Chapter

8. Financial Statistics

Author(s):
International Monetary Fund
Published Date:
July 2008
Share
  • ShareShare
Show Summary Details

Introduction

8.1 This chapter offers a systematic approach to developing a country’s financial statistics, building on the principles described in Chapter VIII of the MFSM and describing the steps for compiling one or more of three levels of financial accounts—basic flow of fund accounts, a 1993 SNA-integrated financial account and corresponding balance sheet, and detailed financial statistics. This chapter is based on international guidelines found in the 1993 SNA and the MFSM and on methodologies developed at the national level.

8.2 The purpose of this approach is to facilitate the compilation of financial statistics in countries that do not currently have such statistics, as well as to assist other countries in improving the quality of their financial statistics with respect to (1) adherence to international accounting and statistical standards for classification, economic sectorization, valuation, and application of other accounting rules for financial assets and liabilities, (2) comprehensiveness and consistency of stock and flow data; and (3) appropriateness of the periodicity and timeliness of their financial statistics.

8.3 Before illustrating the three levels of financial accounts and providing compilation steps, the chapter briefly describes the nature and scope of the financial statistics, compilation methods and source data, and the presentation of the resulting data. (The delineation of economic sectors and classifications and valuation methods for financial assets and liabilities are covered in Chapters 3–5 of this Guide.)

8.4 This chapter describes the recommended framework of the financial statistics—balance sheets and accumulation accounts—with emphasis on the flow of funds accounts. Balance sheets show stocks of nonfinancial and financial assets and liabilities on the date for which the balance sheet is compiled. Accumulation accounts are flow accounts that encompass all changes in the balance-sheet accounts between the beginning and the end of the accounting period. The accumulation accounts consist of the capital account, the financial account, and the other changes in assets account, which is divided into the revaluation account and the other changes in the volume of assets (OCVA) account. Flow of funds accounts—a subset of financial statistics—are transaction accounts that are linked to the other accumulation accounts and the balance-sheet accounts.

8.5 The chapter then describes the presentation of financial statistics, which is in matrices or in time-series tables. The matrices contain sectoral presentations for one or more time period, whereas the time-series tables show the realizations over time for flows and outstanding stocks of individual financial assets and liabilities.

8.6 Success in compiling financial statistics heavily depends on the quality of the source data. Because the availability of source data differs across countries, the chapter examines both core data sources and supplementary data sources. Core data for the financial statistics are available from the Depository Corporations Survey (DCS), the balance of payments statistics, and central government records on a quarterly basis. This Guide recommends that the financial statistics be compiled on a quarterly basis with a time lag of no more than 16 weeks.

8.7 The degree of complexity of compilation on a quarterly basis depends on country circumstances with respect to sectoral and financial instrument coverage and the level of sophistication of the financial statistics being compiled. However, compilation of financial statistics on a less frequent and/or less timely basis (for example, on an annual basis and with a six-month or longer lag) would significantly diminish the policymaking usefulness of the data.

8.8 Compilers are encouraged to apply the valuation and classification guidelines recommended for monetary statistics in this Guide, except for in special cases identified in this chapter. In compiling the financial statistics, several types of estimation procedures are needed. Hence, this chapter describes a diverse collection of estimation and compilation methods.

8.9 At its core, this chapter illustrates three levels of financial statistics:

  • (1) Basic flow of funds accounts that are designed for a developing country that wishes to analyze intersectoral financial flows, but for which data sources are limited;

  • (2) A 1993 SNA -integrated financial account and corresponding balance sheet that fits the needs of an emerging-market country that already has the basic flow of funds accounts and that wishes to enhance the usefulness of the data for policy analysis, through introduction of stock data and flow data that have further disaggregation by economic sector and financial asset category; and

  • (3) Detailed financial statistics that are applicable for a country that has well-developed capital markets and that desires to have financial statistics that contain thorough coverage of all economic sectors and financial instruments specified in the methodology of the 1993 SNA.

8.10 Because countries are at various stages of financial development and capabilities for compiling financial statistics, the chapter recommends that, as a starting point, compilers begin with the basic flow of funds accounts—the simplest form and level of financial statistics. These statistics can be compiled by allocating a minimum amount of resources and by using existing source data only.

8.11 This chapter also describes (1) estimation techniques of missing data and (2) editing, residual calculations, and discrepancies. Numerical examples for the compilation of the financial statistics are provided in some of the tables in this chapter.

8.12 Financial statistics have many benefits, including usefulness in revealing weaknesses in the underlying data for the various sectors. Being able to use financial statistics to identify and quantify discrepancies in the data across economic sectors is one of the most important outcomes for countries that are compiling the financial statistics for the first time.

8.13 Another benefit is that the financial statistics link the financial activities of the nonfinancial sectors to those of the financial corporations (FCs) sector. Data on loans and capital market instruments such as securities show the extent to which countries use FCs and capital markets to obtain funds to finance economic activity. The data offer means for assessing the relative importance of various types of financing and for monitoring the changes in the sources of financing over time. The data indicate the sources of funds to FCs and other sectors. Forms of financial-asset accumulation—deposits, pension and life-insurance reserves, securities, etc.—are also identified. Financial statistics provide a means for examining the contribution of domestic and foreign sources of financing to a country’s current expenditures and capital formation.

8.14 Policymakers use financial statistics to analyze economic and financial developments within countries and to compare economic and financial developments among countries. For example, financial statistics as described in this Guide are an important input to the IMF’s balance-sheet approach to analyzing a country’s vulnerability to external or internal shocks. The financial account shows the flow of funds from net saving sectors to net borrowing sectors, channeled through intermediation in the financial sector or, to a lesser extent, through direct lending between the nonfinancial sectors. The financial statistics record the distribution and redistribution of financial assets and liabilities among the sectors of the economy on a quarter-by-quarter basis.

8.15 A numerical example to illustrate the compilation of financial statistics in practice is provided at IMF.org (www.imf.org/external/pubs/ft/cgmfs/eng/index.htm). An electronic spreadsheet and accompanying documentation describe the linkages between the underlying data (direct source data and estimates) and the complete set of financial statistics.

Definition, Scope, and Framework

Definition and Scope of Financial Statistics

Financial statistics consist of a comprehensive set of stock and flow data on the financial assets and liabilities of all sectors of an economy. The financial statistics are organized and presented in formats designed to show financial flows among the sectors of an economy and corresponding financial asset and liability positions. (MFSM, ¶11)

The scope of the monetary statistics is limited to the assets and liabilities of the financial corporations sector and its subsectors. In contrast, the financial statistics encompass all financial stocks and flows among all sectors of the economy and between these sectors and the rest of the world. (MFSM, ¶405)

8.16 Financial statistics are defined as a comprehensive set of stock and flow data on the financial assets and liabilities of all sectors of an economy. This Guide uses the term financial statistics rather than flow of funds accounts to avoid the ambiguity that results from the various forms of flow data. The set of financial statistics, which is broader than the flow of funds accounts, includes data on stocks, and separate categories of flows in the form of transactions and other types of flows. This Guide uses the term flow of funds accounts to refer to accounts for transactions only.

8.17 The scope of the financial statistics is all financial stocks and flows among all sectors of the economy and between these sectors and the rest of the world. Chapter 3 indicates that, in addition to the FC sector, the total economy consists of the general government sector, nonfinancial corporations sector, and other resident sectors (households and non-profit institutions serving households, or NPISH). Data for the sectors’ financial positions with units in the rest of the world are designated as claims on and liabilities to nonresidents, or as foreign assets and foreign liabilities. Sectors also have subsectors. The general government subsectors are the central government, state and local government, and social security funds (if classified separately). The nonfinancial corporations subsectors are public nonfinancial corporations and other nonfinancial corporations. (The sectors and subsectors are described more fully in Chapter IV of the 1993 SNA.)

8.18 In addition to economic sectors, financial assets are basic building blocks for the financial statistics.

The major categories of financial assets, described in Chapter 4 of this Guide, are monetary gold and SDR holdings (central bank only), currency and deposits, securities other than shares, loans, shares and other equity, insurance technical reserves, financial derivatives, and other accounts receivable/payable.

Framework for Financial Statistics

The balance sheets and accumulation accounts are the recommended framework for financial statistics because they provide an internationally recognized set of guidelines for integrating financial stocks and flows into a complete system of accounts. The balance sheets and accumulation accounts cover the transactions, other flows, and stock positions that are relevant for broad financial analysis. (MFSM, ¶421 and ¶422)

8.19 The framework for financial statistics is the balance-sheet accounts and accumulation accounts in the 1993 SNA. The balance-sheet accounts cover stock positions of assets, liabilities, and net worth. The accumulation accounts consist of the capital account, financial account, and other changes in assets account, subdivided into the revaluation account and OCVA account. Box 8.1 provides a schematic view of the interrelationships among these accounts.

Balance sheets

The balance sheets show stocks of nonfinancial and financial assets and liabilities on the date for which the balance sheet is compiled. The difference between total assets and total liabilities is net worth. For each group of assets and liabilities, and thus net worth, changes between the opening and closing balance sheets result from transactions and other flows recorded in the accumulation accounts. (MFSM, ¶418)

8.20 The balance sheets are a database of time series that are built up from the source data to obtain the matrix of balance sheets for the sectors. For each time series, the database should identify the source, the matrix cell in which the data are included, and the manner in which the data are combined with other time series to produce the cell totals. The stored information usually includes greater detail than is required for compiling the matrix. The database can contain hundreds of categories and subcategories, depending on the level of disaggregation of the source data.

Box 8.1.Relationship Between the Balance Sheets and Accumulation Accounts in the 1993 SNA

8.21 In some areas, however, the source data are less detailed or less timely than is required for the compilation of the financial statistics. In some cases, no direct source data exist, and the data must be estimated or otherwise derived, as described in later sections of this chapter. Most countries have little or no direct source data for other resident sectors (households and NPISH). Depending on the type of transaction, the direct source data for transactions and positions with the rest of the world (nonresidents) may be limited, compared with the data availability for domestic sectors.

8.22 In such circumstances, compilers usually apply either of two techniques—counterpart data collection or residual data—to obtain data from balance sheets. Each technique applies the principle that every asset (other than monetary gold and SDRs at a central bank) must have a matching liability. The information that is stored in balance-sheet databases in greater detail than is required for compiling the matrix facilitates the exercise of control over the allocation of counterpart data and residual data. For example, for most tradable securities denominated in foreign currency, the rest of the world is treated as a residual category. To balance the accounts for each category, countries should nominate for each sector in the matrix a statistician responsible for the methodology. Counterparts and residuals are discussed in greater detail in the Source Data section in this chapter (¶8.66–8.79).

Accumulation accounts

8.23 The accumulation accounts consist of the capital account, financial account, and other changes in assets account. Within the other changes in assets account are the revaluation account and the OCVA account.

Capital account and financial account

The capital account records acquisitions and disposals of nonfinancial assets as a result of transactions with other units or internal bookkeeping transactions linked to production (own account capital formation, changes in inventories, and consumption of fixed capital), and measures the changes in net worth as a result of saving and capital transfers receivable from abroad. The balancing item is net lending or net borrowing, depending on whether saving plus capital transfers is less than the net acquisition of nonfinancial assets. (MFSM, ¶417)

Net lending/Borrowing is the balancing item of the capital account, calculated as net saving plus capital transfers receivable less capital transfers payable less acquisition less disposals of nonproduced nonfinancial assets. The net resources available to an economy or sector from saving and net capital transfers that are not used for capital accumulation are the amount of resources available for net acquisition of financial assets, that is, net lending. Economies or institutional sectors with a surplus of resources (through saving and net capital transfers) over capital accumulation are net lenders. Economies or institutional sectors that have capital expenditures in excess of these resources are net borrowers. Changes in net worth arise from saving and capital transfers. (MFSM, ¶428)

The financial account records the acquisition and disposal of financial assets and liabilities, and shows how net lending or net borrowing from the capital account is reflected in transactions in these financial items. The financial account is the last account in the sequence of accounts recording transactions. (MFSM, ¶417)

The financial account shows financial transactions among institutional units and between institutional units and the rest of the world. Financial transactions cover all transactions involving change of ownership of financial assets, including the creation and liquidation of financial claims. (MFSM, ¶429)

To emphasize the fact that financial transactions can be directly measured, the term net financial investment is used to denote the balancing item of the financial account, calculated as net acquisition of financial assets less net incurrence of liabilities. (MFSM, ¶449)

Net financial investment is always equal in concept to net lending/borrowing. A statistical discrepancy can be shown that represents any difference between the recorded total for saving and capital transfers and recorded total net lending. The discrepancy can arise in practice because of gaps in coverage or mismeasurement of any of the items in the full sequence of accounts. (MFSM, ¶450)

8.24 Compilers use the capital account to derive the statement of sectoral net lending or net borrowing, which is the typical starting point for compiling the financial statistics. Net lending or net borrowing is the balancing item of the capital account, equating saving and capital transfers with a sector’s net acquisition of nonfinancial assets.

8.25 The financial account—the core of financial statistics—shows how the sectors allocate their excess funds (net lending position) or meet their financing requirements (net borrowing position). The financial account shows that part of each sector’s accumulation accounts that consists of transactions in financial instruments—the net acquisition of financial assets and the net incurrence of liabilities. In the financial account, the difference between the net acquisition of financial assets and net incurrence of liabilities is referred to as net financial investment. Although calculated separately by using the data in the financial account, net financial investment is conceptually equivalent to net lending or net borrowing in the capital account. In practice, independent compilation in the capital account and financial account contexts results in a discrepancy. Dealing with the discrepancy is discussed in the Statistical Discrepancies section of this chapter.

8.26 As in assembling balance sheets, compilers assemble financial accounts from a database of time series, which is built up from the source data to obtain a matrix of financial transactions by sector, which constitutes the financial account. For each time series, the database should identify the source data, the sectoral matrix cell in which the data belong, and the manner in which the data are combined with other series to produce the cell total.

8.27Financial transactions measure the net acquisition of financial assets or the net incurrence of liabilities for each type of financial instrument. Some sectors are net lenders, while others are net borrowers. All financial transactions involve (1) simultaneous creation or liquidation of a financial asset and a counterpart liability, (2) conveyance of ownership of a financial asset, or (3) incurrence of a liability. A financial asset of one unit is matched by a liability of another unit (except in the case of monetary gold or SDRs).

8.28 A financial transaction increases the net lending/borrowing by one institutional unit and, by the same amount, decreases net lending/borrowing by another unit. Therefore, the financial account shows how a deficit sector finances its net borrowing by reducing its assets and/or incurring liabilities. Sectors with surplus funds provide financing to other sectors, by acquiring assets (for example, by making loans, adding to deposit accounts or purchasing securities) or by reducing their liabilities.

8.29 The financial account provides a complete set of financial transactions for each sector. In principle, flows into and out of each sector should exactly balance, with net total financial transactions for each sector in the financial account equaling the financial net lending/borrowing for that sector in the capital account. In practice, statistical adjustment items are needed to absorb the errors and omissions in the accounts. Compilers will find that the size of the statistical adjustment items and their period-to-period variation are helpful for assessing the quality of the statistics. Large adjustment items may indicate that new source data and/or improvements in methodology are needed.

8.30 In addition to the capital account and financial account, the accumulation accounts include the other changes in assets account, which contains two sub-accounts—the revaluation account and the OCVA account.

Other changes in assets account: revaluation account and other changes in the volume of assets account (OCVA)

The revaluation account shows changes in net worth arising from holding gains and losses on nonfinancial assets, financial assets, and liabilities resulting from changes in the prices of the various assets and liabilities. (MFSM, ¶417)

The OCVA account shows changes in net worth arising from all factors other than transactions as recorded in the capital and financial accounts and holding gains/losses as recorded in the revaluation account. (MFSM, ¶417)

8.31 The revaluation account and OCVA account complete the linkage between the period-to-period flows and the stocks recorded in the balance-sheet account. That is, the change in the stock of an assets or liability category is the sum of the transactions, revaluations, and OCVA for a given period.

Flow of funds accounts

Flow of funds data, presented in a matrix form showing the financial transactions among all subsectors of an economy, are a particular focus of the financial statistics described in Chapter VIII of this manual. (MFSM, ¶12)

Flow of funds statements sometimes cover both financial and capital transactions, thereby providing a link to the capital account of the 1993 SNA. Parallel stock presentations can also accompany flow of funds statements. (MFSM, ¶408)

Flow of funds accounts are sectoral accounts, and, while these accounts place an emphasis on FCs because of their important role in financial activity, they also attach due consideration to the financial activities of other institutional sectors. Flow of funds accounts had their origin as a separate statistical system but are now commonly linked to the nonfinancial economy by their integration within the national accounting framework, particularly through associating financial data with data on saving and capital formation. Flow of funds are transactions accounts, but they are often linked to balance sheet accounts and are prepared in conjunction with accounts of stocks of financial assets and liabilities of each institutional sector. (MFSM, ¶439)

Flow of funds accounts exist in various forms that differ according to the analytical needs that are being addressed and by the complexity and detail of the accounting presentation and data requirements. The simplest flow of funds accounts identify financial transactions of major importance between sectors at an aggregated level. The most complex flow of funds accounts consist of a three-dimensional matrix that relates the creditor sector, the debtor sector, and the financial asset used in the transaction. The preparation of basic flow of funds accounts is within the capabilities of all countries that have reasonably complete systems of balance of payments, government finance, and monetary statistics. (MFSM, ¶440)

Flow of funds accounts that follow the form of the 1993 SNA financial account can, of course, be fully integrated with capital account transactions and with sectoral and national balance sheets. (MFSM, ¶442)

8.32 The MFSM, which defines the flow of funds accounts as transactions accounts, indicates that flow of funds accounts often are prepared in conjunction with the compilation of the stock data for financial assets and liabilities of each economic sector. In addition, the MFSM notes that the scope and presentation of flow of funds accounts may take various forms. Compilers may present them in a matrix that shows the financial and capital account entries for each sector, which is accompanied by a parallel matrix that shows the stocks of assets and liabilities for the respective sectors. As for scope, the MFSM describes simple accounts (which show the financial transactions between sectors at an aggregated level) and complex accounts (which show three-dimensional flows from creditors to debtors for each type of asset). Flow of funds accounts can and often do follow the form of the 1993 SNA financial account, and compilers are able to integrate the financial account with the capital account and with the sectoral and national balance sheets.

8.33 Though countries use the term flow of funds accounts in practice, the term seldom refers to a separate and distinct system of financial statistics. National compilers, including those in different data-collection agencies, usually integrate the various components of flow of funds accounts with other accounts in the national accounts system, which includes integration with the data for a country’s current accounts and international transactions accounts. The integration is based on the broad framework of the 1993 SNA.

Presentation

8.34 The two most common presentations of financial statistics are matrices with sectors and instruments for a particular time period and sector statements for multiple time periods. In addition, time-series tables can be prepared to show the transactions and positions for individual financial instruments.

8.35 The information potential for a country’s financial statistics presents both opportunities and presentational issues. Striking a balance between providing sufficiently detailed data and efficiently conveying information to various users is challenging, particularly given that different users prefer alternative presentations. Substantial detail in tables may hinder a user’s ability to extract the desired information or to uncover trends and other relationships that may be of interest. However, analytically useful information in more detailed data may be eliminated if the tables are overly condensed.

Matrix Formats

8.36 Matrices can be prepared for (1) transactions, (2) stocks, and (3) other flows. The two-dimensional matrix of the flow of funds can be constructed for one period or several time periods. Less commonly, the flow of funds accounts are presented in a more detailed form that is three-dimensional.

Basic flow of funds account

A basic flow of funds account is a modified form of the flow of funds matrix that employs a reduced number of sector and financial asset categories. The sectors chosen are normally those most important for financial analysis and for which data are available—remaining sectors are placed in a residual category. Countries that prepare macroeconomic accounts covering monetary statistics, government finance data, and the balance of payments can construct the basic accounts. Therefore, countries that have limited statistical resources can nevertheless benefit from compiling a set of interrelated and internally consistent sectoral accounts that are useful for analytic and policy purposes. (MFSM, ¶453)

8.37 Basic flow of funds accounts are shown in modified forms of flow of funds matrices—in Table 8.1 (transactions) and Table 8.2 (stocks). The rows and columns of each matrix show data for a single period.

Table 8.1.Summarized Financial Statistics Matrix: Transactions
Transactions 1st quarter 2003Financial Corporations
General Government
Nonfinancial Corporations
Other Residents
Rest of the World
Total
UsesSourcesUsesSourcesUsesSourcesUsesSourcesUsesSourcesUsesSources
A. Gold and SDRs−9000000−90−9−9−9
B. Currency and deposits–10,3607635,4980–1,31708200251–5,871–5,108–5,108
C. Securities other than shares5,09701125,07814705007−5535,2305,230
D. Shares and other equities370−6013350−640393−275086−202−202
E. Loans–9,683−520−14299–4,9280–4,844−154300–9,538–9,538
F. Insurance technical reserves0894000075201420894894
G. Financial derivatives000000000000
H. Other accounts receivable/payable3,1140005,4570003656,1336,13372
Subtotal–14,2741,0045,9455,0643,813–3,8301,297–4,8446196–2,600–8,661
Net financial investment–15,2788817,6436,1416136,061
Table 8.2.Summarized Financial Statistics Matrix: Stock Positions
Stock Positions End of March 2002Financial Corporations
General Government
Nonfinancial Corporations
Other Residents
Rest of the World
Total
AssetsLiabilitiesAssetsLiabilitiesAssetsLiabilitiesAssetsLiabilitiesAssetsLiabilitiesAssetsLiabilities
A. Gold and SDRs8950000000008950
B. Currency and deposits67,373188,38550,073020,893082,59306,42738,975227,361227,361
C. Securities other than shares139,104013,542110,2712478,805005533,872152,948152,948
D. Shares and other equities11,10652,6142,853035,00414,87514,99503,80127067,75967,759
E. Loans66,11832,18805,9421,91520,540036,44830,3003,21398,33298,332
F. Insurance technical reserves024,452000019,79104,661024,45224,452
G. Financial derivatives622210000001590221221
H. Other accounts receivable/payable8,2579,815003,272190001,0222,54500
Subtotal292,915307,67566,468116,21361,33144,410117,37936,44846,42578,875571,969571,072
Net financial position–14,760–49,74516,92180,931–32,450897

Transactions matrix

8.38 In the transactions matrix, each sector has two columns—one for uses of funds and one for sources of funds. If the transactions matrix contains the full statement of flows for a sector, including investment in nonfinancial assets, the sum of all columns for uses of funds for each sector conceptually equals the sum of all columns for sources of funds. The rows of the transactions matrix are the financial instruments that represent the uses and sources for the respective sectors. A financial instrument may be both a source and a use for an individual sector, for example, when a sector both issues and holds equity shares. For each financial instrument, the sum of sources across sectors conceptually equals the sum of uses across sectors. If the identities do not hold because of data limitations, the matrix would show discrepancies between uses and sources for sectors and instruments.

8.39 The example in Table 8.1 shows that shares and other equities (row D) were both a use and a source for FCs. That is, in the first quarter of 2003, FCs had a net share purchase (uses column) in the amount of 370. In the sources column, FCs had a net issuance of –601, meaning that FCs retired equity in the amount of 601.

Stock positions matrix

8.40 The stock positions matrix summarizes the financial assets and liabilities in the sectors’ balance sheets. Each sector is represented by two columns—one for assets and one for liabilities, corresponding to the uses and sources of funds. The rows of the stock positions matrix are the financial instruments. The example in Table 8.2 shows a heavy reliance of the other residents sectors (households and NPISH) on deposits (row B) relative to investment in financial assets such as equity shares (row D). In the example, nearly half of the FCs’ loans were liabilities of other FCs (row E), and nearly a third of the FCs’ loans were liabilities of nonfinancial corporations.

Other flows matrix

8.41 The other flows matrix is used to record revaluations and OCVA, which bridge the gap between transactions and period-to-period changes in stock positions. For each sector, the other flows matrix has two columns for other changes in assets and liabilities, respectively. The rows of the other flows matrix are the financial instruments. Table 8.3 provides an example of an other flows matrix.

Table 8.3.Summarized Financial Statistics Matrix: Other Flows
Transactions 1st quarter 2003Financial Corporations
General Government
Nonfinancial Corporations
Other Residents
Rest of the World
Total
Changes in assetsChanges in liabilitiesChanges in assetsChanges in liabilitiesChanges in assetsChanges in liabilitiesChanges in assetsChanges in liabilitiesChanges in assetsChanges in liabilitiesChanges in assetsChanges in liabilities
A. Gold and SDRs3700000000373737
B. Currency and deposits642−213000000−213642654654
C. Securities other than shares5,84200−2030005,8425,8425,842
D. Shares and other equities567–6,622−3550–4,0551,080–1,7390411–5,541–5,541
E. Loans−2920000−33400951−282−282
F. Insurance technical reserves000000000000
G. Financial derivatives0970000009709797
H. Other accounts receivable/payable0000762000076200
Subtotal6,796–6,738−355−2–3,293749–1,7390−667,335807807
Net other flows13,534−353–4,042–1,739–7,4010

Uses of the matrix presentation

8.42 The matrix presentation is particularly useful in that, taken together, the transactions, other-flows, and stock-position matrices specify the identities that must be satisfied for each sector and each instrument. The matrix format facilitates the immediate identification of intersectoral inconsistencies that can be investigated and corrected by the compilers. The matrices show the significance of the various categories of financial instruments relative to the overall amount of period-to-period changes in the uses and sources of funds for each sector and the entire economy.

8.43 In addition to having data for a specific period, compilers and users can use the sets of matrices for several time periods to trace the impact of trends such as a sector’s changing preferences for particular financial instruments. That is, the matrix presentation specifies that a new financial instrument must be a substitute for the utilization of existing types of financial instruments, or must be financed from additional resources.

8.44 The two-dimensional matrix does not allow compilers and users to identify counterpart sectors. For example, net purchases of shares and other equity by other resident sectors (Table 8.1, row D) amounted to –275, and the other resident sectors’ holding of shares and other equities (Table 8.2, row D) amounted to 14,995. The stock position matrix shows that all outstanding shares and other equity (Table 8.2, row D) were issued by FCs, non-financial corporations, and the rest of the world. However, the stock position matrix does not reveal the extent to which the other resident sectors’ holdings are separately attributable to equity issuances of FCs, nonfinancial corporations, and the rest of the world.

Detailed flow of funds matrices

The SNA financial account may be expanded into a three-dimensional matrix to track financial transactions between source and user sectors and the financial asset used in the transaction. It therefore shows who finances whom and by means of which financial asset. Because of the symmetrical nature of financial assets and liabilities, a single matrix could be constructed, insofar as one institutional unit’s asset is another institutional unit’s liability. (MFSM, ¶463)

8.45 To overcome the analytical limitations of the two-dimensional format, the matrix presentation can be expanded into a three-dimensional structure that shows the counterpart data for transactions in each financial instrument. Three-dimensional financial statistics correspond to detailed flow of funds matrices, as described along with an example in the MFSM.1 The presentation shows counterpart sectors, which are disaggregated by financial asset and liability categories.

8.46 As the number of financial-instrument categories and subsectors increases, the three-dimensional structure may become cumbersome to maintain and to explain to users. This section describes a method for retaining selected elements of the three-dimensional structure, through separation of various matrices in the detailed flow of funds matrix.

8.47 Some elements of the three-dimensional structure can be retained by combining two-dimensional tables. The concept is shown in Box 8.2. By slicing the three-dimensional structure, or cube, into several rectangles along the axis of own sectors, a set of two-dimensional tables, with an axis for counterpart sectors and an axis for financial instrument categories, can be created.

8.48 An example that retains elements of the three-dimensional structure is shown in Tables 8.4 and 8.5. Table 8.4 shows the FCs’ transactions and positions with the various sectors identified in the columns. Table 8.5 shows the same types of data for nonfinancial corporations’ transactions and positions. The own-sector data for FCs (Table 8.4) and nonfinancial corporations (Table 8.5) are those in the summary table. However, data for the counterpart sectors (for example, those for other resident sectors—households and NPISH) reveal the FCs’ positions with each counterpart sector.

Table 8.4.Financial Corporations Table for Three-Dimensional Structure
TransactionsFinancial Corporations
General Government
Nonfinancial Corporations
Other Residents
Rest of the World
UsesSourcesUsesSourcesUsesSourcesUsesSourcesUsesSources
1st quarter 2003
A. Gold and SDRs−900000000−9
B. Currency and deposits–10,3607635,4980–1,31708200–4,238–10,360
C. Securities other than shares5,097004,9600704000−568
D. Shares and other equities370−6013350−733256−3140111114
E. Loans–9,683−520−2299–5,2620–4,844−351425
F. Insurance technical reserves0894000075201420
G. Financial derivatives0000000000
H. Other accounts receivable/payable3,1140004,957000–4,9573,114
Subtotal–11,4711,0045,8344,9583,206–4,3021,258–4,844–9,293–7,284
Net financial position–12,4758767,5086,102–2,009
Financial Corporations
General Government
Nonfinancial Government
Other Residents
Rest of the World
Stock PositionsAssetsLiabilitiesAssetsLiabilitiesAssetsLiabilitiesAssetsLiabilitiesAssetsLiabilities
End of March 2003
A. Gold and SDRs895000000000
B. Currency and deposits67,373188,38550,073020,893082,593034,82667,373
C. Securities other than shares139,1040096,48308,79700033,824
D. Shares and other equities22,10652,6142,853031,5899,72613,53804,63412,380
E. Loans66,11832,18804,7171,85020,474036,44830,3384,479
F. Insurance technical reserves024,452000019,79104,6610
G. Financial derivatives6222100000022162
H. Other accounts receivable/payable8,2579,815003,1280006,6878,257
Subtotal303,915307,67552,927101,20057,46038,998115,92336,44881,367126,375
Net financial position–3,760–48,27318,46279,475–45,008
Table 8.5.Nonfinancial Corporations Table for Three-Dimensional Structure
TransactionsFinancial Corporations
General Government
Nonfinancial Corporations
Other Residents
Rest of the World
UsesSourcesUsesSourcesUsesSourcesUsesSourcesUsesSources
1st quarter 2003
A. Gold and SDRs0000000000
B. Currency and deposits0–1,31700–1,31700000
C. Securities other than shares7040001470500014
D. Shares and other equities256−73300−6403933809993
E. Loans–5,26229900299–4,928003340
F. Insurance technical reserves0000000000
G. Financial derivatives0000000000
H. Other accounts receivable/payable3,5518,499005,457000–3,551–3,042
Subtotal−7516,748003,813–3,830380–3,118–2,935
Net financial position–7,49907,64338−183
Stock PositionsFinancial Corporations
General Government
Nonfinancial Corporations
Otder Residents
Rest of tde World
AssetsLiabilitiesAssetsLiabilitiesAssetsLiabilitiesAssetsLiabilitiesAssetsLiabilities
End of March 2003
A. Gold and SDRs0000000000
B. Currency and deposits020,8930020,89300000
C. Securities other than shares8,797001972478,80500749
D. Shares and other equities9,72631,5890035,00414,8751,45703,6023,415
E. Loans20,4741,850001,91520,540006565
F. Insurance technical reserves0000000000
G. Financial derivatives622210000001590
H. Other accounts receivable/payable5,8558,983003,27219000190143
Subtotal44,91463,536019761,33244,4111,45704,1133,672
Net financial position–18,622−19716,9211,457440

8.49 In Table 8.4, the net transactions in FCs’ equity shares from the other resident sectors during the period was –314, and the stock position of FCs’ equity shares held by the other residents sector was 13,538. In Table 8.5, the value of nonfinancial corporations shares purchased by the other residents sector was 38, and the value of shares held by the other residents sector was 1,457.

8.50 Despite their usefulness for understanding the financial intermediation by FCs and other financial flows in an economy, the two-dimensional and three-dimensional matrix presentations often become too intricate for practical purposes. In addition, each matrix is limited to a single period.

Time-Series Presentations

8.51 Time-series presentations of financial statistics can be compiled in various formats. Time-series presentations may focus on a single sector or combination of sectors, or on particular financial instruments.

Sector tables

8.52Table 8.6 illustrates a truncated financial statistics account for other resident sectors (households and NPISH) for four time periods—2001, 2002, the fourth quarter of 2002, and the first quarter of 2003. The table shows a sequence of accounts—transactions, other flows (revaluations and OCVA), and end-of-period stocks—for the financial instrument categories of currency and deposits, shares and other equities, and insurance technical reserves.

Table 8.6.Other Residents’ Assets (Transactions, Other Flows, and Stock Positions)
Year/Quarter200120022002: Q42003: Q1
Transactions (quarterly data at quarterly rate, not seasonally adjusted)
1Currency and deposits598766661821
2Bank notes and coins15298139−937
3Other resident sector deposits4466685221,758
4Shares and other equities22258412−276
5Financial corp. shares124615−55−314
6Nonfinancial corp. shares98−316738
7Insurance technical reserves3,5354,8541,451752
8Insurance reserves for residents1,5421,960596444
9Pension reserves1,9922,895855308
10Subtotal4,3546,2051,8461,297
Other flows
11Currency and deposits0000
12Bank notes and coins0000
13Other resident sector deposits0000
14Shares and other equities−111,069−17–1,740
15Financial corp. shares−33832−45–1,846
16Nonfinancial corp. shares2223728106
17Insurance technical reserves0000
18Insurance reserves for residents0000
19Pension reserves0000
20Subtotal−111,069−17–1,740
End-of-period stocks
21Currency and deposits81,00781,77381,77382,594
22Bank notes and coins11,87011,96811,96811,031
23Other resident sector deposits69,13669,80469,80471,562
24Shares and other equities15,35917,01217,01214,996
25Financial corp. shares14,25115,69915,69913,539
26Nonfinancial corp. shares1,1071,3131,3131,457
27Insurance technical reserves14,18519,03919,03919,791
28Insurance reserve for residents3,2745,2345,2345,678
29Pension reserves10,91113,80513,80514,113
30Subtotal110,550117,824117,824117,381

Box 8.2.Concept of Three-DimensionalFlowofFunds Accounts

8.53Table 8.6 shows that other resident sectors held 1,107 in shares of domestic nonfinancial corporations at the end of 2001 (row 26). In 2002, their net reduction of these shares amounted to –31 (row 6). Other flows in 2002—assumed to have resulted from revaluations in a period when equity prices were rising—amounted to 237 in 2002 (row 16). At the end of 2002, the value of shares held (row 26) was 1,313 (= 1,107 – 31 + 237).

8.54 A stylistic feature of the construction of time-series versions of the sector tables is the propensity for mixing of different data frequencies in a single table, such as in Table 8.6, in which both annual and quarterly data are shown. Compilers should take care in specifying the dates to ensure the correct interpretation. Stocks are specified as end-of-period amounts, meaning the amounts outstanding at the close of the markets on the last trading day of the year or quarter presented. Flows refer to those for an entire year or an entire quarter. Alternatively, the flows for a quarter can be multiplied by 4, to express the flows as annualized amounts. Annualizing the flow data enables users to compare the data of different frequencies. In the example, the 2003:Q1 amount of 38 for the other residents sectors’ net purchase of shares could be shown as 152 at an annual “rate.” Users could directly compare the annualized quarterly amount with the corresponding amounts for 2001 and 2002, which would indicate that the other residents sectors’ net purchase of shares in the quarter exceed the average quarterly acquisition in the earlier years.

8.55 Quarterly data for the financial statistics (and economic time series in general) are subject to seasonal variation. It is highly recommended that consideration be given, where possible, to the compilation of both unadjusted and seasonally adjusted data. Statistical techniques and issues related to seasonal adjustment are described in Chapter 6 of this Guide, where the focus is on seasonal adjustment of the monetary aggregates. However, the general principles and techniques also apply to seasonal adjustment of financial statistics.

Instrument tables

8.56 Time-series presentations can also be used to summarize the financial activity in terms of the financial instruments underlying the sources and uses of funds. This type of presentation facilitates users’ analysis of trends in financial markets, given that time-series presentations show changes in the financial instrument composition of sectoral funding and reveal the sectors that are providing particular types of financing. Though the financial instrument tables do not contain complete representations—by financial instrument and by sector—of a particular sector’s financing, the tables offer manageable and highly useful presentations of the sector’s funding and financial investments.

8.57Table 8.7 illustrates a time-series presentation for shares and other equity for the same periods as those in the previous example. The transactions rows show net equity issuance and net equity purchases by sector. Only two sectors—FCs and non-financial corporations—and nonresidents (rest of the world) issue equity, but all sectors are purchasers of equity. Net issuance for the economy as a whole is equal to net purchases for the economy as a whole plus nonresidents’ net purchase of domestically issued shares. The end-of-period stock positions for all issuing sectors must equal the end-of-period stock positions for all holding sectors and nonresidents.

Table 8.7.Shares and Other Equity by Holding Sector (Transactions, Other Flows, and Stock Positions)
Year/Quarter200120022002: Q42003: Q1
Transactions (quarterly data at quarterly rate, not seasonally adjusted)
1Issuing sectors total2,4271,9682,161−201
2Financial corporations1,5432,2321,371−601
3Nonfinancial corporations853−300777393
4Rest of the world3137127
5Holding sectors total2,4162,0472,144−201
6Financial corporations884145562371
7General government18−1400336
8Nonfinancial corporations1,2951,3591,558−641
9Other residents22258412−276
10Rest of the world−398129
Other flows
11Issuing sectors total−165,41459–5,561
12Financial corporations−203,05627–6,623
13Nonfinancial corporations22,360281,061
14Rest of the world2−1141
15Holding sectors total−165,4996–5,561
16Financial corporations51,609−2568
17General government01650−336
18Nonfinancial corporations−352,50224–4,056
19Other residents−111,069−17–1,740
20Rest of the world2515413
End-of-period stock
21Issuing sectors total66,14173,52373,52367,761
22Financial corporations54,54359,83959,83952,615
23Nonfinancial corporations11,36113,42113,42114,875
24Rest of the world237263263271
25Holding sectors total62,46870,01470,01464,252
26Financial corporations8,41410,16810,16811,107
27General government2,8282,8542,8542,854
28Nonfinancial corporations35,83939,70139,70135,004
29Other residents15,35917,01217,01214,996
30Rest of the world27279279291

8.58 In the example, equity holdings of nonfinancial corporations (row 28) are much larger than the nonresident holdings (row 30). In 2003:Q1, the non-financial corporations’ sector net issuance of equity was relatively large (row 3), while the financial corporations sector retired more equity than it issued (row 2). The FCs were net purchasers of equity (row 6), and the nonfinancial corporations were net sellers in the period (row 8).

Compilation Methods and Source Data

Compilation Methods

8.59 Conceptually, source data for the financial statistics can be obtained from both parties to each financial transaction/position. If compilers obtained data from both parties, they would have two independent observations—the creditor’s record and the debtor’s record—for each transaction/position. The two parties’ data should be identical but, in practice, often are not. Data collection costs often outweigh the benefits of data collection from both parties. Thus, compilers often need to rely on the data reported by only one party to a financial transaction/position, using the same data for the counterpart to the transaction/position.

Counterparts

8.60 For example, data for the financial transactions of other resident sectors (households and NPISH) can be obtained from the data reported by sectors that engage in financial transactions with other resident sectors. That is, compilers take advantage of the fact that a sector with many transactors, such as other resident sectors, mainly channel the financial transactions through a sector with relatively few institutional units—in particular, the FC sector. The data for other resident sectors, as derived from FCs’ records, are designated as counterpart data.

8.61 To obtain these data, compilers use the framework in this chapter, basing compilation of financial accounts mainly on balance-sheet data of FCs, non-financial corporations, and government units. These data can be used to obtain the values of financial assets and liabilities for the data reporters and for the counterparts to the transactions.

8.62 Compilers use differencing—subtraction of opening balance-sheet values from closing balance-sheet values—to estimate some transactions and other flows involving financial assets and liabilities. They use other data, such as those for write-offs and holding gains and losses, to distinguish transactions from other flows. In some cases, compilers are able to obtain or estimate transactions data directly from source data.

Residual calculations

8.63 When data cannot be obtained directly or from counterpart data, it may be possible to derive the data residually—using the identity that, for each category of financial instrument (other than monetary gold and SDRs), the sum of the net acquisitions (including those of the rest of the world) must equal the sum of the net incurrences of liabilities. If data are available for all but one sector, the net acquisition or net incurrence of liabilities by the remaining sector can be derived residually, using the identity.

8.64 Estimates of stocks (levels) are made by collecting balance-sheet data from various sources and selecting those that are most reliable. Compilers sometimes must make a choice. For example, the government authorities may report data on most of their lending to state-owned nonfinancial corporations. Liability data for the same loans may be reported by the nonfinancial corporations themselves, as borrowings from the government. The two sets of data sometimes may not agree, because one side of the reporting—the government authorities or the nonfinancial corporations—may not have included data from all state-owned nonfinancial corporations. In this case, compilers can use the most comprehensive and reliable data to estimate both the asset and liability sides of these borrowings.

8.65 In deriving financial transactions using the difference between closing and opening balance-sheet amounts, compilers need to extract the component of the period-to-period change that arose from valuation changes, caused by exchange-rate movements and/or changes in the market prices (or fair values) of financial instruments. For example, compilers can revalue the opening stock of foreign-currency-denominated securities (to be presented in national-currency units), using the market exchange rate at the end of the period. They can subtract the recalculated opening stock from the reported closing stock to obtain an estimate of the value of the total flow (in national currency). They then can subtract the estimated value of transactions—obtained by dividing the total flow by the average exchange rate during the reference period (assuming no OCVA)—to derive the valuation change.2 The compiler then examines the plausibility of these estimates and, if necessary, makes adjustments to them.

Source Data

8.66 The matrix presentation provides a means for understanding the source data and the general compilation procedures for financial statistics. The compiler focuses on a combination of aggregation and estimation of data arrayed by sector (moving vertically within the matrix) and allocation of financial instrument totals among the use (asset) and source (liability) categories of the sectors (moving horizontally within the matrix). Whether moving vertically or horizontally through the matrix, the compiler will need to make use of a variety of data sources.

Main source data

8.67 Using the data described in Table 8.8, compilers of financial accounts most commonly start by determining the stock positions for each sector (FCs, general government, nonfinancial corporations, and other resident sectors) and the rest of the world.

Table 8.8.Types of Main Source Data
SourcePossible Use
1. Balance-sheet data of financial corporations (including counterpart data)Use for stock positions of the financial corporations sector. Obtain control totals for financial instruments such as deposits and loans and allocate them to counterpart sectors.
2. International investment positionUse for stock positions for the rest of the world.
3. Balance-sheet data for nonfinancial corporationsUse to derive the structure of the balance sheet and to obtain data on households’ financing by nonfinancial corporations, such as trade credits.
4. Government debt dataUse as a benchmark in determining control totals for government debt categories, such as treasury bills.
5. Financial market/custodian dataUse for control total of financial instruments such as securities other than shares and shares and other equity, and allocate among issuing/holding sectors.
6. Financial account of balance of payments and government finance statisticsUse for transactions with the rest of the world and general government sectors, including counterpart sectors.
7. Market price indices (for example, share price index)Use to convert book values into market values for financial instruments such as shares and other equity, in order to separate transactions from revaluations.

8.68 Compiling data for the stock positions may or may not be straightforward. If balance-sheet data are available for most major sectors (FCs, general government, and nonfinancial corporations) and for the international investment position (IIP), the stock positions for the other resident sectors (households and NPISH) can be derived residually. However, the actual process is often less straightforward, because balance-sheet data for the general government sector, nonfinancial corporations, and the IIP are unavailable or are not sufficiently disaggregated.

8.69 For countries in which data availability is very limited, compilation of the financial statistics is heavily dependent on the balance-sheet data for FCs, which are used for the counterpart data, as well as for the FC component of the financial statistics. The compilation starts with the preparation of cross-sections from the three-dimensional financial statistics matrices. Basic two-dimensional matrices are obtained by aggregating counterpart breakdowns into major financial instrument categories. This process is described as compilation of Level 2 financial statistics.

8.70 Some countries also have limited availability of balance-sheet data for the nonfinancial corporations sector. Though this source is very useful for obtaining data for financial instruments such as trade credits, the data typically have shortcomings with respect to the coverage of institutional units, the level of detail for financial assets and liabilities, and the frequency and timeliness of the data. When the shortcomings are severe, compilers use the data as benchmarks for estimating the sectors’ stock positions.

8.71 To fill gaps in the data for nonfinancial sectors, compilers can rely on obtaining control totals for financial assets and liabilities—in essence, moving horizontally across the three-dimensional matrix for the financial statistics. Control total refers to the total amount (stocks or flows) of a certain financial instrument issued and held. The compilers can obtain the data for control totals from the following sources:

  • Balance-sheet accounts of the FCs (mainly for non-negotiable instruments such as deposits and loans);

  • Government records for government debt; and

  • Other sources, such as the records of financial market participants or custodians/security registration offices (mainly for negotiable instruments such as securities other than shares and shares and other equity).

8.72 Compilers will find the horizontal approach particularly useful for financial instruments such as securities other than shares and shares and other equity, but they may find that the available data are not sufficient for a full allocation across all sectors. If the financial-markets data are not fully disaggregated by sector, the compilers will need to allocate the aggregate data among issuers and holders, using whatever multiple sources of data and ancillary information are available. Moreover, one sector will need to be designated as residual, thereby ensuring that the total value of the transactions or position in a financial asset/liability category is fully allocated across the entire economy.

8.73 For the flow data, procedures are needed for estimating the amounts of transactions, revaluations, and OCVA. Data from the government finance statistics and the financial account of the balance of payments statistics are useful as direct source data for transactions, or for use in estimating transactions for the general government sector and the rest of the world, respectively. Securities price indices are most often used to estimate valuation changes for securities, when market values of the securities are unavailable. For categories of financial assets and liabilities that are not subject to revaluation and have not experienced OCVA, transactions can be calculated from period-to-period changes in stock positions.

Supplementary source data

8.74 Supplementary source data can be used to improve the data estimates and to fill gaps in the primary source data. Some types of supplementary data are described in Table 8.9. For example, compilers can use government survey data that are applicable to various sectors. Tax records provide useful data in countries in which confidentiality rules do not preclude the use of some such data for statistical purposes. Other useful data sources are trade associations for specific industries and markets, which normally are active in surveying their members and publishing data for their specific markets.

Table 8.9.Types of Supplementary Source Data
TypePossible Use
1. Special surveysData on nonfinancial sectors’ financial activities such as household savings and borrowings, as well as business financing.
2. Tax recordsBalance-sheet data of corporations and nonprofit institutions for use as benchmarks.
3. Trade association publicationsData on activities of other financial corporations (OFCs) and transactions for particular types of financial instruments, such as financial derivatives.
4. Private data vendorsData on activities of particular financial market activities, such as asset securitization.
5. Profit and loss statementsData on OCVA (such as loan write-offs) to be used in separating transactions from OCVA.

8.75 Other private sources of data may include vendors—corporations that compile data for specialized markets or for consulting purposes. Securities rating agencies may be included among these vendor sources. Availability of vendor data to compilers may require specific contractual arrangements with these private sources. Publicly available financial statements—in particular, balance sheets and income statements of large corporations—are useful data sources for OCVA such as loan write-offs, but such data may be available only on an annual or semi-annual basis.

Macro and micro source data

8.76 Source data for the compilation of financial statistics are either macro (for an entire sector or an entire category of financial instrument) or micro (for specific institutional units or specific transactions). An example of macro data is a major category of other depository corporations’ (ODCs’) assets, such as total loans or the total amount of equity shares outstanding for the corporations listed on an exchange. Examples of micro data are those in the balance sheets and income statements of individual nonfinancial corporations and records for the issuance and redemption of individual securities. Compilers need to use estimation techniques to convert micro data into the aggregate data for the financial statistics, especially when micro data are available for only a sample of all units in a sector or for only part of the holdings of a particular type of financial instrument.

Comparability of information

8.77 When a wide variety of source data are used, compilers need to consider the extent to which the data are in accordance with the classification and accounting principles in the preceding chapters of this Guide. Compilers may need to adjust the data for reconciliation across sources; in fact, such adjustments usually are needed. Moreover, the quality of such adjustments has a significant impact on the usefulness of the financial statistics.

8.78Table 8.10 shows several types of adjustments that often are needed in reconciling the source data. The coverage of the source data may not exactly match the units in the sector for which the data are applicable. Data disaggregation with respect to counterpart sectors and/or financial asset and liability categories in source data may differ from those of financial statistics. Also, valuation of the financial assets/liabilities in the source data may not accord with the valuation rules in Chapter 5 of this Guide. Thus, compilers need to adjust the valuations in the source data. In addition, the source data may be recorded on a cash basis, necessitating adjustment so that the data are on an accrual basis.

Table 8.10.Adjustments for Data Comparability
TypeReason for Adjustment
1.CoverageFor sectors, macro data may cover units that are not in a sector or may exclude some units in the sector. Micro data may not cover all transactions or positions or may include some transactions that belong in other categories of financial assets/liabilities.
2.SectorizationSource data may differ from the sectorization used for the financial statistics, for example, when a subsidiary should be classified as an FC but, in the source data, is included in the nonfinancial corporations sector, along with the parent corporation.
3.Classification of financial instrumentsClassification of financial assets and liabilities in source data may differ from the classifications for the financial statistics.
4.Time of recordingIn the source data, transactions may not be recorded on an accrual basis. Time of recording of a transaction may differ for the two parties to the transaction.
5.ValuationSource data (for example, for securities) may not be valued on the basis of current market price or fair value. Valuation of foreign-currency-denominated assets/liabilities may not be based on the appropriate foreign exchange rates.

Discrepancies

8.79 Compilers often must deal with a lack of data, but they sometimes may be confronted with multiple or overlapping sources of data for a financial instrument or a sector. Conflicts arise when one source of data (for example, balance-sheet data) shows amounts that are appreciably different from those in another source (for example, financial markets data). Compilers must assess the relative quality of the data from the alternative sources, resolve the differences, or report the discrepancies.

Dealing with Reporting of Imperfect Data

8.80 In an ideal world, both parties to a transaction would accurately record the transaction at the same time, would use the same initial valuation and later revaluations, and would report the data on a timely basis. Compilers would collect and compile the data, needing only to check to ensure the data match and to eliminate transcription errors, etc. Such a system is clearly not feasible, but is a useful conceptual benchmark. In reality, the data may not match across reporters and may be prone to errors. Compilers often need to deal with data reporting that is inconsistent, partial, or indirect, as well as with the absence of data reporting.

Inconsistent reporting

8.81 When both parties report a transaction, several types of errors may occur and thereby introduce a data inconsistency. The parties may use different methods for the valuation of transactions or positions, different timing for recording transactions, or different classifications by type of transaction or by sector of the counterpart.

8.82 An example is the positions arising from inter-bank transactions. Suppose all ODCs report the amounts of their claims on and liabilities to other ODCs. The net balance—total interbank claims less total interbank liabilities—should be zero, but often is not. Compilers must adjust the data to overcome the discrepancy.

8.83 Another example is the data for local government. ODCs report data for deposits by, and lending to, local government. Suppose the local authorities also provided data on ODC loans and deposits. The figures may not agree for reasons such as time-of-recording differences or more unusual reasons, such as the local government accounts not being compiled on a strictly calendar-quarter basis.

Partial reporting

8.84 For many matrix cells, only one primary source of data exists. In some cases, the primary data—for example, data on net issues of government securities—may be complete and of high quality. In other cases, only partial data may be available, because the data are from a survey that covers only a subset of the units in a sector or the response rate of data reporters is less than 100 percent.

8.85 Errors may occur when data are obtained from sample surveys. Suppose a survey covers only a subset of all financial cooperatives. The compiler’s first task is to identify financial cooperatives. If the financial cooperatives are too numerous, it may be necessary to obtain data from only the largest ones. Of those cooperatives, only a subset may respond on a timely basis. Using whatever data are reported, the procedures for grossing up from the sample data to the estimated data for all financial cooperatives can compound the data measurement errors.

8.86 For the rest of the world, partial data in the form of custodial holdings of securities may be available. These data provide a lower bound (minimum amount) to be used in estimating the total securities holdings.

8.87 Data may also be partial because of the impracticality of requiring that the reporter provide the detailed data that would be ideal for the financial statistics. Compilers may find that, in principle, some types of transactions data should be divided between two or more lines in the financial matrix. However, the additional reporting burden that would arise from requiring the detailed data often cannot be justified.

Indirect reporting

8.88 Indirect reporting occurs in two forms. One form occurs as the absence of data directly from a sector. In this case, compilers obtain more detailed data from the counterpart sector. This case is particularly applicable to data for nonfinancial corporations and households/NPISH sectors. The other form of indirect reporting arises when balance-sheet data for the beginning and end of the period are used as the basis for estimating the net transactions within the period. Factors such as changes in the relevant population, changes in valuations, and OCVA such as write-offs are considered in the estimation.

8.89 Compilers usually derive transactions in deposits and loans from stocks, given that deposits and loans are recorded at book values. For loans and deposits denominated in national currency, the transactions are equal to the period-to-period changes in the stocks less OCVA. For loans and deposits denominated in foreign currency, period-to-period changes in the stocks must be decomposed into separate estimates of transactions and valuation changes arising from variations in exchange rates. Compilers may need to make assumptions about the proportions of loans or deposits denominated in various foreign currencies and the average exchange rate at which the transactions denominated in these currencies took place.

No reporting

8.90 No direct or indirect data may be reported for some types of financial transactions. Hopefully, information will occasionally become available to provide some indications of the outstanding stocks or flows for these financial assets/liabilities. Examples are transactions in trade credit or loans from the nonfinancial corporations sector to households and transactions between other resident sectors (households and NPISH) and the rest of the world. In such cases, compilers may need simply to omit the transactions from the accounts and admit to a lack of coverage in the data.

Main sources and residual sectors

8.91 The major data sources contain much of the data needed for the financial statistics, but do not cover the transactions that do not pass through the financial system—in particular, transactions for which both parties and counterparties are nonfinancial corporations or other resident sectors (households and NPISH) and transactions of these sectors with the rest of the world.

8.92 To incorporate some of the transactions conducted with the rest of the world, a survey of major nonfinancial corporations may be implemented to obtain data on balance of payments flows such as those from direct foreign investment.

8.93 To ensure that asset transactions equal liability transactions across each row, compilers may designate one or more sectors as the residual sector(s). Any errors and omissions for a particular row are allocated to the residual sector(s). If more than one residual sector exist, each of these sectors is allocated a proportion of the total residual amount, based on the best information available to compilers.

Systematic Development

8.94 Countries need a wide range of source data and methods to compile the full range of financial statistics. Given differences in source data availability across countries, a single level and presentation of financial statistics is not applicable to all countries. This chapter describes compilation methods and issues for three levels of accounts.

  • Basic flow of fund accounts, which are designated as Level 1 financial statistics. Developing countries that have limited sources of data may use these statistics for analyzing intersectoral financial flows.

  • The 1993 SNA -integrated financial account and financial balance sheet, which are designated as Level 2 financial statistics. Emerging market countries are likely to develop these statistics after they already have the Level 1 statistics and wish to enhance the analytical content by incorporating interactions among a larger group of sectors and a more detailed set of financial instruments.

  • Detailed financial statistics matrices, which are designated as Level 3 financial statistics. Countries that have well-developed capital markets may wish to expand their financial statistics to delineate each sector and each category of financial instruments in the 1993 SNA. The Level 3 financial statistics can be integrated with other components of the national accounts statistics, such as the production and income accounts.

8.95Table 8.11 describes the basic differences in the stock/flow data in financial statistics at Level 1, Level 2, and Level 3. The Level 1 statistics are flow data only. The source data for Level 1 may include reported data for transaction flows, but period-to-period changes in stocks are usually used to estimate the transactions. Level 2 statistics contains both flow and stock data, which are compiled with heavy reliance on stock data for the calculation of flows and, where possible, with separate estimates of transactions, valuations changes, and OCVA. Level 3 statistics provide a more detailed decomposition and reconciliation of stock/flow relationships, using data on changes in market values (or fair values) of financial assets/liabilities and OCVA data.

Table 8.11.Levels of Financial Statistics
CharacteristicsLevel 1Level 2Level 3
1.Use of flow and stock dataShows flows only; relies on changes in stocks where flow data are not available.Shows both flows and stocks; relies heavily on period-to-period changes in stock positions for compilation of flows.Shows both flows and stocks; reconciles stock and flow data using accounts for revaluations and OCVA.
2.Accounting entriesDouble entries arranged in one column: source (inflow) is positive, and use (outflow) is negative.Quadruple entries arranged in two columns: both sources and uses shown for each sector.Quadruple entries arranged in two columns: both sources and uses shown for each sector.
3.Sector detailFour sectors: central government, depository corporations (DCs), private sector, and rest of the world.Expanded domestic sector coverage to include state and local government, social security funds, nonfinancial corporations, and other residents (household and NPISH). Financial sector expanded to include DCs and other financial corporations (OFCs).Full set of sectors in line with the 1993 SNA and MFSM, including separation of households and NPISH and expanded detail for financial intermediaries.
4.Instrument detailNone. Only total transactions shown.Most basic categories, including currency and deposits, securities other than shares, shares and other equity, loans, financial derivatives, and insurance technical reserves.A wide range of traditional instruments plus details of financial derivatives (forward-type and option-type) and new financial instruments such as structured financing products.
5.Source dataRelies almost exclusively on aggregate data from the DCS and balance of payments statistics.Relies substantially on balance-sheet data for FCs, supplemented with data from the general government, IIP, and capital market sources.A wide variety of sources, including (but not limited to) reports from government and regulatory agencies, capital market and trade publications, and special surveys of households and corporations.

8.96 The accounting entries for inflows and outflows among sectors in Level 1 differ from the entries for Level 2 and Level 3. Level 1 uses a double-entry system and a single column for each sector. The outflows and inflows in the single column are distinguished by using positive entries to represent increases in financial resources and negative entries to represent decreases in financial resources. Level 2 and Level 3 use a quadruple-entry system and two columns for each sector, thereby distinguishing the uses from the sources. Each transaction results in four entries to the system. For example, if a household purchases a newly issued bond directly from a governmental unit, entries are made to show (1) an increase in other resident sectors’ bond holdings, (2) a reduction in other resident sectors’ cash holdings, (3) an increase in the general government sector’s cash holdings, and (4) an increase in general government sector liabilities in the form of bonds.

8.97 Moving from Level 1 to Level 3, the presentation is enhanced by progressively more disaggregation by sector and financial instrument. The example of Level 1 statistics, in the next section, contains four sectors—central government, private sector, depository corporations (DCs), and rest of the world. In particular, separate sectors for nonfinancial corporations and other resident sectors (households and NPISH) are not specified, and the data are not disaggregated by financial asset/liability category. For the Level 2 example, the financial statistics are expanded to include separate identification of additional sectors and subsectors—central bank, ODCs, OFCs, state and local government, social security funds, public nonfinancial corporations, other nonfinancial corporations, and other resident sectors. Level 2 also incorporates detail for financial assets/liabilities categories. Level 3 shows all sectors that are present in the full integration of the financial statistics and the nonfinancial components of the national accounts statistics; additional subsectors and financial instruments are separately specified in the example for Level 3.

8.98 The source data need to be expanded substantially in moving from the Level 1 statistics to the higher levels. For Level 1, only aggregate data from the nonfinancial national accounts, central government, balance of payments statistics, and the DC component of the monetary statistics are used. For Level 2, comprehensive balance-sheet data for FCs, which include disaggregation by counterpart sector and financial asset/liability category, are used. The data in the IMF’s standardized report forms are the source for the balance sheets of the subsectors of the FC sector, as described in Chapter 7 of this Guide. The Level 2 statistics also make use of IIP data and capital-markets data for government securities and corporate shares, as supplementary sources. For Level 3, balance-sheet data for FCs are still a basic source, but the source data are augmented by capital-markets data and balance-sheet data for the nonfinancial sectors.

Structure

Basic Flow of Funds Account (Level 1 Financial Statistics)

A basic flow of funds account is a modified form of the flow of funds matrix that employs a reduced number of sector and financial asset categories. The sectors chosen are normally those most important for financial analysis and for which data are available—remaining sectors are placed in a residual category. (MFSM, ¶453)

Overview

8.99Table 8.12 shows a matrix of a basic flow-of-funds account (Level 1 financial statistics). The Level 1 financial statistics have relatively few sectors and transactions components.3

Table 8.12.Level 1 Financial Statistics Matrix
Sectors
Domestic economy
TransactionsCentral gov.Private sectorDepository corp.Rest of the worldSum
1.Net lending (+)/borrowing (–)NLNBgNLNBp0NLNB0
2.External financing
2.1. Foreign direct investment+FDIp+FDIodcFDI0
2.2. Increase in external liabilities+NFBg+NFBpNFB0
2.3. Increase in net external assets–ΔNFANFA0
2.4. Of which: net international reserves of the central bank– ΔNIRNIR
3.Domestic financing
3.1. Change in domestic creditDCgDCp– ΔDC0
3.2. Change in broad money– ΔBMBM0
3.3. Other domestic financing+ODFODF0
4.Net errors and omissions0+OINp+OINdc+OIN0
5.Sum0000
Note: Inflow and outflow of funds are shown in a single column for each sector. The plus and minus signs indicate inflows (increases in resources) and outflows (decreases in resources), respectively. The sum in each column or each row of the matrix is always equal to zero.
NLNBNet lending/borrowingThe difference between net saving and net investment; that is, the excess of funds available to lend if saving exceeds investment, or the amount borrowed if investment exceeds saving.
FDIForeign direct investmentNet change in inward and outward foreign direct investment.
NFBNet foreign borrowingsNet change in foreign borrowings.
ΔNFANet foreign assetsNet acquisition or disposal of foreign assets of DCs.
ΔNIRNet international reservesChange in net international reserves.
ΔDCDomestic creditChange in loans and other credit extended by DCs.
ΔBMBroad moneyChange in deposits and similar liabilities of DCs.
ODFOther domestic financingChange in securities, loans, and other credit between the private sector and the central government.
OINNet errors and omissions (Other items net)Balancing item that is the difference between NLNB and the sum of external financing and internal financing.
Note: Inflow and outflow of funds are shown in a single column for each sector. The plus and minus signs indicate inflows (increases in resources) and outflows (decreases in resources), respectively. The sum in each column or each row of the matrix is always equal to zero.

8.100 The economy is divided into domestic sectors and the rest of the world. The domestic sectors are disaggregated into the central government sector, private sector, and DC subsector. The private sector is the residual sector, which includes the households, NPISH, nonfinancial corporations, OFCs, and state and local government.

8.101 The Level 1 statistics may also include non-financial economic activities—using data from the 1993 SNA current account and capital account and referred to as above-the-line components—along with financial activities, or below-the-line components. Table 8.12 could include above-the-line flows such as disposable income, final consumption expenditures, net capital formation, exports, and imports. These data are used to obtain a measure of each sector’s net lending or net borrowing. This more complete version of Level 1 financial statistics could be compiled on an annual basis only, unless above-the-line data for net lending/borrowing were available on a quarterly basis. However, the analytical focus of the Level 1 statistics is on explaining the sectors’ allocations of their net lending/borrowing through changes in their financial assets/liabilities in the below-the-line items.

8.102Table 8.12 illustrates that the Level 1 account is a zero-sum matrix, meaning that each row and each column sums to zero. For the sector identities in the columns, a positive entry represents an increase in financial resources. A positive entry in line 1 (net lending) indicates that the sector increased its financial resources from the production process. A positive entry in the lower part of the table (domestic and external financing) indicates that the sector obtained financial resources by reducing its financial assets or increasing its liabilities.

8.103 Negative entries indicate reductions in sectors’ financial resources. A negative entry in line 1 (net borrowing) means the sector’s spending exceeded its net income from production. A negative entry below line 1 shows that the sector provided funds by increasing its financial assets or reducing its liabilities. The rows in the table also sum to zero. This is ensured by using contra-entries when direct source data are not available. In that way, all totals are fully allocated among the sectors.

8.104Table 8.12 shows only flow data compiled from existing sets of macroeconomic statistics—national accounts statistics, as well as balance of payments and monetary statistics (specifically, from the DCS). Monetary statistics are reported as stocks, whereas the other source data usually are flow data. Flows for the monetary statistics components are calculated as period-to-period changes (denoted by Δ in the table) in the stock positions.

8.105 Financing alternatives are shown in lines 2–4 of Table 8.12, as follows:

  • Data for external financing (lines 2.1–2.4)—financing from the rest of the world—are available from the balance of payments statistics. External financing is divided into (1) foreign-direct-investment transactions between the rest of the world and the private sector (line 2.1); (2) net increases in liabilities of the central government and private sectors to the rest of the world (line 2.2); and (3) the increase in net foreign assets of the DC subsector with the rest of the world (line 2.3). Line 2.2 shows a net increase/decrease, meaning that the changes in financial assets held by the private and central government sectors are subtracted from the changes in liabilities incurred. Line 2.3 is also a net amount, meaning that foreign liabilities of DCs are subtracted from the foreign assets acquired by DCs. Line 2.4 shows that part of the change in net foreign assets arises from changes in the central bank’s holdings of net international reserves.

  • Data for domestic financing (lines 3.1–3.3), which are mostly obtained from the DCS, consist of three major components. The change in domestic credit (line 3.1) consists of changes in loans and other types of credit that DCs provide to the central government and private sectors. The change in broad money (line 3.2) is composed of changes in funds supplied by the private sector (money holders) to the DCs. The change in other domestic financing (line 3.3) consists of changes in loans, equity shares, and other credit from OFCs and capital markets (mainly, capital-market transactions between the central government sector and the private sector).

Compilation steps

8.106 A description of the five steps for compiling a matrix for Level 1 financial statistics follows.

Step 1

8.107 Step 1 involves the use of the above-the-line data from the national income and product accounts to obtain the net lending/borrowing in line 1. Net lending/borrowing is calculated as the difference between net saving and net capital formation, including net investment in inventories. For the rest of the world, net lending/borrowing corresponds to the current account balance plus capital transfers in the balance of payments statistics. Net lending/borrowing is a control total; that is, the sum of external and internal financing must match the net lending/borrowing. Net lending/borrowing for the DCs sector is assumed to be zero.

Step 2

8.108 Step 2 involves the insertion of the data for external financing, which are available from the balance of payments and monetary statistics. Foreign direct investment (line 2.1), both inward and outward, is assumed to arise exclusively from transactions between the rest of the world and the private sector (even though foreign direct investment by DCs may be significant in some countries).

8.109 Estimation of the change in net external liabilities (line 2.2) is based on the balance of payments statistics for individual sectors. NFBg and NFBp denote the data for the central government and private sector, respectively. The sum, NFB (= NFBg + NFBp), differs from the total change in external liabilities in the balance of payments, because of the exclusion of foreign liabilities of the DCs sector. Foreign liabilities of the DCs sector are included in line 2.3, using the data from the DCS. ΔNFA is equal to the net change in foreign assets less the net change in foreign liabilities for the DCs sector. Line 2.4 shows the change in net international reserves of the central bank.

Step 3

8.110 Step 3 uses DCS data to obtain the change in domestic credit (line 3.1) and the change in broad money (line 3.2).

8.111 The ΔDC in line 3.1 is the period-to-period change in domestic credit extended by the DCs, where ΔDCg refers to net domestic claims on the central government less central government deposit holdings. The illustration in Table 8.12 shows the central government sector as a recipient of funds (positive sign). Similarly, the table shows changes in domestic credit to the private sector (ΔDCp)—based on the period-to-period change in claims on the private sector—as a recipient of funds.

8.112 The change in broad money (ΔBM) in line 3.2, also obtained from the DCS and shown with a positive sign, represents an increase in financial resources in the DCs sector and a reduction in resources in the counterpart sectors—that is, the private sectors that are money holders.

Step 4

8.113 Step 4 involves the computations for the other domestic financing in line 3.3. For the central government sector, other domestic financing is given by the difference between the sector’s net lending/borrowing position and the previously estimated financing components, that is,

  • ODF = (SI)gNFBg – ΔDCg,

where (S – I)g denotes saving minus investment of the central government.

8.114 The same absolute amount, but with the opposite sign (–ODF), is the other domestic financing for the private sector, implying that domestic financing from outside the DCs sector is in the form of an exchange of resources between the central government and the private sectors.

Step 5

8.115 Step 5 involves the calculations for line 4, which contain the statistical discrepancy between net lending/borrowing and the sum of external and domestic financing. Conceptually, the results in Line 4 should be zero for each sector, but such results in general are not obtained in practice. For the central government sector, the discrepancy is zero, because the financing is forced to equal net lending/borrowing. However, a nonzero statistical discrepancy can be expected for other sectors. The statistical discrepancy for the DCs sector is equal to changes in other items (net) in the DCS. The statistical discrepancy for the rest of the world is the same as “net errors and omissions” in the balance of payments statistics.

Comments on Level 1

8.116 Level 1 financial statistics have advantages and shortcomings. Among the advantages, a Level 1 table identifies the broad sectors of the economy that provide funds for investment. Despite its simplicity, the Level 1 framework also shows how those funds were transformed and used by other sectors. Over several periods, the table provides a glimpse of how the financing of the economy is evolving. It shows the growth of the use of financing by the private sector relative to use by the central government and the importance of financing supplied by the rest of the world relative to financing by DCs.

8.117 Despite the relative ease of compilation, Level 1 financial statistics have significant shortcomings, including the limited information about the types of financing provided by FCs or capital markets.

8.118 A further limitation is that the Level 1 framework is for flow data only, even though economy-wide data on stock positions have become increasingly important for financial analysis and policymaking. Moreover, the single-column presentation does not allow users to analyze the components of changes in sectoral balance sheets. The Level 1 statistics show only that the increase in a sector’s financial resources has arisen from a increase in liabilities or a decrease in financial assets, or that a reduction in a sector’s resources has been the result of a decrease in liabilities or an increase in financial assets.

8.119 Shortcomings of the Level 1 statistics also include the lack of separate data for nonfinancial sectors—in particular, the other resident sectors (households and NPISH) and nonfinancial corporations—which masks the divergent behavior of financing by these sectors.

8.120 Use of data for the central government would facilitate the calculation of the financial flows between the central government and the private sector, while improving the calculation of the residuals described earlier in Step 4.

8.121 Finally, the Level 1 framework account does not facilitate an independent check that more detailed financial statistics could provide for the amount of net lending/borrowing. Financial accounts that embody more detailed data for FCs and capital markets more fully specify the amount of net financial investment, which can be used to evaluate the accuracy of the data for net lending/borrowing.

The SNA-Integrated Financial Account and Financial Balance Sheet (Level 2 Financial Statistics)

The SNA integrated financial account (presented in Table 8.3 and in the financial part of Table 8.6) represents further development of flow of funds beyond the sectoral and financial asset detail provided in the basic accounts. The integrated financial account is a two-dimensional matrix that covers all institutional sectors and financial asset categories. (MFSM, ¶460)

8.122 This section presents the five steps for compiling Level 2 financial statistics, which are designated as the 1993 SNA -integrated financial account and financial balance sheet. For illustrative purposes, it is assumed that the country has a statistical system that produces the DCS and FCS and that the country’s capital market is at a formative stage in which equity shares and debt securities are beginning to be traded.

Overview

8.123 The Level 2 financial statistics contain FC balance-sheet data that increase the level of detail of the statistics, relative to those at Level 1. The accounts include both stock and flow data (transactions and other flows). For each sector, financial assets (or uses, in the case of flows) and liabilities (sources, in the case of flows) are shown. This Guide recommends that Level 2 financial statistics be compiled on a quarterly basis. The Level 2 statistics make use of below-the-line data only and therefore can be compiled in the absence of quarterly data for the nonfinancial components of the national accounts statistics.

8.124 A major objective of Level 2 financial statistics is to incorporate data for sectors’ net financial investment, which is defined as the difference between net acquisition of financial assets and net incurrence of liabilities. Conceptually, net financial investment is equal to net lending/borrowing, as derived from the above-the-line data. Therefore, Level 2 financial statistics facilitate the evaluation of the above-the-line data in the capital account in the nonfinancial section of the national accounts statistics. Quarterly data for the Level 2 statistics provide an estimate of the bottom line in the above-the-line data in the absence of a full set of quarterly national accounts statistics.

8.125 The Level 2 financial statistics incorporate supplementary data from the capital markets—specifically, data for government securities and equity shares from the perspectives of both issuing and holding sectors. At Level 2, counterpart data in the FCs’ balance sheet are used to obtain the stock positions for domestic sectors and the rest of the world.

8.126 A critical difference between the financial statistics and the DCS and FCS presentations is that, in the financial statistics, transactions and stock positions of nonfinancial and financial sectors are presented on a gross basis, whereas the DCS and FCS presentations focus to some extent on FCs’ net positions and net flows. Level 2 financial statistics present transactions and positions across the FC sub-sector on a gross basis.

8.127 Many transaction flows in Level 2 financial statistics, like those in Level 1 statistics, are derived from changes in stock positions. Directly reported data for transactions among sectors, except those from balance of payments statistics, are likely to be sparse. The compilation of transactions data requires inputs from other data sources or estimation procedures to separate transactions from valuation changes and OCVA.

8.128Table 8.13 shows a representative list of sectors and financial instrument categories for the Level 2 financial statistics. More sectoral disaggregation is included than at Level 1. In the Level 2 statistics, the FC data are divided into separate sets for the DC and OFC subsectors, and DCs are further disaggregated to separate the central bank and ODC data. Subsectors are also specified for the nonfinancial corporations and general government sectors. Also incorporating the other resident sectors (households and NPISH) and the rest of the world, the Level 2 statistics contain all major sectors and subsectors, as described in Chapter 3 of this Guide.

Table 8.13.Examples of Sectors and Financial Instrument Categories in Level 2 Financial Statistics
Sectors
Financial corporations
Depository corporations
Central bank
Other depository corporations
Commercial banks
Building societies
Trust companies
Other financial corporations
Finance companies
Life insurance corporations
Non-life insurance corporations sector
Pension funds
General government
Central government
State and local government
Social security funds
Nonfinancial corporations sector
Public nonfinancial corporations
Other nonfinancial corporations
Other residents
Rest of the world
Financial Instrument Categories
Gold and SDR
Gold (central bank only)
SDR holdings (central bank only)
Currency and deposits
Currency notes and coins
Interbank deposits (central bank and ODCs)
Nonbank financial institutions’ deposits
Central government deposits
State and local government deposits
Social security funds deposits
Public nonfinancial corporations deposits
Other nonfinancial corporations deposits
Other resident deposits
Foreign notes and coins
Deposits with/from nonresidents
Securities other than shares
Treasury bills
Treasury bonds
State and local government securities
Financial corporations securities
Public nonfinancial corporations securities
Other nonfinancial corporations securities
Securities issued by nonresidents
Shares and other equity
Financial corporations shares
Quoted
Unquoted
Nonfinancial corporations shares
Quoted
Unquoted
Foreign shares
SDR allocation
Loans
Central bank loans
Loans to banks other than central bank loans
Loans to nonbank financial institutions
Loans to central government
Loans to state and local government
Loans to public nonfinancial corporations
Loans to other nonfinancial corporations
Mortgage loans
Other loans
Loans to other residents
Mortgage loans
Other loans
Loans to/from nonresidents deposits
Insurance technical reserves
Insurance reserves for residents
Insurance reserves for nonresidents
Pension reserves
Financial derivatives
Other accounts receivable/payable
Other accounts with residents
Other accounts with nonresidents

8.129 As shown in Table 8.13, the Level 2 statistics contain financial instrument categories that are consistent with the financial asset classifications in the MFSM, providing much more detail than the Level 1 statistics. It is recommended that the Level 2 framework be built around the major categories of financial instruments in the MFSM. Further disaggregation of the these categories can be introduced in the national context. The Level 2 framework, as illustrated in Table 8.13, contains the full range of financial instruments—currency, deposits, loans, securities other than shares, shares and other equity, and other categories. As shown in Table 8.13, the framework can incorporate further disaggregation by type of financial asset/liability, if data for the more narrowly defined categories are available. For example, loans can be disaggregated into separate categories for mortgage loans and other loans, and other loans can be further disaggregated into commercial loans and consumer loans—subject to data availability, cost constraints on data compilation, and whether the data are sufficiently analytically useful to justify disaggregation.

Compilation steps

8.130 The five steps for the compilation of the Level 2 financial statistics are:

  • Step 1. Balance-sheet data for FCs are used to construct FC-sector and FC-subsector totals.

  • Step 2. Counterpart data from the FCs’ balance sheets are used to commence the compilation of the data for the nonfinancial sectors.

  • Step 3. Data from the capital markets and other sources are used to enhance the data disaggregation by financial instrument for the nonfinancial sectors to complete the accounts for the nonfinancial sectors.

  • Step 4. IIP data are used to construct the financial statistics for the rest of the world.

  • Step 5. Flow data from the FCs, balance of payments statistics, and price indices for securities are used to construct separate data for transactions and other flows.

8.131 At each step, compilers must exercise judgment about the source data to be used and the estimation procedures to be employed in the absence of source data. If more than one data source for a component is available, compilers must determine the most reliable source data to use as a control total.

8.132 Compilers need to compile the sector accounts on a gross basis, given that data netting could result in the loss of significant analytical content. Suppose a financial instrument is a financial asset of an FC and a liability of another FC. The compilers would record both the asset and liability positions in the FC accounts. Using these data for gross positions, the amount of FC lending to FCs and the amount of FC borrowing from FCs could be compared, which would not be the case if inter-FC positions were netted out.

Step 1

8.133 Step 1 is based on the balance-sheet data for FCs, which are used in the DCS and FCS. The matrix in Table 8.14 illustrates the stocks of financial assets and liabilities in the FC sector and its subsectors for the central bank, ODCs, and OFCs. The table displays the row summations for financial assets and liabilities for each subsector, without netting of inter-sectoral positions.

Table 8.14.Stock Positions of Financial Corporations
Stock PositionsFinancial CorporationsDepository CorporationsCentral BankOther Depository CorporationsOther Financial Corporations
End of March 2003AssetsLiabilitiesAssetsLiabilitiesAssetsLiabilitiesAssetsLiabilitiesAssetsLiabilities
Gold and SDRs895089508950000
Currency and deposits67,373188,38653,824186,04630,09875,03723,726111,00913,5502,340
Securities other than shares139,1040124,742071,12853,614014,3620
Shares and other equities11,10752,6151,14444,0356,3631,14437,6729,9638,580
Loans66,11832,18863,07328,6434,79927,02058,2741,6233,0463,545
Insurance technical reserves024,4520000024,452
Financial derivatives6222162221159626200
Other accounts receivable/payable8,2579,8165,3666,8333,2151,0102,1505,8222,8912,983
Subtotal292,916307,678249,105265,779110,135109,590138,970156,18843,81241,900
Net financial position0−14,7620−16,6745450−17,21901,912

8.134 At Step 1, compilers need to ensure that important equalities hold. For example, the amount of ODC deposit liabilities in the central bank accounts must equal the amount of the corresponding deposit holdings in the aggregate ODC accounts. In some cases, the equality may not hold, because the central bank’s time of recording of such deposits may differ from the time of recording at ODCs. In such cases, the central bank’s data are used for both the central bank’s liabilities and the ODCs’ assets, overriding the data obtained from the sectoral balance sheet of the ODCs. Central bank data usually are expected to be more accurate, given that these data come from a single source, rather than from aggregation of data from multiple sources (that is, data reported by individual ODCs), and central banks usually have reliable recording procedures. In addition, the central bank data usually include the deposit liabilities to banks in liquidation, which may be excluded from the set of individual banks that are data reporters.

8.135 The upper row of Table 8.15 shows that, prior to data adjustment, the total ODC assets in the form of deposits at the central bank did not equal the central bank’s deposit liabilities to ODCs. The lower row shows the data after an adjustment in which the ODC asset total was replaced by the central bank total, based on information that some deposits of banks in liquidation were not included in the asset data reported for ODCs and that this omission was the most likely source of the discrepancy. Depending on the type of discrepancy and the information available, alternative methods of data adjustment may be appropriate.4

Table 8.15.Adjustment for Conflicting Data
Depository CorporationsCentral BankOther Depository Corporations
ODC depositsAssetsLiabilitiesAssetsLiabilitiesAssetsLiabilities
Before adjustment16,13515,21615,21616,135
After adjustment15,21615,21615,21615,216

8.136 Compilers should ensure that OFCs’ deposit holdings equal the liabilities for such deposits in the FC sector. If they are not equal, the data for DCs’ deposit liabilities usually are used as the control total. Based on additional information, compilers may choose to distribute any discrepancy among the OFC subsectors, possibly in proportion to the reported deposit holdings of the various categories of OFCs.

Step 2

8.137 In Step 2, counterpart data in the balance sheet of the FCs are used to estimate the stocks of the financial assets and liabilities of the nonfinancial sectors. For each financial asset/liability category in the FC sector, the amount in the financial asset or liability category of the FCs is entered in the corresponding liability or asset category, respectively, of the nonfinancial sector counterpart.

8.138 Initially, compilers may find it helpful to assume that units in the nonfinancial sectors transact only with FCs, which is clearly the case for deposits and insurance and pension reserves. The assumption is less likely to hold for other financial instrument categories, but is a useful approximation for developing countries in which the FC sector is the dominant counterpart for financial transactions of the nonfinancial sectors.

8.139 Availability of counterpart data may depend on the type of financial instrument. Compilers should retain all such data from the collection process. If information on counterparties is not available, compilers may estimate totals or may construct sectoral allocations for some categories of financial asset holdings and liabilities, using information about the financial instruments in a category. Compilers might need to assume that most corporate securities were issued by other nonfinancial corporations, or that all subordinated debentures were issued by FCs.

8.140 When the data are not sufficient to identify the securities holder or issuer, the compiler should attempt to obtain supplementary information directly from FCs. Without such direct information, the compiler would need to distribute securities holdings to the most likely counterpart sector(s), documenting the distribution method for future evaluation.

8.141 Numerous other issues are likely to arise, owing to insufficient counterpart data. Experience indicates that loan data obtained from OFCs are likely to be inadequately sectorized by counterpart (borrower). Supplementary information may be available for estimation of a reasonable allocation across counterpart sectors. If not, judgment methods must be used in making the sectoral allocations. In determining the financial instrument categories, for instance, compilers should consider differences in the availability of counterpart data.

8.142 Even when available, counterpart data may not be in a form that can be readily inputted into the accounts. For example, FCs sometimes categorize loan data by industry rather than by economic sector. In such cases, a special survey may be needed to obtain information for allocating the business loan data, given that an industry category can apply to both public and other nonfinancial corporations and sole proprietorships. In the absence of a special survey, compilers may need to confer with lenders in allocating the data to the appropriate economic sectors.

8.143 Alternatively, reported data for FC lending may be classified by purpose of borrowing, rather than by industry or sector. In such cases, the sector of the borrower is sometimes revealed by the type of loan. Most home mortgage loans are liabilities of the household sector, and most commercial mortgage loans are liabilities of other nonfinancial corporations. It may be appropriate to separate mortgage loans from other loans, as is shown in Table 8.13, because the allocations among counterparts are somewhat different.

8.144 Counterpart data rarely exist for currency notes and coins issued by central banks (or, in some cases, central governments). Compilers need to distribute currency holdings across sectors, using rules of thumb (such as in proportion to sectors’ deposit holdings). In most developing countries, other resident sectors (mainly households) hold a large share of the notes and coins in circulation.

Step 3

8.145 In Step 3 for the Level 2 financial statistics, data from the capital markets and other sources are used to provide greater detail and to improve the quality of the financial asset/liability data for nonfinancial sectors. Such data may be scarce in the early stages of a statistical program of a developing country, and cost constraints may impede the development of new source data. Nonetheless, additional data may become available from market sources, such as equity and bond market exchanges, trade associations, government regulatory bodies, or custodian and registration agents. Compilers of Level 2 financial statistics should search for such sources of data. An active program to develop the financial statistics may spur the supplying of such data, given that market participants and policymakers have incentives for obtaining financial statistics for their analytical purposes.

8.146 Supplemental data for government securities and equity shares listed on exchanges often are available at a relatively early stage of financial statistics development. The government may provide data on the counterpart sectors (holders) for the securities that the government issues. However, the government data may not be identical to the balance-sheet data for FCs. Compilers must decide if data consistency can be enhanced by using only the FCs’ balance-sheet data as the source for FCs’ holdings of government securities. The government data could be used on a secondary basis to distribute government securities holdings by sectors other than the FC sector.

8.147 Compilers are able to use stock exchange data to obtain an estimate of the total market value of outstanding corporate shares issued by financial and nonfinancial corporations. However, data on share holdings by sector may be available only for FCs’ holdings of equity shares, as reported by the FCs. Compilers may assume that the remainder of the domestic holdings outside the FCs sector are held by other nonfinancial corporations.

Step 4

8.148 Step 4 involves insertion of IIP data—stock data for the rest of the world—into the Level 2 matrix. In countries where IIP data are available only on an annual basis, compilers must estimate the quarterly data, using quarterly flow data in the balance of payments statistics. Data on domestic counterparts to nonresidents’ financial asset/liability positions (other than those with FCs) are likely to be limited. Compilers need to overcome the data deficiencies by using estimation methods that are based on the IIP data and any available supplementary data or qualitative information.

8.149Table 8.16 shows a mapping of IIP components into the corresponding stock positions in the financial statistics. IIP data for equity shares—direct investment in equity shares, including reinvested earnings, and portfolio investment in equity shares—are the inputs to the shares and other equity (foreign) component of the financial accounts (see Table 8.16, line 1 and line 3). Most equity shares issued by the rest of the world are financial asset holdings of other nonfinancial corporations, but FCs in some countries also hold some of these shares. Data for FCs’ holdings are obtained from the balance-sheet data reported by FCs. The difference between FCs’ holdings and the IIP total can be allocated to the other nonfinancial corporations, using the assumption that FCs and other nonfinancial corporations are the only domestic holders of foreign equity shares. The same assumption is applied to sectorization of foreign debt securities in portfolio investment (see Table 8.16, line 4).

Table 8.16.IIP Mapping into Level 2 Financial Statistics
IIP ComponentFinancial Instrument Classification in the Financial Statistics
Assets
Direct investment abroad
1.Equity capital and reinvested earningsShares and other equity (foreign shares): liabilities of the rest of the world and assets of other nonfinancial corporations (input the amount that exceeds FCs’ balance sheet data).
2.Other capitalLoans: liabilities of the rest of the world and assets of other nonfinancial corporations.
Portfolio investment abroad
3.Equity securitiesShares and other equity (foreign shares): liabilities of the rest of the world and assets of other nonfinancial corporations (input the amount that exceeds FCs’ balance sheet data).
4.Debt securitiesSecurities other than shares (securities issued by nonresidents): liabilities of the rest of the world and assets of other nonfinancial corporations (input the amount that exceeds FCs’ balance-
5.Other investmentOther foreign assets: liabilities of the rest of the world and assets of other nonfinancial corporations (input the amount that exceeds FCs’ balance data).
6.Reserve assetsData from the central bank’s balance sheet
Liabilities
Direct investment in reporting economy
7.Equity capital and reinvested earningsShares and other equity (domestic shares of nonfinancial corporations): assets of the rest of the world and liabilities of other nonfinancial corporations.
8.Other capitalLoans: assets of the rest of the world and liabilities of other nonfinancial corporations.
Portfolio investment in reporting economy
9.Equity securitiesShares and other equity (domestic shares of nonfinancial corporations): assets of the rest of the world and liabilities of other nonfinancial corporations.
10.Debt securitiesSecurities other than shares (government bonds or securities issued by other nonfinancial corporations): assets of the rest of the world and liabilities of the central government or other nonfinancial corporations.
11.Other investmentOther foreign liabilities: assets of the rest of the world and liabilities of other nonfinancial corporations (input the amount that exceeds FCs’ balance

8.150Table 8.16 (line 2) also shows the mapping of other capital assets in direct investment into the stock position in loans in the financial statistics. These loans are assets of the other nonfinancial corporations sector and liabilities of the rest of the world—in addition to the liabilities for loans extended by the FC sector. Other capital liabilities in the IIP (see Table 8.16, line 8), which are liabilities of nonfinancial corporations and asset holdings of the rest of the world, are also classified as loans in the financial statistics, and are in addition to the loan liabilities of FCs and the government sector.

8.151 The mapping of other investment assets from the IIP (Table 8.16, line 5) requires care to avoid double counting of balance-sheet data for FCs. The IIP data are the amounts of deposits, loans, and other foreign liabilities of the rest of the world. In the absence of sectoral disaggregation in the IIP, the positions are assumed to be foreign assets of ODCs and OFCs. Differences between these other investment assets in the IIP and the balance-sheet data for ODCs and OFCs are recorded as other foreign assets of the other nonfinancial corporations. The same treatment is applied to other investment liabilities in the IIP (Table 8.16, line 11).

8.152 Reserve assets in the IIP (Table 8.16, line 6) are not used in the financial statistics matrix, because these data are obtained from the central bank’s balance sheet. The central bank data and the IIP data for reserve assets should agree unless the central government holds some of the reserve assets (which are entered in the financial statistics, classified by type of financial instrument, in the asset section for the central government).

8.153 The IIP data for direct investment in equity capital and reinvested earnings (Table 8.16, line 7) are used to estimate the shares and other equity holdings of the rest of the world. However, it is unlikely that the IIP data are included in the total market capitalization of the shares on the exchange, which the compiler uses as control data for total domestic shares. Compilers incorporate the direct investment component of the IIP data in the total value of domestic shares. In the absence of other information, the compiler may assume that, on the liability side, the shares have been issued by other nonfinancial corporations.

8.154 The market capitalization of exchange-traded shares is likely to include portfolio investment in equity, which is included in the IIP data (Table 8.16, line 9). Therefore, the value of foreign holdings of shares does not need to be added in calculating the total market capitalization. These asset holdings of the rest of the world are deducted from the total issuance to obtain the holdings of other nonfinancial corporations. In the absence of other information, compilers may assume that all of the liabilities are claims on other nonfinancial corporations.

8.155 For debt securities (Table 8.16, line 10), the IIP shows asset holdings of the rest of the world, but compilers need to make assumptions about the most plausible asset distribution by type of financial instrument. Particular attention should be given to reconciling the government bond data, as reported by the government and as contained in the IIP.

8.156 A country that does not have IIP data is likely to estimate foreign direct investment by accumulating flow data from the balance of payments statistics. For a country that compiles Level 2 statistics, compilers may find that positions between nonfinancial sectors (typically, the other nonfinancial corporations sector) and the rest of the world are less significant than positions between the FCs and the rest of the world. Nevertheless, the amount of foreign direct investment by other nonfinancial corporations could be significant. Compilers need to explore the availability of stock data for foreign direct investment.

Step 5

8.157 Step 5 is used to distinguish between transactions and other flows. A compiler can request that data be provided directly by FCs—for example, with reference to (1) the profit-or-loss accounts as source data for holding gains and losses arising from financial asset/liability revaluations, for extraordinary items to be classified as OCVA, etc.; and (2) information about reclassification of assets/liabilities and other events affecting the financial statements of FCs. Similar information can be requested from non-financial corporations.

8.158 The compiler should obtain transactions data for the rest of the world and its counterpart sectors from the balance of payments statistics, while being mindful that differences between the balance of payments data and period-to-period changes in IIP data may arise from differences in the estimation of financial asset/liability valuations and the treatment of OCVA.

8.159 Compilers will find that balance of payments statistics generally contain more disaggregation than the IIP data—by type of financial instrument (for example, for the category of other investment) and by economic sector. As a result, the mapping of the balance of payments data into the financial statistics is more straightforward than the mapping of IIP data.

8.160 Compilers can obtain transactions data for the general government from government finance statistics, while being aware that the government finance statistics often have shortcomings in terms of the coverage (for example, the exclusion of nonbudgetary central government activities, state and local government, and social security funds), disaggregation by financial instrument category, and availability of counterpart data.

8.161 For shares and other equity, compilers in countries with organized stock exchanges should have access to share price indices (SPIs), which can be used in the estimation of equity revaluations. The transactions data are derived as the period-to-period changes in the stock positions less the revaluations.5

8.162 Compilers for countries that do not have active secondary markets for debt securities will find it difficult to obtain price indices for securities other than shares. If share price indices are unavailable, period-to-period changes reported by FCs and other sectors may need to be used to estimate the transactions in securities other than shares. A security-by-security database provides the most relevant data, but such databases are unavailable for many countries.

Comments on Level 2

8.163 Level 2 financial statistics provide many advantages over Level 1 statistics. In particular, the Level 2 framework facilitates more thorough analysis of financing in the economy, given that Level 2 includes data for individual economic sectors and specifies the linkages between financial flows and stocks. The data for various categories of financial assets/liabilities allow users to monitor growth and developments in specific financial markets. The Level 2 data are also more useful for making international comparisons.

8.164 More generally, the expansion to two columns for each sector, showing both assets and liabilities, offers a clearer picture of the financial activities of the sectors. Level 1 data do not show the types of transactions that led to changes in net financial investment, whereas the Level 2 data show the extent to which the changes in such investment resulted from changes in liabilities and changes in assets, respectively. For example, the data might show an improvement in the financial position of the nonfinancial corporations sector, resulting from both an increase in financial assets and a reduction in liabilities.

8.165 Level 2 financial statistics have significant shortcomings, given that a major subset of the Level 2 data is based on counterpart data from the FCs’ balance sheets and given that the flow data may be estimated primarily from the stock data. Economies can be expected to experience financial-market development and financial innovations that result in the introduction of additional types of financial instruments and new channels for financial intermediation.

8.166 Countries may consider an expansion of the financial statistics by migrating to the Level 3 framework, which shows the positions and transactions between domestic nonfinancial sectors. Given that the Level 2 framework focuses on the calculation of transactions from period-to-period changes in stocks, revaluation of securities and write-offs of loans are sometimes ignored or are not fully represented, thereby distorting the data for financial flows. Such distortions become increasingly significant as a country’s financial system is liberalized and its capital markets develop.

Detailed Financial Statistics Matrices (Level 3 Financial Statistics)

While a financial account flow of funds provides a great deal of sectoral detail, it is only at the two-dimensional level, that is, it shows net incurrence of liabilities by sector and net acquisitions of assets by sector. To address the three-dimensional issue of which sectors finance other specific sectors through the use of specific financial assets, it is necessary to develop more elaborate flow of funds matrices. (MFSM, ¶462)

8.167 This section discusses issues that are encountered in upgrading from Level 2 financial statistics to the detailed financial statistics matrices at Level 3.6 The Level 3 statistics usually are applicable to countries that have money and capital markets that are highly developed. In these countries, corporations and households have wide choices of financial instruments, including financial derivatives and structured-finance products. A prerequisite to the development of the Level 3 statistics is that a country have a formal statistical system that yields high-quality data from a wide variety of sources. Even if the country’s formal statistical system is well developed, additional data or supplementary information for refining the Level 3 statistics may emerge from private-sector sources such as trade associations, exchanges and financial markets, and privately sponsored surveys of particular types of financial activity.

8.168 The channels and methods for collecting the data for the Level 3 statistics can be expected to vary across countries, reflecting the unique elements of each country’s statistical and financial systems. However, the countries’ Level 3 statistics are similar to the extent that, in each country, the statistics facilitate the tracing of financial intermediation through various channels in the economy.

8.169 A fundamental objective of Level 3 financial statistics is the integration of financial statistics and the nonfinancial components of the national accounts statistics. The financial statistics include the accumulation accounts—the capital account, financial account, revaluation account, and OCVA account—and the balance sheets of the sectors specified in the 1993 SNA. The Level 3 framework can be viewed as a refinement of Level 2 statistics with respect to collection/estimation of flow data—the separate categories of transactions, revaluations, and OCVA—and the reconciliation of the flow and stock data.

8.170 The Level 3 framework contains more detailed derivation of sectors’ net financial investment, which must be reconciled with net lending/borrowing in the capital account, as derived from the nonfinancial components of the national accounts statistics.

8.171 The structure of Level 3 financial statistics contains detailed sector and subsector accounts for net purchases of financial assets and net incurrence of liabilities, as well as accounts for sector and sub-sector holdings of financial assets and liabilities. Accompanying the accounts for transactions and balance-sheet positions are sector and subsector accounts for revaluations and OCVA.

8.172 A separate set of steps for compilation of Level 3 financial statistics is not presented. In fact, the five-step procedure described for the Level 2 statistics also applies at Level 3. The Level 3 framework can be viewed as an extension of the Level 2 statistics, resulting from refinements in the data collection and estimation techniques and inclusion of additional accounts for a larger set of financial subsectors and a more disaggregated set of financial assets/liabilities. In addition, Level 3 contains more financial information about the subsectors of the nonfinancial sectors and their transactions.

8.173 For subsectors in the FCs sector, Level 3 includes money market mutual funds (MMMFs)—a subsector of the DCs sector—and mutual funds other than MMMFs, which are a subsector of the OFC sector. The insurance corporations and pension funds subsectors also have more detail than in Level 2. Other subsectors are possible, depending on the structure of the financial markets and institutional units in the country. Table 8.17 also includes NPISH, as an additional nonfinancial subsector.

Table 8.17.Examples of Disaggregated Data Categories at Level 3
Sectors
FCs
Of which: Public FCs
Depository corporations
Central bank
Other depository corporations
Commercial banks
Building societies
Trust companies
Money Market Mutual Funds (MMMF)
Other FCs
Other financial intermediaries
Finance companies
Mutual funds other than MMMF
Specialized financial institutions
Special purpose companies
Funding corporations
Insurance corporations and pension funds
Life insurance corporations
Non-life insurance corporations
Reinsurance corporations
Corporate pension funds
Other private pension funds
Financial auxiliaries
General government
Central government
State and local government
Social security funds
Nonfinancial corporations
Public nonfinancial corporations
Other nonfinancial corporations
Households
Nonprofit institutions serving households (NPISH)
Rest of the world
Asset/liability categories
Nonfinancial assets
Reproduced assets
Non-reproduced assets
Of which: Land
Gold and SDR
Gold
SDR holdings
Currency and deposits
Bank notes and coins
Bank deposits
Nonbank financial institutions deposits
Central government deposits
Local government deposits
Social security funds deposits
Public nonfinancial corporations deposits
Other nonfinancial corporations deposits
Other resident deposits
Foreign notes and coins
Deposits with/from nonresidents
Securities other than shares
Treasury bills
Treasury bonds
Local government securities
FCs securities
Public nonfinancial corporations securities
Other nonfinancial corporations securities
Structured financing products
Commercial paper
Securities issued by nonresidents
Shares and other equity
Financial corporations shares
Nonfinancial corporations shares
Foreign shares
Mutual fund shares
SDR allocation
Loans
Central bank loans
FC loans to banks other than central bank loans
FC loans to nonbank financial institutions
FC loans to central government
FC loans to state and local government
FC loans to public nonfinancial corporations
FC loans to other nonfinancial corporations
Mortgage loans
Other loans
FC loans to other residents
Mortgage loans
Consumer credit
Other loans
Financial leases
Government loans
Nonfinancial corporations loans
Loans to/from nonresidents
Insurance technical reserves
Insurance reserves for residents
Insurance reserves for nonresidents
Pension reserves
Financial derivatives
Forward-type
Option-type
Other accounts receivable/payable
Trade credits
Other

8.174 Compilers typically are able to classify instruments under the broad categories in Table 8.17, which follow the recommendations in the MFSM and this Guide. For example, among financial instruments, securities other than shares include commercial paper—issued by the largest and most creditworthy corporations, as a substitute for bank financing. Another subcategory is structured-finance products such as asset-backed securities. The list of financial instruments in Table 8.17 includes mutual fund shares, consumer credit, financial leases, government loans, nonfinancial corporation loans, and financial derivatives disaggregated by type of contract. Compilers need to be prepared to introduce new classifications, arising from continuing development and innovation in financial markets that expand the list of financial instruments. Given that some financial instruments may be unique to a particular country, this Guide is not prescriptive with respect to the classification of financial instruments within the broad categories of financial assets/liabilities; Table 8.17 is only illustrative.

Enhancing detail for nonfinancial sectors

8.175 Level 3 financial statistics are distinguished from those at Level 2 by the extent of the detail shown for nonfinancial sectors. Whereas Level 2 statistics acquire value by exploiting the balance-sheet data for FCs, Level 3 enhances the information for the general government sector, nonfinancial corporations sector, and NPISH.

8.176 A specific area for improvement relative to Level 2 financial statistics is the incorporation of financial positions between the government sector and nonfinancial sectors. These positions include government loans to corporations, households, and NPISH. The Government Finance Statistics Manual 2001 (GFSM 2001) offers examples from balance sheets of the government sector.

8.177 Data for nonfinancial corporations may allow identification of financial positions with other domestic sectors. Nonfinancial corporations’ loans and trade credits are examples of the types of data that can be introduced into the Level 3 statistics.

8.178 An objective of the Level 3 financial statistics is to develop separate data for the NPISH sector. Few countries have such data, though several countries continue to work toward that end. A benefit from separate treatment of the NPISH sector is more accurate data for households; otherwise, NPISH are combined with households in the other resident sector, which is treated as the residual sector. Countries that are progressing toward distinguishing the NPISH data from the household data use special surveys, tax and registration records, and counterpart data.

8.179 Finally, compilers should work toward the development of surveys for balance-sheet data that contain enough detail to show the full range of financial assets and liabilities of nonfinancial corporations. Unlike countries for which Level 2 statistics are broadly adequate, more-developed economies have sizable financial positions that are not covered by counterpart data in the FCs’ balance sheets.

Special purpose entities

8.180 Data collection for special purpose entities (SPEs) can be challenging. Level 3 financial statistics need to include the data for these entities, which are created for issuance of securities backed by financial assets or for other purposes. Such data are also needed for Level 2 financial statistics in which net financial investment is compiled for all sectors.

8.181 The SPEs’ balance sheets usually have few accounts. An SPE’s assets may be limited to an asset portfolio acquired for securitization, and most or all of an SPE’s liabilities may be in the form of the securities issued with the backing of the asset portfolio. However, SPEs often are unregulated, and formal means for collecting data on their assets and liabilities may be unavailable. The securities issued by SPEs sometimes are asset holdings within sectors that have very limited data reporting, or are totally outside the data collection of the statistical system.

8.182 SPE data sometimes are available through the agencies where the securities are registered. However, the data may be limited to total amounts of SPE securities issuance. Industry sources such as brokers and dealers who specialize in securitization may also be able to provide data on new issuances, repayments, and outstanding amounts of the securities. Market prices for asset-backed securities often are quoted in the capital markets. The period-to-period changes in market value are components of the revaluation accounts of the SPEs and the sectors that hold the securities.

Effect of secondary market transactions

8.183 Institutional units may trade outstanding securities several times in secondary markets, frequently leading to the transfer of the securities from the balance sheet of the original-purchaser sector to the balance sheet of another sector. Intrasectoral transactions are less problematic, given that the transactions do not affect the aggregated balance sheet of the particular sector. For economies with well-developed secondary markets and substantial intersectoral securities trading, data on securities holdings by sector of original purchaser can quickly become outdated for the purpose of estimating current securities holdings by sector and subsector. For example, central governments often provide data on purchases of new issues of government securities—that is, primary-market purchase data. However, through an active secondary market, the ownership of a single set of government securities may be transferred several times across sectors and subsectors prior to the maturity of the securities—for example, through the secondary market, from a securities dealer (OFC) to a commercial bank (ODC), from the commercial bank to an insurance corporation (OFC), and from the insurance corporation to an individual (other resident sectors).

8.184 The existence of well-developed secondary markets necessitates the development of source data beyond those provided for the primary markets. In some cases, the data are directly reported by investors such as depository corporations, insurance corporations, pension funds, and securities dealers. Typically, other resident sectors is treated as the residual category of securities holders, if the total amount of outstanding securities and the individual holdings of all other sectors and subsectors are known.

8.185 If data are missing for more than just the other resident sectors, the holdings of some sectors/subsectors must be estimated with the use of data sources such as securities registration offices or trusts or other custodians of their clients’ securities. However, difficulties may be encountered in disaggregating the securities by sector of holder, using the data as reported by registration offices and custodians.

Compiling OCVA and revaluation accounts

8.186 At Level 3, the flow data for financial statistics are disaggregated into transactions, revaluations, and OCVA. Table 8.18 summarizes the steps for estimating the stock and flow components for shares and other equity. The method is practical, but deficient in some respects. The market value for unquoted shares and other equity must be estimated in Step 2. International agreement on the estimation method does not exist. The method described in Table 8.18 also assumes that each sector holds a representative portfolio of equity shares that is roughly proportional to the weights in the market price index and that each sector has holding gains/losses that are proportional to their share holdings. However, the timing of transactions is throughout the holding period, resulting in differences between the estimated and actual holding gains/losses of the sectors. Refinements in the method could be built around the use of additional data that are sector specific—for example, data on holding gains/losses in income statements of insurance corporations and pension funds.

Table 8.18.Estimating Transactions, Revaluations, and OCVA for Shares and Other Equity
StepsPossible Data SourcesComments
1.Transactions: Estimate net issuance of shares and secondary market transactions by each sector, at transaction prices.Government registrations, market exchanges, and private industry sources.Data are required for both gross issuance and retirement of shares. Information on the industry and/or type of corporation facilitates estimation by sector.
2.Stocks: Estimate the market value of total shares outstanding and the amount held by each sector.Sectoral balance-sheet data and stock exchange sources.Other resident sectors (primarily households) typically is designated as the residual holder. Data from household surveys, if available, may be directly used or may provide a check on the plausibility of the data obtained residually.
3.Other flows: Calculate the period-to-period change in the market value of shares held by each sector. The difference between the total change in value of shares held and net issuance (from Step 1) is the sum of holding gains/losses and OCVA.Calculated from the flow and stock data in Steps 1 and 2.Refinements include use of sector data, where available—for example, profit-or-loss data for holding gains/losses.
4.Separation of revaluations and OCVA: The amount calculated in Step 3 is disaggregated into holding gains and losses, using price indices, and OCVA.The price indices should be as broad as possible. In some cases, specialized indices can be used for some sector holdings.

8.187 For securities other than shares, estimates of holding gains/losses may be less reliable than the estimates for holding gains/losses for equity. The data sources for the estimation of transactions in bonds and similar financial assets may be less developed, and corporations may report the holdings at original purchase prices rather than at subsequent market values. Nonetheless, data on outstanding amounts are likely to be available, and securities price indices may be used for estimation of holding gains and losses on an aggregated basis.

8.188 For loans, period-to-period changes in stock positions consist of transactions (net lending less repayment) and OCVA that are primarily in the form of loan write-offs and write-downs. Data for OCVA arising from loan write-offs and write-downs can be obtained from the expense records of FCs and nonfinancial corporations.

8.189 All foreign-currency-denominated assets and liabilities need to be translated into national currency units, using market exchange rates. The total holding gain/loss for each category of financial assets and liabilities is the sum of the holding gains/losses arising from the change in the market price or fair value (denominated in the foreign currency) and the holding gain/loss from translation into national currency units.

Comments on Level 3

8.190 Compilation of Level 3 financial statistics requires considerable judgment and often provides substantial leeway in choosing the best data from alternative sources that each have data deficiencies. Estimation techniques for missing data must be devised, and residual calculations may be numerous. Nonetheless, Level 3 financial statistics provide a substantial set of data for policy analysis and research. The main reasons for development of Level 3 financial statistics are to obtain a better understanding of the financial interactions among all sectors of the economy and with nonresidents and to be able to relate these interactions to current performance, development, and growth of the economy.

Estimation Techniques for Missing Data

8.191 Compilers of financial statistics rely on various techniques to estimate incomplete data. To maintain a timely release calendar for financial statistics, compilers need to estimate missing data that will become available later, sometimes after protracted data reporting lags. In addition, techniques are available using annual data to estimate quarterly data.

8.192 Two broad categories of estimation techniques are described in this section. The first category is techniques for obtaining quarterly estimates that are based on annual data. The second category is techniques for obtaining estimates of quarterly data that, though not yet available, will be provided in the future.

Estimation of Quarterly Figures from Annual Sources

Sliding level method

8.193 The simplest and most common method for estimating missing values of quarterly stock data is based on the annual data for the immediately preceding year. The amount at the end for each of the first three quarters of year t is specified as the amount at the end of the year t–l, denoted by Yt–1. The amount for the fourth quarter of year t is specified as the amount for the end of year t (premised on the availability of data for Yt). That is,

  • y1q = Yt–1

  • y2q = Yt–1

  • y3q = Yt–1

  • y4q = Yt.

For the fourth quarter data, the estimation “slides into” the current-year data. The sliding level is particularly useful for data series that are irregular, showing neither trend nor seasonality.

Complementation method

8.194 If the changes in the annual data appear to be spread fairly evenly over the quarters of each year, the quarterly data can be estimated as follows:

  • y1q = Yt–1 + [(YtYt–1)/4]

  • y2q = y1q + [(YtYt–1)/4]

  • y3q = y2q + [(YtYt–1)/4]

  • y4q = Yt,

where [(YtYt–1)/4] represents the “complementation” in the estimates for the first three quarters of year t.

Constant ratio method

8.195 Annual data sometimes indicate that a quarterly time series can be approximated as a stable proportion of an annual time series. For example, annual data may indicate that bond holdings of pension funds can be treated as a constant proportion of total bond holdings. In that case, total bond holdings in year t is given by Xt = Y1,t + Y2,t…. + YN,t, where Yj,t (j = 1,…,N) are the bond holdings of the N subsectors. The pension funds are denoted as sub-sector i. The pension-fund proportion of total bond holdings is pi = (Yi, t–1/Xt–1), and the quarterly estimates for the bond holdings of the pension funds are:

  • yi, 1q = pi Xt

  • yi, 2q = pi Xt

  • yi, 3q = pi Xt

  • yi, 4q = pi Xt.

If quarterly data are available for total bond holdings, these data can be used in estimating the quarterly data for the bond holdings of the pension funds:

  • yi, 1q = piX1q

  • yi, 2q = piX2q

  • yi, 3q = piX3q

  • yi, 4q = Yt.

Flow increment method

8.196 It may be necessary to estimate quarterly stock data when only annual stocks and quarterly flows are available. Using the flow increment method, the accumulation of quarterly flows, fjq (j = 1, 2, 3), is added to the previous annual amount to obtain the estimates for the first three quarters:

  • y1q = Yt–1 + f1q

  • y2q = Yt–1 + f1q + f2q

  • y3q = Yt–1 + f1q + f2q + f3q

  • y4q = Yt.

Proportional method

8.197 A data series sometimes can be estimated by relating the time series to the quarterly growth of another time series. For example, quarterly amounts for a particular category of OFCs’ financial assets (for which only annual data are available) may be related to changes in the same category of ODC assets (for which quarterly data are available). Denoting the quarterly changes in the ODC series by xjq (j = 1, 2, 3), the quarterly estimates for the OFC data are:

  • y1q = Yt–1(x1q/Xt–1)

  • y2q = Yt–1(x2q/Xt–1)

  • y3q = Yt–1(x3q/Xt–1)

  • y4q = Yt.

Smoothing method

8.198 Several methods can be applied to annual data for the estimation of a quarterly series that is relatively smooth. These estimation methods attempt to retain the cyclical and trend patterns implied in the annual data. One such method is shown below.

  • y1q = Yt–1 + Ft × Yt–2/(Yt–2 + Yt–1 + Yt + Yt+1), where Ft = YtYt–1

  • y2q = Y1q + Ft × Yt–1/(Yt–2 + Yt–1 + Yt + Yt+1)

  • y3q = Y2q + Ft × Yt/(Yt–2 + Yt–1 + Yt + Yt+1)

  • y4q = Yt.

Estimation of Quarterly Figures When Data are Missing or Incomplete

Sliding level method

8.199 The sliding level method can be applied to the estimation of recent observations for a quarterly data series for which the quarterly data are available, but only with a lag. The basic form of the sliding level method is applied when the data do not exhibit seasonality; it is assumed that the amount for the current quarter is the same as that for the immediately preceding quarter:

  • yt = yt–1

If the data exhibit a seasonal pattern, the sliding level method is modified by relating the amount for the current quarter to that in the same quarter of the previous year:

  • yt = yt–4.

Substitution method

8.200 The substitution method has both additive and multiplicative forms. In the additive form, the period-to-period changes in another series, xt, are used in the estimation of the missing quarterly data:

  • yt = yt–1 + (xtxt–1).

In the multiplicative form, the formula for estimation of the missing quarterly data is:

  • yt = yt–1(xt/xt–1).

Trend method

8.201 Using the trend method, it is assumed that changes in the unobserved time series are additively or multiplicatively related to past changes in the time series. The formulas for a data series that does not exhibit seasonal behavior are:

Additive: yt = yt–1 + (yt–1yt–2).

Multiplicative: yt = yt–1(yt–1/yt–2).

The formulas for a data series that has seasonal patterns are:

Additive: yt = yt–4 + (yt–4yt–8)

Multiplicative: yt = yt–4/yt–8).

Extrapolation methods

8.202 Extrapolation methods can be used for missing data series that are trended. The extrapolation formulas for the missing data are based on the latest available data and a weighted average of data for earlier periods. A formula that places the greatest weight on the most recent observations is:

yt = yt–1([(3/6)(yt–1/yt–2)] + [(2/6)(yt–2/yt–3)] + [(1/6)(yt–3/yt–4)]).

A similar formula for data that have a seasonal pattern is:

  • yt = yt–1([(3/6)(yt–1/yt–5)] + [(2/6)(yt–2/yt–6)] + [(1/6)(yt–3/yt–7)]).

Regression method

8.203 Regression models can be used to incorporate the influence of various economic variables on the estimate of the missing data. Suppose the missing data are the amount of securities other than shares issued by nonfinancial corporations. Estimation of the missing data could be formulated so as to take account of the effects that interest rates and nonfinancial variables (for example, nonfinancial corporations’ expenditures on capital goods) have on their issuances of securities. Though analytically appealing, the regression method has drawbacks, including the relatively large amount of effort that must be devoted to specification and estimation of the regression models. Having estimated the regression model, the results must be monitored to ensure that the relationship continues to hold over time or that the regression model needs to be reformulated and re-estimated.

Editing, Data Checking, and Statistical Discrepancies

8.204Table 8.19 describes the relative reliability that is typical for various data collected or estimated for a financial statistics matrix. The data shown as highly reliable are those that usually can be directly obtained from FCs’ reported data and the IIP and balance of payments statistics.

Table 8.19.Reliability of Data Estimates
HighMiddleLow
Financial CorporationsGeneral GovernmentNonfinancial CorporationsOther Resident SectorsRest of the World
Depository corporationsOther FCsPublicn on financial corporationsOther nonfinancial corporations
AssetLiabilityAssetLiabilityAssetLiabilityAssetLiabilityAssetLiabilityAssetLiabilityAssetLiability
Currency and deposits
Currency
Deposits
Loans
Securities other than Shares
Central government Securities
Other securities
Structured-financing Instruments
Shares and other equities
Financial derivatives
Insurance technical Reserves
Other accounts receivable/payable

8.205 The data shown as moderately reliable are those that involve estimation, but for which some source data are available on an annual or less frequent basis, or are contained in surveys. Estimates for trade credit and some types of financial asset holdings of government may be in this category.

8.206 The data shown as having low reliability are those for which few, if any, source data exist. Many estimates of data in this category are based on residual calculations. These include data for the financial assets of other resident sectors (households and NPISH), as well as data for miscellaneous categories of financial assets and liabilities of nonfinancial sectors.

8.207 The unshaded cells in Table 8.19 are those for which liabilities do not exist. For example, households and NPISH do not issue currency, and only the government issues government securities.

Editing and Checking Data

8.208 Careful examination of the data in each row and column of the financial statistics matrix is the simplest and most direct approach to identifying data problems. This examination can be followed by the construction of charts and tables of the time-series data that may reveal data outliers that may need to be verified or corrected.

8.209 For plausibility testing of the aggregate data, compilers should have some expectation as to the reasonableness of data for each cell in the financial statistics matrix. Unexpected movements in the data should be explainable in terms of economic behavior, if not attributable to data collection, estimation, or compilation errors. For this purpose, basic underlying relationships among macroeconomic data need to be understood. Compilers’ knowledge may need to be complemented by consultation with experts outside the unit that compiles the financial statistics.

8.210 Data problems are often more apparent in a time-series context than through examination of a financial statistics matrix for a single period. The matrix format provides a framework for checking that the data meet adding-up requirements and broad plausibility tests, but time series presented in tables or charts—possibly in differenced or percentage-change form, or as ratios of two time series—are highly useful for identification of outliers that need to be investigated.

Statistical Discrepancies

8.211 Statistical discrepancies arise when two data sets provide different results for a particular data category. A problem that often appears in financial statistics is the statistical discrepancy between net lending/borrowing (NLNB) derived from the capital account and net financial investment (NFI) in the financial account. From the capital account, the basic identity for net lending/borrowing is:

  • NLNB =Net SavingNet Capital Formation.

From the financial account, the basic identity for net financial investment is:

  • NFI = Acquisition of Financial AssetsIncurrence of Liabilities.

Conceptually, NLNB equals NFI. In practice, the separate compilation of the data for the capital account and financial account leads to a statistical discrepancy.

8.212 An international consensus on the treatment of this discrepancy does not exist. One approach is to “eliminate” the discrepancy through use of a residual calculation (referred to as a “balancing item”). Alternatively, the amount of the discrepancy can be distributed across one or more items in the capital account, the financial account, or both—treating the discrepancy as a transaction or valuation change or, more likely, as an element of OCVA.

8.213 The motivation for removal of discrepancies by incorporating them in the accounts is that, by providing “balanced” accounts, an element of ambiguity is eliminated for the users of the financial statistics. However, presentation of data for discrepancies assists users in gauging the relative magnitude of errors and the overall quality of the data. Compilers can provide users with differing data for NLNB and NFI, while informing users that the one set of data (possibly, those for NFI and its components) is relatively more reliable than the other set.

8.214 Regardless of how the discrepancies are treated in the data that countries release and publish, the record of data discrepancies provides compilers with valuable information for identifying areas that need improvement in the data collection, estimation, and compilation system for the national accounts statistics and, in particular, the financial statistics.

See the Detailed Flow of Funds section and Table 8.9 in the MFSM, Chapter VIII.

On the use of this estimation technique in the presence or absence of OCVA, see the section on Estimation of Transactions and Valuation Changes from Exchange Rate Movements in Chapter 5, ¶5.22–5.33, and Annex 5.1.

Tables 8.7 and 8.8 of the MFSM and Table 8.1 of this Guide are also basic flow-of-funds accounts that can be used as a reference. However, the Level 1 account in this Guide has the advantages of relative simplicity and significant analytical usefulness.

Compilers are likely to seek information from agencies that collect other types of statistics (for example, regulatory and supervisory agencies), as well as from industry sources.

Enhanced procedures for Level 3 financial statistics are shown in Table 8.18.

Chapter VIII of the MFSM focuses on the flow of funds matrices, whereas this section covers both flow of funds and balance-sheet matrices.

    Other Resources Citing This Publication