Improved Annual Industry Accounts for 1998-2003: integrated annual input-output accounts and gross-domestic-product-by-industry accounts

Survey of Current Business, June, 2004 by Brian C. Moyer, Mark A. Planting, Paul V. Kern, Abigail M. Kish

ON June 17, 2004, the Bureau of Economic Analysis (BEA) released the initial results of the comprehensive revision of the annual industry accounts. This release marks a major advance in the timeliness, accuracy, and consistency of the annual input-output (I-O) accounts and the gross-domestic-product-(GDP)-by-industry accounts as a result of significant improvements in BEA's estimating methods. The new methodology combines source data between the two accounts to improve accuracy; it prepares the newly integrated accounts within an I-O framework that balances and reconciles industry production with commodity usage. For the first time, the annual I-O accounts and the GDP-by-industry accounts are released concurrently--accelerating the release of the annual I-O accounts by 18 months--and they present fully consistent measures of gross output, intermediate inputs, and value added by industry for 1998-2002. In addition, this release includes GDP-by-industry estimates for 2003 that are based on BEA's recently developed, accelerated methodology. (See the box "Accelerated Gross-Domestic-Product-by-Industry Estimates for 2003.")

The methodology used to integrate the annual I-O accounts and the GDP-by-industry accounts draws the best information from the rich sources of industry-based data that are available in the United States by ranking these source data by quality on an industry-by-industry basis. (1) As a result, the accuracy of both sets of accounts is improved. For example, for the industries for which the Bureau of the Census provides relatively complete data for intermediate inputs, value added is computed as the difference between gross output and intermediate inputs. For the industries for which less data on intermediate inputs are available, BEA estimates value added using alternative data sources, including data on wages and salaries from the Bureau of Labor Statistics and data on corporate profits before tax from the Internal Revenue Service.

The integration methodology then combines the estimates of gross output by industry, intermediate inputs by industry, value added by industry, and the final uses of commodities in a balanced I-O framework. The estimates of intermediate inputs by industry are computed as the difference between gross output by industry and value added by industry, which are then adjusted through an iterative row-and-column balancing process to ensure that the use of commodities equals their supply, that the sum of value added and intermediate inputs by industry equals gross output by industry, and that the sum of final uses equals GDP.

The integration methodology results in industry accounts that provide a more accurate and consistent picture of the U.S. economy than the previously published accounts. For example, the balanced I-O framework ensures that the estimate of gross output for the construction industry is consistent with both the estimate of investment in structures in the national income and product accounts (NIPAs) and the purchases of construction services by all industries. In the GDP-by-industry accounts, the previously published estimates of gross output for this industry did not reflect the NIPA estimates of investment in structures.

Using this methodology also reduces the portion of real GDP growth that is "not allocated by industry." Because the new methodology imposes greater consistencies, the differences between economy-wide real value-added growth in the GDP-by-industry accounts and real GDP growth in the NIPAs are reduced. For example, before integration, the growth rate in economy-wide real value added was 4.7 percent in 1998-99, and real GDP growth was 4.5 percent, so the "not allocated by industry" was 0.2 percentage point. After the integration, the growth rate in economy-wide real value added was revised down to 4.5 percent, the same growth rate as real GDP. Similarly, for 1999-2000, integration reduced the economy-wide real value-added growth rate from 3.8 percent to 3.7 percent, the same as the growth rate in real GDP. Integration does not eliminate these differences for all years, but a comparison of published estimates from the GDP-by-industry accounts for 1998-2001 with the estimates in the integrated accounts indicates that on average, the "not allocated by industry" was reduced 0.2 percentage point.

The integration of the annual I-O accounts and the GDP-by-industry accounts is the most recent in a series of improvements to the industry accounts. It marks an important step towards BEA's longer term goal of the integration of all the industry accounts, including the benchmark I-O accounts, and the integration of the industry accounts with the NIPAs. (2) In the future, as a result of the integration with the NIPAs, a feedback loop from the annual I-O accounts to the NIPAs could improve the commodity composition of GDP, and it could further improve the accuracy of the industry accounts and the NIPAs. This integration will also incorporate expanded data on intermediate inputs by industry from the 2002 Economic Census and from the annual surveys that are currently being collected and tabulated by the Bureau of the Census.