Measuring the Government Sector of the U.S. Economic Accounts (1998)

The international SNA is intended for use by countries with widely varying economic and political structures. Hence, it provides for flexibility in implementation of the overall guidelines. In applying the SNA guidelines to the U.S. accounts, numerous decisions have to be made regarding the classification of individual types of government receipts and expenditures and the appropriate use of imputation and valuation techniques. Several such issues, ranging from the sector classification of Part B of Medicaxe to the appropriate techniques for estimating government production in constant prices, were subsequently discussed at the workshop.

Another set of issues concerns the implementation process: that is, how to explain and document changes so that the changes are well understood by users, confidence in the integrity of the accounts is maintained, and the accounts retain their value for their traditional uses and for a wide and varied group of users.

It is clear to us that certain general principles pertaining to the transition to the SNA are critical. These include the need for full documentation of changes; the need to relate national account measures to other measures of government finances, such as the unified budget; the desirability of reasonable consistency of practice among different sectors of the accounts; and the need to facilitate ease of use.

Complete documentation is important to maintaining public confidence in the integrity and reliability of the national accounts. The need for changes in

accounting procedures has to be fully explained, and the data needed to reconstruct the traditional accounting categories and to relate the national accounts to other measures of government finances has to be made available.

The changes introduced in 1996 already have affected such highly publicized measures as the overall government surplus or deficit, the national savings rate, and the amount of domestic fixed investment. It is important that the public understand that the changes in these numbers are neither arbitrary nor politically motivated, but rather result from carefully considered, long-planned implementation of internationally agreed upon guidelines that are in use throughout the world.

The ability to reassemble the government sector estimates on their traditional basis will be important to many users. Even the most knowledgeable users of the accounts, who may be strongly supportive of the new system, may want to know how certain estimates would have looked on the old basis. Economists involved with macroeconomic analysis and short-term forecasting may have a particular need to reconstruct estimates on the traditional basis. For example, the immediate impact of government activity on the economy may be best measured by total current government purchases, including investment purchases, and this concept is incorporated into formal and informal forecasting models used by many economists.

Although there are serious gaps in the existing statistical information about government finances, users often feel overwhelmed or confused by the abundance of information that is available. Presentation of this information in ways that are objective, comprehensible, and reconcilable with one another is a major challenge. Simply to list some of the most important sources of information about government finances illustrates the magnitude of the task.

In addition to the national accounts, major sources of information about federal government finances include current service budget estimates prepared by OMB and the Congressional Budget Office (CBO), the President's budget requests, congressional enactments of authorizations and appropriations, Treasury Department statements of actual receipts and expenditures, and Census Bureau statistical series on federal spending by type by city and county. Each of these data sets serves an important purpose, but they vary from one another in many aspects of timing, definition, and classification. Some are estimates of actual spending and receipts; others represent projections or proposals. Spending and receipt estimates may be on a cash, accrual, or obligational basis. Loan and loan guarantee programs present their own, complex set of accounting problems.

State and local governments, of course, maintain their own sets of accounts. Fiscal year definitions and accounting practices may vary. Intergovernmental

cooperative arrangements, such as transportation authorities and water districts that obtain financial support from participating governments, complicate the picture. Through its annual surveys and quinquennial censuses, the Bureau of the Census collects financial data from state and local governments and uses standardized definitions to present this data on a consistent basis. This information in turn becomes the primary input for the NIPA estimates of state and local government finances.

Tables published annually in the NIPAs reconcile the NIPA estimates of federal government receipts and expenditures with the federal unified budget and the NIPA estimates of state and local government receipts and expenditures with the Bureau of the Census government finance data.⁵ In addition, BEA each year publishes estimates of the President's budget requests on a NIPA basis: these include quarterly as well as annual estimates so that the timing of receipts and expenditures can be better understood and analysis can be performed on either a fiscal or calendar year basis (see, e.g., Beall and Northwood, 1995).

Adoption of the SNA framework is in some ways increasing the definitional and timing differences between the national account government sector and other measures of government finances; this will increase the importance of maintaining accounts and accompanying explanations that reconcile the national accounts with other principal measures.

The definition, classification, and valuation of government and government-sponsored activities present extraordinarily difficult issues. Difficulties arise from the sheer variety of government activities, the differing practices of different levels of government, the mixed public-private nature of many activities, and the special difficulties of valuing output not sold at market prices. Consistency of treatment between the government sector and other sectors of the economy or among specific spending programs within the government sector will not always be attainable or desirable. When consistency is possible, however, it offers the advantages of more fully comparable values for output and accounts that are easier to understand. Some examples of the difficulties and the advantages of consistency are discussed below, including differing views and factors for consideration in making a decision.

Because so little of government output is sold at market prices, the NIPAs value government output at the cost of production. However, prior to 1996, the

costs of the services of government-owned capital was not included. The SNA also values government output at cost of production, but it incorporates the costs of consumption of fixed capital (Dobbs, 1993). Thus the SNA, and the NIPA changes introduced in 1996, move the valuation of government output closer to a full-cost valuation. However, the accounts still do not incorporate a return to capital, or what might be called a profit, in the valuation. The valuation method remains inconsistent with that used for most private output, which is valued at market prices and which thereby includes return to capital.

Adoption of the SNA does not achieve consistency of valuation methods among all sectors of the accounts. Complete consistency may be neither feasible nor desirable, but workshop participants expressed interest in continued research on incorporating market price information into the valuation of government output. This is discussed further below, in ''Going Beyond the System of National Accounts.''

As discussed above, the SNA treats fully funded government employee pension plans consistently with private pension plans, classifying employer contributions as income of the household sector and the net equity of pension funds as household assets. This approach seems reasonable for state and local government pension plans. The operations of state and local plans are closely analogous to their private counterparts. The plans operate with a substantial degree of independence from their parent governments and invest largely in private and federal assets, and their assets and earnings are not normally a factor in the budgetary decisions of their parent governments (Dobbs, 1993).

Federal employee pension plans operate somewhat differently. They invest exclusively in federal securities, and the federal government has considerable power to change the benefits promised by the plans and to modify cost-of-living adjustments. Employee pension plan receipts and benefits are shown in the federal budget as current federal receipts and expenditures (Dobbs, 1993). Their special characteristics cast doubt on the extent to which prospective benefits are true household assets and raise questions about the adoption of the SNA treatment for federal plans. Yet pension entitlements of federal employees are, in general, secure, and there are analytic advantages of the SNA approach: the benefits of consistent treatment of all fully funded pension plans and the advantages of international comparability.

The SNA uses a broader definition of government enterprises than does the current NIPAs. The NIPA criteria for classifying an entity as a government enterprise require that it cover a "substantial portion of its operating cost" through

sales to the public (Bureau of Economic Analysis, 1988:6). The corresponding SNA criteria are that the entity charge "economically significant" prices for its output and be able to maintain business credit and finance some capital formation without assistance from the owning government. Both the NIPA and the SNA criteria are general in nature and require judgment in their application to specific entities. In Dobbs (1993), the SNA criteria are interpreted as adding publicly owned hospitals and colleges and universities, as well as some federal lending activities, to the groups of entities included as government enterprises.

Colleges and universities offer an example of the difficulties inherent in deciding where the line should be drawn between government enterprises and general government. Many state universities largely share the characteristics of private universities. They operate with a large degree of independence from the state government and cover a significant portion of their costs through private tuition payments, contributions, and endowment income. Not all public colleges and universities have these characteristics, however. Community colleges typically charge little tuition and have little private financial support. They often would not meet the SNA criteria for classification as government enterprises. In the Dobbs (1993) estimates of the impact of the SNA, however, all colleges, including community colleges, were classified as government enterprises. According to Dobbs, the data necessary to separate community colleges from other colleges and universities are not available. In this example, as in many others, improved data will be needed if the SNA is to be fully implemented.

The NIPAs place private colleges and universities (and most other private not-for-profit entities) in the household sector. The SNA would move them to the private business sector. If this were done and public colleges and universities were also moved to the business sector (by classifying them as government enterprises), one type of consistency would be achieved by placing all colleges and universities in the same sector. However, there is disagreement about the classification of colleges and universities as government enterprises: the advantages of having public education at all levels grouped together in the same sector may outweigh the advantages of enterprise classification for colleges and universities, which would argue for keeping all public education in the general government sector.

A further consideration is the SNA's recommendation that national accounts contain a separate private not-for-profit sector. As noted, the SNA would place private colleges and universities in the business sector. However, an earlier report of the Committee on National Statistics (Slater and David, 1994) recommended that private not-for-profit educational institutions be placed instead in the new not-for-profit sector. Achieving reasonable consistency in sectorial definitions in the accounts requires an interrelated set of decisions with respect to various account sectors, not simply a set of isolated decisions about the government sector.

Hospitals present classification questions similar to those for higher educa-

tion. The hospital industry is divided among private for-profit, not-for-profit, and public components. Pricing policies of public hospitals vary widely.

Neither the present NIPA classifications nor the SNA will result in all components of either the education or the health care industry being assigned to the same sector. These are only two of many examples of complex classification questions to which there is no single solution. Cross classifications of government and government enterprise activities by industry or function comparable to private industry classifications (for profit and not for profit) will be desirable for many types of analysis. The SNA makes provision for such cross-classification.

A paper prepared for the workshop (Popkin, 1993) reviewed the SNA's proposed Classification of Functions of Government (COFOG) and concluded that its functional concepts are well defined and comparable to private-sector market concepts, although updating of COFOG in the area of environmental activities would be desirable.

The national accounts are used for many purposes and by a broad spectrum of users, who vary in their familiarity with the accounts and their degree of economic sophistication. In implementing the SNA, care should be taken that new complexities do not diminish the usefulness of the accounts for users.

Good documentation and explanation of changes, clear delineation of the steps involved in moving from one system to another, and reasonable consistency of definitions and practice all will contribute to public understanding and ease of use. For some decisions, simplicity and ease of use might well be the deciding factor.

One example of such a decision is whether to treat the Supplementary Medical Insurance Trust Fund (Medicare, Part B) as a social insurance fund, an issue that was discussed at the workshop. Participation in Part B of Medicare is voluntary, but elected by most eligible persons. It is financed about one-quarter by insurance premiums and three-quarters by transfers from general government (Dobbs, 1993). Viewed in isolation, Part B does not meet the SNA criteria for a social insurance fund; viewed as an integral part of Medicare, it would. While conceptual arguments could be advanced for including only Part A of Medicare in the social insurance subsector of the accounts, we believe that these arguments are clearly outweighed by the convenience to users of presenting the entire Medicare program in same sector of the accounts.

Implementation of the SNA is a complex undertaking. The changes being made to the presentation of the government sector are of widespread public interest. Implementation decisions should reflect adherence to a set of principles stemming from the basic need to maintain and enhance the usefulness of the accounts.

RECOMMENDATION 2. A consistent set of principles should be developed and followed in making the many decisions that implementation of the SNA requires. These principles should reflect the need to maintain and enhance the usefulness of the accounts for a variety of applications and by a wide and varied group of users and should include:

• provision of sufficient information on the derivation and composition of changes to permit reconstruction of the government sector of the accounts on its traditional basis;
• presentation of the supplementary information needed to compare government activities by industry with private components of the same industry and to relate the government sector of the accounts to other measures of government receipts and expenditures;
• maintenance of reasonable consistency of the government sector with other sectors of the national income and product accounts with respect to definitions, classifications, and estimation practices, including valuation and imputation techniques; and
• preservation of sufficient clarity and simplicity of basic presentations.