Gross domestic income: revisions and source data.

Holdren, Alyssa E. ; Grimm, Bruce T.

GROSS domestic product (GDP) is the Bureau of Economic Analysis featured measure of the economy's output and is widely used by policymakers, academics, investment professionals, the media, and others to assess the condition of the U.S. economy. BEA also produces other measures, notably gross domestic income (GDI), that can also be used to assess the economy. Indeed, the Business Cycle Dating Committee of the National Bureau of Economic Research cited GDI in its recent determination that a business cycle peak occurred in December 2007. (1)

While GDP measures output as the sum of goods and services sold to final users plus the change in private inventories, GDI calculates the value of output as the sum of income payments and other costs incurred in the production of goods and services. In principle, GDI should equal GDP because they are measuring the same output. In practice, they differ because they are estimated using different source data.

GDI is also a major component in the calculation of gross national income (GNI), which is the most comprehensive measure of the nation's income and is calculated as the sum of GDI and net income receipts from the rest of the world. In other words, GNI measures all income from current production accruing to U.S. residents regardless of where the production occurs. In contrast, GDI measures all income generated from domestic production, whether accruing to U.S. or foreign residents.

In the February 2006 SURVEY OF CURRENT BUSINESS, BEA published an article that discussed the source data used to calculate the earliest quarterly GDP statistics and the subsequent annually revised statistics. (2) This article presents a similar analysis of GDI statistics.

BEA prepares quarterly and annual estimates of GDI. The "current quarterly" estimates of GDI are published in a series of releases. Table 1 shows the release dates for the full set of estimates for the third quarter of 2005. Except for the fourth-quarter estimates, the initial estimates of GDI are released with the "preliminary" GDP estimates near the end of the second month after the quarter ends. Revised estimates are released with the "final" GDP estimates, near the end of the third month after the quarter ends. (3) For the fourth quarter of each year, the estimates of GDI and some related series are released only with the final GDP estimates.

In addition, when the preliminary estimates of GDP for the most recent quarter are released, BEA releases revised estimates for the previous quarter for two components of GDI--private wages and salaries and employer contributions for government social insurance. These revised estimates incorporate the most recently available wage and salary data from the Bureau of Labor Statistics (BLS) Quarterly Census of Employment and Wages (QCEW). (4) The data from the QCEW are more comprehensive than the monthly data on employment, hours, and earnings from the BLS Current Employment Statistics survey that are used to derive the earlier estimates. The QCEW data include the pay of supervisors and irregular pay, such as bonuses and gains from the exercise of stock options.

BEA also releases three "annual revision" estimates of GDI and its components. These statistics provide updated quarterly estimates for the most recent calendar year and the preceding 2 years. The first annual revision estimates for a given year are usually released in July of the following year. (5) The first annual revision estimates for 2007, for example, were released in July 2008. The second and third annual revision estimates follow in successive years. The estimates released in July 2008 also included second annual revision estimates for 2006 and third annual revision estimates for 2005.

After the third annual revision, the GDI statistics--both quarterly and annual--are generally not revised until the next comprehensive revision. Comprehensive revisions occur about every 5 years. (6)

Source data types

To estimate GDI and related measures in the national income and product accounts (NIPAs), BEA relies on a wide range of source data, including measures of employment and wages, pension contributions and employer-provided health insurance data, information from tax returns, financial statement data for private corporations, and estimates of government outlays.

Some of the source data are used as indicator series to interpolate of extrapolate quarterly of annual estimates. Generally, monthly of quarterly source data are not as comprehensive as annual source data. Thus, for some estimates, the more frequent but less comprehensive source data may be used as an indicator of the movements of the component series rather than as a measure of the absolute levels of the series. Specifically, for the periods for which annual estimates are available and the quarterly estimates must be forced to average these totals, the quarterly pattern is estimated by interpolation. For periods not yet covered by annual estimates (such as the current quarter), the quarterly estimates are made by extrapolation. (7) For example, the tax return data used to prepare the second annual revision estimates of corporate profits are only available annually, so quarterly estimates are interpolated using data from Census Bureau quarterly financial reports, regulatory agency reports, and publicly available corporate financial statements. In other cases, interpolation and extrapolation are based on trends.

For many components of GDI, estimates are derived from "value data,' which already reflect quantity and price information. For example, the QCEW wage and salary data used to prepare revised estimates of private wages and salaries reflect both the number of people employed (quantity) and the wages paid to the employees (price).

For estimates that are not derived flora value data, BEA may combine separate quantity and price data. (8) For example, to calculate the initial current quarterly estimates of private wages and salaries, for each industry, BEA simply takes the product of employment times average hourly earnings times average hours worked.

In other cases, BEA uses variations of a "stock of assets/liabilities times an effective rate of interest" method. For example, the current quarterly estimates of net domestic monetary interest are partly derived by multiplying a stock of interest-bearing assets by an effective rate of interest.

Source data categories and successive estimates

In this article, BEA categorizes the source data that are used to prepare the GDI estimates according to quality, availability, and use. The four categories areas follows:

Comprehensive data. These data provide comprehensive coverage of the relevant population and are consistent, or can be adjusted to be consistent, with national accounts concepts. These data include census data and survey data with comprehensive coverage as well as tax or regulatory data that have been adjusted to correct for misreporting, differences in coverage, and differences in accounting concepts. For example, the annual revision estimates of government wages and salaries reflect the incorporation of QCEW wage and salary data. These data are more comprehensive than the monthly data that are used to extrapolate the earlier estimates.

Direct indicator data. These source data are used to prepare indicator series. While these data are closely related to the comprehensive data that are ultimately incorporated into the estimates, they may be less detailed or less comprehensive initially. For example, the current quarterly and first annual revision estimates of corporate profits are extrapolated using data on corporate profits from Census Bureau quarterly financial reports, flora regulatory agency reports, and from publicly available corporate financial statements. These data are less comprehensive than the data from corporate tax returns that are used to prepare the second annual revision estimates.

Indirect indicator data. These source data are used to prepare indicator series; they include two main types: (1) volume or activity indicators; for example the current quarterly estimates of private wages and salaries are extrapolated using BLS data on employment, hours, and earnings. And (2) other NIPA estimates; for example, the current quarterly estimates of employer contributions for government social insurance are extrapolated using NIPA estimates of wages and salaries as indicator series.

Trend-based data. These data are typically calculated flora previous estimates and trends, using judgment by BEA economists. For example, the preliminary quarterly estimates of net domestic imputed interest paid by commercial banks are judgmentally extrapolated using historical trends in the series.

Table 2 and chart 1 summarize the shares of the GDI estimates for successive vintages that incorporate the four types of source data. (9) The preliminary and final current quarterly estimates are based on data from all four categories, and the largest share of the estimates (about 58 percent) is derived using indirect indicator data. (The estimates of private wages and salaries account for nearly two-thirds of the total share of the estimates that incorporates indirect indicator data.)

For the preliminary quarterly estimates, about 12 percent of the estimates are based on source data used to prepare direct indicators, and about 28 percent of the estimates are trend based. For the final quarterly estimates, about 14 percent of the estimates are derived using direct indicator data, and about 26 percent of the estimates are trend based. The 2 percent shift from trend-based estimates to estimates based on direct indicator data primarily reflects the incorporation of data from the Federal Reserve Board on commercial banking activity, which are used to derive estimates of imputed interest; these data are not usually available until the final quarterly estimates.

As noted above, estimates of private wages and salaries, which reflect the incorporation of QCEW wage and salary data, are released with the preliminary GDP estimates for the most recent quarter. These data are more comprehensive than the monthly employment, hours, and earnings data that are used to construct indirect indicators to derive the earlier estimates. When the QCEW data are included, the percent of the estimates that incorporate comprehensive data increases from about 2 percent for the preliminary and final estimates to about 39 percent.

The first and second annual revision estimates of GDI include revisions to the quarterly estimates for the previous 3 years in addition to new annual estimates. By the time the first and second annual revision estimates are calculated and released, much of the source data used to calculate the earlier quarterly estimates have been replaced with more comprehensive or more reliable data.

Many of the quarterly estimates in the first and second annual revisions are based on source data that cover a full year (annual data) and are either interpolated or extrapolated using monthly or quarterly source data. (10) The monthly or quarterly data are often the same as those used to calculate the current quarterly estimates, with revisions, if necessary, to make them consistent with the annual data and with updated seasonal adjustments.

About 52 percent of the first annual revision estimates incorporate comprehensive data. About 18 percent of the estimates are derived from direct indicator data, about 9 percent are based on indirect indicator data, and about 22 percent of the estimates are trend base&

By the second annual revision, more than 99 percent of the estimates incorporate comprehensive data. The increase in the share of the estimates that incorporate comprehensive data from the first to the second annual revision primarily reflects the incorporation of tax return data compiled by the Internal Revenue Ser vice (IRS). These data are used to prepare several estimates, including corporate profits, nonfarm proprietors' income, net domestic monetary interest, and capital consumption allowance. Before the second annual revision, these estimates are extrapolated using director indirect indicators or are trend based, as noted above.

The third annual revision estimates incorporate more newly available or revised annual source data as well as revised monthly and quarterly data. In addition, BEA releases a comprehensive revision about every 5 years. (11) Comprehensive revisions incorporate

even more detailed source data from various economic censuses. In years when comprehensive revisions are released, no annual revision is released.

GDI components

The estimates of private wages and salaries, net domestic imputed interest, and nonfarm proprietors' income illustrate how source data change from the preliminary estimates to the second annual revision estimates (table 3).

The preliminary and final quarterly estimates of private wages and salaries are extrapolated using an indirect indicator constructed with monthly data on employment, hours, and earnings. The revised estimates for the previous quarter (released with the preliminary GDP estimates for the most recent quarter) and the first and second annual revision estimates incorporate QCEW quarterly data. (12)

The preliminary quarterly estimates of net domestic imputed interest paid by commercial banks are extrapolated using historical trends in the series. (13) The final quarterly estimates are extrapolated using a direct indicator that incorporates quarterly data from commercial bank income statements and balance sheets and are not affected by the incorporation of QCEW data. The first and second annual revision estimates incorporate quarterly commercial banking data that are more detailed than the data used to compile the final quarterly estimates.

The preliminary and final quarterly estimates and the first annual revision estimates of nonfarm proprietors' income are extrapolated using other NIPA estimates and several volume and activity indicators. For example, the income of proprietors engaged in manufacturing is extrapolated using the value of manufacturers' shipments. Proprietors' income for the real estate industry is extrapolated using a NIPA estimate of real estate brokers' commissions. Estimates of nonfarm proprietors' income for the previous quarter are not revised when QCEW data are incorporated. The second annual revision estimates incorporate newly available data on sole proprietorship and partnership tax returns compiled by the IRS. These data are more reliable than the data used to extrapolate the earlier estimates. However, the tax return data are only available annually; therefore, for the second annual revision, the quarterly estimates are interpolated using the same indirect indicator data that are used to prepare the earlier estimates.

Current quarterly and annual revision comparison

The current quarterly estimates receive the most attention from policymakers and business analysts because they are the first estimates published for a quarter. Except for the fourth quarter of each year, the preliminary estimates are the first available estimates of GDI. For the fourth quarter of every year, the final estimates are the first available. From 1983 to 2007, the current quarterly estimates of GDI growth in current dollars ranged from 0.0 percent to 14.2 percent.

To assess the magnitude of revisions to current-dollar GDI growth estimates, the two current quarterly estimates were compared, and the current quarterly estimates were compared with the first and second annual revision estimates. (14) These comparisons allow for ah assessment of the impact of source data changes because the first and second annual revision estimates incorporate higher quality source data, mainly newly available or revised annual data and revised monthly and quarterly data. For example, the second annual revision estimates of corporate profits, nonfarm proprietors' income, and net domestic monetary interest incorporate newly available tax return data from the IRS that are more comprehensive than the data used to derive the earlier estimates. In addition, the effects of revisions due to changes in definitions or methodologies affect some of the annual revision estimates.

Overall, the magnitude of the revisions to the growth rates of GDI is small and roughly the same as the magnitude of the revisions to GDW. (15) Furthermore, the revisions from the current quarterly GDI estimates to the latest estimates, which are considered to be the most accurate, are similar to the revisions from the annual revision estimates to the latest estimates (table 4).

Mean revisions. Although the annual revision estimates incorporate more comprehensive source data than the current quarterly estimates, the average revisions from the current quarterly estimates to the annual revision estimates are small and negative. The mean revisions from the preliminary and final estimates to the first annual revision estimates are both -0.03 percentage point. The revisions from the three earlier vintages of estimates to the second annual revision estimates range from -0.05 percentage point to -0.07 percentage point.

The revisions from the current quarterly estimates and the two annual revision estimates to the latest estimates are positive, ranging from 0.03 percentage point to 0.15 percentage point. This shift from negative to positive average revisions reflects the effects of comprehensive revisions that are fully incorporated in the latest estimates.

Mean absolute revisions (MARs). The mean revision without regard to sign from the preliminary estimates to the final estimates of GDI is 0.29 percentage point. The MAR from the preliminary GDI quarterly estimates to the first annual revision estimates is 0.98 percentage point, and the MAR from the final quarterly estimates to the first annual revision estimates is 0.88 percentage point. The MARs from the two current quarterly estimates to the second annual revision estimates are nearly identical--1.07 percentage points and 1.08 percentage points, respectively. In contrast, the MAR from the first annual revision estimates to the second annual revision estimates is 0.80 percentage point. The MARs to the latest estimates decline steadily across successive vintages, from 1.21 percentage points for the preliminary estimates to 0.90 percentage point for the second annual revision estimates.

Karla L. Stanley-Allen prepared this article.

(1.) See National Bureau of Economic Research, Business Cycle Dating Committee, "Determination of the December 2007 Peak in Economic Activity" at www.nber.org/cycles/dec2008.html.

(2.) See Bruce T. Grimm and Teresa L. Weadock, "Gross Domestic Product: Revisions and Source Data," SURVEY 86 (February 2006): 11 15.

(3.) Unlike GDP, ah "advance" estimate of GDI is not released near the end of the first month after the quarter ends because data on domestic profits and data on foreign, or rest-of-world, payments and receipts of income used to derive estimates of net interest of domestic industries are not available. For the fourth quarter, these data are not available until the "final" estimate.

(4.) Revised estimates for the previous quarter, which reflect the incorporation of QCEW data, were first published for the first quarter of 2002. Other series that are revised because of the incorporation of data from the QCEW include the statistical discrepancy, GNI, national income, personal income, disposable personal income, personal saving, gross (national) saving, compensation of employees, personal current taxes, and gross value added of corporate business.

(5.) Before the release of the first annual revision estimates, BEA releases early annual estimates of GDI and related series. For most of the components of GDI, early annual estimates are released with the preliminary GDP estimates for the fourth quarter, usually in February of the following year. Early annual estimates of GDI, corporate profits, and the series that are affected by the release of data on foreign payments and receipts of income are released with the final GDP estimates for the fourth quarter, usually in March of the following year. Early annual estimates of GDI, private wages and salaries, and employer contributions for government social insurance are revised to reflect the incorporation of fourth-quarter QCEW data, usually in May of the following year.

(6.) After the release of the comprehensive revision in July 2009, BEA will move to "flexible annual revisions" that will allow BEA to incorporate changes on a flow basis rather than delaying incorporation for up to 5 years. For more information, see "Taking Account" SURVEY 87 (July 2007): iv.

(7.) For more information on the use of interpolation and extrapolation to prepare NIPA estimates, see Concepts and Methods of the U.S. National Income and Product Accounts (Introductory Chapters 1 4) (July 2008): 4-6-4-7, www.bea.gov/national/pdf/NIPAhandbookch1-4-pdf.

(8.) For more details, see "Updated Summary of NIPA Methodologies," SURVEY 88 (November 2008): 8-25.

(9.) The third annual revision estimates are not included in table 2, because the shares of the estimates that incorporate each type of source data do not change from the second annual revision estimates to the third annual revision estimates. However, the third annual revision estimates incorporate revised source data that are more reliable than the data used to derive the earlier estimates.

(10.) In the interpolation process, the annual estimates are interpolated using quarterly of monthly source data that generally retain patterns of the annual source data; the quarterly estimates are interpolated in ways that retain the annual totals.

(11.) As mentioned previously, after the release of the comprehensive-revision in July 2009, BEA will move to "flexible annual revisions."

(12.) BEA adjusts data from the QCEW to conform to NIPA concepts and definitions and to fill gaps in coverage. Adjustments are also made to account for nonreporting and under reporting of wages by employers and to ensure that the proper geographic coverage is maintained.

(13.) For more information on the definition and estimation of imputed interest, see Dennis J. Fixler, Marshall B. Reinsdorf, and George M. Smith, "Measuring the Services of Commercial Banks in the NIPAs: Changes in Concepts and Methods," SURVEY 83 (September 2003): 33-44.

(14.) The comparisons are not quite complete, because there are no preliminary estimates of fourth-quarter GDI after 1994. Because the revised estimates for the previous quarter, which reflect the incorporation of QCEW data, only begin in 2002, they are excluded from the comparisons.

(15.) See Dennis J. Fixler and Bruce T. Grimm, "The Reliability of the GDP and GDI Estimates," SURVEY 88 (February 2008): 16-32.

Table 1. Release Schedule for the Estimates of Gross
Domestic Income for the Third Quarter of 2005

Estimate

Preliminary estimate November 2005
Final estimate December 2005
Revised previous quarter estimate February 2006
First annual revision estimate July 2006
Second annual revision estimate July 2007
Third annual revision estimate July 2008

Table 2. Shares of Source Data for the Successive
Gross Domestic Income Estimates

[Percent]

 Pre- Revised
Sources liminary Final previous
 estimates estimates quarter
 estimates

Comprehensive data 2.1 2.1 39.3
Direct indicator data 12.4 14.1 14.1
Indirect indicator data 57.8 57.8 20.6
Trend-based data 27.7 26.0 26.0

 First Second
Sources annual annual
 revision revision
 estimates estimates

Comprehensive data 51.7 99.4
Direct indicator data 17.7 0.5
Indirect indicator data 8.6 0.0
Trend-based data 21.9 ...

NOTE. Shares were calculated using annual data for 2005; quarterly
data are not available for a number of the detailed components of GDI.
Shares may not sum to 100 because of rounding.

Table 3. Sources for the Successive Estimates of Quarterly Gross
Domestic Income

 Preliminary Final
 Component estimates estimates

Compensation of employees,
 paid
 Wage and salary accruals
 Disbursements
 Private domestic
 industries Indirect indicator Indirect indicator
 data data
 Government Indirect indicator Indirect indicator
 data data
 Wage accruals less
 disbursements Indirect indicator Indirect indicator
 data data
 Supplements to wages and
 salaries
 Employer contributions
 for employee pension
 and insurance funds
 Pension and profit
 sharing
 Private pension and
 profit sharing Trend-based data Trend-based data
 Publicly
 administered
 government
 employee
 retirement plans
 Federal Comprehensive data Comprehensive data
 State and local Trend-based data Trend-based data
 Private insurance
 funds Trend-based data Trend-based data
 Employer contributions
 for government social
 insurance Indirect indicator Indirect indicator
 data data

Taxes on production and
 imports
 Federal Comprehensive data Comprehensive data
 State and Local
 State sales taxes
 General Direct indicator Direct indicator
 data data
 Other Trend-based data Direct indicator
 data
 Other Trend-based data Trend-based data

Less: Subsidies
 Federal Comprehensive data Comprehensive data
 State and local Trend-based data Trend-based data

Net operating surplus
 Private enterprises
 Net interest and
 miscellaneous
 payments, domestic
 industries
 Net interest
 Domestic monetary
 interest, net Indirect indicator Indirect indicator
 Domestic imputed data data
 interest, net
 Commercial banks Trend-based data Direct indicator
 data
 Other Trend-based data Trend-based data

 Federal rents and
 royalties Comprehensive data Comprehensive data
 State and local rents
 and royalties Trend-based data Trend-based data
 Business current
 transfer
 payments, net Trend-based data Trend-based data
 Proprietors' income
 with IVA
 Farm Indirect indicator Indirect indicator
 data data
 Nonfarm Indirect indicator Indirect indicator
 data data
 Rental income of
 persons Indirect indicator Indirect indicator
 data data
 Corporate profits
 with IVA,
 domestic
 industries Direct indicator Direct indicator
 data data
 Current surplus of
 government
 enterprises
 Federal Trend-based data Trend-based data
 State and local Trend-based data Trend-based data
Capital consumption
 adjustments plus
 consumption of fixed
 capital (1) Trend-based data Trend-based data

 Revised previous First annual
 Component quarter estimates revision estimates

Compensation of employees,
 paid
 Wage and salary accruals
 Disbursements
 Private domestic
 industries Comprehensive data Comprehensive data

 Government Indirect indicator Comprehensive data
 data
 Wage accruals less
 disbursements Indirect indicator Indirect indicator
 data data
 Supplements to wages and
 salaries
 Employer contributions
 for employee pension
 and insurance funds
 Pension and profit
 sharing
 Private pension and
 profit sharing Trend-based data Direct indicator
 Publicly data
 administered
 government
 employee
 retirement plans
 Federal Comprehensive data Comprehensive data
 State and local Trend-based data Trend-based data
 Private insurance
 funds Trend-based data Trend-based data
 Employer contributions
 for government social
 insurance Indirect indicator Comprehensive data
 data

Taxes on production and
 imports
 Federal Comprehensive data Comprehensive data
 State and Local
 State sales taxes
 General Direct indicator Direct indicator
 data data
 Other Direct indicator Direct indicator
 data data
 Other Trend-based data Trend-based data

Less: Subsidies
 Federal Comprehensive data Comprehensive data
 State and local Trend-based data Trend-based data

Net operating surplus
 Private enterprises
 Net interest and
 miscellaneous
 payments, domestic
 industries
 Net interest
 Domestic monetary
 interest, net Indirect indicator Indirect indicator
 Domestic imputed data data
 interest, net
 Commercial banks Direct indicator Comprehensive data
 data
 Other Trend based data Direct indicator
 data
 Federal rents and
 royalties Comprehensive data Comprehensive data
 State and local rents
 and royalties Trend-based data Trend-based data
 Business current
 transfer
 payments, net Trend-based data Trend-based data
 Proprietors' income
 with IVA
 Farm Indirect indicator Comprehensive data
 data
 Nonfarm Indirect indicator Indirect indicator
 data data
 Rental income of
 persons Indirect indicator Direct indicator
 data data
 Corporate profits
 with IVA,
 domestic
 industries Direct indicator Direct indicator
 data data
 Current surplus of
 government
 enterprises
 Federal Trend-based data Comprehensive data
 State and local Trend-based data Trend-based data
Capital consumption
 adjustments plus
 consumption of fixed
 capital (1) Trend-based data Trend-based data

 Second annual
 Component revision estimates

Compensation of employees,
 paid
 Wage and salary accruals
 Disbursements
 Private domestic
 industries Comprehensive data

 Government Comprehensive data

 Wage accruals less
 disbursements Indirect indicator
 data
 Supplements to wages and
 salaries
 Employer contributions
 for employee pension
 and insurance funds
 Pension and profit
 sharing
 Private pension and
 profit sharing Comprehensive data
 Publicly
 administered
 government
 employee
 retirement plans
 Federal Comprehensive data
 State and local Comprehensive data
 Private insurance
 funds Comprehensive data
 Employer contributions
 for government social
 insurance Comprehensive data
Taxes on production and
 imports
 Federal Comprehensive data
 State and Local
 State sales taxes
 General Comprehensive data

 Other Comprehensive data

 Other Comprehensive data

Less: Subsidies
 Federal Comprehensive data
 State and local Comprehensive data

Net operating surplus
 Private enterprises
 Net interest and
 miscellaneous
 payments, domestic
 industries
 Net interest
 Domestic monetary
 interest, net Comprehensive data
 Domestic imputed
 interest, net
 Commercial banks Comprehensive data

 Other Comprehensive data

 Federal rents and
 royalties Comprehensive data
 State and local rents
 and royalties Comprehensive data
 Business current
 transfer
 payments, net Comprehensive data
 Proprietors' income
 with IVA
 Farm Comprehensive data

 Nonfarm Comprehensive data

 Rental income of
 persons Direct indicator data

 Corporate profits
 with IVA,
 domestic
 industries Comprehensive data

 Current surplus of
 government
 enterprises
 Federal Comprehensive data
 State and local Comprehensive data
Capital consumption
 adjustments plus
 consumption of fixed
 capital (1) Comprehensive data

(1.) The total of capital consumption adjustments and consumption of
fixed capital is based on capital consumption allowances which are
estimated from tax data.

NOTE. For more information on the source data and estimating methods
used to prepare national income and product account estimates, see
Concepts and Methods of the National Income and Product Accounts
(Introductory Chapters 1-4) (July 2008); www.bea.gov/national/pdf
NIPAhandbookchl-4.pdf.

IVA Inventory valuation adjustment

Table 4. Mean and Mean Absolute Revisions to the Successive
Estimates of Gross Domestic Income for 1983-2007

[Percentage points]

 First Second
 Final annual annual Latest
 estimates revision revision estimates
 estimates estimates

Mean revisions
 Preliminary
 estimates 0.01 -0.03 -0.05 0.03
 Final estimates ... -0.03 -0.07 0.05
 First annual
 revision
 estimates ... ... -0.06 0.08
 Second annual
 revision
 estimates ... ... ... 0.15
Mean absolute revisions
 Preliminary
 estimates 0.29 0.98 1.07 1.21
 Final estimates ... 0.88 1.08 1.16
 First annual
 revision
 estimates ... ... 0.80 1.07
 Second annual
 revision
 estimates ... ... ... 0.90

Chart 1. Shares of Source Data for the Quarterly GDI Estimates

PRELIMINARY

Trend-based data, 27.7%
Indirect indicator data, 57.8%
Comprehensive data, 2.1%
Direct indicator data. 12.4%
Indirect indicator data, 57.8%

FINAL

Trend-based data, 26.0%
Comprehensive data, 2.1%
Direct indicator data, 14.1%

REVISED PREVIOUS QUARTER

Trend-based data, 26.0%
Indirect indicator data, 20.6%
Comprehensive data, 39.3%
Direct indicator data, 14.1%

FIRST ANNUAL REVISION

Trend-based data, 21.9%
Comprehensive data, 51.7%
Indirect indicator data, 8.6%
Direct indicator data, 17.7%

SECOND ANNUAL REVISION

Comprehensive data, 99.4%
Direct indicator data, 0.5%

NOTE. Shares may not sum to 100 because of rounding.

U.S. Bureau of Economic Analysis

Note: Table made from bar graph.