use in purchases of food or other commodities, with subsequent redemption by government, be classified as subsidies or social assistance grants to households?

Government issuance and redemption of coupons, such as food stamps, to households raises several statistical questions.

•If households are allowed to choose some of the products for which the coupons are to be used, should the overall operation be viewed as a government purchase of goods and services or as an unrequited government transfer? Proposed establishment of a category of individual consumption financed by sectors other than households would appear to affect this question in a way similar to that of government reimbursement for medical costs.

•If such operations are classified as transfers, should they be regarded as subsidies to enterprises or as social assistance grants to households? Subsidies are defined as all unrequited current transfers to enterprises that, by definition, permit them to charge lower prices for their sales. It is for this reason that subsidies are added to the market price to arrive at factor cost. That enterprises are required to provide households with goods and services in exchange for the coupons makes the payments the enterprises receive from the government requited, rather than unrequited, from the enterprises' point of view. Hence, such payments would be inappropriate for classification as subsidies permitting market prices below factor costs.

•Although cash-basis GFS data can register the operations only when payment is made to enterprises redeeming the coupons, should accrual-based SNA data register the operation when the coupons are issued to households, when the households use the coupons to make purchases and receive delivery, when enterprises present their claim for payment from the government, when such payment is due some interval after presentation for payment, or when payment is actually made? The answer depends at least in part on the actions in the household or enterprise sectors that registration of the government's operation is meant to parallel and portray.

Welfare Work Assignments

Should payments to welfare recipients that require performance of work assignments be classified as social assistance grants or as payments for factor services?

In some instances, to prevent malingering, to introduce welfare recipients to the work ethic or to particular kinds of work, or to provide particular services to the community, recipients of government welfare payments are required to perform work assignments. Whether such work should be included as an addition to GDP, and its performance classified as provision of factor services, may depend on whether it is viewed as contributing primarily to improving the character, motivation, or skills of the individual performing the work, or to the remainder of the community, which could be regarded as consuming the product. With the growing reform of welfare systems in some countries to include such work requirements, guidelines for their consistent classification take on greater importance.

Measurement questions arise also from the symmetrical, but opposite, phenomenon of government employees hired for income-maintenance purpose without the performance of work assignments. These questions are usually resolved by adding the value of such employees' wages and salaries to GDP, since government output, in the absence of sales proceeds, is valued as the sum of its inputs.

Functions of Government: Revision of COFOG

COFOG was produced in the late 1970s, published in 1980, and is now applied by many countries. COFOG replaces SNA Table 5.3. It is essentially a United Nations classification, but it has also been adopted as it stands in the Fund's GFS and in the ESA. Data on the breakdown of government expenditure by function are much in demand and widely used.

Although COFOG was in many ways an improvement over SNA Table 5.3, applying it has highlighted some problems that should be examined in the SNA revision process. They relate to (1) the treatment of administrative, regulatory, and research activities; (2) the functional allocation of subsidies; and (3) relationships with the ISIC. These issues are discussed in the following subsections.

It may seem unfortunate to be proposing changes for such a classification so shortly after its introduction, but it should be borne in mind that the SNA revision will only affect data in the late 1990s. Therefore, if there are problems with COFOG, the sooner they are identified the better.

Treatment of Administrative, Regulatory, and Research Expenditures

COFOG differs from the former SNA Table 5.3 in that COFOG distributes all expenditure on general administration, regulation, and research among the headings of the classification served, down to the most detailed three-digit level.

First, and most important for the revision of the SNA, this practice creates difficulties in attempting to isolate individual consumption expenditure of general government to add to private consumption in the calculation of total or enlarged consumption of the population. The reason is that government administration is of two types:

  • Local administration within units that produce services (hospitals, for example) that could be rendered by comparable private sector units (this expenditure should be included in the value of the services considered as individual consumption in order to make valuation as comparable as possible)
  • General administration, regulation, and research that cannot be found in private sector institutions providing individual services (this expenditure should be considered collective consumption).

Functional Allocation of Subsidies

Interest in this question first arose from proposals to include the value of certain subsidies in total consumption of the population, but the question also seems to have more general aspects.

For calculating total consumption of the population, some exercises have added the value of certain social or consumption subsidies to the values, according to SNA, of household consumption and the “individual” consumption expenditure (mainly education and health) of general government and private nonprofit institutions. The reason is that these “social” subsidies are considered to have an effect similar to social benefits or direct provision of the good or service by government.

In fact, for calculating a total consumption aggregate within the main SNA accounts, the Expert Group Meeting on the Household Sector in September 1987 concluded that consumer subsidies should not be introduced as a form of final expenditure; instead, a detailed classification of subsidies by type and purpose should be developed and included as part of the SNA analysis of government expenditure. Even if adding certain subsidies to consumption in future only has the role of an alternative analysis, greater clarity of the allocation of subsidies by function is desirable.

COFOG, which is used to delimit the individual consumption expenditure of government, makes special mention of several consumption subsidies, moving them from the “economic” to the “social” categories. For example, under subgroup 07.11, “Housing affairs and services,” it says that

Subsidies, grants or loans for increasing, improving or maintaining the housing stock other than rent subsidies paid to households are considered a form of income assistance and are classified to subgroup 06.15 (Family and child allowances) or 06.16 (Other social assistance to persons) as appropriate.

The precise meaning is not clear because the terminology is rather loose, since strictly a subsidy can only be paid to a market producer unit. If one retains the words “paid to households,” then the text is saying either that “rent subsidies paid to households” should be considered as social benefits, not as subsidies at all, or that they should be restricted to subsidies to owner-occupiers to “charge” themselves a lower imputed rent.

Slightly clearer is the section on “Distributive trade affairs and services including storage and warehousing” (13.11), which says that

Food and other subsidies applicable to particular population groups or individuals (for instance, those applied to milk for babies) are considered welfare and are classified to the appropriate subgroup of 06.

The inclusion of this sentence under the distributive trades function may imply a deliberate intention to restrict this treatment to subsidies that are on goods, not services, and that are paid at the distribution stage. Alternatively, it may be intended to represent a principle applicable to other areas.

From the text of COFOG, one cannot tell whether its authors, in mentioning these cases, were simply asserting the functional nature of the classification or were anticipating a definition of consumption subsidies based on the social functions in the same way as on individual consumption. In the latter case, the problem is to determine exactly which subsidies should be singled out for such treatment, or whether a general definition of them could be given.

A problem does, however, arise in applying this solution when one comes to compile total consumption not just in aggregate but by function. For example, to show total consumption of the population on, say, transport, one needs to bring transport subsidies back out of welfare.

In fact one can ask quite generally what is the proper functional classification of a subsidy: does it depend on (1) the ministry that pays it, (2) the industry that receives it, or (3) some perhaps more subjective assessment of its purpose (for example, welfare)?

Surely the aim and spirit of a functional classification favors the last of these. Classification by receiving branch already exists elsewhere in the national accounts; classification by the ministry that pays may also be interesting for budgetary analysis, but it is not “functional.” The function served is, however, not unambiguous and may ultimately be rather a matter of subjective judgment. Subsidies do, in general, bring benefits to both producers and consumers as well as in some cases to producers' employees, underdeveloped or declining regions, and so on.

This problem is recognized in the introduction to COFOG section I.C (and repeated in the GFSM on page 144), which attempts to make a distinction between purpose and function by means of certain examples. For example, a subsidy to the shipbuilding industry may have as its purpose maintaining facilities to build warships in time of need, but its function, according to COFOG, is manufacturing affairs and services.

Relationship with the ISIC

COFOG stipulates that “the units of classification are, in principle at least, individual transactions,” but that for many types of outlays related to production of goods and services “COFOG codes will have to be assigned to agencies, offices, program units, bureaux and similar units within government departments” (pages 2-3). Exactly what those units might be in practice is not clear.

The SNA (paragraph 5.31) states that they are establishment-type units and recommends that they be the same as for the activity (ISIC) breakdown of producers of government services. This, it concludes, would allow “transferring data classified according to kind of economic activity of the producers into data classified according to the purpose they serve.”

Even if the statistical units are the same, however, the structure and the detail of the two classifications (ISIC and COFOG) hardly seem to have been developed with this aim in mind. Some detailed improvements have been made in the recent revision of ISIC (rev. 3); for example, military schools and hospitals are now classified the same way in both the ISIC and COFOG. But the general structure of the two systems remains very different. One feature is the different treatment of general administration (ministries), which in the ISIC is separated from the activity the ministries serve.

At a more detailed level the correspondence is not good at all: for instance, education and health services are both broken down rather differently in the two systems. Further detailed study would undoubtedly reveal other problems. Whether convergence of these two classifications is desirable and whether a detailed study should be undertaken in this area are matters for further consideration.


United Nations, The Classification of the Functions of Government, Studies in Methods, Series M, No. 70 (New York, 1980).


This issue does not arise in GFS, which eliminates in consolidation transactions between departmental enterprises and producers of government services and calculates only the cash operating surplus or deficit of departmental enterprises’ sales to the public.


United Nations, International Standard Industrial Classification of All Economic Activities, Statistical Papers, Series M, No. 4, Rev. 3 (New York, 1986).


EUROSTAT, European System of Integrated Economic Accounts—ESA,2d ed. (Luxembourg, 1979).


United States, U.S. Department of Health and Human Services, Social Security Programs Throughout the World—1983, Research Report 59 (Washington: Government Printing Office, 1984), p. viii.


United Nations, Handbook of National Accounting: Public Sector Accounts, Studies in Methods, Series F, No. 50 (New York, 1988).


Chapter IX of the SNA (pp. 222–25 of the English version) suggests a set of accounts for production, income and outlay, and capital finance for this “Public Sector”; separate (nonconsolidated) production accounts are shown for general government and nonfinancial public enterprises and public financial institutions, but the other accounts are shown on both a consolidated and nonconsolidated basis.


The SNA, however, does not follow the GFS reroutings through the financial institutions sector of government transactions involving performance of monetary authorities’ functions or the acceptance of demand, time, or savings deposit liabilities by units without the authority both to acquire financial assets and to incur liabilities in the capital market.


T.P. Hill, Price and Volume Measures for Non-Market Services (Luxembourg: EUROSTAT, April 1975).


United Nations, Manual on National Accounts at Constant Prices, Statistical Papers, Series M, No. 64 (New York, 1979).


United Nations, Handbook of National Accounting: Accounting for Production—Sources and Methods, Studies in Methods, Series F, No. 39 (New York, 1986).


The SNA net lending concept, of course, represents the net result of government lending and repayment of previous lending, and government borrowing and amortization of previous government borrowing.


Organization for Economic Cooperation and Development, Revenue Statistics of OECD Member Countries, 19651986 (Paris: OECD, 1987).

    Other Resources Citing This Publication