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

21 Flow-of-Funds Accounts, A System of National Accounts, and Developing Countries

Vicente Galbis
Published Date:
September 1991
  • ShareShare
Show Summary Details
John C. Dawson

The purpose of this paper is to survey flow-of-funds (FOF) or national financial accounting in the context of the forthcoming revision of the United Nations' A System of National Accounts (SNA), with special emphasis on the potential of FOF analysis for developing countries. The task has been divided into three main sections.

Section I considers the analytical uses of the FOF accounts, using simple data matrices of the sort that could be available in most of the countries of the world. A key scheme for the analysis of financial fluctuations—the saving-investment process—is presented and related to the data matrix. The data are also related both to financial policy problems that are commonly faced by developing countries and to more sophisticated types of projections and programming. In all, the substantial usefulness of FOF data in the formation and execution of public policies in the financial area is emphasized.

In Section II the conceptual relations between the FOF accounts and the SNA are examined by considering how certain key features of the SNA standard accounts have evolved during successive revisions of the SNA, with particular emphasis on the representation of finance in the SNA and how financial data should be portrayed in the forthcoming SNA revision (labeled SNA 1993).Section II also addresses the closely related matter of harmonization between the Fund's statistical systems and the SNA as well as the technical problem of relating financial flows and balances in the accounts.

Section III considers the procedures to be used in the estimation of FOF accounts in the developing countries. It is demonstrated that an effective FOF system can now be implemented quickly and easily by most of these countries. The data problems that these countries face, however, are seen to have implications for the design of SNA 1993.Section IV presents conclusions.

I. Analytical Uses of the Flow-of-Funds Accounts

The broad purpose of the FOF accounts, or the sector capital finance accounts as they appear in the SNA, is to facilitate analysis of the operation of the financial system. That is, these accounts aim to interpret the transactions in financial markets carried out by the various economic sectors and to relate these transactions to the behavior of the nonfinancial economy. To this end, the sector accounts record for each sector its capital formation, its gross saving (that is, its operating surplus), and its acquisitions of various types of claims (financial uses of funds) and incurrence of various types of liabilities (financial sources of funds). Financial institutions are separated in the sectoring scheme in order to distinguish their functional role. Finally, the whole can be presented as a technically interlocking matrix with balancing cross and down totals. Tables 1 and 2 present simple examples of this type of matrix format that could be currently produced by most of the nations of the world.

Table 1.Financial Flow Matrix I
mentCentral BankBanksSectorRest of WorldAncvTotal
Gross capital famuiion271304332
Cross saving−245325197332
Net Lending (+). burrowing (−)−5152−5397
Currency and deposits771293144−14156156
Bills, bonds, and loans:
Centra] government debt1599214165417176176
Central government loans109109109109
Central bank advances27236−22727
Other loans and advances80808080
Other domestic debt141038334343
Foreign assels6−27−12−9−21−21
Olher claims and discrepancy, net66−2−2022
Tolal sources, uses1421429292117117464046487870902902
Note: “U” indicates uses of funds; “5” indicates sources of funds; a dash (—) indicates that the entry is zero.
Note: “U” indicates uses of funds; “5” indicates sources of funds; a dash (—) indicates that the entry is zero.
Table 2.Financial Flaw Matrix II
Govern-Govern-Enter-CentralMercialIrtsriru-DomesticResl ofDiscrep-
Cross capital formation27331201151332
Crass saving−24−142153124397332
Net lending (+), borrowing (−)−51−47−995219297
Currency and deposits−167712931138−14156156
Bills, bonds, and loms:
Central government debt15992143135417176176
Central government loins109198811109109
Central bank advances27236−22727
Other loans and advances139308668888
Other domestic debt171−24102345335252
Foreign assess6−27−13−21−21
Other claims and
discrepancies, net620106−23−52270
Total source, uses14214239391361269292117117141430230287870919919
Note: See Table 1.
Note: See Table 1.

The analytical power of the FOF system of accounts—like that of the income and product accounts—stems fundamentally from the interlocking character of the system—from the cross and down totals that balance for every period. Social accounting consistency requires that a flow change in any matrix cell be accompanied by corresponding changes in at least three other cells. For example, if increased government capital formation is to be financed by government debt issues, some sector must absorb the issues. To do so that sector must have larger sources of funds or must reduce other acquisitions. By making use of this feature in various forms, it is often possible to trace the impact of each sector's financial behavior on the others and eventually on the nonfinancial economy, or vice versa.

In the following four subsections, the intent is to indicate the range of different forms of FOF analysis. The level of detail naturally depends on the sector and the transaction detail available in the accounts, but the focus here is on condensed matrices of the sort that many countries are able to construct. Although the emphasis is on structures that are best revealed by the matrices, substantive work would be largely expressed as time-series analysis.

Saving-Investment Process Analysis

Perhaps the most frequently used structure in FOF analysis is the saving-investment process, by which saving is transformed into ultimate lending, whence it passes through various financial channels to ultimate borrowing, and then into real investment. An important feature of this scheme is the insertion of financial institutions in the center between ultimate lending and ultimate borrowing in order to portray their role as financial intermediaries.

The way in which the details of the process are worked into the scheme will vary with the analyst. One such scheme is shown in Figure 1, which is designed to reveal the domestic financing process.1 Private saving appears at the right of the figure and, as the arrows indicate, is either placed internally into real investment or into ultimate lending (that is, used to acquire claims). The claims acquired may take any of three major forms: claims on financial institutions, thus placing funds with these institutions; private credit instruments, thus passing funds directly through these markets to finance private ultimate borrowing; or federal securities, thus advancing funds in this market. In turn, financial institutions (bank or nonbank), in lending out the funds placed with them, must acquire either federal securities or private claims. Finally, private ultimate borrowing assembles the borrowing from banks, nonbanks, and ultimate lenders, which together with internal financing provides the funds to finance real investment.

Figure 1.Saving-Investment Process

Note: Codes for all flows refer to cells in Table 3.

a (C1) + (D1) + (E1).

b (A2) − (C1) − (D1) − (E1).

c (E4) − (C3) − (D3).

Federal ultimate borrowing in Figure 1 also assembles the lending of the various sectors and provides the financing for the federal deficit. Here, however, there is also a “feedback” arrow (at the top of the diagram). This is a somewhat awkward expression of the fact that governments often borrow not to finance the government deficit, narrowly defined, but to on-lend the funds. In developing countries this is a key form of finance for government enterprises. In this way general government becomes a kind of financial intermediary.

Note that, in effect, in Figure 1 the saving-investment account is split down the middle, with the financial system inserted between. As a result, the placement of saving and the financing of investment appear to be two distinct activities. But they necessarily articulate: whatever saving is placed with the financial system necessarily provides the financing of investment. Further, because saving and real investment are both nonfinancial variables, the financial markets are linked to the real economy at each end of the scheme.

This scheme is tied closely to FOF data because the various flow arrows can be measured by the cells in a FOF matrix (Table 3). The codes on the various arrows in Figure 1 refer to the rows and columns of Table 3. Each of the major flows (italic) in the matrix is represented in the scheme. The saving-investment process scheme thus provides a means of interpreting the flow movements in the matrix over time.

Table 3.Condensed Flow-of-Funds Matrix
A. Saving142.
B. Real investment141.00.30.4141.7
C. Intermediary claimsc42.
D. Private claims4.651.71.916.828.451.731.7
E. Federal securities2.
F. Miscellaneous & discrepancies, net4.
G. Total194.4194.47.87.820.920.933.033.0256.1256.1
Note: See Table 1. italic numbers indicate major flows.

Households, nonprofit, enterprises, provincial and local government, and rest-of-world.

Central bank, commercial banks, and treasury monetary funds.

Currency and demand deposits, time and saving accounts, and saving through life insurance and pension funds.

Note: See Table 1. italic numbers indicate major flows.

Households, nonprofit, enterprises, provincial and local government, and rest-of-world.

Central bank, commercial banks, and treasury monetary funds.

Currency and demand deposits, time and saving accounts, and saving through life insurance and pension funds.

In using the scheme, the analyst often finds that causation moves from left to right, counter to the arrows. For example, fluctuations in real investment or the federal deficit will result in fluctuations in various forms of borrowing. Figure 1 indicates how the impact of these borrowing fluctuations on financial institutions and markets can be traced. The advantage of having available in the FOF accounts the immediate forms of borrowing by each sector, not just the ultimate “financing” of the saving figures, becomes apparent. With added detail the scheme can become a very effective vehicle for understanding how capital formation or the federal deficit is financed.

Policy Problem Analysis in Developing Countries

Some key financial planning problems regularly faced by developing countries are posed in this subsection, and their relationship to the FOF accounts is briefly indicated. Will there be an adequate amount of foreign exchange? In the first instance, of course, this question implies an analysis of the rest-of-world account. Given the balance of payments current deficit (rest-of-world saving), the foreign exchange balance is protected by foreign borrowing. To judge the extent of foreign borrowing, an analysis of the domestic borrowers' financing is needed (private and public enterprises, financial institutions, and general government). How will the central government deficit be financed? The government deficit (gross capital formation less gross saving) plus its lending will be financed by the issue of government debt. The placement of these issues is analyzed by moving horizontally across the FOF matrix. Estimating the other sectors' capacities to absorb the issues requires sector analysis of the major debt buyers. The absorption by the banking system may imply further money supply expansion or the crowding out of private borrowers. How will the major government enterprises be financed, and by whom? Government enterprise capital formation will be financed first by the enterprises' own saving (that is, their operating surpluses). Beyond this, their major sources of funds will be government loans, borrowing from abroad, and borrowing from commercial banks. The volume of the last of these is of special interest because of the possibility, again, of crowding out private borrowers.

For all three of these policy questions, the answers require an impact analysis for various sectors, types of transaction, or both. It is the articulation of the accounts within the FOF matrix that enables one to seek and judge the answers.

Financial Projections and Programming

The essence of a FOF projection is a matrix for some future time period. Such projections can be used for a variety of purposes. One would be to test the consistency of separately prepared sector or market forecasts. Another would be to develop a consistent view of the financial flow implications of a particular set of assumptions about future events. A third might be used by the planners in a developing country to examine the implications of parts of an economic plan.

For example, consider a projection of Table 1. A developing country might have sector capital formation and saving data from a tentative plan formulation. It might then impose certain constraining balance of payments estimates. The interior of the matrix could be projected under these constraints in order to work out the implications for domestic finance. The projection might well be able to identify certain types of credit shortages or the amount of necessary monetary expansion.

The projection of a more elaborate matrix—of some 10 sectors and 20 or more financial transaction types—has for many years been a regular part of monetary and credit policymaking at the U.S. central bank. In a series of iterations, the judgmental forecasts of numerous financial market and national account specialists in the research division are used to modify an extrapolated matrix. At the same time, monetary policy implementation and impact are gradually worked into the projection. In the end the exercise provides a detailed statistical picture of future financial flows that is closely integrated with the prepared policy position to be placed before the governing board.

A sophisticated illustration can be drawn from the type of financial programming engaged in by the Fund in the design of adjustment programs. The policy packages of these programs aim primarily to establish a viable balance of payments position over the medium term. An important element of this type of programming is an analysis that traces major credit demands to a focus on the banking system, even on the central bank itself. First, a balance of payments target is established, and the implications of projected imports, exports, and foreign borrowing for foreign exchange reserves are calculated. The pathway of balance of payments impact is through the movement in the foreign exchange reserves to the banking system. Because the central bank is assumed to be the residual buyer of government debt, public finance needs can ultimately also be traced to the banking system. Given an independent estimate of the appropriate currency and deposit expansion, the credit extension available to the private domestic sector can be calculated residually, and its adequacy can be judged in the light of financing needs. Iterations of this analysis lead to the firming of various policy objectives. Until now, financial programming has generally been done without the help of formal FOF projections, but it is evident that such projections would be a useful complement to the analysis.

Long-Run Development of Financial Institutions

FOF projections need not be merely short-run; they can cover 5-year plan periods or even 15- to 20-year financial developments. In this latter case the projection may reveal, or be required to take into account, various institutional developments in the financial system. Thus, shifts in the main sectors carrying out capital formation will be accompanied by changes in the types and composition of claims that the financial system will be asked to absorb. Similarly, as some households become more affluent and sophisticated, the supply of funds moving into the various financial markets will alter. Wealthy households may wish to deal regularly in equity share markets or to place funds into life insurance and pension saving. These changing borrower and lender needs will be forces for change in the financial institutions and the markets that are their financial channels. An efficient group of financial institutions can comfortably absorb the various instruments offered to it by borrowers and can offer a suitable variety of claims to meet the needs of lenders. The long-run changes in these borrower and lender needs will thus suggest appropriate institutional growth, and policies can be designed for its promotion.

One or two examples from developing countries may be illustrative. Well-organized open markets for financial instruments may not be common in developing countries. A country may wish to place more of its government debt issues domestically but less with its banking system. The situation may be appropriate for the design of a small-saver instrument that can be widely distributed and that would carry a very high rate of return in the interest of mobilizing private nonbank saving. Or a country may wish to store liquid business deposits in foreign assets. Such accumulation represents lending to the rest of the world and may be part of a pattern in which the rest of the world is being treated as a financial intermediary, accepting short-term funds and lending back long-term funds. It may be desirable to attract the short-term flow to the domestic economy. What is suggested is the need for a secure, liquid financial instrument—for example, a treasury bill—with a return attractive enough to compete with foreign instruments. The FOF projection can incorporate such new growth patterns into its flows and can assist one in judging their impact on other markets and sectors.

Given all these promising applications, why has FOF analysis not become more prominent during the past 30 years? On the one hand, FOF analysis—unlike input-output analysis—seems not to have developed a strong academic tradition. On the other hand, many applied economists have apparently not sensed how they might make use of the system. Or perhaps much FOF analysis is so informal as to remain hidden, as in much of the Fund's work. Nevertheless, given the growing importance of financial problems for most of the countries of the world, the system's potential remains great.

II. Conceptual Relationship of Flow-of-Funds Accounts to the SNA

In this section, the relation of FOF accounts to the SNA is examined.

Finance and the Evolution of the SNA

As background to a consideration of the relationship between FOF accounts and the SNA, it will be useful to review briefly the evolution of the SNA itself over the past 35 years: the so-called standard accounts as they first appeared in the 1953 version of the SNA, then as they appeared in the 1968 version, and finally as they might appear in the revision under consideration (SNA 1993). Several key features of each phase of the evolution will be discussed in turn.

The 1953 SNA

The standard accounts as initially laid out in the 1953 SNA focus on the circular flow of product and income. Accounts 1 and 2 present on the right side the sale of the final product and on the left side the income and other payments generated in producing the product.2 Accounts 3 through 6 are the accounts that receive the income and, taking account of the various transfer and financing operations, use it in purchasing the product. This portrayal of the income circuit in the accounts thus separates the activity of production from all other activities engaged in by the institutional sectors of the economy—the activities that later came to be associated with income or outlay and capital finance. The foundation of the 1953 SNA system is a cross-classification of the economy: once by activity and once by institution. This first key feature is an obvious and familiar one.

The second key feature is that these standard accounts consist of a set of balancing and interlocking accounts. One source from which the analytical power of the system stems is that each account balances conceptually, that each is a selection of receipt items providing the finance for its corresponding group of disbursement items; each is a balancing statement for sources and uses of funds. Another source of the system's analytical power is its interlocking character. That is, each disbursement (use of funds) item in one account can be matched to an equal receipt (source of funds) item in a different account. The system is technically a quadruple entry system: a change in any one item implies a change in at least three other items. These are characteristics that greatly facilitate the tracing of intersectoral economic effects.

In connection with this interlocking feature, it is necessary to consider Account 6, External Transactions (rest-of-the-world account). The treatment of this account in the 1953 SNA— and in the 1968 SNA as well—is somewhat ambiguous with respect to the point of view represented by the account.

On the one hand, the title and wording of the account imply that it is an assembly of the domestic sector records of international transactions—as in a balance of payments account. Thus, we have an external transactions account, an account derivable by consolidating all the domestic sector accounts—Accounts 1 through 5. Such an account records transactions from the viewpoint of the domestic sectors. In this view the account reflects one kind of domestic sector activity, and the rest of the world is not an institutional sector in a system of national accounts.

On the other hand, if the accounts are to be considered as a complete system of interlocking accounts, Account 6 must be viewed as recording the transactions of the rest of the world (for example, a domestic sector receipt in Accounts 1 through 5 must be matched by a rest-of-world disbursement in Account 6). In the 1953 SNA, this view is implied both by the interlocking coding of account items and by the consistency with which the accounts elsewhere record sources of funds on the right and uses of funds on the left. In this view the right side of Account 6 should be labeled (Rest-of-world) Receipts rather than (Domestic Sector) Disbursements. It would seem desirable to remove this ambiguity in SNA 1993. And if the interlocking feature of the system is as important as we assert, it would seem that the External Transactions account should become unambiguously a rest-of-world account and the rest of the world should become an institutional sector in a closed social accounting system.

A third key feature of the 1953 SNA is the analytical significance of the system. This feature was unquestionably a central motivation for the statistical development of the accounts. First, Western economic analysis had universally accepted the circular flow of product and income as a central organizing concept. Further, GDP (or some variant of it) was becoming widely accepted as an important measure of overall economic performance. In addition, the Keynesian hypothesis that GDP is determined by aggregate demand remains even today as the key practical hypothesis in macroeconomic analysis. In the 1953 SNA aggregate demand is prominently set out in Account 1. These accounts, particularly the sector accounts, also contain a judicious selection of other important variables—for example, household income and saving, the general government current deficit (saving), and the economy's external surplus on current account. Moreover, as was noted above, all the variables are related through the balancing and interlocking feature. Finally, although the 1953 SNA accounts are somewhat cluttered with items of minor importance for many countries, the scheme is simple enough to be analytically manageable. The list of such characteristics could be extended, but suffice it to say that the scheme was nicely designed and timed to catch on as a tool for economic analysis.

Finally, since the ultimate purpose here is to relate the SNA to finance, it is necessary to consider the portrayal of finance in these standard accounts. At first sight, Account 3 appears to be a capital formation account for the whole economy, showing domestic capital formation together with its methods of finance. But items 3.7 and 3.8, despite their labels as “finance,” are imputed items measured as types of real capital formation. If one consolidates the three capital reconciliation accounts (for households and the like, general government, and the rest of the world) with Account 3, however, one obtains an account for capital formation together with the elements of sector saving that ultimately finance it (items 4.6, 5.5, and 6.6; see the 1953 SNA, Standard Table V). Because this consolidated “saving and investment” account was in practice substituted for Account 3 and the reconciliation accounts were never developed, the best sense of the 1953 SNA is from the viewpoint of finance to see the system as presenting the sector surpluses that ultimately finance real investment but as not presenting any information about the flows in financial assets and liabilities by which the saving moves through the financial system into investment. These flows in financial claims have in effect been consolidated out.

An alternative view is to see the capital reconciliation accounts as primitive capital finance accounts (items 4.12 and 5.13 being converted into capital formation components) and Account 3 (by netting items 3.7 and 3.8) as applying only to corporate enterprises. This approach, in effect, deconsolidates the Domestic Capital Formation account. When this is done, the financial claim items (items 3.6, 4.18, 5.19, and 6.11) are retained, and finance remains explicitly in the system. It will be appropriate to pursue this view in discussing the 1968 SNA.

The 1968 SNA

Three different systems of standard accounts are apparent in the 1968 (current) version of the SNA. The first consists of the four consolidated accounts for the nation (Accounts 1, 3, 5, and 6). This system is then deconsolidated in two quite separate ways. One method deconsolidates the GDP and expenditure account (Account 1) in the direction of detailed production analysis. Because the concern here is finance, this path will not be pursued. The other deconsolidates the national income and outlay account (Account 3) and the national capital finance account (Account 5) into five domestic sectors each. When these 10 accounts are put together with the GDP and external accounts (Accounts 1 and 6), one has a 12-account version of the 1968 SNA. This is the standard system we shall examine as the 1968 SNA.

How have the various key features that were sketched for the 1953 SNA fared in this revision? The 1968 SNA focuses even more sharply on the circular flow of income and product than did its predecessor. Note that the first two accounts in the previous system, the Domestic Product and National Income accounts, have here been merged into a single GDP and expenditure account (Account 1). Thus, all of the GDP expenditure flows are on one side of Account 1, and all of the equal income flows—here called gross domestic income—are on the other side of the same account.

With respect to the balancing and interlocking features, the former is clearly present. But in the sector accounts the parenthetical references that specified the interlocking character of the previous system have disappeared. The 1968 SNA is nevertheless interlocking, but in a somewhat different sense. It should be understood as a matrix of accounts with balancing cross totals (total sources and uses for each transaction type), as well as balancing down totals (total sources and uses for each account). For example, the sum of the final consumption expenditure items as disbursements equals, conceptually, final consumption expenditure as receipts; similarly, total property income receipts should equal property income disbursements. The analytical power of the quadruple entry system is in this way preserved.

From the viewpoint of financial analysis, this 12-account system takes two crucial steps forward. Although the consolidated capital finance account (Account 5) remained similar to the earlier capital formation or saving-investment account in that it presented only the ultimate origins of the finance of investment, the deconsolidated sector capital finance accounts each present, as well, the immediate financial sources and uses. That is, these accounts show details of the net incurrence of liabilities and the net acquisition of financial assets as well as the sectors' saving (income and outlay surplus) and capital formation. Furthermore, these financial flows are for the first time defined (other than residually) and classified. As indicated earlier, such data make possible a very useful and straightforward analysis of how a sector finances itself and how it contributes to the finance of other sectors.

The second crucial step forward occurs by virtue of the separate sectoring of financial institutions in the 1968 SNA domestic sector breakdown. As pointed out in the 1953 SNA,3 the separate sectoring of financial institutions is necessary for a meaningful analysis of the financial flows. This is because of the intermediary role played by financial institutions in providing finance to governments and to the enterprise and household sectors. This role is characteristic of developing as well as developed economies; almost all countries have important central bank and commercial banking institutions.

The matter of analytical significance raises questions about the 12-account system of the 1968 SNA: it appears to be far more complex than the 6-account system of the 1953 SNA. How much more analytically effective is the newer system? Is the added detail worth the effort to collect it? At least some sector detail is necessary to facilitate intersectoral impact analysis, and separate sectoring of financial institutions is essential for effective financial analysis. Thus, a meaningful system should accommodate, it would seem, no fewer than four domestic sectors.4

But proliferation of sectors is not in fact the main source of added complexity, which stems primarily from added detail of transaction type in the income and outlay accounts and the capital finance accounts.5 These classification schemes could be substantially condensed with little analytical loss for the purposes of standard national accounts. Furthermore, with respect to the detail on financial transactions presented in the sector capital finance accounts, effective financial analysis requires not only less detail but detail of a different sort, which will be proposed in the discussion of SNA1993. Suffice it to say here that a standard system containing well-designed sector capital finance accounts would be capable of producing financial system analysis far beyond the capability of the 1953 SNA.

Types of National Accounts Data Currently Submitted

Thus far the UN schemes of what was desirable in social accounting data have been discussed. There is also a need to observe what has been happening in practice over the past 35 years. One way of doing this is to look at what types of data the countries of the world are able to submit for the UN's compendium of national accounts.6Table 4 summarizes a recent survey of this kind.

Table 4.Major Types of National Accounts Data Reported in UN Compilation for 1983
Percentage of Countries
CurrentCurrent or
Itemdataback data
1.1Expenditures on the gross domestic5398
product (current prices)
1.2Expenditures on the gross domestic4873
product (constant prices)
1.3Cost components of the gross domestic4684
1.4General government current receipts and2458
expenditures, summary
1.7External transactions on current account,3873
1.8Capital transactions of the nation,3762
summary (saving and capital formation)
1.10Gross domestic product by kind of activity4994
(current prices)
1.11Gross domestic product by kind of activity4976
(constant prices)
1.12Relations among national accounting5082
Source: United Nations, National Accounts Statistics: Main Aggregates and Detailed Tables, 1983 (New York, 1986).Note: All tables for which more than half of the countries (146 countries reporting in accordance with the SNA) reported either current or back data are included here.
Source: United Nations, National Accounts Statistics: Main Aggregates and Detailed Tables, 1983 (New York, 1986).Note: All tables for which more than half of the countries (146 countries reporting in accordance with the SNA) reported either current or back data are included here.

One can see in Table 4 strong evidence that the 1953 SNA scheme has taken hold in practice. Almost all of the 146 reporting countries are producing a GDP account (GDP, 1.1, and income, 1.3), although it is true that for many countries these data are not yet current. Most also provide the GDP by industry (1.10) and in constant prices (1.2 and 1.11). More than half are able in time to add current accounts for general government (1.4) and the rest-of-world (1.7), as well as a capital formation (saving-investment) account (1.8). These four accounts are four of the five needed for the closed 1953 SNA system referred to above, in which the capital reconciliation accounts are merged with the domestic capital formation account to form a saving-investment account. About a quarter of the reporting countries are able to add a household income and outlay account (not shown in table) and form the complete system.

Table 4 reports 1983 data, but its source compendium was published in 1986—more than 15 years after the publication of the 1968 SNA. Do we find evidence of its impact? The reporting scheme does not include the 1968 SNA's Account 3 (National Disposable Income and Its Appropriation—the consolidated income and outlay account) or Account 5 (Capital Finance—the consolidated capital finance account). So, apparently, the four “Consolidated Accounts for the Nation” as a standard scheme did not take hold. As regards the deconsolidated 12-sector scheme, only a handful of countries—perhaps 10 percent, largely in Europe—are able to report sector income and outlay accounts at the level of detail of the 1968 SNA, and a still smaller handful—perhaps 5 percent—are able to report detailed sector capital finance accounts. One thus must conclude that the 1968 SNA is not widely used as a system of standard accounts.

Implications for SNA 1993

In formulating SNA 1993, it is important to consider why the 1968 SNA has had so little impact. The data suggest an important clue: whereas 57 percent of the countries reporting could submit a summary account of general government current receipts and expenditures, only 11 percent could submit one at the level of detail required by the standard system; for the external account, the corresponding figures are 73 percent and 11 percent. One might infer that the 1968 SNA is too complicated and detailed for most countries to produce at present. It is this author's view that the 1968 SNA is too complex and contains too much insignificant detail for straightforward policy analysis. These suggestions underscore the objective in the current revision process of presenting a simpler standard set of accounts.

Flow-of-Funds Representation in the SNA

What, then, should inform a vision of the about-to-be-created SNA 1993 standard accounts? If the main lesson of the 1968 SNA is to be heeded, the new accounts should not be far ahead of the current statistical accomplishment of most countries. To be serviceable, the new accounts should be next steps in feasible statistical development. In addition, the analysis has stressed the need to maintain sector accounts in the standard central system. In addition, sector capital finance accounts and the separate sectoring of financial institutions are needed for effective financial analysis. Are these objectives compatible? It appears so.

In sketching a suggested new system, let us begin where a good many countries are now: they have a gross product and income account; income and outlay accounts for the rest-of-world, general government, and households (including within it nonprofit institutions); and a consolidated capital formation (saving-investment) account. Let us move along the lines of the 12-account scheme of the 1968 SNA but keep in mind the need for simplification.

First, envision a vertical extension of the rest-of-world account to derive its capital finance account and a similar extension for general government. The development of full balance of payments and government finance data for most countries during the past few decades should make this a relatively easy step. (As discussed further below, however, the balance of payments data must be in local currency.)

Three capital finance accounts remain: for nonfinancial enterprises, for financial institutions, and for households and the like (including nonprofit institutions). Imagine them lumped together into a single residual capital finance account. For analytical purposes it is essential to estimate and break out an account for at least a large share of financial institutions. For many countries this can be done through an account for the monetary and banking system alone, and these basic data are available for almost all countries. As noted below, it is analytically less essential to break nonfinancial enterprises away from households and the like, but such a step would in any event have to await the compilation of balance-sheet data for substantial groups of enterprises. If the SNA 1993 wishes to emphasize feasibility, a first-stage version of it might contain, first, a capital finance account for the monetary and banking system and, second, a catch-all capital finance account for other domestic sectors.

Consider also the types of financial transactions to be used. As suggested earlier, much of the difficulty in producing sector capital finance accounts may stem from the complexity of the classification scheme for financial claims. For SNA 1993, a greatly condensed scheme consisting of four broad financial categories is suggested: (A) foreign assets; (B) currency and deposits; (C) bills, bonds, and loans; and (D) other claims.

The existing 13-claim categories can, of course, be allocated to these headings.7 The purpose here is only in part simplification. In many applications foreign assets (category A), properly defined, will consist largely of net foreign exchange. In addition, currency and deposits (category B) are closely related to money, broadly defined. So these categories are likely to be of analytical use.

To understand why category C as a whole might not be very useful, consider such an item in the various 1968 SNA capital finance accounts. Each of the sectors might accumulate and each of them might issue these instruments. In these circumstances it would generally not be possible to identify who is acquiring from whom (that is, who is financing whom). And narrowing the category to long-term bonds would probably not improve matters. But consider the category of central government bond issues. Here only one sector issues the instrument. In the absence of secondary market transactions, the acquiring sector is financing the government. Even with secondary market transactions, there is a much better chance with government bond issues of being able to identify who is financing whom—and that is the central objective in financial flow tracing. To generalize, the most useful financial transaction types will be those designed so as to have only one or two sectors issuing or only one or two sectors acquiring the instrument. Category C needs to be broken down.

This discussion of the most useful categories of financial transaction types for a FOF matrix is related to Table 24 in the 1968 SNA. The same 13 types of claim that appear in the sector capital finance accounts appear there, but within each of the 13 types also appears institutional sector detail to indicate sectors bearing the liabilities for that particular claim.8 The line for a subcategory such as 5.iii (long-term bonds, liability of the central government) would have a single issue figure under central government off to the right and figures for the various sector acquisitions off to the left. Given such detail, it is clear that the category of central government debt could be assembled as the sum of 4.iii, 5.iii, 7.iv, and 8.iii—that is, the sum of all the central government liability components that one wishes to define as constituting central government debt. Alternatively, if one wishes to build up data for the 13 basic claim types, one would have to reverse this process, allocating central government debt to the Table 24 sub-categories. The Table 24 analysis thus shows how the 13 basic claim types can be conceptually related both to basic data sources and to analytically significant categories.

Perhaps the simplest effective breakdown for the bills, bonds, and loans that make up category C would be into central government debt, bank loans, and other, but most countries should consider at least the following:

C. Bills, bonds, and loans

  • 1. Central government debt

  • 2. Central government loans

  • 3. Central bank advances

  • 4. Financial institution loans and advances

  • 5. Other debt.

This or a similar list should be adopted for the SNA 1993 standard accounts to suggest the type of detail that countries would find analytically useful. These are the types of claims shown in Tables 1 and 2 that were used to indicate intersectoral tracing of financial flows.

In summary, producing the accounts as presented in the 1968 SNA sector capital finance accounts requires a high level of financial detail—detail at the level of the subcategories of SNA Table 24, which is generally available only in the most statistically advanced countries. Further, the transaction types in these accounts are not very effective in financial flow analysis: they provide little sector-to-sector identification of flows. The alternative classification presented above is simpler and, as will be shown in Section III, can be derived directly from data available in almost all countries. In addition, it is demonstrably useful in financial flow analysis.

In conclusion, the FOF accounts should be represented in SNA 1993 by a closed set of condensed capital finance accounts for the rest-of-world, general government, and monetary and banking sectors, and the other domestic sector. To form a standard system, a gross product and income account and three income-outlay accounts (rest-of-world, general government, and households and the like, including nonprofit institutions) could be added to these four capital finance accounts.9 These eight accounts could form a closed, interlocking system—a system that in its condensed form is achievable for most countries in the next decade or so and that could be a standard first stage leading to a later, more detailed standard system.

If financial analysis is to have an improved social accounting data base, increased statistical effort must focus on the sector capital finance accounts. To help in this effort, these accounts must have a prominent place in formulation of the simplest, central, UN standard accounts.

Flow-of-Funds Accounts, the Fund's Statistical Systems, and Harmonization

The view taken here is that the hope for an early statistical realization of sector capital finance accounts rests in large part on early development of such accounts for the government, the rest-of-world, and the monetary and banking sectors. To serve the social accounting purpose, such accounts must, of course, fit into a matrix. This means that at the technical level there must be consistent sector coverage and consistent accounting treatment of transactions within and (especially) across sectors. This is what harmonization means. Harmonization among the three Fund systems—government finance statistics (GFS), balance of payments statistics (BOPS), and money and banking statistics (MBS)—and between each of these and SNA1993 becomes an issue because GFS, BOPS, and MBS (or closely related country data) are likely to be basic source material for the new FOF accounts.

Each of the three Fund systems has had international comparability as a goal, and each has provided useful compilations of data. In this process some member countries have altered their accounts in the direction of the respective Fund conventions. To the extent that the Fund's standard systems are not harmonized, such alterations may make the local articulation of these accounts more difficult. There is to some extent an inherent conflict between international comparability and local articulation of sector accounts, but this is especially so if the international standards are not harmonious.

If GFS, BOPS, and MBS had been developed for systematic social accounting purposes, their instruction manuals might well now provide uniform and consistent conventions of measurement. These systems did not, however, come about in this way. They were created within the Fund over a considerable period, each guided by the analytical needs of a particular group of specialists. It is true that steps toward harmonization were taken as the later systems were developed and the earlier ones revised. But differences remain, and now that the differing conventions of measurement have taken hold among analysts both within and outside the Fund, changing them will be difficult. Perhaps over time the analytical advantages of a social accounting approach may be persuasive in moving toward harmonization: after all, the problems dealt with by the Fund usually deal with the interrelations among these three sectors. At present, however, the differing needs of each system's primary users may continue to impede true harmonization.

SNA 1993, in contrast, is constrained to build full consistency into its system. In this situation, what is to be done? Of course, one should harmonize to the extent possible. But wisdom may suggest stressing the construction of standard reconciliation or bridge tables linking the key time series in the four systems to permit analysts to move from one system to another in both estimation and analysis. Such an effort, which is under way, should be given urgent priority because such tables could greatly facilitate the construction of SNA accounts by individual countries.

Curiously, the single most important harmonization issue for the SNA seems likely to be overlooked entirely. To form a closed system, all the sector accounts of that system must be in the same unit of account, the same currency. Thus, it is essential to the SNA that the rest-of-world account be expressed in the local currency. But this in turn means that there must be a balance of payments account in the local currency as a data source. The increasing trend throughout the world is to do balance of payments analysis in some international currency such as U.S. dollars. The loss of any analytical connection with domestic sector data is apparently less important than the fact that payments to abroad must be made in an international currency. In countries without balance of payments data in local currency, or where such data are of low quality, the Fund could aid the development of the SNA accounts by ensuring the early compilation or improvement of these data.10

Dealing with the many technical aspects of the harmonization issue is outside the scope of this paper, but an attempt will be made to offer an approach to these problems, one that suggests a rough ordering of priorities. Attention should first be focused on the definitions of sector coverage for the major domestic sectors and the rest-of-world sector. The remaining issues here are relatively minor, and—with compromise—true consistency should be achievable. In any case, complete harmony on the definitions of major sectors would greatly simplify the conversion of BOPS, GFS, or MBS data into the corresponding SNA capital finance accounts.

The second priority for harmonization would be toward standardization of the financial-nonfinancial line in each sector account for the four systems. Conceptually, entries indicating the exchange of goods, services, and transfers lie above this line in each sector account, and entries recording the exchange of claims (including cash balances) lie below this line. In the 1968 SNA the line is marked by the net lending item in each capital finance account, which, broadly, equals the difference between gross saving and gross capital formation for each sector.11 In the harmonization, the attempt would be to standardize the definition of this deficit or surplus of each sector; conceptually, the algebraic sum of these items across all sectors (including the rest-of-world sector) should be zero. Drawing this line is an essential feature of FOF accounts.

For example, in the GFS, although lending minus repayments is separated out from other government expenditure and revenue, it is for analytical purposes located above the overall deficit or surplus line (that is, as part of the net aggregate that is to be financed by government debt issue). Although this practice cannot be accepted by the SNA, the separate identification of the item makes the shift of these financial transactions to below the deficit line simple. This applies, of course, providing that the technical specifications for the total of lending minus repayments are in harmony with those for the other sectors. This harmonization is an example of the second-priority focus.

Finally, the complexities of the harmonization of the various types of financial transactions should be addressed. Here, too, one can take a “closing in” strategy. First, standardization should be attempted across sectors of broad groups of claims, such as the four groups we suggested earlier, by way of simplifying the capital finance account (see the preceding subsection). In this way there are far fewer borderlines to define. Beyond this, one can only recommend the wisdom of the view that in these matters it is necessary to have some agreed convention; which particular convention is less important. Where agreement is not possible, one should take on the obligation of providing reconciliations.

Consider just one key example in this category. It is hoped that harmonization can be effected in the area of foreign exchange and, in particular, the area of international reserves. The flow in these net foreign assets appears only in the monetary sector and the rest-of-world sector accounts. The analytical purpose of the identification of these reserves—a key part of the means of payment for international transactions—is broadly the same for users of both BOP and MBS. Furthermore, both estimates generally rely on the same (monetary sector) sources for data. To a practitioner there would seem to be no reason for differences in these flows where they are expressed in the local currency unit. At a minimum, there should be a clear reconciliation.

Balances, Flows, and Valuation Changes

About a decade ago the standard accounts of the 1968 SNA were extended in the Provisional Guidelines12 to add a system of balance-sheet and reconciliation accounts to the capital finance accounts. The construction of national and sectoral balance sheets and reconciliation accounts is in general envisioned as a stage beyond the construction of sectoral capital finance accounts (that is, FOF accounts). Nevertheless, because flow estimates are often derived from balance-sheet increments, the two areas appear to have a substantial overlap, and it is this overlap that will here be explored.

From an accounting viewpoint, asset and liability balances change either as a result of transactions, which are encompassed in the flow measure, or as a result of revaluations—that is, write-ups or write-downs of the asset or liability valuations. This principle is embodied in the definition of the reconciliation accounts, which include all the differences between opening and closing balance-sheet accounts that are not covered by the flows in the capital finance accounts. That is, the reconciliation accounts will consist largely of revaluations of balance-sheet items.13

In general, the Provisional Guidelines call for nonfinancial assets on the balance sheets to be valued at current market values. This rule makes large annual revaluations of tangible assets highly likely; because the companion entry to every revaluation is a corresponding change in the transactor's net worth, large and frequent revaluations in sector equities are assured. The situation differs dramatically, however, when one considers financial assets. With only three exceptions, the Provisional Guidelines call for these assets to be valued at nominal or face values. These financial assets—so-called fixed-claim assets–are similarly valued as liabilities. The upshot of these rules is that much of the array of claims in the FOF matrix is not subject to regular revaluations owing to asset price changes.

The exceptions to the use of nominal values for financial assets are made in the cases of (1) gold, (2) long-term bonds, and (3) corporate equity securities. Current market valuations are proposed for all three of these. Market valuations are also proposed for long-term bonds and corporate equity securities on the liability side of the balance sheet. Thus, revaluations are to be expected in these three items. The Provisional Guidelines also allow for another important type of revaluation. The foreign assets category will be subject to revaluation if the local balance-sheet convention is to show the local currency equivalent of the foreign exchange that is receivable or payable on the claim. In all, one is alerted to three types of FOF claims in which revaluations are likely to play a role in the scheme for balance-sheet and reconciliation accounts: (1) corporate equity securities, (2) long-term bonds, and (3) foreign assets (including gold).14

Let us now return to the situation of the implementation not of balance sheets and reconciliation accounts but merely of the capital finance accounts themselves. The problem will be reduced to removing revaluations that are picked up when flows are estimated from balance-sheet increments. To consider it, one must anticipate some of the results reached below in the discussion of estimating procedures. Of the main sectors, at first only the accounts for the financial sectors will be estimated from balance sheets. The valuation conventions for these balance sheets will be near enough to face values so that the increments will be good approximations of the actual flows for most items. But if long-term bonds are a substantial portfolio item, and if the balance sheets value them at market values, removal of the revaluations will be in order. Finally, there may be difficulty in reconciling the flow in net foreign exchange to the corresponding flow as shown in the balance of payments. Valuation changes may be at the heart of this matter. This last is the harmonization issue referred to above.

At a later stage in the implementation, balance sheets of enterprises—both government and private—will be used as sources. Revaluations of fixed assets as well as other items on the balance sheet are to be found here and can probably be avoided in flow estimates only by a case-by-case scrutiny of the accounts. In addition, large revaluations in the corporate stock account are so likely that this item should not be used to estimate the cash inflow derived from net cash stock issues if at all possible. In most cases, this flow is small and, luckily, plays a minor role in the FOF matrix.

Overall, this review of the problems involved in the overlap of FOF estimation and the revaluation of balance-sheet items has indicated that the overlap is not large. Experience indicates that at present most of the errors that are due to unidentified revaluations being mixed into the flows are small errors in relation to classification problems and other data shortcomings.

III. Feasible Flow-of-Funds Accounts and the Developing Countries

Most of the 160 or so countries of the world are developing countries. One of the features of a developing country is a developing statistical system. Unfortunately, the 1968 SNA did not take into account that a great many countries have only the most limited capacity for producing new and improved data. One surmises that this was a key reason that the 1968 SNA has been, as concluded above, too complicated and detailed for most countries to produce during the past 20 years; the goals were set too far ahead of the possibilities.

In presenting its full set of standard accounts and supporting tables, the only concession the 1968 SNA made to development was a suggested order in which the full accounts should be implemented.15 The developing countries were not offered any intermediate targets leading up to the full system. In particular, they were not offered simpler versions that, although within their statistical capabilities, would offer the analytical advantages of a social accounting system (that is, of a closed set of interlocking accounts). The objective in this section is to demonstrate a simplified system in the FOF area that is currently feasible for almost all countries and that can lead in steps to the 1968 SNA system of sector capital finance accounts or to a complete FOF system. It is, in effect, a design for a series of intermediate systems leading toward complete implementation; it thus embodies its own suggestions for statistical priorities.

Getting the Initial Estimates

The method of presentation will be to describe the statistical implementation of a simplified FOF matrix by using the sort of statistical base found in the developing countries in, let us say, the 1970s. This will enable an enumeration of the bare statistical necessities for such an undertaking. Then the successive stages of statistical elaboration of the matrix will be described in the order of movement toward a full system. For this exercise we shall assume that the Fund's BOPS, GFS, and MBS systems are not available to use as data sources.

Table 1 earlier presented an illustration of the format used for the earliest stage of a FOF matrix.16 It contains a very simple sectoring scheme of five sectors: central government, central bank, commercial banks, rest-of-world, and other domestic (a catch-all sector that includes enterprises, households, and a variety of smaller sectors including provincial and local government and other financial institutions). The classification for types of financial claims follows closely the scheme advocated above. In Table 1, financial institution loans and advances is divided between central bank advances and other loans and advances, which in this simple sectoring become entirely commercial bank loans and advances. Note that several of the key claims selected for enumeration are claims attributable to the sectors that have been selected, so that if data are available for these sectors they will in general be available for these claims. Note also that this layout of sector accounts does not suggest government as a whole or financial institutions as a whole. Each of these more aggregative sectors contains subsectors for which data are not likely to be available at this earliest stage.

As a primary matter, the time reference that is to be standard for all sectors must be considered. Local data sources may deal with the central government on a fiscal year basis, whereas the banking data and balance of payments are on calendar years. Which time reference is used will depend on the ease of conversion. If an insufficient amount of quarterly or semiannual data is available to permit conversion and, hence, uniformity where it does not exist, then FOF implementation should probably be deferred. Financial flows are simply too volatile for the average of two calendar years to be representative of a fiscal year, or vice versa.

A second primary requirement is the existence of balance of payments data on a local currency basis. As suggested above, moving from dollar or SDR figures to a local currency basis with a single exchange ratio can sometimes give unsatisfactory results. However, the most critical data are those for the current account balance and the items above it. Because local currency figures here are necessary for the income and product accounts, it is likely that even for developing countries a local currency balance of payments will be available or can be estimated. If it is not possible to obtain it, however, the FOF effort should be deferred.

The question of how current the basic data are is a final preliminary. The data are likely to be reasonably current in the directly estimated sectors because they are all tied to areas of current public policy decision making. At the start, one need not aim to assemble a very recent year. This compiler would be well-advised to move back in time to a year in which definitive rather than preliminary or revised data are available. Once the system can be successfully produced with a lag of a few years, one can use the more recent, more preliminary, and more fragmentary sources to carry the system toward current data production.

One may now proceed to the matrix itself, referring to Table 1. The totals data at the right of the table for the first two lines—gross capital formation and gross saving—will come from the national income and product (NIP) data. It is on these first two lines that the FOF construction ties in with the income and outlay accounts that lie above the capital finance accounts in the SNA. The key breakdown of capital formation between central government and other sectors is usually available in the NIP source. The saving breakdown is generally more troublesome. The NIP source will have a rest-of-world saving figure that is (minus) the surplus of the nation on current account, but NIP is unlikely to have central government saving. This figure may well have to be derived as an operating surplus from the central government data. Given the domestic saving breakdown between central government and other sectors, one has the surplus or deficit of each sector (shown in Table 1 as net lending),17 which is to be analyzed into the incurrence of liabilities (financial sources of funds in Table 1) and the acquisition of financial assets (financial uses in Table 1) in the lower section of the matrix.

Now consider the initial derivation of each of the directly estimated sector accounts from its basic sources. For the central government, two such sources are generally available: the data associated with the budget-making process and the data associated with the auditing process. Each will contain past actuals for income, outlay, and financing—the budget source often being more up to date but less detailed. The quality of either source may be much better than the other. It is better to stick to one source if possible, since each reflects a whole system of accounts and tables.

The worksheet task is to transcribe the balancing account and supporting detail so that the account can be rearranged and condensed to yield a first approximation to the Table 1 sector account. The focus will first be on deriving a nonfinancial surplus or deficit (net lending) from the revenue and expenditure figures. The main difficulty is the removal of financial transactions from the expenditures and revenues.18 Expenditures are likely to contain new loans or equity advances (or both) made and also repayments of government debt principal; receipts will probably contain some types of debt issues and repayments of principal on loans. These financial entries are to be assigned below the line to complete the estimates for the net increases in government debt and government loans. It is useful to keep the balance in the account during this process of netting, reclassification, and condensation, making use of a miscellaneous (or “other sources”) net item. Finally, if necessary, the nonfinancial surplus control figure from the government finance data can be combined with the capital formation figure from the income and product data to obtain a residual estimate for central government gross saving.

As this process proceeds, classification decisions will be made. If ways and means advances from the central bank can be identified in these data, they can be assigned either to the bank loans or government debt category according to local policy usage. For FOF purposes, where they are assigned is not critical; the key point is to make the same assignment in the central bank account.

In broad outline, the estimation of the rest-of-world account from the balance of payments data is a simple affair.19 The debits and credits are reversed to shift to a rest-of-world viewpoint, and the current account is netted down to a single gross saving figure. In the transforming of the capital account, it is essential that the basic data separate asset items from liability items so that financial uses can be separated from sources. The capital flow to the official sector is an estimate of the rest-of-world's acquisition of government debt, and that to the private sectors is an estimate of its acquisition of other domestically issued debt. All the rest-of-world liabilities are assembled into the claim category of foreign assets. To the extent that these are held by the central bank or the commercial banks, they become foreign exchange; in general, other domestic sector holdings are small. It is sometimes analytically desirable to present the foreign exchange of the two banking sectors on a net basis, and that convention is followed here. In principle, then, the rest-of-world's foreign exchange entry is the sum of the net foreign exchange held by commercial banks and by the central bank. These entries are the bulk of the foreign assets line.

The flows for both the central bank and the commercial banks in Table 1 are derived as balance-sheet increments. The basic process is again straightforward. Balance-sheet transcriptions are taken from the basic banking statistics; they are reclassified into FOF format, and increments are taken. To ensure that the currency and deposits liabilities of the banking sectors reflect the entire quantity of money, a small accounting adjustment is sometimes made (as in MBS), transferring any monetary accounts of the central government into the central bank account.

Whenever financial flows are being estimated by balance-sheet increments, as here, the question arises whether these increments reflect only the transactions in the instrument or also reflect book write-ups or write-downs.20 Conceptually, the revaluations should be removed so that the flow portrays only the transactions. Practically, the matter hinges on the size of the revaluation in relation to the flow and, indeed, on whether the data availability makes the removal possible. Fortunately, for developing countries, experience indicates that the problem is not serious in the banking sectors as far as the domestic assets and liabilities are concerned.21 As noted earlier, it is most likely to arise in foreign assets and with respect to net foreign exchange. For example, it has become somewhat conventional for central bank balance sheets to revalue their monetary gold stocks for balance-sheet purposes. Because the international transactions in gold are likely to be small, the flow will be overwhelmed in the balance-sheet increment by a revaluation of the stock, and a removal adjustment will be essential.

With respect to commercial banks, the sector coverage will be governed by the basic aggregation in the banking statistics. It is very important for effective sector articulation that the data used be as of the actual year end and not, for example, as of the end of the last business week of the year. In the account for this sector, it is useful to accept for the time being the definition of bank loans that is found here (interbank loans are eliminated). Further, the bank loan portfolio figures are taken as the statistical control total for the bank loans (other loans and advances) category. With the Table 1 matrix, the only institutional sector breakdown of bank loans that is required is that between central government and other, and the usual balance sheet will provide it. But, to anticipate, a bank loan classification by institutional sector (loans to local governments, government enterprises, other financial institutions) will at a later stage be very useful.

Once preliminary sources and uses of funds accounts have been completed for the four sectors, the assembly of the other domestic sector in Table 1 is simply a matter of calculating the residuals to balance the cross totals. In this process, any items in the discrepancy columns not involving the other domestic sector should be taken as given. In fact, the discrepancy columns should be treated as if they comprise an institutional sector, with a residual estimate at the bottom to force the total net sources to zero.

The matrix at this stage is a fairly rough affair. It is essential now to examine and tighten the articulation of the sector accounts. This is best done by focusing “cross-wise” on the transaction type accounts or the level version of these accounts, which may be called debt distribution tables. For example, suppose that the banking statistics offer a table that breaks down the government debt holdings by sector, or that data are available for portions of the government debt such as treasury bills or government bonds. The assembly of such tables may well have involved adjusting the technical basis of the various sector records into uniformity and may thus enable the compiler to do so as well.22 Most important, this process should enable one to make progress on what are likely to be the main inconsistencies at this stage—the coverage of the types of transaction. Borderline transactions need to be treated in the same way by all sectors across the matrix. For example, if the treasury carries an overdraft, it must consistently be assigned either to deposits or to bank loans.

In the end, the compiler will have a matrix of substantial usefulness both as a way of conceptualizing important relationships and as a source of substantive analysis. A number of types of analysis that are available even with so condensed a matrix were indicated in Section I. If it is urged that a residually estimated “other domestic sector” is not a very firm basis for economic policy, then the reply must be that these estimates are inferred from data for the central government, the banking system, and the balance of payments that are routinely used as the basis for policy. One may as well see clearly what such data imply for the other domestic sector. Finally, as it is hoped this description has made clear, one has an instrument that is feasible and inexpensive to create—easily done by one social accountant in, perhaps, two months.

Tightening the System

One of the great advantages of the initial FOF implementation described above is its launching of a creative stage for the statistical system of a developing country. Broadly speaking, the condensed FOF matrix provides the incentive and points out the direction for several substantial improvements in financial statistics. The nature of these developments will now be indicated.

Table 1 might have very few entries in the discrepancy columns. If this is the case, it would not be because of the high quality of the estimates, but rather because of the necessity for residual estimates for the other domestic sector. In effect, many of the discrepancies will be transferred to and incorporated in the estimates for this sector. These discrepancies can be conceived of as arising from the existence of two differing estimates for one FOF matrix cell—estimates from two differing statistical sources. For example, figures for bank reserve deposits usually appear in the balance sheets of both the central bank and the commercial banks. If the figures differ and if, rather than selecting one as the basis for both bank reserve flow estimates, one retains both flow estimates—each in its respective sector—then one has identified a discrepancy that will appear in the discrepancy column. It is very important in FOF work to make such identifications wherever possible and to have them appear in the discrepancy columns: if the discrepancies are large and involve key policy variables, they should be out in the open so they can be dealt with.

The estimates assembled so far have involved three (largely) separate data sources. Thus it is likely that this sort of discrepancy will already have been encountered and that the differences will turn out to be recalcitrant (that is, the estimates cannot be reconciled by modest efforts). Five possibilities for such discrepancies might well be significant: (1) central government borrowing from abroad (central government versus balance of payments data), (2) central government cash balance (central government versus banking data), (3) change in foreign exchange reserves (balance of payments versus banking data), (4) central government gross saving (central government versus national income and product data), and (5) government debt issue (central government data versus data compiled in the central bank).

To gain a sense of why the exploration of these discrepancies has creative possibilities, visualize the typical statistical “system” of many developing countries. The data sources considered here will to some extent correspond to different data-producing organizations. Perhaps there will be a bureau of statistics producing the national income and product data. Probably the central bank will supervise banking data and the assembly of the balance of payments data, and the treasury (or auditor general) will generate the central government data. In a developing country, these organizations may well generate their respective data with a minimum of technical communication with respect to statistical methods or detailed breakdowns. Coordination and cooperation among governmental agencies may not yet have developed. The point is that a detailed reconciliation effort of the sort required to reduce the FOF discrepancies requires the kind of technical consultation that can become an important force for coordination.

Organizationally, FOF analysis also needs a home. Some single agency needs to be responsible for its generation; an interagency committee will not have sufficient authority to be successful. Probably the central bank will have closest access to the necessary data sources and, eventually, to analytical users. The latter is desirable because users will press for continuous, current generation of the data and for needed types of further detail.

Expanding the System

The strategy for expanding the condensed matrix, either toward the 1968 SNA's sector capital finance accounts scheme or toward a still fuller FOF system, is the same simple one: to break down the “other domestic sector”—that is, to proceed to make direct estimates for subsector accounts and in turn break each subsector account out of the other domestic sector, each time leaving a smaller and more homogeneous, residually estimated other domestic sector. This procedure implies that added sector detail is more useful analytically than transaction detail at this point. This will continue to be true until the main parts of the enterprise sector can be separately shown.

If a country has a substantial system of economic planning, an appropriate statistical direction would be toward a public sector account (that is, to estimate accounts for provincial and local government and for government enterprises). Provincial and local governments themselves (or at least the large ones) should be able to provide income, outlay, and financing data in budget documents. If necessary, earlier actuals can be carried forward on the basis of budget figures. Or it may be possible to assemble a synthetic account from data from the planning ministry if the plan covers provincial and local government capital formation and financing. In addition, perhaps the banking data can provide figures for bank loans to this sector. As this sector is entered into the matrix, the components of provincial and local government debt issues will be defined and allocated. There may be security issues (perhaps assigned into other domestic debt), borrowing from the central government (into central government loans), bank loan borrowing (into other loans and advances), and, perhaps, borrowing from abroad (into other domestic debt).

Next, nonfinancial government enterprises will be considered. Suppose that the central government or planning data do not provide the detail needed to assemble a capital finance account, and so a new compilation must be made. The initial task will be to transcribe the balance-sheet and appropriate income statement items for each enterprise from published annual reports or reports filed with regulatory agencies or with the registrar of companies.23 If there are many such enterprises, a sample of the largest may be advisable; often, a surprisingly small number of government enterprises can account for 80 percent or more of the sector. Given the transcriptions, a standard balance-sheet format must be designed—one as close as possible to the FOF classification but still one into which all the enterprise statements can be transformed in order to permit compilation. If the accounts have been audited by auditing companies, the balance-sheet formats will be surprisingly similar. At this point in the derivation of the FOF account from these data, rough estimates of some items will no doubt be necessary. Finally, either the sample or blown-up estimates for the universe may be used for the FOF sector.

Another subsector that will be quite easy for most developing countries to assemble is that of the nonbank financial institutions. In the early development of a financial system, there are usually few such institutions. The procedure is as outlined above for government enterprises.

The expansions of the condensed matrix are illustrated by Table 2 (above). The sectors in the matrix at this stage are similar to the full FOF scheme, except that private corporate enterprises are not yet separated out of the other domestic sector.24 For many developing countries, this will be as far as the FOF implementation can be taken, given a modest commitment of resources. An assembly of basic financial data for private corporations is not likely to exist, and such a compilation may be beyond the capability of a small statistical shop. Thus, the final stage in this case will await the compilation from tax files, official financial reports, or company reports of aggregate balance sheets and income statements for a significant sample of private companies. Still, these data are important for many uses, and the task of compilation will not be an enormous one.25

Throughout this description of breaking down the other domestic sector, little attention has been paid to the expansion of detail for transaction types. But the coverage of financial transactions has been slowly expanding. Each new directly estimated subsector picks up new claims in its liabilities or its portfolio that were not previously in the matrix because, in effect, they were washed out when the subsector was a consolidated part of the other domestic sector. That sector is de facto a consolidated one because of its residual derivation. Thus, perhaps the greatest increase in the system's coverage of private claims will occur when private corporations can be finally separated from households.

A final commentary relating SNA 1993 to the implementation process as it has been described here should be made. As the developing countries set out to disaggregate the national capital finance account into sector capital finance accounts, their data sources do not permit them first to estimate such an account for general government as a whole and then break it down further into central, provincial, and local government, or to estimate such an account for all financial institutions and then break this down for the central bank, commercial banks, and other financial institutions. As we have seen, the process is rather the reverse. Accounts are first estimated for the key subsectors—those of policy importance—and these are later added to in order to assemble government or financial institutions or nonfinancial enterprises as a whole. This suggests that perhaps SNA 1993 should present a sequence of standard systems—a developmental sequence that would guide countries toward the implementation of a full SNA system.

IV. Conclusions

The conclusions of this paper may be summarized as follows. First, a survey of the applications of FOF accounts concluded that analysis based on these data could be of considerable use to many nations. These accounts are an especially effective data organization for the key public policy problems in the financial area. Moreover, the techniques can vary according to need: from crude flow tracing, to the use of the saving-investment process scheme, to highly sophisticated programming models. If the case for the usefulness of FOF accounts is correct, one may infer that these accounts should have a central place in the standard SNA.

Several more specific conclusions emerge from an examination of the representation of finance in the 1953 SNA and the 1968 SNA and in the forthcoming revision SNA 1993, as follows.

•For the proper articulation of national accounts as a system, the rest of the world needs to be treated clearly as an institutional sector.

•The path taken in the 1968 SNA of deconsolidating the national capital finance account and separately sectoring financial institutions was correct. In effect, the comprehensive set of sector capital finance accounts becomes the FOF system of accounts.

•However, the 13-claim financial transaction classification in the 1968 SNA is not well designed for analysis. It should contain at least key elements of the “three-dimensional classification” identifying creditor and debtor sectors, as is completely done in Table 24 of the 1968 SNA. At the same time, the standard classification needs to be much simpler, at least for developing countries.

•From the viewpoint of many developing countries, the standard accounts should emphasize three main capital finance accounts—those for the rest of the world, for general government, and for the monetary and banking system. This may imply, at least initially, a catch-all “other domestic sector” account.

Because the Fund's statistical systems (BOPS, GFS, and MBS) will provide basic financial data for the estimation of the key capital finance accounts, for SNA purposes these systems need to be harmonized. The issue is a difficult one because of the independent life of these systems within the Fund. The compromise position is to provide for reconciliation among the systems. If harmonization fails, reconciliation should be made an urgent priority. For harmonization, the effort should concentrate first on sector articulation and second on a uniform definition of the sector deficit or surplus.

The relations among flows, balances, and valuation changes must also be considered. In the estimation of flows from balances in order to derive capital finance accounts, the difficulties posed by valuation changes involve only the estimates for (1) corporate equity securities, (2) long-term bonds, and (3) foreign assets (including gold). In practical terms, especially for developing countries, these difficulties do not appear to be great.

A simple but analytically effective form of the FOF system can now be implemented quickly and easily, even by most developing countries. To demonstrate this statistical feasibility, the paper described a process by which these accounts can be derived. Showing the movement of estimates from the simplest loose matrix toward a tighter and more detailed formulation also shows the developmental or evolutionary path of implementation.

Several elements of the analysis have suggested that the design of SNA 1993 should take such an evolutionary view of the development of national accounts. Much of the 1968 SNA has not been implemented, and there are indications that the 1968 SNA is too complex to be a feasible next step in statistical development for many countries. Moreover, the actual process of FOF estimation necessarily has an evolutionary character. The new SNA might wish, therefore, to present both simpler and more complex standard systems, in a sequence that would provide developing countries with a useful immediate system and, at the same time, with guidance in achieving the full SNA system.

Rest-of-world saving, lending, and borrowing could easily be shown separately in another variant.

If accounts 1 and 2 are consolidated, one obtains gross national product on the right and an income total on the left that may be referred to as gross national income.

Page 14, paragraph 285: “No further classification of these [net changes in financial assets and liabilities] flows is undertaken in the system since, to produce meaningful results this would require a greater number of sectors including, in particular, a separate sector for banks and other financial intermediaries.”

Private nonprofit institutions serving households need not be a separate sector in the standard accounts.

When social accountants try too hard for conceptual neatness, complexity is sometimes the result.

United Nations, National Accounts Statistics (New York, annual).

The allocation is straightforward except for category A. Foreign assets consist of all rest-of-world liabilities held by domestic sectors as financial assets. In the cases of the central bank and the commercial banks, however, it is somewhat customary to assign to this category (by netting against their foreign assets) certain of their short-term deposit liabilities held by the rest of the world. In this way, the foreign asset holdings of these two banking sectors become net foreign exchange—the balances used to effect international transactions.

To simplify this discussion, the subclasses indicating sectors holding the particular claim are neglected here.

A simplification would be not to have separate income-outlay accounts for nonfinancial enterprises or financial institutions. The latter account is analytically insignificant, and the former greatly overlaps in content the gross product and income account. In effect, these two income-outlay accounts could be merged with the larger account without loss of analytical power.

Technical readers will be aware that, for countries with multiple exchange rates, the conversion of the account from local currency to dollars or vice versa cannot be satisfactorily done with the use of a single conversion ratio.

The minor items are neglected here: purchases of land, net; purchases of intangible assets n.e.c., net; and capital transfers, net.

United Nations, Provisional International Guidelines on the National and Sectoral Balance-Sheet and Reconciliation Accounts of the System of National Accounts, Statistical Papers, Series M, No. 60 (New York, 1977).

The reconciliation entries will also include some nonrevaluation elements, such as changes that are due to asset items not included in the capital finance accounts and statistical discrepancies and discontinuities.

In principle, one may also find revaluations that are due to exchange rate response in the case of debts held by foreigners, but such cases seem uncommon.

Note that the chapter in the 1968 SNA devoted to developing countries was concerned with how the full system, presumably once that system was achieved, could be adapted to the development situation.

Use of this format is recommended not only for countries with the least-developed statistical systems but also as the starting FOF format for countries at a more advanced statistical stage.

As a first approximation, these lines for the banking sectors may be assumed to be zero.

Other technical problems may involve the removal of internal transaction entries and the adjustment of accrual items.

It is assumed here that the figures for the balance of payments current account surplus from the balance of payments data and the rest-of-world saving from NIP data are equal, apart from sign reversal.

The issue here is not affected by whether the gain or loss is realized but, rather, by whether balance-sheet valuations reflect the gain or loss, realized or not. The sale of a financial asset that realizes and records a gain should be viewed as a book revaluation followed by a transaction.

It is assumed here that the commercial banks’ long-term bond holdings are small.

The major origins of discrepancies may be listed as differences in (1) sector coverage, (2) transaction classification, (3) basis of valuation, (4) timing of recording, and (5) degree of netting or of consolidation.

Gross saving can be approximated by the sum of retained profit plus the depreciation charge from the income statement. Gross capital formation can be approximated by the increment in gross fixed assets.

The other domestic sector also still includes local government and the government enterprises and nonbank financial institutions lying outside their respective samples.

Far off on the horizon, the question might arise whether to split a noncorporate business sector away from a remaining household sector, as is done in some full FOF systems. Even if data are available, such a split would be highly arbitrary in the institutional context of most countries. It would seem wise for SNA 1993 to leave noncorporate business sectored within households, except for the quasi-corporate portion.

    Other Resources Citing This Publication