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

29 Treatment of Output in the Banking Industry

Vicente Galbis
Published Date:
September 1991
  • ShareShare
Show Summary Details
Abul Siddique

There are different ways of looking at what the banking industry produces, although the concept of input is intermediate purchases as in other industries. According to one view, the output of banks consists of services to depositors rather than to borrowers because, in national income and product accounting, interest payments are regarded as output of the paying rather than the receiving industry. This view, however, is grounded in the difficulties of allocating banking services among the borrowing industries. A comprehensive definition of banking output should, therefore, be based on a closer examination of banking activity and banking costs. Such an examination would show that financial services (rather than deposits or loans) are the products of banking. These services can be grouped into three main categories:

  • Providing the medium of payments

  • Intermediation between borrowers and lenders

  • Specialized financial services such as trust department and foreign exchange services, investment banking services, and so on.

Medium of payments services are provided to demand depositors; intermediation services are rendered to both depositors and borrowers; and other specialized services are provided to depositors, borrowers, and other customers.

All these banking activities involve interest receipts and payments, and explicit and implicit service charges. The crucial problem in defining value added is in determining the part of interest payments that is for intermediation services provided by banks and the part that is for the depositor's liquidity preference, since depositors are paying to the banks implicitly by accepting interest rates lower than what they could earn from higher-yielding assets. Another important issue concerns the net exports of banking services by a country. In the present version of the United Nations' A System of National Accounts (SNA), all imputed bank service charges are allocated to domestic industry as intermediate demand and, as such, do not affect gross domestic product (GDP). A deficiency of nonallocation of imputed bank service charges between intermediate and final demand is that the GDP of countries that are net exporters of banking services would be understated, whereas the GDP of countries that are net importers of banking services would be overstated. Without resolving these issues, a straightforward application of national accounting concepts to the calculation of value added in the banking industry results in very small or negative value added in the banking industry for reasons that are explained below.

The rest of the paper will take the present SNA treatment of this “banking income anomaly” and the various proposed revisions as points of departure for suggesting some practicable procedures for computing sectoral origins of the banking product for intermediate and final consumption. Section I outlines the present SNA treatment of the banking income anomaly and reviews the various proposals made at the Organization for Economic Cooperation and Development (OECD) Meeting of National Accounts Experts in May 1986.1 This section also discusses the allocation of imputed banking services between intermediate consumption and final demand, with an illustration using data for Luxembourg. Section II briefly outlines the work done in the IMF's Bureau of Statistics on sectoral imputations and allocation of banking output.2 Finally, Section III makes suggestions for further work, and, in particular, discusses briefly issues relating to an appropriate deflator for banking sector output, including the question of measuring productivity in this sector.

I. Present SNA Practice and Review of Proposals for Change

As pointed out above, the application of the usual procedure for estimating value added for the banking sector produces very small or negative value added. This is because value added by banks comprises net interest paid, employee compensation, and other income net of provisions for loan losses; of these, the largest item is usually net interest paid, which is commonly a negative figure.3 This is clearly inconsistent with the generally evident growth of the banking sector in national economies.

Present SNA Practice

The present SNA has tried to get around this anomaly by imputing a charge for the undercharged or “free” services that banks usually provide to their customers. This imputed amount represents the difference between the income received by financial institutions on loans and investments made from deposits and the interest paid on these deposits. These imputed charges are considered to have been paid by various industries for banking services and are treated as intermediate consumption of those industries. To avoid double counting and because it is extremely difficult to attribute these imputed charges to the various industries, the aggregate imputed value is deducted from GDP as a single item representing the negative value added of a national financial institution.

Proposals for Change

A review of the proposals for change, prepared by the OECD, suggests two basically different ways of solving the banking income anomaly in the present SNA. One set of proposals suggests a redefinition of interest transactions as the sale and purchase of services for which receipts and payments are explicit and, as such, no imputation is necessary. The other set of proposals suggests that the services provided by banks to depositors are not explicitly paid for, but a service charge is implicit in the interest payments made to depositors. These two approaches have different effects on the measures of total national product and of output originating in nonfinancial industries.

If interest transactions were redefined as the sale and purchase of services, the debtor sector's value added would be reduced by the amount of interest paid, the banking sector's value added would be increased by interest received, and the nation's output would remain unchanged. With appropriate imputation to depositors, banking sector output would increase to the extent that the imputations constitute intermediate and final demand, the borrowing sector's output would remain unchanged, and the national output would increase to the extent that the imputations represent final demand.

Some percentage estimates of this allocation of imputed bank service charges, based on input-output tables of three countries, are4

United States (1982)41.858.2
Australia (1978–79)79.620.4
Japan (1980)95.05.0

With the proliferation of international and offshore banking centers, it can be reasonably assumed that banking services provided to nonresidents may have increased significantly. Many of the countries concerned, however, probably do not have input-output tables relating to the recent past, and such estimates in those cases will be impracticable. For those countries for which net exports or imports of banking services are quantitatively important and data are available, an attempt should be made to make the allocation of imputed bank service charges not only between intermediate and final demand but also between domestic final demand and external final demand.

Illustration with Data from Luxembourg

The rapid growth of banking output and of imputed banking services provided to nonresidents during the period 1975–83 in Luxembourg can be seen from Table 1. The output of banks has been estimated as the sum of actual banking charges (that is, net revenues from commissions, exchange operations, and miscellaneous banking services) and the imputed bank service charge. The latter has been calculated as the difference between the property income received by banks on investments and the interest paid by them on deposits. The export of banking services are estimates from the Central Office of Statistics and Economic Studies (STATEC) of Luxembourg and are derived on the basis of proportions of total deposits and credit attributed to nonresidents.

Table 1.Banking Output in Luxembourg: Components and Allocation, 1975–83 (In billions of Luxembourg francs)
Banking charges (A)2.2424.3735.8354.9507.5745.13411.92211.12912.969
Inputed bank service charge (B)15.77519.59122.71725.74927.08531.20841.37967.00480.123
Output of banks (A + B), of which:18.01723.96428.55230.69934.65936.34253.30178.13393.092
Intermediate demand9.81710.46412.65213.29918.65919.94237.00154.13367.192
Source: Computed from STATEC basic data (Central Office of Statistics, Luxembourg), except for exports of banking services as estimated by STATEC.
Source: Computed from STATEC basic data (Central Office of Statistics, Luxembourg), except for exports of banking services as estimated by STATEC.

The definition of bank output in this illustration is the same as in the present SNA. However, instead of treating the entire imputed bank output as intermediate demand (the SNA treatment), this illustration treats part of the output as exports, thus as final demand. This has the effect of increasing GDP by the additional final demand originating from banking services provided to nonresidents. The significant differences between GDP excluding the export of banking services and GDP including the export of banking services during the period 1975–85 are shown in Table 2.

Table 2.Gross Domestic Product (GDP) in Luxembourg, Excluding and including Export of Banking Services, 1975–85 (In billions of Luxembourg francs)
GDP excluding export of banking
Export of banking servicesa8.213.515.917.416.016.416.324.025.926.528.2
GDP including export of banking
Source: STATEC (Luxembourg).

The proportion of banking services allocated to exports has been calculated on the basis of total bark deposits and credit belonging to nonresidents.

Source: STATEC (Luxembourg).

The proportion of banking services allocated to exports has been calculated on the basis of total bark deposits and credit belonging to nonresidents.

It has not been possible to estimate the proportions of imputed bank service charge that should be attributed to the household and the government sectors because of the unavailability of data on deposits received from and credit provided to those sectors. Thus, GDP is still understated to the extent that the imputed bank service charge attributable to households and government is final demand rather than intermediate consumption. The additional final demand of the imputed bank service charge from these two sectors can in principle be calculated on the basis of opportunity cost or other economic criteria.5 This element should be relatively more important for countries that are not significant exporters of banking services but have a sizable banking sector catering to the domestic economy.

Although the imputation discussed above is not necessary in the “commodity-type service” treatment, it would make the contribution of capital to the value added of a firm depend on whether the capital was borrowed or owned.6 Economic reasoning would suggest that own capital also has cost because it could have earned interest as bank deposits or other types of marketable assets.

II. New Perspective on Sectoral Imputations and Allocations of Banking Output

From a review of the different proposals, it would appear that it is insufficient to impute banking services to depositors only; the imputation should also be allocated, according to conceptually valid procedures to all groups benefiting from it. The paper by Zee on the concept and sectoral imputations and allocations of banking output makes such an attempt.7 It is based on the Mehl variant of the interest-as-composite-payment concept of banking output8 and allocates banking services to both borrowers and depositors, thus highlighting the intermediation function of the banking industry. Although imputations are necessary in the determination of the value of services sold implicitly to depositors, they are based on such economic concepts as opportunity costs of deposits and liquidity premiums, as illustrated for the United States in Zee's paper. Instead of making arbitrary imputations, as in the present SNA, Zee has used data from the Internal Revenue Service, the Federal Reserve System, and other U.S. agencies.

It is suggested that these procedures be considered for adoption in the revised SNA. The application of these procedures would increase the value added in the banking industry and consequently its savings. The savings of the borrowing industries would be reduced, and the entries in the SNA capital accumulation account would change correspondingly.

III. Suggestions for Further Work, with Special Reference to a Deflator for Banking Output

An unresolved question, which has not been discussed in any of the proposals, is that of an appropriate deflator or extrapolator for the current price output of the banking sector. The United Nations' Manual on National Accounts at Constant Prices suggests that some kind of “neutral” index, such as the consumer price index or the index of wholesale prices, be used as a deflator.9 Theoretical questions have action hypothesis in deriving deflators for banking sector output. It has been shown that the deflator generated by the liquidity approach, which assumes that banks produce money to hold, tends to approximate the product of an index of interest rates and an index of the general price level.10 In contrast, the deflator derived from the transaction approach, which regards banks as producers of money to spend, tends to approximate an index of the general price level adjusted by the shifting relationships between an index of interest rates and the inverse of an index of the velocity of circulation.11 Empirical work on this question remains scanty. Further theoretical and empirical work could try to find out whether, taking the same general price index, an index of interest rates on deposits or an index of wages and salaries of bank employees, or a combination of both, is a better deflator for the banking sector's output.

There is a further question that concerns the homogeneity of banking services. Even if there is agreement on the appropriate measures of gross output and value added, there is a need to review the scope for a more detailed commodity classification of banking services from the point of view of identifying homogeneous product groups that are suitable for decomposition into prices and quantities. This question should be addressed in the course of a discussion of appropriate deflators and of the kind of information that would be required. When extrapolation techniques are used (as would be the case in the real input approach), there will be a serious problem of quality change that would need to be addressed because of the substantial changes in the technology available to banks in conducting their business that have taken place in many financial markets in recent years. There is also the question of what indicators to use in estimating value added when balance-sheet and income statements of banks are not available. Because deposits and advances account for the bulk of the banking sector's income-generating transactions, a weighted average of these two could be considered.

To reduce errors and inaccuracies in estimating banking industry output, the SNA classification of the related products needs to be revised. Any revised classification should ensure that the products in the Central Product Classification (CPC) are consistent with the products measured and accounted for in the national accounts. Although national accounts estimates might in most cases be made as broad aggregates, these should be consistent with an implicit aggregation of the separate products as identified in the classification.

The OECD proposals were originally contained in a paper by the Committee on Financial Markets, “Measuring the Output of Financial Institutions in the National Accounts” (Paris: OECD, December 1985).

Presented in a paper by Howell Zee, “Determination of the Output of the Banking Industry in National Income Accounts: A Critical Survey of Concepts” (unpublished; Washington: IMF, Bureau of Statistics, August 1981).

Ibid., p. 6.

United Nations Statistical Office (UNSO), “Imputations for Financial Services,” paper presented at the Meeting of National Accounts Experts, Paris, May 14–16, 1986.

See Zee, “Determination of the Output of the Banking Industry.”

P. Sunga of Statistics Canada has proposed that the gross output of banks would consist of the interest they receive plus any other service charges levied, and their intermediate consumption would include any interest that they pay to their depositors and lenders. Thus, there would be no need for imputation.

Zee, “Determination of the Output of the Banking Industry.”

G.M. Mehl, “A Reconsideration of the Commercial Banking Imputation in the Official U.S. National Accounts” (doctoral dissertation; State University of New York at Binghamton, 1976).

United Nations, Manual on National Accounts at Constant Prices, Statistical Papers, Series M, No. 64 (New York, 1979), p. 92.

The gross national product (GNP) deflator is not a suitable indicator for this purpose because the deflator for the imputed portion of bank output is a component of the overall GNP deflator, which cannot be derived until the commercial bank deflator has been settled. Besides, the GNP deflator is a Paasche-type index, and it represents not only changes in prices but also changes in the relative proportions of goods and services purchased. Either the wholesale price index or the consumer price index can be used, but the latter is to be preferred because it includes services that can show significant difference in price movements compared with goods.

See John A. Gorman, “Alternative Measures of the Real Output and Productivity of Commercial Banks,” in Production and Productivity in the Service Industries, ed. by Victor R. Fuchs (New York: Columbia University Press for the National Bureau of Economic Research, 1969).

    Other Resources Citing This Publication