Printer Friendly

Gross product in industry, 1977-88: a progress report on improving the estimates.

Gross Product by Industry, 1977-88: A Progress Report on Improving the Estimates

This article reports on the first phase of BEA's work to improve the estimates of constant-dollar gross product by industry. In 1989, BEA suspended publication of these estimates in response to criticism of their accuracy and outlined its plans to improve them.(1) Revised constant-dollar estimates for 1977-87 and new estimates for 1988, the results of the first phase of this work, are presented in this article.

The picture of U.S. industries shown by the gross product estimates is not altered much by the revision. In terms of growth rates for broad industry groups, the general pattern is the same both before and after the revision: Wholesale and retail trade and services show the highest constant-dollar growth rates, and government enterprises and the agriculture, mining, and construction group show the lowest (chart 4). The average annual rate of growth of manufacturing industries--of special interest to many users--was estimated at 2.7 percent for 1977-87 before the revision, and it is now estimated at 2.5 percent. Several of the improvements in methodology that are incorporated into the revised estimates had surprisingly little impact on most industries.

The first part of the article provides an overview of the gross product by industry series; it describes the estimating methods used by BEA, the improvements incorporated into the revised estimates, how these improvements address the earlier criticisms, and the further work to be done to improve the estimates. The second part discusses the impact of the revisions on industry growth rates. A "Technical Note" describes the methodologies used for the previously published estimates and the revised estimates. The estimates are presented in a set of tables at the end of the article.

Gross Product by Industry--An Overview

Gross product, or gross product originating (GPO), by industry is the contribution of each industry--including government and the rest of the world--to GNP.(2) An industry's GPO, often referred to as its "value added," is equal to its gross output (sales or receipts and other operating income, plus inventory change) minus its intermediate inputs (consumption of goods and services purchased from other industries or imported).

In concept, GNP measured as the sum of GPO in all industries is the same as GNP measured in two other ways: (1) As the sum of expenditures (consumer spending, investment, net exports, and government purchases) and (2) as the sum of costs incurred (compensation of employees, net interest, indirect business taxes, etc.) and profits earned in production. In practice, BEA implements these three ways using less than perfectly consistent source data, and the resulting totals are not the same. For current-dollar estimates, BEA measures what it refers to as "GNP" as the sum of the expenditure components, and it uses "charges against GNP" to refer to the sum of costs incurred and profits earned. The difference between GNP and charges against GNP is the statistical discrepancy. In constant dollars also, GNP is measured as the sum of the expenditure components. Constant-dollar charges against GNP is not estimated, however, because price indexes for deflation cannot be associated with income measures as they can be with the goods and services that make up the expenditure measures.(3)

GPO estimates in current dollars are prepared as distributions by industry of the components of charges against GNP. Thus, the sum of the current-dollar GPO estimates also differs from current-dollar GNP by the statistical discrepancy. To prepare constant-dollar GPO estimates, a different strategy--described in the next section--is used. The sum of the constant-dollar GPO estimates differs from constant-dollar GNP by the constant-dollar statistical discrepancy plus an additional discrepancy, titled the "residual."

Estimating methods for constant-dollar GPO

The strategy for estimating constant-dollar GPO by industry is to work with the measures of goods and services--gross output and intermediate inputs--that are used to define GPO. The industry estimates can then be derived using one of three methods: Double deflation, extrapolation, or direct deflation. The method chosen usually depends on the availability of source data.

In the double-deflation method, constant-dollar GPO is derived as the difference between constant-dollar gross output and constant-dollar intermediate inputs. In practice, there are at least three variants of this method, and the variant chosen depends on the availability of source data.

(1) When complete and consistent

gross output and inputs series are

available, these series can be

deflated, and constant-dollar GPO

can be measured as the difference

between them. (2) When only complete and

consistent gross output and GPO

series are available, a

current-dollar input series can be derived

and then deflated for use in the

output-minus-input calculation. (3) When only consistent but

incomplete gross output and inputs

series are available, the available

series can be used to derive a GPO

deflator to apply to the appropriate

current-dollar GPO estimate. (See the "Technical Note" for more information.)

In the extrapolation method, constant-dollar GPO is derived by extrapolating the base-year value of GPO (for which the current-dollar value equals the constant-dollar value) by an indicator series. The indicator series usually is constant-dollar gross output, the number of employees, or the number of hours worked.

In the direct-deflation method, constant-dollar GPO is derived by deflating current-dollar GPO. The index used for deflation usually is gross output prices or earnings.

Generally, double deflation is the conceptually preferred method because it measures GPO in the same way that GPO is defined. Moreover, assuming the availability of appropriate source data, double deflation is preferred because it allows for changes over time in the relationships between gross output and inputs. The extrapolation method will yield the correct results only if the rates of change in constant-dollar gross output and inputs are the same. The direct-deflation method will yield the correct results if the deflators for both constant-dollar gross output and inputs are the same.

Double deflation is not the preferred method for the two industries--private households and general government--for which gross output and GPO are defined as employee compensation and for the industry--the rest of the world--for which GPO is defined as net flows of factor incomes. For private households and general government, the most appropriate method is extrapolation by an indicator of labor input that reflects changes in productivity. For the rest of the world, the most appropriate method is direct deflation using a composite deflator calculated from the GPO of the industry and country in which the income was earned.

Improvements in the revised estimates

The revised constant-dollar GPO series reflect both the wider use of double deflation and improvements in the source data underlying the industry estimates derived using double deflation. In the revised estimates, double deflation was used in 51 industries, which accounted for 87 percent (in 1987) of the total GPO for industries for which it is the preferred method, compared with 25 industries, which accounted for 43 percent, in the previously published estimates (table 1).(4) Further, the double-deflation method was applied more consistently in the revised estimates; the same variant of double deflation was used for 49 of the 51 industries.

For most of the 25 industries for which GPO had previously been estimated using the double-deflation method, the source data were improved in three major ways.(5) The estimates of gross output were improved by benchmarking them to information from BEA's 1977 and 1982 input-output (I-O) tables to provide a more complete and consistent framework for estimating gross output and input and by preparing the constant-dollar estimates in greater detail. The estimates of inputs were improved by using information on their composition from the I-O tables for 1977 and 1981-85 to replace the estimates for 1982, which had been used for all years. Finally, the prices of inputs were improved by introducing prices of imported goods and services and by introducing more appropriate prices for service inputs.

For 26 of the 36 industries for which GPO had previously been estimated using either extrapolation or direct deflation but for which double deflation is the preferred method, the double-deflation method was introduced. The introduction of double deflation was made possible by the development of new estimates of gross output, input composition, and input prices incorporating the source-data improvements referred to in the preceding paragraph.

Addressing the earlier criticisms

The criticisms of the accuracy of BEA's previously published constant-dollar GPO estimates focused on four major areas: The adjustments BEA made so that the GPO estimates would be more consistent with the constant-dollar GNP estimates, the use of BEA's price index for computers, the use of indexes that covered only domestic prices and that only roughly approximated service prices to estimate constant-dollar inputs, and the use of out-of-date information on the composition of inputs.(6) The last two areas are addressed by the improvements incorporated into the revised estimates; the first two are discussed briefly in the following paragraphs.

When BEA published revised 1947-84 estimates in 1985, it adjusted the constant-dollar GPO of manufacturing and nonmanufacturing industries for 1947-81. The adjustments were criticized as being too subjective.(7) BEA made the adjustments to reduce the residual, which it views as a measure of the error in the constant-dollar GPO estimates. This view is supported by the fact that the source data available to prepare these estimates are less accurate and comprehensive than the source data available to prepare constant-dollar GNP (from expenditure estimates). Most of the adjustments were applied to manufacturing. They lowered the 1947-72 estimates of manufacturing by about the same percentage each year and lowered the estimates for 1973-82 by successively smaller percentages. The effect of the adjustments was to lower the initial constant-dollar 1972 estimate by $60 billion and the 1977 estimate by $25 billion, thus raising the growth rate of GPO in manufacturing from 1972 to 1982. In the revised estimates, the residuals are relatively small, and adjustments were not deemed necessary. (See line 82 of table 2.) Nevertheless, BEA still views the residual as a measure of error in the constant-dollar GPO estimates and the adjustment of GPO estimates to reduce the residual, when necessary, as an appropriate procedure.

In 1985, BEA introduced a new price index for computers. The index declined at an average annual rate of 15 percent from 1977 to 1987, and its steep decline contributed to the growth of both constant-dollar GNP and manufacturing GPO. The criticism was that the index, which is used to prepare the constant-dollar estimates, is not consistent with the resource-cost concept of capital that BEA uses in measuring other types of capital goods.(8) As explained in an article in the Survey of Current Business, BEA does not agree with this criticism and will continue to use the index for measuring constant-dollar GNP and GPO.(9) In the criticism, it was noted that the use of the index made the estimates much more sensitive to the choice of the base year in calculating constant-dollar measures. BEA plans to introduce estimates with alternative base years in the next comprehensive revision of the national income and product accounts (NIPA's) to help deal with this problem.(10)

Further work to improve the estimates

The 1987 and 1988 current- and constant-dollar estimates presented in this article are not fully consistent with the presently published GNP estimates; they do not yet incorporate source data used in the July 1990 revisions of the NIPA's. Revised estimates for 1987 and 1988 and estimates for 1989 consistent with the published estimates will be published in the April 1991 Survey.

BEA will continue to address the shortcomings of the revised GPO estimates, and it plans to incorporate additional improvements into the estimates to be released following the next comprehensive NIPA revision. The further work will focus on improving procedures for deriving the composition of inputs for years not covered by I-O tables, on developing and incorporating improved service prices, and on extending the use of double deflation. Improved estimates will also be prepared for years before 1977.

Long-term improvement in the estimates of both GPO and GNP will require additional research apart from the GPO improvement work. Research on the prices of services is one area. The service prices incorporated into the revised GPO estimates improve on the previous ones because they cover a far wider range of services and because they represent prices of inputs rather than implicit deflators of GPO. However, the formidable statistical problems of measuring prices of many services are still present in the new estimates; only a substantial research effort over many years holds any promise of overcoming these statistical problems. Research on the definition of output of several industries--most notably, financial services--is another area in which some research is under way.

The Impact of the Revisions on Industry Growth Rates

Chart 4 summarizes the impact of the revisions by showing, for seven broad groups of industries, the average annual rate of growth of constant-dollar GPO from 1977 to 1987 before and after revision. The chart shows that the general pattern of high- and low-growth industries is the same both before and after the revision. Wholesale and retail trade and services show the highest constant-dollar growth rates, and government enterprises and the agriculture, mining, and construction group show the lowest. The average annual rate of growth of manufacturing industries, previously estimated at 2.7 percent, is now estimated at 2.5 percent.

Impact of changes in methodology.--One reason for the similarity of growth rates before and after the revision is that several of the changes in methodology had surprisingly little impact on most industries. For example, introducing import prices in the deflation of intermediate inputs raised manufacturing growth from 1977 to 1987 by an annual average rate of less than 0.1 percent. From 1979 to 1985, when the exchange value of the U.S. dollar rose, the effect of introducing import prices was to lower manufacturing growth by 0.2 percent per year; from 1977 to 1979, the effect was strongly in the opposite direction. Outside of manufacturing, the effects of import prices were small in all years.

The methodological change that provided for changes in the composition of intermediate inputs rather than for use of estimated 1982 composition for all years also had a minor impact on most industries. From 1977 to 1985 (the latest year in which the composition changes in the revised estimates), the impact on total manufacturing was to raise manufacturing growth by an average of 0.2 percent per year: Growth was raised before 1982 and lowered thereafter. The revision largely reflected the use of variable input composition and its impact on the inputs of the nonelectrical machinery industry; the computer portion of this industry has used increasing quantities of inputs accompanied by falling prices of inputs. The use of variable composition reduced the importance of falling input prices in the years before 1982 and therefore lowered the growth of inputs and raised the estimated growth of GPO. After 1982, these effects were in the opposite direction.

The methodological change that introduced improved prices for services also had a minor overall impact. The impact was minor because the new service prices, as did the previous ones, rise much more rapidly, than do the prices of goods. A comparison of two deflators illustrates the difference: The implicit deflator of gross output in the service industries that are double deflated rose at an average annual rate of 7.6 percent from 1977 to 1987, while the implicit deflator of gross output in the manufacturing industries rose at a rate of only 4.3 percent. Omitting the petroleum refining and nonelectrical machinery industries because prices in them are heavily influenced by special factors raises the manufacturing increase to 4.7 percent, still well below the increase in services.

The largest industry revisions.--The largest revisions in the broad industry groups shown in the chart are mainly due to the combined impact of two factors: The extension of double deflation to many additional industries and the rapid rise of service prices relative to goods prices. Other changes--including revisions in gross output, new price series, and revisions in current-dollar gross product--have sizable impacts on a few individual industries but not on the broad groups.

For wholesale and retail trade, which displays the largest upward revision in the chart, the replacement of extrapolation by double deflation introduces into the calculations input prices that, on average, increase rapidly because of the importance of intermediate services in these industries (inputs purchased by trade industries do not include the goods to be resold, because gross output is defined as the margin between sales and purchases of goods). The rapid increase in prices of inputs depresses the growth of constant-dollar inputs, which, in turn, increases the growth of constant-dollar GPO. To a considerable extent, the upward revision in the growth of transportation, communication, and public utilities reflects the same forces.

The downward revision in the services industries also reflects the increased use of double deflation and the behavior of service prices relative to goods prices, but these factors interact in a different way in the services industries. Here, the increased use of double deflation introduces intermediate input prices that tend to rise less than the price of gross output, because in the services industries--in contrast to the trade industries--gross output reflects services and inputs reflect both goods and services. Estimated constant-dollar inputs therefore tend to rise relatively rapidly, which, in turn, lowers the growth of GPO.

Revisions in manufacturing.--The small downward revision in the growth of manufacturing industries is entirely due to nonelectrical machinery. In part, it is the result of dropping an adjustment that had been made in 1977 to reduce the residual; as the previous section notes, the revised estimates have not been adjusted to reduce the residual. In addition, the revision in nonelectrical machinery is primarily the result of departing from the assumption of fixed input composition. This departure had the effect of increasing the estimated growth of constant-dollar inputs, which, in turn, lowered the growth of constant-dollar GPO.

Technical Note

BEA has incorporated major improvements in methodology--i.e., in estimating methods and source data--into the revised estimates of constant-dollar GPO. No changes in methodology have been made to the previously published current-dollar GPO estimates; the 1986 and 1987 estimates were revised only to reflect changes introduced in the July 1989 annual NIPA revision, and several estimates for earlier years were revised to eliminate industry-allocation errors uncovered during the review of the revised estimates.

The first section of this note reviews the methodology for the previously published estimates of constant-dollar GPO by industry. The second section presents the methodology for the revised estimates.

Methodology for the previously published estimates

In the previously published constant-dollar estimates of GPO by industry, double deflation was used for farms; construction; manufacturing industries except petroleum and coal products; railroad transportation; electric, gas, and sanitary services; and the nonfarm housing services part of the real estate industry. The estimates were prepared using one of three variants--the choice depending on the availability of source data.

(1) For farms and nonfarm housing services, where complete and consistent gross output and intermediate inputs series are available, constant-dollar GPO was measured as the difference between constant-dollar gross output and constant-dollar inputs.(11)

(2) For manufacturing industries except petroleum and coal products, where only a gross output series consistent with the current-dollar GPO series is available, the derived input variant was used. Current-dollar inputs were derived as the difference between current-dollar gross output and current-dollar GPO, and constant-dollar inputs were calculated by deflating the derived current-dollar series using a composite input deflator that reflected the estimated composition of inputs in 1982 (based on information from the 1977 input-output (I-O) table) and the price indexes for domestic inputs.(12) For goods inputs, either producer price indexes or other appropriate price indexes were used. For service inputs, GPO deflators for the
COPYRIGHT 1991 U.S. Government Printing Office
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 1991 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Leeuw, Frank de; Mohr, Michael; parker, Robert R.
Publication:Survey of Current Business
Date:Jan 1, 1991
Previous Article:National income and product accounts.
Next Article:Personal income continued to grow slowly in third quarter 1990.

Related Articles
Gross state product by industry, 1963-86.
Annual input-output accounts of the U.S. economy, 1985.
Annual input-output accounts of the U.S. economy, 1986.
Gross national product by industry, 1987-89.
Revised constant-dollar inventories, sales, and inventory-sales ratios for manufacturing and trade.
Gross product by industry, 1991-92.
Bias in aggregate productivity trends revisited: aggregate productivity trends were revised upward; it is now less clear whether bias exists, but...
Gross domestic product by industry for 2002.
Suggested reading. (Appendix B).
Annual industry accounts advance estimates for 2004.

Terms of use | Copyright © 2017 Farlex, Inc. | Feedback | For webmasters