BEA logo

From the January 2000 SURVEY OF CURRENT BUSINESS



Annual Input-Output Accounts
of the U.S. Economy, 1996

By Sumiye O. Okubo, Ann M. Lawson, and Mark A. Planting

In december 1999, the Bureau of Economic Analysis (BEA) released the 1996 annual input-output (I-O) accounts for the U.S. economy. These accounts are based on an update of the 1992 benchmark I-O accounts, and they reflect the recent comprehensive revision of the national income and product accounts (NIPA's)./1/ The I-O accounts were prepared using 1996 estimates of industry and commodity output and the 1996 estimates of gross domestic product (GDP) from the NIPA revision.

This presentation of the 1996 annual I-O accounts marks the resumption of the regular preparation of annual I-O accounts and the refocusing of the resources that had been used to speed up the preparation of the 1992 benchmark I-O accounts. The last set of annual accounts, which presented estimates for 1987, was published in the April 1992 issue of the SURVEY OF CURRENT BUSINESS. The annual I-O accounts for 1997 are scheduled for release in the fall of 2000.

The annual I-O accounts provide estimates of the domestic production, the export and import of commodities (goods and services), the use of commodities by each industry, the commodity composition of GDP, and the industry distribution of value added. The annual I-O accounts are used in a variety of analytical and statistical contexts, including studies of interindustry relationships within the economy and as the basis for developing satellite accounts on particular aspects of the economy.

This article is presented in two parts. The first part describes the 1996 annual I-O tables, explains how the accounts can be used, and identifies some of the highlights. The second part describes the methodology that was used to prepare the 1996 annual I-O accounts.

1996 Annual I-O Accounts

The annual I-O tables

The 1996 annual I-O estimates are presented in five tables, beginning on page 48. These tables consist of a make table, a use table, a direct requirements table, and two total requirements tables. In addition, alternative make and use tables that are based on a classification of industries that more closely relates to the 1987 Standard Industrial Classification system have been prepared (see the box "Alternative Make and Use Tables" on page 43 and the box "Data Availability" on page 46).

The presentation of the annual I-O tables is generally the same as that of the benchmark I-O tables, but the information is less detailed. The annual I-O tables present summary estimates for 97 industries, while the benchmark I-O tables present more detailed estimates for 498 industries. The annual use and total direct requirements tables present estimates of total value added by industry, while the corresponding benchmark tables present detailed estimates of value added for compensation of employees, indirect business tax and other nontax liability, and other value added. In addition, the presentation of the annual I-O tables has been changed to incorporate the definitional and classificational changes, such as the change in the treatment of business and government expenditures for software, that were introduced in the 1999 comprehensive revision of the NIPA's./2/

The make table shows the commodities that are produced by each industry (table 1), and the use table shows the inputs to industry production and the commodities that are consumed by final users (table 2). As discussed in the section on methodology, the estimates of commodity output and industry output in the make and use tables, and the estimates of final uses in the use table, are based on new source data. Most of the other estimates are based on updated relationships from the 1992 benchmark I-O accounts./3/

The three requirements tables are derived from the make and the use tables. The direct requirements table shows the amount of a commodity that is required by an industry to produce a dollar of the industry's output (table 3). The two total requirements tables show the production that is required, directly and indirectly, from each commodity (table 4) and from each industry (table 5) to deliver a dollar of a commodity to final users.

The uses of the annual I-O accounts

The annual I-O accounts are an important tool for economic analysis because they show the interdependence among the producers and the consumers in the U.S. economy. The accounts can be used to estimate the direct and indirect effects of changes in GDP expenditures for final uses on industries and commodities. For example, the accounts can be used to estimate the effects of a change in Federal Government consumption and investment on industry and commodity output, and, supplemented with additional information, they can be used to estimate the effects of an increase in U.S. exports on employment.

The I-O accounts are used in several ways to prepare other economic statistics. For example, in the 1999 comprehensive NIPA revision, estimates from the 1996 annual I-O tables were used to estimate the 1996 commodity distribution for most of the components of personal consumption expenditures (PCE) for goods,/4/ and estimates from the 1992 benchmark I-O accounts were used to prepare the estimates of final expenditures./5/ Detailed information from the 1996 annual I-O accounts will be used to update the 1992 transportation satellite accounts, the 1992 travel and tourism satellite accounts, and the regional I-O multiplier estimates./6/

Highlights from the 1996 annual I-O accounts

These highlights are drawn from several analytical tables that are based on the 1987 and 1992 benchmark I-O accounts and the 1996 annual I-O accounts. Estimates of changes in current-dollar commodity output, exports, imports, and domestic supply provide insight on the changing structure of the U.S. economy and particularly on the increasing role of trade in recent years./7/ The growing importance of exports and imports is a factor for both rapidly growing, high-tech commodities—such as computer and office equipment and audio, video, and communications equipment—and some slower growing or declining basic-consumer commodities—such as apparel and footwear, leather, and leather products (table A).

As the average annual growth in output for all commodities increased from 5.8 percent in 1987–92 to 6.8 percent in 1992–96, the commodity composition of growth changed significantly (table B). Of the 10 fastest growing commodities in 1992–96, only computer and data processing services was also among the top 10 in 1987–92. In 1992–96, five of the other fastest growing commodities are also considered high-tech commodities—electronic components and accessories; radio and TV broadcasting; special industry machinery and equipment; audio, video, and communication equipment; and computer and office equipment.

For four of the fast growing commodities—computer and office equipment; special industry machinery and equipment; electronic components and accessories; and audio, video, and communication equipment—increases in exports accounted for over one-fourth of the increase in domestically produced output in 1992–96. For all commodities, exports accounted for 6.6 percent of the increase in domestically produced output.

Of the 10 slowest growing commodities in 1992–96, tobacco products and water transportation were among the faster growing commodities in 1987–92 (table C). Declines in output of two commodities—ordnance and accessories and aircraft and parts—partly reflected declines in national defense spending. In addition, a drop in exports accounted for 39 percent of the decline in aircraft and parts and for 6 percent of the decline in ordnance and accessories. In coal mining, a decline in output primarily reflected a decline in coal prices, and it partly reflected slower growth in electric service utilities, which are major users of coal.

Of the 10 commodities with the fastest growth in domestic supply in 1992–96, only electronic components and accessories and "computer and data processing services, including own-account software," were also in the top 10 in 1987–92 (table D). Four of the top 10 commodities are durable goods—farm, construction, and mining machinery; materials handling machinery and equipment; metalworking machinery and equipment; and truck and bus bodies, trailers, and motor vehicles parts. These commodities, which are produced by "heavy" industries, grew relatively slowly in 1987–92, but they rebounded in 1992–96 as the result of the overall growth in the economy after the 1990–91 cyclical contraction.

In 1992–96, increases in imports accounted for 62 percent of the increase in the domestic supply of computer and office equipment, for 34 percent of the increase in the domestic supply of electronic components and accessories, and for 36 percent of the increase in the domestic supply of metalworking machinery and equipment. For all commodities, imports accounted for 8.5 percent of the increase in the domestic supply.

Of the 10 slowest growing commodities in domestic supply in 1992–96, tobacco products, scientific and controlling instruments, ophthalmic and photographic equipment, aircraft and parts, and water transportation were among the faster growing in 1987–92 (table E).

Changes in final uses PCE, private investment, and government consumption expenditures and gross investment—also affected the relative rates of growth in domestic supply. In 1992–96, reductions in national defense spending contributed directly to the declines in the domestic supply of ordnance and accessories, aircraft and parts, and scientific and controlling instruments. The slow growth in scientific and controlling instruments also reflected the decline in aircraft and parts, because it is an important intermediate input in the production of aircraft. The step-up in private investment contributed directly to the strong increases in "computer and data processing services, including own-account software"; computer and office equipment; farm, construction, and mining machinery; special industry machinery and equipment; materials handling machinery and equipment; and metal working machinery and equipment.

Methodology
for the 1996 Annual I-O Accounts

The 1996 annual I-O accounts are based on both the 1992 benchmark I-O accounts and on the most recently revised NIPA's. The 1996 estimates incorporated the definitional, classificational, and statistical changes that were introduced in the 1999 comprehensive revision of the NIPA's, including a definitional change in the treatment of business and government expenditures for software. Business and government purchases of software (except software embedded in other equipment) that were previously treated as intermediate purchases by business or as government consumption expenditures are now treated as investment. The costs—the intermediate inputs and the value-added inputs—that are associated with the production of own-account software are also treated as investment./8/ In the I-O accounts, these costs are added to the gross output of the computer services industry (industry 73A)./9/ These and other changes were incorporated into the 1996 annual I-O estimates to make them more consistent with the NIPA's.

The methodology used to prepare the 1996 annual estimates is similar to that used for the 1992 benchmark estimates, but the annual estimates are based on less comprehensive and less detailed source data. For the annual estimates for which data were unavailable, the relationships from the 1992 benchmark accounts were extrapolated to 1996.

The annual I-O estimates are prepared in five steps: (1) The output total for each industry and commodity is calculated; (2) the commodity composition of intermediate inputs for each industry is estimated; (3) the domestic supply of each commodity is estimated; (4) the commodity compositions of the GDP expenditure components for PCE, gross private fixed investment, and government consumption and investment expenditures are derived; and (5) the table is balanced. In the rest of this section, for each of these steps, the procedures and source data used to prepare the 1996 annual estimates are compared with those used to prepare the 1992 benchmark estimates.

Industry and commodity output totals

For most industries, source data are available to estimate 1996 industry output at the same level of detail as that used for the 1992 benchmark accounts. For manufacturing, trade, and most service industries, the source data for the 1996 estimates are based on sample surveys, whereas the source data for the benchmark estimates were based on more complete and detailed data from the quinquennial economic censuses. For agriculture, insurance, and government enterprises and for major parts of transportation, communications, utilities, finance, and real estate, the source data used for the 1996 estimates are comparable to those used for the benchmark estimates. For the industries for which annual source data at the benchmark level of detail are not available, aggregated industry source data are used to extrapolate the industry output estimates. Table F summarizes by major industry division the number of I-O industries and the number of industry extrapolators available for the 1996 estimates; table G shows the data sources for these estimates.

For most commodities, source data are available to estimate 1996 commodity output at the same level of detail as that used for the 1992 benchmark accounts, and the data used for these estimates are from the same sources as those used to estimate industry output. For commodities without a commodity extrapolator, the commodity output is estimated using the industry output extrapolator and the 1992 benchmark commodity composition of industry output. This procedure is based on the assumption that the proportions of commodities produced by industries were constant from 1992 to 1996.

Commodity composition of intermediate inputs

The 1996 estimates of the composition of intermediate inputs used by each industry are based on 1992 benchmark relationships, with adjustments for changes in relative prices and other factors. First, each industry's 1996 output, valued in 1992 dollars, is estimated using an industry price index that is calculated by weighting commodity price indexes with the commodity composition of each industry's output. Generally, the number of price indexes available for commodities is fewer than the number of commodities; for commodities for which a price index is unavailable, an aggregate price index is applied to multiple commodities (tables F and G). /10/

Second, each industry's 1996 output, valued in 1992 dollars, is multiplied by that industry's 1992 direct requirements per dollar of output to obtain 1996 intermediate inputs valued in 1992 dollars. This procedure is based on the assumption that the 1996 composition of an industry's inputs per dollar of its output valued in 1992 constant dollars is unchanged from that in the 1992 benchmark accounts. The results are then reflated to current-dollar values using commodity price indexes. For the benchmark estimates, source data for the commodity composition of intermediate inputs for each industry were available, primarily from the quinquennial economic censuses.

Finally, commodity taxes, transportation costs, and trade margins for each input are estimated. Commodity taxes are added to raise inputs in basic prices to producers' prices. Transportation costs and trade margins are estimated to provide the producer-value inputs of these commodities by industries, using rates based on 1996 commodity output and 1992 relationships.

Domestic supply

Domestic supply—the total value of goods and services available for consumption as intermediate inputs by industries or as PCE, gross private fixed investment, and government consumption and investment expenditures—is calculated as domestic commodity output plus imports less exports less the change in private inventories. Exports and imports in both the annual and benchmark I-O accounts are based on Census Bureau foreign trade statistics and the BEA international transactions accounts./11/ The 1996 change in private inventories by industry are from the NIPA's, and the commodity composition of inventories held by industries are based on 1992 benchmark relationships.

Commodity composition of final uses excluding trade and change in private inventories

The 1996 annual estimates of the major expenditure components of final uses for PCE, gross private fixed investment, and government consumption and investment are based on the procedures used to estimate GDP in the NIPA's./12/

The major differences between the source data and the estimating procedures used for the 1992 benchmark estimates and those used for the 1996 NIPA estimates by major GDP expenditure component are as follows:

The initial estimates of the 1996 commodity composition of PCE and gross private fixed investment are based on the commodity-flow method. The initial estimates for government expenditures are extrapolated using 1992 benchmark relationships.

Balancing the table

For each commodity, the initial estimates of the commodity distribution of domestic supply to all intermediate industries, PCE, gross private fixed investment, and government consumption and investment expenditures are adjusted so that these shares of domestic supply are similar to the shares in the 1992 I-O benchmark accounts. These estimates are then further adjusted to reflect the 1996 estimates of final expenditure categories from the 1999 NIPA comprehensive revision. Value added by industry is estimated by subtracting the sum of intermediate inputs by industry from industry output.

Box: Acknowledgments

Footnotes:

1. For an overview of the I-O accounts, see Ann M. Lawson, "Benchmark Input-Output Accounts for the U.S. Economy, 1992: Make, Use, and Supplementary Tables," SURVEY OF CURRENT BUSINESS 77 (November 1997): 36–82; and "Benchmark Input-Output Accounts for the U.S. Economy, 1992: Requirements Tables," SURVEY 77 (December 1997): 22–47.

For information on the 1999 comprehensive revision of the NIPA's, see Eugene P. Seskin, "Improved Estimates of the National Income and Product Accounts for 1959–98: Results of the Comprehensive Revision," SURVEY 79 (December 1999): 15–43.

2. See Brent R. Moulton, Robert P. Parker, and Eugene P. Seskin, "A Preview of the 1999 Comprehensive Revision of the National Income and Product Accounts: Definitional and Classificational Changes," SURVEY 79 (August 1999): 7–20.

3. "Final uses" in the input-output accounts are the same as the "product-side" components of GDP in the NIPA's.

4. Brent R. Moulton and Eugene P. Seskin, "A Preview of the 1999 Comprehensive Revision of the National Income and Product Accounts," SURVEY (October 1999): 9.

5. Leon W. Taub and Robert P. Parker, "Preview of Revised NIPA Estimates for 1992 From the 1992 I-O Accounts," SURVEY 77 (December 1997): 11–13. The differences between the 1992 benchmark I-O accounts and the revised 1992 estimates from the 1999 comprehensive revision of the NIPA's largely reflect the definitional change in the treatment of software in the NIPA's and statistical changes, including the use of economic census data on inventories for construction and for mineral industries and the use of newly available source data, primarily final tabulations of State and local government expenditures from the 1992 Census of Governments.

6. The satellite accounts are based on the 1992 benchmark I-O accounts: See Bingsong Fang, Xiaoli Han, Ann M. Lawson, and Sherlene K.S. Lum, "U.S. Transportation Satellite Accounts for 1992," SURVEY 78 (April 1998): 16–27; and Sumiye Okubo and Mark A. Planting, "U.S. Travel and Tourism Satellite Accounts for 1992," SURVEY 78 (July 1998): 8–22. For a description of the regional input-output multiplier estimates, see U.S. Department of Commerce, Bureau of Economic Analysis, Regional Multipliers: A User Handbook for the Regional Input-Output Modeling System (RIMS II), Third Edition (Washington, DC: U.S. Government Printing Office, 1997).

7. Domestic supply is calculated as domestic commodity output plus imports less exports less the change in private inventories.

8. Moulton, Parker, and Seskin, 8.

9. In the I-O accounts, these costs are "redefined"—subtracted from the inputs of businesses that produce own-account software and from government consumption expenditures and added to the inputs of the computer services industry. Own-account construction is treated similarly.

10. Slightly different estimates of intermediate consumption by industry results would have been obtained if the 1996 output and the resulting real intermediate consumption estimates had been reflated using chain-type price indexes like those used for most other BEA estimates. However, any such difference would not affect the allocation between final uses and intermediate inputs, because the allocations for final uses is based on the 1996 current-dollar NIPA estimates and the actual 1996 current-dollar commodity output.

11. Net exports in the I-O accounts and the NIPA's are equivalent.

12. See "Updated Summary NIPA Methodologies," SURVEY 78 (September 1998): 14–35 and Seskin, 15–43.

13. For a description of the retail-control method, see U.S. Department of Commerce, Bureau of Economic Analysis, Personal Consumption Expenditures, Methodology Paper No. 6 (Washington, DC: U.S. Government Printing Office, 1990): 41; for a description of the commodity-flow method, see Lawson, 39.

14. See "Updated Summary NIPA Methodologies," 20–21.

DCSIMG