Benchmark input-output accounts for the U.S. economy, 1987.

Lawson, Ann M. ; Teske, D.A.

THIS ARTICLE presents the 1987 benchmark input-output (I-O) accounts for the U.S. economy.(1) The first part of the article addresses the 1987 benchmark; it discusses the steps taken to speed up the benchmark's completion and then describes some improvements that have been made in the tables. The second part describes the concepts and methods underlying the U.S. I-O accounts and illustrates how the I-O tables are used.

The 1987 I-O estimates presented here are in summary form; that is, they are aggregated to 95 I-O industries from 480-industry detail. The make (production) of commodities by industries is shown in table 1, the use (consumption) of commodities by industries in table 2.1, and the components of value added by industries in table 2.2. The following summary I-O tables will be presented in the May Survey of Current Business: Commodityby-industry direct requirements per dollar of industry output; commodity-by-commodity total requirements, direct and indirect, per dollar of delivery to final use; and industry-by-commodity total requirements, direct and indirect, per dollar of delivery to final use. All of the summary tables, as well as the detailed tables, are available on diskette (see the box on page go).

This article includes supplementary tables that relate the I-O accounts to the national income and product accounts (NIPA's); these tables permit more extensive analyses of the I-O estimates. The article also contains two appendixes: Appendix A provides a list of selected SURVEY articles about the I-O accounts; appendix B provides a concordance between the industry codes used in the I-O accounts and the 1987 Standard Industrial Classification (sic) codes.

The 1987 benchmark I-O estimates will be incorporated into the NIPA's during the next comprehensive NIPA revision, which is tentatively scheduled for release in late 1995.

The 1987 Benchmark Accounts

In recognition of user needs--expressed, for example, by the interagency Working Group on the Quality of Economic Statistics--the Bureau of Economic Analysis (BEA) has developed a program to speed up the availability of I-O accounts.(2) For I-O benchmarks, which are prepared primarily from the Census Bureau's quinquennial economic censuses, the long-term goal is to make the I-O tables available within 5 years of a census year and within 1 year after release of all economic census data.

For the 1987 benchmark, BEA devised a set of procedures that captured the most important parts of the 1987 economic census data, but that abbreviated the normal time-consuming process of assembling a wide variety of other data for constructing components not based on economic census data. These procedures enabled BEA to complete the 1987 tables faster than otherwise would have been the case and to turn its resources toward the 1992 benchmark at the earliest possible time.

Procedures for the 1987 benchmark

In preparing benchmark I-O accounts, BEA relies heavily on economic census data covering mining, construction, manufacturing, wholesale trade, retail trade, transportation, and selected services. The data are released by the Census Bureau as they are completed, over a period of time that usually begins about 1 year after the end of the census year and continues for about 30 months. (For example, the planned release dates for the 1992 census year extend from early 1994 through late 1996.) To estimate outputs and inputs and to allocate commodities across industries and final users, BEA must augment the economic census data with data from hundreds of other sources, such as the U.S. Department of Agriculture, U.S. Department of Transportation, U.S. Department of Treasury, Office of Management and Budget, and other government agencies and private organizations.

In preparing the i987 benchmark I-O accounts, BEA used standard I-O procedures for the estimates of industry and commodity output, except for new construction (see table A). For previous benchmarks, approximately 50 construction industries were analyzed and estimated separately. For the i987 benchmark, the economic census total for construction output was distributed among only five industries--four related to mining and one "all other" category, which covers the remaining industries within new construction and maintenance and repair construction.

[TABULAR DATA A OMITTED]

BEA also used standard I-O procedures for the estimates of industry intermediate inputs where hard data were readily available--primarily for material inputs from the economic censuses. In previous benchmarks, the standard procedure has been to supplement these economic census data with estimates of other intermediate inputs from hundreds of other information sources. For the 1987 benchmark, BEA estimated these intermediate inputs by first extrapolating 1982 benchmark estimates to 1987 based on the change in industry output, and then by adjusting the extrapolated estimates to be consistent with--or to balance--commodity and industry outputs (see table B).

[TABULAR DATA B OMITTED]

Value added components were prepared using the same procedures as in the past.(3) Data for compensation of employees and for indirect business tax and nontax liability are from the U.S. Department of Treasury, Office of Management and Budget, Bureau of Labor Statistics, and Census Bureau; NIPA estimates are also used.

For most final use components--personal consumption expenditures, gross private fixed investment, change in business inventories, exports of goods and services, and imports of goods and services--BEA used the same data and procedures as in the past.(4) Most estimates of personal consumption expenditures and gross private fixed investment were prepared with the commodity-flow method.(5) Inventories held by industries were based on economic census and Internal Revenue Service data. Exports and imports of goods and services were based on data from the Census Bureau and the U.S. balance of payments accounts.

For Federal Government and State and local government final use components, a combination of new and old procedures was used. Total expenditures by type of purchase, for Federal Government and for State and local governments, were obtained from the NIPA's, as in the past. Government purchases by I-O commodity were estimated using 1982 benchmark I-O estimates as weights, a new procedure for the 1987 estimates.

Some procedures used to prepare the i987 benchmark I-O accounts suggest certain caveats. First, the technology represented by the relationships of commodity inputs to industry outputs in the use table (as well as in the commodity-by-commodity and industry-by-commodity total requirements tables) is a hybrid of that in 1987 and that represented in the 1982 benchmark I-O accounts. Second, other value added was derived as a residual for most industries after subtracting total intermediate inputs, compensation of employees, and indirect business tax and nontax liability from total industry output.(6) (For a few industries, estimates of other value added were available from other data sources; for example, other value added estimates for agriculture are from the U.S. Department of Agriculture.) As a result, the other value added component includes estimating errors from other parts of the I-O accounts. For studies requiring comparisons of value added components, users may find BEA's estimates of gross product originating by industry more useful.(7)

Improvements and other changes

The 1987 benchmark I-O tables differ from previous tables in several respects. The summary 1987 benchmark tables, which begin on page 98, cover 95 I-O industries instead of the 85 I-O industries used previously. For the new summary tables, 14 I-O industries were aggregated into 7, and 12 I-O industries were disaggregated into 30.(8) With one exception, the aggregations involved small, declining industries; new construction and repair and maintenance construction were aggregated because of the abbreviated procedures used for the 1987 benchmark. The disaggregations involved large, growing industries. Appendix B shows the new aggregations and disaggregations of I-O industries. (The disaggregated industries are designated with an alphabetical suffix to the 1982 benchmark I-O industry number.)

The industry classification of the I-O accounts is now based on the 1987 sic; the 1982 benchmark tables and subsequent annual tables were based on the 1972 SIC. In addition, the 1987 benchmark tables incorporate all of the 1991 comprehensive NIPA revisions, including the change from gross national product to gross domestic product (GDP).(9)

Introduction to the U.S. I-O Accounts

The I-O accounts for the U.S. economy show the production of commodities by each of nearly 500 industries, in the "make" table, and the consumption of commodities by these industries, in the "use" table. Chart 1 illustrates the make and use tables in matrix form in, respectively, the upper and lower panels. The commodity composition of GDP and the industry distribution of value added are also shown in the use table.

BEA prepares benchmark I-O accounts primarily from data that the Census Bureau collects every 5 years in its economic censuses for mining, construction, manufacturing, wholesale trade, retail trade, transportation, and selected services, as well as in its census of governments. Data from the U.S. Department of Agriculture, U.S. Department of Transportation, U.S. Department of Treasury, and other government agencies and private sources are also used.

The I-O accounts show compactly the relationships between all industries in the economy and all the commodities they produce and use. Estimates for commodities are typically shown at producers' prices.(10) When producers' prices are used, transportation costs and wholesale and retail trade margins are treated as commodities that are separately produced and used by industries (see the section "Definitions and conventions for valuation").

The I-O accounts consist of five basic sets of tables: (i) Make, (2) use, (3) commodity-by-industry direct requirements, (4) commodity-by-commodity total requirements, and (5) industry-by-commodity total requirements.(11) For the 1987 benchmark, details for the value added components of the use table and of the commodity-by-industry direct requirements table are contained in separate tables. Only the make and use tables are presented in this article. The remaining three tables and their descriptions will be published in the May Survey.

The make table

The make table (table 1), in the upper panel of chart 1, shows the dollar value, in producers' prices, of each commodity produced by each industry. In each row, there is one "diagonal" cell that shows the value of production of the commodity for which the corresponding industry has been designated the "primary" producer. Entries in the other cells in the row show the value of production of commodities for which the industry is a "secondary" producer.(12) For example, the newspapers and periodicals industry (row 26A) is the primary producer of the newspapers and periodicals commodity (column 26A). It is also a secondary producer of the following commodities: Paper and allied products, except containers (column 24); other printing and publishing (column 26B); rubber and miscellaneous plastics products (column 32); miscellaneous manufacturing products (column 64); and advertising (column 73D). The sum of all entries in a row is the total output by the industry.

The entries in each column of the make table represent the production by both primary and secondary producers of the commodity named at the head of the column. For example, computer and data processing services (column 73A) includes the output by the primary producer--the computer and data processing services industry (row 73A)--and by the following secondary producers: Computer and office equipment (row 51); audio, video, and communication equipment (row 56); scientific and controlling instruments (row 62); finance (row 70A); and other business and professional services, except medical (row 73c). The sum of all entries in a column is the total output of the commodity. An industry's share of the production of a commodity can be calculated from the values in the make table by expressing the entries in a given colunm as a percentage of the column total. From the 1987 benchmark, for example, column 62 in table 1 shows that the production of scientific and controlling instruments (commodity I-O 62) totaled $86 billion, of which the scientific and controlling instruments industry (industry I-O 62) produced $8o billion, or about 93 percent of the total.

The industry and commodity output totals for this table are estimated primarily from the quinquennial economic censuses, conducted by the Census Bureau (see table A). The economic census data, which are on an sic basis, cover most establishments with payrolls. Information from other government and private sources is used for I-O industries not covered by the economic census data, such as finance, insurance, real estate, utilities, and schools and religious organizations. Data from other government agencies are also used to supplement the economic census data for some industries.

BEA makes two adjustments to the economic census data. First, it adds estimates of the output for establishments without payrolls that are not covered by the economic census data. Second, BEA adjusts for misreported tax return information; this adjustment is necessary because in some cases, the Census Bureau data for expenses and receipts reflect tax return records rather than information collected directly from survey reports.(13)

BEA also adjusts the economic census data based on the sic to the I-O industry classification system to attain greater homogeneity in the input structures for commodities produced by an I-O industry. This type of adjustment is discussed in the section "Definitions and conventions for classification."

The use table

The use table (table 2) is presented in two parts: Table 2.1 shows the dollar value, in producers' prices, of each commodity used by each industry and by each final user; table 2.2 shows detail, in producers' prices, on the value added components used by each industry in table 2.1 to produce its output. In table 2.1, entries in a row show the use of the commodity named at the beginning of the row by each industry or final user named at the head of the column. For example, the commodity radio and TV broadcasting services (row 67) is used by the industries radio and TV broadcasting (column 67) and advertising (column 73D), as well as by persons--that is, as part of personal consumption expenditures (column 91).

In table 2.2, industries are shown in the rows, and total output, total intermediate inputs, and the components of value added are shown in the columns. For example, the total output for the radio and TV broadcasting industry (row 67) was $29 billion, of which $10 billion was labor compensation, $1 billion was indirect business tax and nontax liability, $3 billion was other value added, and $16 billion was intermediate inputs. The column totals for industries in table 2.1 equal the right-hand row totals in table 2.2. For example, the column total for the radio and TV broadcasting industry in table 2.1 equals the row total for that industry in table 2.2, or $29 billion. (The relationship between value added and other parts of the use table is depicted in the bottom panel of chart 1.)

In table 2.1, industry uses sum to total intermediate use, shown in the right-hand column of the industries portion, and the final uses SUM to GDP, shown in the right-hand column of the final uses portion. The total output of each commodity is the sum of all intermediate uses of the commodity by industries and all sales to final users. The total output of each industry is the sum of all intermediate inputs consumed by the industry--that is, the raw

materials, semifinished products, and services that the industry purchases--and of the value added by the industry. For the economy as a whole, the total of all final uses of commodities equals the total value added by all industries, or GDP.

The rows in table 2.1 show the wide variation in the proportion of commodity output that is sold directly to final users. For example, the i987 use table shows that some commodities, such as apparel (the primary product of industry I-O 18), were sold almost entirely to final users; therefore, the demand for these commodities is affected primarily by changes in the buying patterns of final users. Other commodities, such as industrial and other chemicals (I-0 27A), were used almost entirely as intermediate inputs. For these commodities, the connection between production and final uses is primarily indirect and can be traced mainly through industrial users' sales of commodities to final users.

The rows also show the wide variation in the direct usage of commodities by industries. For example, the i987 use table shows that paper and allied products, except containers (I-O 24), with $81 billion of commodity output, were used by nearly all industries. The largest user was other printing and publishing (I-O 26B), which used $15 billion, or 18 percent of total commodity output. In contrast, metal containers (I-O 39), with $12 billion of commodity output, were used by only 20 industries. The largest user was food and kindred products (I-O 14), which used $9 billion, or 74 percent of total commodity output.

The rows in table 2.2 show the wide variation in the use of value added inputs by industries to produce their outputs. For example, the real estate and royalties industry (I-O 71B) required $28o billion of value added inputs, or 74 percent of its total output; of this, $27 billion was for labor compensation, $53 billion was for indirect business tax and nontax liability, and $200 billion was for other value added. In contrast, the livestock and livestock products industry (I-O 1) required $15 billion of value added inputs, or 17 percent of its total output; of this, $3 billion was for labor compensation, $1 billion was for indirect business tax and nontax liability, and $11 billion was for other value added.

BEA estimates intermediate inputs in the use table through a number of processes. The economic censuses are the primary source for data on intermediate inputs; however, BEA must supplement these data to cover establishments without payrolls and industries not covered by the economic censuses. BEA also separates information for some broader categories of purchases into I-O commodities; for example, BEA separates data on purchases of office supplies into purchases of postal service, paper, envelopes, etc., using commodity-shipment proportions and other available information. BEA also uses related information that is available to make I-O estimates of inputs for which there is little hard data. For example, fees paid by industries for accounting services are estimated on the basis of industry employment. (Table B shows the principal methods and sources used for the 1987 benchmark.)

BEA estimates the final uses of commodities either by incorporating data into the I-O accounts directly from other sources after minor adjustment, or--for personal consumption expenditures and producers' durable equipment--by employing the commodity-flow method. An example of source data incorporated directly with only minor adjustments is exports of goods, which is obtained from the balance of payments accounts.

In the commodity-flow method, an estimate is first developed for the total supply of a commodity for domestic use. Then either a fixed percentage of total supply is attributed to final users, or the total supply is adjusted for intermediate purchases and the residual is attributed to final users.(14)

An example of commodity flow using the fixed percentage method can be illustrated by examining its use in estimating personal consumption expenditures for polishes and sanitation goods; in this case, approximately 40 percent of total output is allocated to personal consumption expenditures. An example of commodity flow using the residual method can be illustrated by examining its use in estimating personal consumption expenditures for wheat flour. First, an estimate is made for the total domestic supply of wheat flour: Total wheat flour sales by domestic firms, minus wheat flour exports, plus wheat flour imports. Next, an estimate is made for total consumption of wheat flour by intermediate users, including food manufacturers--of bread, cookies, crackers, and frozen bakery products--and restaurants. The wheat flour consumed by all intermediate users is then subtracted from domestic supply; government purchases of wheat flour are also subtracted. The residual is then assumed to be the wheat flour purchased by persons and is included in personal consumption expenditures.

The components of value added (see footnotes 3 and 6) are estimated using different methods. Compensation of employees by industry is estimated directly from source data. Indirect business tax and nontax liability by industry is either estimated directly from source data or is extrapolated based on the 1982 benchmark. For most industries, other value added is derived as a residual after subtracting total intermediate inputs, compensation of employees, and indirect business tax and nontax liability from total industry output (that is, industry sales receipts). For a few industries, estimates of other value added were available from other data sources; for example, other value added estimates for agriculture are from the U.S. Department of Agriculture.

Uses of the I-O accounts

The I-O accounts have a variety of statistical and analytical uses. For example, they can provide an economic framework to assess data quality and completeness, and they can be used as an analytical economic tool to study industry production. This section describes some uses of the I-O accounts in preparing economic statistics and in studying interindustry relationships within the economy, as well as some of the assumptions analysts must make when they use I-O accounts as an economic tool.

The use of I-O accounts requires certain simplifying assumptions. Among these is the assumption that interindustry relationships established in the I-O accounts for a benchmark year will remain stable over time and through a range of output levels. Users of I-O tables generally must make the assumption that changes in interindustry relationships occur only gradually--for example, that the interindustry relationships represented in the 1987 benchmark are applicable for a band of years surrounding 1987. Also, I-O accounts implicitly assume that all adjustments to a change in final demand are achieved instantly and without price changes. For analyses that require different assumptions, other economic tools may be more appropriate.

Statistical uses.--The I-O accounts are used in several ways to prepare economic statistics. For NIPA comprehensive revisions, they are the single most important regular source for estimating the expenditure components of GDP and for parts of several income components. Because the I-O accounts have an internally consistent framework that tracks the input and output flows in the economy, any estimating weaknesses in the national economic accounts become readily apparent when they are compared with the I-O accounts. For the NIPA revision, the NIPA estimates of personal consumption expenditures and producers' durable equipment are based on the final use components of the I-O benchmark accounts, with additional adjustments to reflect the definitional, classificational, and statistical changes incorporated into the NIPA's since completion of the I-O accounts.(15)

The I-O benchmark accounts are also used as a framework to weight and calculate index numbers for price, volume, and value. For example, BEA uses the I-O-based detailed estimates of producers' durable equipment to weight producer price indexes for calculating the constant-dollar NIPA estimates of producers' durable equipment.

Analytical uses.--The I-O accounts are an important analytical tool because they show the interdependence among various producers and consumers in the economy. Because of their industry detail, the I-O accounts can be used for analyzing a wide range of related empirical issues. The main contribution of the I-O accounts to economic analysis is that they permit analysts to measure the repercussions that changes in final uses have on industries and commodities, both directly and indirectly. For example, an increase in consumer demand for motor vehicles will initially have a direct effect that will increase the production of cars, which in turn will have indirect effects, including increased steel production. Increased steel production will in turn require more chemicals, more iron ore, more limestone, and more coal. Increased car production will also require more upholstery fabrics, and the increased production of these fabrics will require more natural fibers, more synthetic fibers, and more plastics. Further, increased production of synthetic fibers will require more electricity and containers, and so on.

These repercussions are only a few in the continuing chain resulting from the initial increase in consumer demand for motor vehicles. Through I-O analysis, it is possible to trace this chain throughout the economy, measuring the direct and indirect effects on the output of each industry and commodity. Within the I-O accounts, these effects are quantified in coefficient tables. These tables can be used, for example, to determine the impact of a disaster on the economy or, when supplemented with additional information, to compute the effect on employment of an increased demand for U.S. exports. The Federal Emergency Management Agency, the U.S. Department of Defense, and the Census Bureau, among others, have found the I-O accounts to be useful for such studies.

When the U.S. I-O accounts are augmented with regional data, they can show economic impacts by region. For example, a State Government agency has used regional I-O accounts to estimate the economic effects of a high-speed intercity rail project on the State's economy, and a private consulting group has used regional I-O accounts to analyze the impact of a sports stadium on the local economy. BEA's Regional Economic Analysis Division helps planners and analysts estimate the regional impacts of project and program expenditures by industries.(16)

Definitions and conventions for classification

The I-O accounts use two classification systems, one for industries and another for commodities, but both classification systems generally use the same I-O numbers and titles. In the I-O industry classification system, output typically represents the total output of all establishments in each industry, regardless of whether the commodities produced are primary to the industry (that is, make up the largest proportion of the establishment's output) or are secondary (that is, primary to another industry). In the I-O commodity classification system, output represents the total output of the product or service, regardless of the classification of the establishments that produce it. This section discusses first the I-O industry classification system and then the I-O commodity classification system.

The I-O industry classification system is based on the SIC system, which classifies establishments into industries based on their primary products or services.(17) Establishments are defined as economic units that are generally at a single physical location where business is conducted or where services or industrial operations are performed. Establishments are classified into an SIC industry on the basis of their primary products or services.(18)

The I-O industry classification system adjusts the SIC system primarily to attain a greater degree of homogeneity in the structure of inputs to the commodities produced by an I-O industry. The adjustments, which affect I-O-defined primary and secondary production, are called, in I-O terminology, redefinitions and reclassifications.(19) The I-O system also provides for other industries and "special" industries that the SIC does not; these are discussed later in this section.

In a redefinition, the input purchases and the output sales receipts for a particular secondary product or service are moved from the SIC-defined industry to the I-O-defined industry. The input structure of the redefined product or service is assumed to be the same as that for the I-O industry in which the product or service is primary; this assumption is called, in I-O terminology, the commodity-based technology assumption.(20)

An example of a redefinition involves restaurants located in hotels. Both inputs and outputs of these restaurants are moved from the hotels and lodging places industry (the industry of the establishment where the product or service occurs) to the eating and drinking places industry (the industry where the product or service is primary). The input structure related to the output of restaurants located in hotels is assumed to be similar to that for the eating and drinking places industry.

Redefinitions are used in the following cases:

* Construction work (both new construction

and maintenance and repair) performed by

all industries is redefined to the construction

industries. Construction work performed

by and for nonconstruction industries is

referred to as "force-account construction."

* Manufacturing in trade and service industries

is redefined to the appropriate

manufacturing industries.

* Retail trade in service industries is redefined

to the retail trade industry. Services in the

trade industries are redefined to service industries.

Some services are also redefined

within service industries.

* Manufacturers' wholesale sales of purchased

goods (resales) are redefined to the wholesale

trade industry.

* Rental activities of all industries are redefined

to the real estate and rental industries.

* The preparation of meals and beverages in

most industries is redefined to the eating and

drinking industry.

Redefinitions affect a number of industries; however, for most industries, the total output involved is small. Examples of industries with large dollar amounts of redefinitions of secondary products or services out of or into the industry are automobile and repair services (I-O 75), with $131 billion of total industry output, of which $40 billion has been redefined out to a number of other industries and $1 billion has been redefined in from a number of other industries; eating and drinking places (I-O 74), with $209 billion of total industry output, $34 billion out and $1/2 billion in; wholesale trade (I-O 69A), with $424 billion of total output, $7 billion out and $69 billion in; and retail trade (I-O 69B), with $421 billion of total output, $25 billion out and $46 billion in.

In a reclassification, the I-O system creates a secondary product or service from an SIC-defined primary product or service. For these reclassified products and services and for an other SIC-defined secondary products and services that are not redefinitions, the I-O system moves the output receipts from the SIC-defined product or service class to the I-O-defined primary product or service class within the same I-O industry. In this case, total output for the affected industry remains unchanged; however, output for each affected commodity group changes.

An example of a reclassification involves the newspaper industry. The SIC defines the primary product or service classes of this industry as newspaper subscriptions and sales and newspaper advertising. The I-O system considers the primary product or service of the newspaper industry to consist of newspaper subscriptions and sales. It considers the advertising component to be secondary and, therefore, moves advertising receipts or output to the advertising commodity group. Total output for the I-O newspaper industry remains unchanged, but output for the newspaper commodity is reduced, and output for the advertising commodity is increased.

Reclassifications affect about 70 commodities; however, for the most part, the dollar values involved are not very large. Examples of industries with large dollar amounts of reclassified sales receipts are the newspapers and periodicals industry (I-O 26A), for which $20 billion of its $36 billion total commodity output is moved to the advertising commodity (I-0 73D); and the crude petroleum and natural gas industry (I-O 8), for which $12 billion of its $8o billion total commodity output is moved to the gas production and distribution (utilities) commodity (I-O 68B).

When the total requirements tables are calculated, inputs and outputs of each I-O-defined secondary product or service are moved to their particular I-O-defined commodity groups. The input structures of secondary products or services are assumed to be similar to those for the industries in which the products or services are primary; this assumption, in I-O terminology, is called the industry-based technology assumption (see footnote 20).

As mentioned earlier, the I-O system also provides for other industries and "special" industries that the SIC does not. The I-O system replaces the SIC-defined government-owned establishments with two industries to cover government enterprises as defined in the NIPA's--Federal Government enterprises (I-0 78) and State and local government enterprises (I-0 79). The I-O system also provides "special" industries, such as general government (I-O 82), in which output and value added are defined as general government compensation of employees, and the inventory valuation adjustment (I-O 85), which is a NIPA adjustment to derive GDP (see appendix B for a complete listing of I-O special industries).

The I-O commodity classification system is closely related to that for industries. Each commodity receives the code of the industry in which the commodity is the primary product. This code is then used to group production of the commodity in the industry in which it is the primary product with its production in other industries in which it is a secondary product.

In several cases, the I-O commodity classification differs from that specified by the industry classification. If the same commodity is the primary product of more than one SIC industry, all of the I-O commodity is assigned the I-O commodity number that corresponds to the I-O industry that is the largest producer of the commodity. This results in there being no commodity output for the following I-O commodity groups: Forest products (commodity 2.0701); knit outerwear mills (commodity 18.0201); knit underwear and nightwear mills (commodity 18.0202); knitting mills, not elsewhere classified (commodity 18.0203); fertilizers, mixing only (commodity 27.0202); cold-rolled steel sheet, strip, and bars (commodity 37.0104); steel pipe and tubes (commodity 37.0105); secondary nonferrous metals (commodity 38.0600); Federal electric utilities (78.0200); State and local government passenger transit (commodity 79.0100); and State and local government electric utilities (commodity 79.0200).

Definitions and conventions for valuation

Transactions in commodities are typically valued in I-O accounts at producers' prices, which exclude distribution costs (transportation costs and wholesale and retail trade margins), but include excise taxes collected and paid by producers. Transportation costs and trade margins are shown as separate purchases by the users of the commodities. The sum of the producers' value, transportation costs, and trade margins equals the purchasers' value.

The I-O tables do not trace actual flows of commodities to and from wholesale trade and retail trade. If trade were shown as buying and reselling commodities, industrial and final users would make most of their purchases from a single source--trade. To show the relationship between the production of commodities and their purchase by intermediate and final users, commodities are shown as if they move directly to users, bypassing trade. The margin associated with a commodity is shown as a separate purchase of the commodity from wholesale trade and retail trade by users. Transportation costs are the freight charges paid to bring the commodity from the producer to the user, either intermediate or final. All transportation costs are included in the transportation rows (rows 65A-E) of the use table.

Wholesale trade has one primary product--distributive services for the sale of goods to final users other than for personal consumption expenditures. Examples of distributive services provided by wholesalers include merchandise handling, stocking, selling, and billing.

Wholesale trade output is measured one way for merchant wholesalers, agents, and brokers and another way for manufacturers' sales branches. For merchant wholesalers, agents, and brokers (on own account), wholesale margin is measured as wholesale sales receipts less the cost of goods sold plus taxes collected by the distributor. For manufacturers' sales branches, it is measured as expenses plus taxes collected by the sales branches.

Nonmargin output occurs when the wholesale trade service is purchased separately from the commodity. Nonmargin output includes, for example, a sales commission paid to a wholesaler acting as a broker. Nonmargin output is measured as the sum of expenses on goods sold by manufacturers' sales offices, commissions on goods sold by agents and brokers, and customs duties. Wholesale trade output--both margin and nonmargin--is included in the wholesale trade row (row 69A) of the use table.

Retail trade has one primary product--distributive services for the sale of goods to persons. Retail output is defined as the retail margin, which is measured as retail sales less the cost of goods sold plus the taxes collected--if any--by retail trade establishments.

Retail trade margins also apply to some purchases of goods by other final users; for example, retail trade margins apply to some purchases of personal computers by business and are included in gross private fixed investment. All retail trade margins are included in the retail trade row (row 69B) of the use table.

Imports of goods and services, a component of final uses, are treated in one of two ways, depending on whether or not they are comparable to U.S. commercially produced goods and services. Those that are comparable are included in the use table along with the distribution of the output of their domestic counterparts. The U.S. domestic port values of imported commodities are shown as negative entries in the imports of goods and services column of final use (column 95), so that the row total for a commodity equals the domestic output of that commodity. Other imported goods and services--those not comparable to U.S. commercially produced goods and services, and those purchased and used abroad by U.S. residents--are shown in the use table row for noncomparable imports (row 8o). Examples of noncomparable imports are coffee beans and parakeets; an example of goods purchased and used abroad by U.S. residents is food purchased by U.S. military personnel stationed abroad. The total value of all noncomparable imports is shown as a single negative entry in the imports of goods and services column (column 95).

Imports of goods by commodity (the entries in column 95) are valued at U.S. domestic port values plus duties. Imports of services are valued at producers' values. The entries for transportation imports and for trade imports include adjustments that convert the total of all commodity imports of goods and services to a foreign port value equivalent. This adjustment is made for conceptual consistency between the I-O accounts and the NIPA's and the balance of payments accounts.

Exports of goods and services--both by commodity and as a total--are valued in U.S. producers' prices, which are considered to be equivalent to U.S. domestic port values. Exports are also a component of final uses.

Inventory change, another component of final uses, represents the change in inventory of each commodity, wherever held, over the benchmark year. It is stated at book value--that is, at its original cost--in the use table. The inventory valuation adjustment, which converts inventory change from book value to replacement cost, is shown as a single entry for the total of all commodities (row 85, column 93).

Supplementary tables

Four supplementary tables, which can be used with the five basic sets of I-O tables, are provided with this article. Three tables (tables C-E) cover the I-O commodity composition of NIPA final demand, of NIPA personal consumption expenditures, and of NIPA producers' durable equipment; a fourth table (table F) reconciles I-O exports of goods and services and imports of goods and services with NIPA estimates.

The commodity composition tables are necessary as bridges between the I-O accounts and the NIPA's because the two sets of accounts are based on different valuations and definitions. In the I-O accounts, final use categories are expressed in producers' prices; in the NIPA's, final demand categories are expressed in purchasers' prices. Also, the definitions of I-O final use categories differ from those of the NIPA final demand categories. Before the I-O total requirements tables can be used to measure and analyze the changes in commodity or industry output requirements arising from changes in the level or composition of NIPA final demand, NIPA final demand categories must be converted to equivalent I-O final use categories. That is to say, the analysis should be consistent with I-O final use commodities that are valued at producers' prices for the I-O year, with separate entries for transportation costs and trade margins.

Table C shows the I-O commodity composition in 1987 of each NIPA category of final demand in producers' and purchasers' prices. It provides a bridge between I-O commodities in producers' prices and NIPA final demand categories in purchasers' prices. For each I-O commodity within a category of NIPA final demand, the table shows the transportation costs and trade margins included in the purchasers' prices.

[TABULAR DATA OMITTED]

Table D shows the I-O commodity composition in 1987 of each NIPA category of personal consumption expenditures (NIPA table 2.4) in producers' and purchasers' prices. It provides a bridge between I-O commodities in producers' prices and NIPA personal consumption categories in purchasers' prices. For each I-O commodity within a NIPA category, the table shows the transportation costs and trade margins included in the purchasers' prices.

Table E shows the I-O commodity composition in 1987 of each NIPA category of producers' durable equipment purchases (NIPA table 5.8) in producers' and purchasers' prices. It provides a bridge between I-O commodities in producers' prices and NIPA producers' durable equipment categories in purchasers' prices. For each commodity, the table shows the transportation costs and trade margins included in the purchasers' prices. This table is useful for analyses relating the effects of changes in investment on industry and commodity output.

[TABULAR DATA E OMITTED]

Table F reconciles the I-O estimates of exports and imports of goods and services with those in the NIPA's. The same adjustments are made for both exports and imports; therefore, there is no net effect on total GDP. The adjustments are necessary because the NIPA's--unlike the I-O accounts--include in imports the U.S. merchandise that is returned to the United States from other countries and in exports the foreign merchandise that is reexported from the United States to other countries.(21) The NIPA's also exclude definitional and statistical revisions to the balance of payments accounts between NIPA comprehensive revisions.

Table F.--Relation of Exports and Imports in the Input-Output
Accounts to the National Income and Product
Accounts, 1987 Benchmark
 [Millions of dollars]
 1987
Exports of goods and services, NPA 363,952
Less: U.S. merchandise returned 6,781
 Reexports 8,875
Plus: Statistical revisions, BPA 276
Equals: Exports of goods and services, I-O 348,572
Imports of goods and services, NIPA 507,050
Less: U.S. merchandise returned 6,781
 Reexports 8,875
Plus: Statistical revisions, BPA -952
Equals: Imports of goods and service, I-O 490,442
NIPA National income and product accounts
BPA Balance of payments accounts
I-O Input-output accounts

Appendixes A and B and tables 1 and 2 follow.

Appendix A.--Chronological List of Selected Survey of Current Business Input-Output Articles

1. Morris R. Goldman, Martin L. Marimont, and Beatrice N. Vaccara, "The Interindustry Structure of the United States: A Report on the 1958 Input-Output Study," November 1964.

2. "Industrial Impact of the 1966 Housing and Commercial Building Decline," November 1966.

3. "Input-Output Structure of the U.S. Economy: 1963," November 1969.

4. Allan H. Young and Claiborne M. Ball, "Industrial Impacts of Residential Construction and Mobile Home Production," October 1970.

5. Beatrice N. Vaccara, "An Input-Output Method for Long-Range Economic Projections," July 1971, Part I.

6. Philip M. Ritz and Eugene P. Roberts, "Industry Inventory Requirements: An Input-Output Analysis," November 1973.

7. "The Input-Output Structure of the U.S. Economy: 1967," February 1974.

8. Irving Stern, "Industry Effects of Government Expenditures: An Input-Output Analysis," May 1975.

9. Philip M. Ritz, "The Input-Output Structure of the U.S. Economy, 1972," February 1979.

10. Philip M. Ritz, Eugene P. Roberts, and Paula C. Young, "Dollar-Value Tables for the 1972 Input-Output Study," April 1979.

11. "The Input-Output Structure of the U.S. Economy, 1977," May 1984.

12. "Benchmark Input-Output Accounts for the U.S. Economy, 1982," July 1991.

13. "Annual Input-Output Accounts of the U.S. Economy, 1987," April 1992.

Appendix B

[TABULAR DATA OMITTED]

Data Availability

The estimates from the 1987 benchmark I-O accounts are available on diskette at two-digit (95 I-O industries) and six-digit (480 I-O industries) levels. They can be ordered for "transactions," for "total requirements," or for "all." "Transactions" includes the six-digit make table, use table, direct requirements coefficients table, and estimates by commodity of transportation costs and of wholesale and retail trade margins. "Total requirements" includes six-digit industry-by-commodity or commodity-by-commodity coefficients. Products specifying "all" contain all above data, but for the two-digit I-O industry level only. Each product includes information on the mathematical derivation of the coefficients tables. The BEA accession numbers and the prices for these products are listed below.

For further information about I-O products or when ordering by MasterCard or Visa, call the Interindustry Economics Division at (202) 606-5585. To order by mail, write to the Public Information Office, Order Desk, BE-53, Bureau of Economic Analysis, U.S. Department of Commerce, Washington, DC 20230. Specify the item, accession number, and price of the product(s) being ordered. For foreign shipment, add 25 percent to the total amount of the order. A check or money order payable to "Bureau of Economic Analysis" must accompany all written orders. Be sure to include a return address.

BEA's 1987 benchmark I-O accounts, at both the two-digit and six-digit levels, will also be available on CD-ROM through the Commerce Department's National Economic, Social, and Environmental Data Bank (NESE-DB) CD-ROM. The NESE-DB is produced quarterly in February, May, August, and November. Call the Office of Business Analysis at (202) 482-1986 for more information or to place an order. The NESE-DB is also available for public use at over 900 Federal Depository Libraries.

[TABULAR DATA OMITTED] (1) Earlier benchmarks covered 1947, 1958, 1963, 1967, 1972, 1977, and 1982. BEA also has produced annual I-O accounts based on less comprehensive source data. The most recent annual accounts, for 1987, were presented in the April 1992 Survey of Current Business. (2.) See "Improving the Quality of economic Statistics: The 1992 Economic Statistics Initiative," Survey 71 (March 1991): 4-5. (3.) Value added equals gross output (sales or receipts and other operating income, plus inventory change) minus intermediate inputs (consumption of goods and services purchased from other industries or imported). It includes compensation of employees, indirect business tax and nontax liability, and other value added. (4.) In the I-O accounts, change in business inventories covers commodities wherever held; capital purchases--producers' durable equipment and structures--are included in gross private fixed investment; and imported commodities are included with domestically produced commodities in both final use and intermediate use. (5.) The commodity-flow method generally begins with an estimate of the total supply of a commodity available for domestic uses; it then either attributes a fixed percentage of supply to final users, or it adjusts for intermediate purchases and attributes the residual to final users. For more information, see U.S. Department of Commerce, Bureau of Economic Analysis, Personal Consumption Expenditures, Methodology Paper Series MP-6 (Washington, D: U.S. Government Printing Office, June 1990): 31-34. (6.) For most I-O industries, other value added includes consumption of fixed capital, proprietors' income, corporate profits, and business transfer payments. For banking and for credit agencies other than banks, other value added also includes net interest. For owner-occupied dwellings and for real estate agents, managers, operators, and lessors, it also includes rental income. For the six industries covering the Federal Government and State and local government enterprises, it also includes current surplus less government subsidy payments. (7.) See Robert P. Parker, "Gross Product by Industry, 1977-90," SURVEY 73 (May 1993): 33-54; and Robert E. Yuskavage, "Gross Product by Industry, 1988-91," Survey 73 (November 1993): 33-44. (8.) The net addition of industries resulting from the aggregations and disaggreptions of 1982 I-O industries is 11. In addition, the rest of the world is no longer technically considered to be an industry because of the change from GNP to GDP as the primary measure of final demand. Thus, there is a net increase of 10 industries in the 1987 benchmark. (9.) The 1991 NIPA revision was described in the following SURVEY articles: "A Preview of the Comprehensive Revision of the National Income and Product Accounts: Definitional and Classificational Changes," September 1991; "A Preview of the Comprehensive Revision of the National Income and Product Accounts: New and Redesigned Tables," October 1991d "The Comprehensive Revision of the U.S. National Income and Product Accounts: A Review of Revisions and Major Statistical Changes," December 1991 (10.) Estimates for commodities in purchasers' prices can be derived by adjusting for transportation costs and for wholesale and retail trade margins; these costs and margins are included on the diskettes that can be ordered for the 1987 benchmark I-O (see the box on page 90). (11.) In the designation of I-O tables, the row is referred to first and the column second. Thus, tables in which commodities appear in the rows and industries in the columns are designated "commodity-by-industry" tables, and tables in which industries appear in the rows and commodities in the columns are designated "industry-by-commodity" tables. (12.) Primary and secondary products and the classification of industries are discussed further in the section "Definitions and conventions for classification." (13.) See Robert P. Parker, "Improved Adjustments for Misreporting of Tax Return Information Used to Estimate the National Income and Product Accounts, 1977," Survey 64 (June 1984): 17-25. (14.) See Personal Consumption Expenditures, pages 31-34. (15.) For more information on the I-O accounts and their relationship to the NIPA's, see Personal Consumption Expenditures, pages 17 and 31-34. (16.) A typical I-O table in the Regional Input-Output Modeling System is derived mainly from two data sources: (1) The U.S. benchmark I-O accounts and (2) BEA's four-digit SIC county wage-and-salary data. For more information, see U.S. Department of Commerce, Bureau of Economic Analysis, Regional Multipliers: A User Handbook for the Regional Input-Output Modeling System (RIMS II), Second Edition (Washington, DC: U.S. Government Printing Office, 1992). (17.) The I-O two-digit and six-digit industry categories and their composition in terms of the 1987 SIC codes are given in appendix B. (18.) For a discussion of the SIC system, see Office of Management and Budget, Executive Office of the President, Standard Industrial Classification Manual: 1987, (Springfield, Virginia: National Technical Information Service 1987): 11-18. (19.) Fewer I-O adjustments to SIC-defined industries may be necessary for the 1997 and subsequent benchmark I-O accounts when the North American Industry Classification System (NAICS) is completed. The proposed NAICS is expected to be a common international system--covering the United States, Canada, and Mexico--for grouping establishments by similarity of production process. For a discussion, see Jack E. Triplett, "Economic Concepts for Economic Classifications," SURVEY 73 (November 1993): 45-56. (20) The I-O commodity-based and I-O industry-based technology assumptions are important when estimating the total-requirements tables. The significance of the assumptions is discussed elsewhere in the economic I-O literature. See, for example, United Nations, System of National Accounts, 1993, prepared under the auspices of the Inter-Secretariat Working Group on National Accounts (New York: United Nations, 1993): chapter 15, in particular pages 367-70; and Ronald E. Miller and Peter D. Blair, Input-Output Analysis: Foundations and Extensions (Englewood Cliffs, New Jersey: Prentice-Hall, Inc., 1985): 149-99. (21). U.S. merchandise returned consists of domestically produced goods that were previously exported to other countries for processing or assembly, or both, and then returned to the United States. An example would be articles of metal that are manufactured in the United States, then exported for further processing abroad, and then returned to the United States for more processing. Reexports consists of commodities of foreign origin that were previously imported into the United States and then exported from the United States in substantially the same condition as when imported. An example would be imported foreign-made monitors that are purchased by U.S. personal computer manufacturers, joined with U.S.-made consoles, and then exported to a third foreign country.

Ann M. Lawson, Chief of the Interindustry Economics Division, directed the preparation of the 1987 benchmark input-output study and coauthored the article with D.A. Teske. Mark A. Planting, Acting Assistant Division Chief, planned and coordinated division efforts to produce the estimates. Belinda L. Bonds, Chief of the Goods Branch, and Karen Horowitz, Chief of the Services Branch, assisted in the planning and implementation of the study and in the estimation, review, and finalization of the data. Brian D. Kajutti designed the data processing system and coordinated the computer programming and processing efforts.

Staff contributors were William A. Allen, Timothy D. Aylor, Alvin D. Blake, Cheryl Carlson, Esther Carter, Jeffrey W. Crawford, Sergio Delgado, Gary T. Fee, Kara Gordon-Palley, Carole Henry, David Huether, Greg M. Key, Myles J. Levin, Fritz Mayhew, William McCarthy, Donna McComber, Clinton P. Mccully, Rhonda E. Monroe, Ted Morgan, Diane E. Nisson, Robert S. Robinowitz, Brooks B. Robinson, Timothy F. Slaper, Patricia A. Washington, Raquel Watson, and Diane Young.