Gross domestic product and gross domestic income: revisions and source data.

Holdren, Alyssa E.

THE NATIONAL income and product accounts (NIPAs) produced by the Bureau of Economic Analysis (BEA) provide a timely, comprehensive, and reliable picture of the U.S. economy. The featured measures-gross domestic product (GDP) and gross domestic income (GDI)-provide two ways to measure the value of U.S. output. GDP measures output as the value of goods and services sold to final users plus the change in private inventories, while GDI captures the value of the same output as the sum of income payments and other costs incurred in the production of those goods and services. In principle, GDP should equal GDI; however, they differ in practice because each is estimated using different, and largely independent, source data.

This article analyzes the source data used to prepare the GDP estimates and the GDI estimates according to a set of criteria that reflects the quality, the availability, and the use of the data. This analysis allows for a better understanding of the differences between the source data that underlie the GDP estimates and those that underlie the GDI estimates and illustrates how the incorporation of the increasingly detailed and comprehensive source data leads to revisions to the GDP estimates and to the GDI estimates. (1)

The revision cycle

BEA produces quarterly and annual estimates of GDP and GDI in a series of vintages-current quarterly estimates, annual revision estimates, and comprehensive revision estimates. BEA revises its estimates in successive vintages in order to improve the accuracy of the estimates and to provide a more detailed picture of the economy by incorporating the most complete and reliable source data available at a given time. (2) Revisions are not errors. Revisions are mainly driven by the incorporation of more complete and revised source data, and the release of the revised estimates is determined by the availability of these data. Source data may take the form of indicators that have been benchmarked to more comprehensive and detailed annual surveys, or in some cases, they may represent comprehensive source data that take the place of judgemental trends or indicators used to prepare the previously published estimates. Table 1 shows the release dates for the full set of estimates for the third quarter of 2009.

Current quarterly estimates provide an "early read" of economic activity and are produced using a variety of source data, including preliminary survey results, various indicators, and trade source data. Annual revision estimates incorporate newly available and revised source data that are more complete, more detailed, and more reliable than those incorporated for the current quarterly estimates, such as data from mandatory annual surveys and from administrative records. Annual revision estimates may also reflect the introduction of methodological improvements. Comprehensive revision estimates often reflect the introduction of conceptual and major methodological changes and benchmarking to the BEA benchmark input-output accounts in addition to the incorporation of newly available and revised source data, such as data from Census Bureau quinquennial economic censuses.

Current quarterly estimates

The current quarterly estimates of GDP are published in a series of releases-advance, second, and third--approximately 30, 60, and 90 days, respectively, after the end of the reference quarter. Each current quarterly estimate for a quarter incorporates increasingly comprehensive and improved source data. Unlike GDP, an advance current quarterly estimate of GDI is not released. (3)

At the time that the second 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)

Annual revision estimates

BEA releases at least three annual revision estimates of GDP, GDI, and their components. Generally, annual revisions provide updated quarterly and annual estimates for the most recent calendar year and for at least the preceding 2 years. (5) The first annual revision estimates for a given year are usually released in July of the following year. The second and third annual revision estimates follow in successive years. For example, the first annual revision estimates for 2009 were released in July 2010. The estimates released in July 2010 also included second annual revision estimates for 2008 and third annual revision estimates for 2007. Beginning with the 2010 annual revision, BEA adopted a flexible annual revision policy that allows for the incorporation of methodological changes to the accounts on a more timely basis and for a longer time span than the typical 3 years, rather than delaying the incorporation of these changes until the next comprehensive revision. (6)

Comprehensive revision estimates

Comprehensive revisions are conducted about every 5 years, following the release of data from the Census Bureau quinquennial economic censuses. They encompass a wider range of improvements than most annual revisions-including major changes in definitions, in classifications, and in presentations-and may result in revisions that extend as far back as 1929, the first year for which annual estimates are published. The most important source of information for comprehensive NIPA revisions is the BEA benchmark input-output accounts, which, in turn, reflect the incorporation of data from the economic censuses. The input-output accounts are used to establish the level of GDP for the census year and provide critical information for estimating GDP for periods after the census year. For most components of GDI, no such benchmarking process exists.

Source Data Categories

This section categorizes the source data that are used to prepare the GDP and GDI estimates according to quality, availability, and use into five categories: comprehensive data, adjusted comprehensive data, direct indicator data, indirect indicator data, and trend-based data.

Comprehensive data

These data provide comprehensive or nearly comprehensive coverage of the relevant population. They are either consistent with national accounts concepts, coverage, and timing or can be made to be consistent with only minor adjustments. For example, 3 months of Census Bureau data on trade in goods are incorporated into the estimates of exports and imports of goods, beginning with the second current quarterly estimates. The Census Bureau data provide nearly comprehensive coverage of the movement of goods into and out of the United States, and only minor adjustments for coverage and valuation are necessary to make the data conform to national accounts concepts.

Adjusted comprehensive data

These data provide comprehensive or nearly comprehensive coverage of the relevant population; however, these data require substantial adjustments in order to conform to national accounts concepts. These data include survey data as well as tax or regulatory data that are adjusted to correct for misreporting, differences in coverage, and differences in accounting concepts. For example, data from Internal Revenue Service (IRS) tabulations of corporate tax returns are incorporated into the second and third annual revision estimates of corporate profits. These annual data are more comprehensive than the indicators used to extrapolate the earlier estimates; however, they must be substantially adjusted, so that they conform to national accounts concepts. (7)

Direct indicator data

These data are used to indicate the movements of a series rather than the levels of a series (see the box "Extrapolation and Interpolation With Source Data Indicators"). These data are closely related to the comprehensive data that are ultimately incorporated into the estimates; however, they are less detailed or do not provide comprehensive coverage of the relevant population. For example, quarterly services survey (QSS) data from the Census Bureau are incorporated into the third current quarterly estimates of consumer spending for health care services. However, these data are less comprehensive than the Census Bureau service annual survey (SAS) data that are used to prepare the annual revision estimates.

Indirect indicator data

Indirectly related data are used to prepare indicator series when more suitable data are not available. These data include volume or activity indicators and other NIPA estimates. For example, the advance current quarterly estimates of business investment in research and development are extrapolated using BLS data on employment, hours, and earnings.

Trend-based data

Trends are used when no data are available. Trends are typically based on previous estimates, using moving averages, regressions, growth factors, or judgment. For example, for the advance and second current quarterly estimates of private fixed investment in software, no suitable source data are available, and the estimates are therefore based on moving averages of Census Bureau QSS data for prior quarters.

The incorporation of different categories of source data as estimates progress through the revision cycle is shown in tables 3 and 4.

Source Data for Successive Vintages

Table 2 and charts 1 and 2 present the shares of the five categories of source data that are incorporated in the successive vintages of the GDP and GDI estimates. This section summarizes the primary reasons for the changes in these shares as the estimates are revised with each vintage in order to incorporate more complete and more reliable source data.

GDP estimates

Current quarterly estimates

The current quarterly estimates of GDP incorporate data from each of the five categories. For the advance current quarterly estimates, about one-third of the estimates incorporate indirect indicator data, about one-quarter of the estimates incorporate direct indicator data, about one-quarter incorporate comprehensive data, about 10 percent of the estimates incorporate direct indicator data, and less than 1 percent incorporate adjusted comprehensive data (table 2 and chart 1).

The major difference between the advance current quarterly estimates and the second current quarterly estimates is that comprehensive data replace BEA assumptions (trend-based data) for the third month of the quarter for several key components, including private and state and local government investment in structures and exports and imports of goods. (8) In addition, Census Bureau monthly retail sales data (comprehensive data) replace advance monthly retail sales data (direct indicator data) for the third month of the quarter for the estimates of consumer spending for goods and for food services.

Between the second current quarterly estimates and the third current quarterly estimates, the biggest change is the incorporation of Census Bureau QSS data (direct indicator data) into the estimates of consumer spending for services and of private and government investment in software that replace estimates extrapolated either with BLS data on employment, hours, and earnings (indirect indicator data) or with judgmental trends (trend-based data).

Annual revision estimates

For the first annual revision estimates, the shares of the estimates based on comprehensive data and adjusted comprehensive data both increase, while the share of the estimates based on each of the other three categories of source data decreases (table 2 and chart 2). These changes primarily reflect the following:

* SAS data (comprehensive data) replace QSS data (direct indicator data) and other indicators for the estimates of consumer spending for services and of private and government investment in software.

* BEA annual survey data (comprehensive data) replace indicators for the estimates of exports and imports of services.

* BLS QCEW data (comprehensive data) replace BLS employment, hours, and earnings data (indirect indicator data) for the estimates of general government wages and salaries.

[GRAPHIC 1 OMITTED]

For the second annual revision estimates, the continued increases in the shares of the estimates based on comprehensive data and adjusted comprehensive data primarily reflect the following:

* Census Bureau annual survey of manufactures data (comprehensive data) replace Census Bureau monthly survey of manufacturers' shipments, inventories, and orders data (direct indicator data) for the estimates of private fixed investment in equipment.

* National Science Foundation survey data (comprehensive data) replace company financial report data (direct indicator data) for the estimates of private fixed investment in research and development.

* Fiscal year census of governments/government finances (COG/GF) data (comprehensive data) replace judgmental trends for the first half of the year for estimates of state and local government spending. (9)

For the third annual revision estimates, the major difference is that COG/GF data (comprehensive data) replace judgmental trends for the second half of the year for the estimates of state and local government spending.

Comprehensive revision estimates

For the comprehensive revision estimates, the major difference is that decennial census of housing data on housing stocks and average rents (comprehensive data) replace annual survey data on housing stocks and BLS consumer price index data on rents (direct indicator data) for the estimates of rental of tenant-occupied nonfarm housing and of imputed rental of owner-occupied nonfarm housing.

GDI estimates

Current quarterly estimates

The current quarterly estimates of GDI incorporate data from four of the five categories of source data; adjusted comprehensive data are not incorporated into any of the estimates until the second annual revision. For the second and third current quarterly estimates, just over half of the estimates incorporate indirect indicator data, about a third of the estimates incorporate trend-based data, a little over 10 percent incorporate direct indicator data, and less than 1 percent incorporate comprehensive data (table 2 and chart 1).

When the BLS QCEW data are incorporated into the revised estimates of private wages and salaries for the previous quarter (released with the second current quarterly estimates of GDP), the share of the GDI estimates that incorporates comprehensive data increases from less than 1 percent to about 35 percent.

Annual revision estimates

For the first annual revision estimates, the shares of the estimates based on comprehensive data and on direct indicator data increase, while the shares based on the other categories decrease (table 2 and chart 2). These changes primarily reflect the following:

* BLS QCEW data (comprehensive data) replace BLS employment, hours, and earnings data (indirect indicator data) for the estimates of government wages and salaries.

* Federal agency data (comprehensive data) replace indicators and trends for the estimates of supplements to wages and salaries.

* Annual survey data on housing stocks (direct indicator data) replace housing stocks extrapolated with Census Bureau data on housing starts (indirect indicator) for the estimates of rental of tenant-occupied nonfarm housing and of imputed rental of owner-occupied nonfarm housing within rental income of persons.

For the second annual revision estimates, the major change is the incorporation of IRS tax return data (adjusted comprehensive data) into the estimates of corporate profits, nonfarm proprietors' income, net domestic monetary interest, and capital consumption allowance; these data replace indicators and trends used for the earlier estimates. In addition, COG/GF data (comprehensive data) replace judgmental trends for the first half of the year for the estimates of several categories of state and local government receipts, including local government taxes on production and imports, rents and royalties, and the current surplus of government enterprises.

For the third annual revision estimates, the share of the estimates covered by comprehensive data increases, while the shares covered by direct indicator data, by indirect indicator data, and by trend-based data decrease. These changes primarily reflect the following:

* Pension benefit guaranty corporation (PBGC) plan-level tabulations of IRS form 5500 data (comprehensive data) replace PBGC aggregate data (direct indicator) for the estimates of employer contributions and of imputed interest on plans' claims on employers for private pension plans.

* Data from a complete sample of state and local government systems' financial and actuarial reports (comprehensive data) replace data from an incomplete sample of reports (direct indicator data) for the estimates of employer contributions and of imputed interest on plans' claims on employers for state and local government pension plans.

* COG/GF data (comprehensive data) replace judgmental trends for the second half of the year for the estimates of state and local government receipts.

Comprehensive revision estimates

For the comprehensive revision estimates, the one difference is that decennial census of housing data on housing stocks and average rents (comprehensive data) replace annual survey data on housing stocks and BLS consumer price index data on rents (direct indicator data) for the estimates of rental of tenant-occupied nonfarm housing and of imputed rental of owner-occupied nonfarm housing within rental income of persons.

Extrapolation and Interpolation With Source Data Indicators

Generally, monthly or quarterly source data that are used to prepare the current quarterly estimates are not as comprehensive as the annual source data that are used to prepare the annual revision estimates. Thus, some estimates are prepared through extrapolation or interpolation, using the more frequent but less comprehensive data as indicators of the movements of the series rather than as measures of the levels of the series. (1)

Specifically, for periods not yet covered by annual estimates (such as the current quarter), the quarterly estimates are extrapolated from the latest annual estimate that is based on comprehensive annual source data, using the pattern of the less detailed quarterly source data. Similarly, for periods for which the annual estimates are based on comprehensive annual source data, the quarterly estimates are interpolated between these annual estimates; that is, the quarterly estimates are controlled to these annual estimates while preserving the movements in the quarterly source data.

For example, the annual tax return data used to prepare the estimates of corporate profits are only available for a given year, beginning with the second annual revision estimates for that year. Therefore, prior to the second annual revision, changes in the annual and quarterly estimates are extrapolated using a variety of industry-specific indicators, based on Census Bureau tabulations of quarterly financial reports, on regulatory agency reports, and on publicly available corporate financial statements. Beginning with the second annual revision estimates, the annual levels of corporate profits are based on the tax return data. The interpolation process ensures that the pattern of change in the quarterly estimates follows the movements in the quarterly indicators that were used to extrapolate the earlier estimates.

For components where more frequent source data are not available, indicators may be based on trends.

(1.) 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 (February 2014): 4-6-4-7.

(1.) For an earlier analysis of the source data used to prepare the successive vintages of the GDP estimates, see Bruce T. Grimm and Teresa L. Weadock, "Gross Domestic Product: Revisions and Source Data," Survey of Current Business 86 (February 2006):11-15. For an earlier analysis of the source data underlying the GDI estimates, see Alyssa E. Holdren and Bruce T. Grimm, "Gross Domestic Income: Revisions and Source Data," Survey 88 (December 2008):14-20.

(2.) The successive revisions do not substantially change the general picture of economic activity. For more information, see Dennis J. Fixler, Ryan Greenaway-McGrevy, and Burce T. Grimm, "Revisions to GDP, GDI, and Their Major Components," Survey 91 (July 2011): 9-31.

(3.) At the time of the advance estimates, data on domestic profits and on net interest of domestic industries are not available. For fourth quarters, these data are not available for the advance estimates or for the second estimates.

(4.) The QCEW data 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. Revised estimates, which reflect the incorporation of QCEW data, were first published for the first quarter of 2002. Other NIPA series that are revised because of the incorporation of the data from the QCEW include GDI, the statistical discrepancy, gross national income, 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 GDP and GDI. Early annual estimates of GDP are initially released with the advance estimates for the fourth quarter, usually in January of the following year, and the estimates are then revised for the second and third estimates for the fourth quarter. Early annual estimates of GDI are initially released with the third estimates for the fourth quarter, usually in March of the following year. In addition, early annual estimates of GDI are revised to reflect the incorporation of fourth-quarter QCEW data, usually in May of the following year.

(6.) For more information, see "Taking Account," Survey 87 (July 2007): iv.

(7.) For more information on the adjustments to the IRS corporate tax return data, see Concepts and Methods of the U.S. National Income and Product Accounts, Chapter 13: Corporate Profits (February 2014): 13-22-13-38.

(8.) For many of the key series used to prepare the advance current quarterly estimates of GDP-including retail sales, unit automobile and truck sales and inventories, manufacturers' shipments of nondefense capital goods, manufacturers' inventories of durable goods, and federal defense spending-source data are available for all months of the quarter. For the following series, source data for the third month of the quarter usually are not available in time for inclusion in the advance current quarterly estimates, and BEA makes assumptions for these source data: new nonresidential, new single-family, and new multifamily construction spending; change in private inventories for nondurable-goods manufacturing and for merchant wholesale and retail industries other than motor vehicles and equipment; U.S. exports and imports of goods based on the international transactions accounts; and state and local government construction spending. These assumptions are published in the technical note that accompanies each monthly GDP release.

(9.) Because the fiscal years for many state and local governments do not coincide with calendar years, the incorporation of fiscal year data may affect the estimates for more than 1 calendar year. For more information, see Concepts and Methods of the U.S. National Income and Product Accounts, Chapter 9: Government Consumption Expenditures and Gross Investment (February 2014): 9-12.

Table 1. Release Schedule for the Estimates of Gross Domestic Product
(GDP) and Gross Domestic Income (GDI), Third Quarter of 2009

Estimate                                  GDP             GDI

Advance estimate                     October 2009        n.a.
Second estimate                      November 2009   November 2009
Third estimate                       December 2009   December 2009
Revised previous quarter estimate        n.a.        February 2010
First annual revision estimate         July 2010       July 2010
Second annual revision estimate        July 2011       July 2011
Third annual revision estimate         July 2012       July 2012
Comprehensive revision estimate        July 2013       July 2013

n.a. Not applicable

Table 2. Shares of Source Data for the Successive Estimates of
Gross Domestic Product and Gross Domestic Income
[Percent]

                              Advance   Second   Third   Revised
                                                         previous
                                                         quarter

Gross domestic product

Comprehensive data              25.5     36.8    38.5        ...
Adjusted comprehensive data      0.0      0.0     0.0        ...
Direct indicator data           12.2      9.7    27.8        ...
Indirect indicator data         34.4     33.1    21.5        ...
Trend-based data                27.9     20.3    12.2        ...

Gross domestic income
Comprehensive data               ...      0.9     0.9       34.2
Adjusted comprehensive data      ...      ...     ...        ...
Direct indicator data            ...     11.8    12.9       12.9
Indirect indicator data          ...     56.3    55.9       22.5
Trend-based data                 ...     31.0    30.3       30.3

                              First    Second   Third
                              annual   annual   annual

Gross domestic product

Comprehensive data             73.5     84.3     87.6
Adjusted comprehensive data     0.5      0.9      0.9
Direct indicator data          15.3      9.0      9.0
Indirect indicator data         4.0      2.7      2.5
Trend-based data                6.7      3.0      0.0

Gross domestic income
Comprehensive data             53.0     58.3     64.8
Adjusted comprehensive data     ...     31.6     31.6
Direct indicator data          15.2      7.8      3.2
Indirect indicator data        13.4      0.5      0.3
Trend-based data               18.3      1.8      ...

                              Comprehensive
                                  revision

Gross domestic product

Comprehensive data                    96.7
Adjusted comprehensive data            0.9
Direct indicator data                  0.0
Indirect indicator data                2.3
Trend-based data                       0.0

Gross domestic income
Comprehensive data                    68.0
Adjusted comprehensive data           31.6
Direct indicator data                  ...
Indirect indicator data                0.3
Trend-based data                       ...

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

Table 3. Major Source Data Incorporated for the Current
Quarterly Estimates of Gross Domestic Product and Gross
Domestic Income by Category of Source Data

Gross domestic product              Gross domestic income

Comprehensive data

Personal consumption                Private wages and
  expenditures (PCE) for goods        salaries--Bureau of Labor
  and for food services--Census       Statistics (BLS) quarterly
  Bureau monthly retail trade         census employment and wages
  survey (MRTS) data (2 months        (QCEW) data (revised previous
  for the advance estimates and       quarter estimates).
  3 months for the second and
  third estimates).

PCE for new motor vehicles and      Imputed interest paid and
  private fixed investment in         received by commercial
  new autos and new light             banks--Federal Reserve Board
  trucks--Ward's Automotive           (FRB) Call Report data (second
  Reports data on unit sales,         and third estimates).
  J.D. Power and Associates data
  on registrations (2 months for
  advance estimates and 3 months
  for second and third
  estimates), and R.L. Polk data
  on average prices.

PCE for prescription
  drugs--Intercontinental
  Medical Statistics Health,
  Inc. data on prescription drug
  sales to consumers.

PCE for tobacco--Department of
  the Treasury data on tobacco
  consumption (1 month for
  advance estimates, 2 months
  for second estimates, and 3
  months for third estimates)
  and BLS consumer price index
  (CPI) data for tobacco and
  smoking products.

PCE for net foreign
  travel--quarterly
  international transactions
  accounts data (second and
  third estimates).

Implicit services of
  commercial banks--FRB Call
  Report data (second and third
  estimates).

Private fixed investment and
  state and local government
  gross investment in struc-
  tures--Census Bureau monthly
  construction spending
  data. (1)

Exports of goods and imports
  of goods--trade in goods and
  services data. (1)

Exports of services and
  imports of services--quarterly
  international transactions
  accounts data (second and
  third estimates).

Federal government
  spending--Monthly Treasury
  Statement data and Department
  of Defense financial reports
  on disbursements (third
  estimates).

Adjusted comprehensive data

Change in private inventories       Not applicable
  for retail industries other
  than motor vehicles--MRTS
  data. (1)

Direct indicator data

PCE for goods and for food          State government general sales
  services--advance MRTS data         taxes--a sample of state
  (1 month for advance                monthly tax collections data
  estimates).                         (advance and second
                                      estimates).

PCE for nondiesel gasoline--        State and local government
  monthly Energy Information          taxes on production and
  Administration (EIA) data on        imports--preliminary data from
  quantities of gasoline              the Census Bureau quarterly
  supplied (1 month for advance       tax survey (third estimates).
  estimates, 2 months for second
  estimates, and 3 months for
  third estimates) and BLS data
  on average prices.

PCE for services and private        Corporate profits--Compustat
  fixed investment in software-       data and Census Bureau
  -Census Bureau quarterly            Quarterly Financial Report
  services survey (QSS) data          data (second and third
  (third estimates).                  estimates).

PCE for electricity and gas
  services--monthly EIA usage
  and unit value data (1 month
  for advance estimates, 2
  months for second estimates,
  and 3 months for third
  estimates).

Private fixed investment in
  petroleum and natural gas
  structures--American Petroleum
  Institute data on footage
  drilled and Baker Hughes data
  on rotary rig counts.

Private fixed investment in
  equipment--Census Bureau
  advance monthly survey of
  manufacturers' shipments,
  inventories, and orders (MSM)
  data (1 month for advance
  estimates) and MSM data (2
  months for advance estimates
  and 3 months for second and
  third estimates).

Private fixed investment in
  research and development--
  company financial report data
  (second and third estimates).

Change in private inventories
  for durable-goods
  manufacturing industries--
  advance MSM data (1 month for
  advance estimates) and MSM
  data (2 months for advance
  estimates and 3 months for
  second and third estimates).

Change in private inventories
  for nondurable-goods
  manufacturing industries--MSM
  data. (1)

Change in private inventories
  for merchant wholesale trade
  industries--Census Bureau
  monthly wholesale trade survey
  data. (1)

Gross domestic product              Gross domestic income

Indirect indicator data

PCE for nondiesel gasoline and      Private wages and salaries and
  for electricity and natural         government wages and salaries-
  gas services--EIA ShortTerm         -BLS current employment
  Energy Outlook data (2 months       statistics (CES) data.
  for advance estimate sand 1
  month for second estimates).

PCE for services--CES data          Employer contributions for
  (primarily advance and second       government social insurance--
  estimates).                         estimates of wages and
                                      salaries from the national
                                      income and product accounts
                                      (NIPAs).

Rental of tenant-occupied           State gasoline sales taxes--
  nonfarm housing and for             EIA gasoline supplied data
  imputed rental of                   (advance and second
  owner-occupied nonfarm              estimates).
  housing--Census Bureau housing
  starts data and CPI data on       Local general sales taxes--
  average rent.                       NIPA estimates of state
                                      general sales taxes.

PCE for cellular telephone          Monetary interest, net--NIPA
  services--cellular telephone        estimates of personal interest
  company revenue data (second        income, net interest paid by
  estimates).                         government, and personal
                                      interest payments.

Private fixed investment in         Nonfarm proprietors' income--
  residential improvements--          various activity indicators
  advance MRTS data (1 month for      and NIPA estimates of wages
  advance estimates), MRTS data       and salaries.
  (2 months for advance
  estimates and 3 months for        Farm proprietors' income--
  second and third estimates),        U.S. Department of Agriculture
  and BLS occupational earnings       data.
  data (2 months for advance
  estimates and 3 months for
  second and third estimates).

Private fixed investment in         Rental of tenant-occupied
  own-account software and in         nonfarm housing and for
  research and develop-               imputed rental of owner-
  ment--CES data (advance             occupied nonfarm housing
  estimates).                         within rental income of
                                      persons--Census Bureau housing
                                      starts data and CPI data on
                                      average rent.

Private fixed investment in
  entertainment, literary, and
  artistic originals--CES data
  (advance and second estimates)
  and QSS data (third
  estimates).

General government wages and
  salaries--CES data.

General government employer
  contributions for government
  social insurance-- NIPA
  estimates of wages and
  salaries.

Government consumption of
  fixed capital--NIPA estimates
  of gross investment and of
  prices.

Trend-based data

PCE for tobacco (2 months for       Employer contributions for
  advance estimates and 1 month       employee pension and insurance
  for second estimates).              funds.

PCE for services (primarily         Federal taxes on production
  advance and second estimates).      and imports.

Implicit services of                State and local government
  commercial banks (advance           taxes on production and
  estimates).                         imports.

Private fixed investment and        Government subsidies.
  state and local government
  gross investment in               Imputed interest paid and
  structures.                         received by commercial banks
                                      (advance estimates).

Private fixed investment in         Business current transfer
  software (advance and second        payments (net).
  estimates).

Change in private inventories       Current surplus of government
  for nondurable-goods                enterprises.
  manufacturing industries, for
  merchant wholesale industries,
  and for retail industries
  other than motor vehicles. (1)

Exports of goods and of
  imports of goods. (1)

State and local government
  spending other than for wages
  and salaries, employer
  contributions for government
  social insurance, and
  consumption of fixed capital.

(1.) For several of the key series used to prepare the
advance current quarterly estimates of GDP source data are
only available for the first 2 months of the quarter, and
assumptions are made for the third month.

Note. In this table, "advance," "second" and "third" refer
to the advance, second, and third current quarterly
estimates, respectively. The references to the current
quarterly estimates in parentheses indicate for which of the
estimates the data are incorporated. Where no reference is
included in parentheses, the data are incorporated for each
of the current quarterly estimates. In addition, for
estimates for which data are not available for all 3 months
of the quarter, information may be included in parentheses
to identify the sequence in which the data become available.

Table 4. Major Source Data Incorporated for the Annual
Revision Estimates of Gross Domestic Product and Gross
Domestic Income by Category of Source Data

Gross domestic product              Gross domestic income

Comprehensive data (1)

Personal consumption                State and local government
  expenditures (PCE) for goods        wages and salaries--Bureau of
  and for food services--Census       Labor Statistics (BLS)
  Bureau monthly retail trade         quarterly census employment
  survey (MRTS) data (first           and wages (QCEW) data.
  estimates) and annual retail
  trade survey (ARTS) data
  (second and third estimates).

PCE for services--Census Bureau     Employee contributions for
  service annual survey (SAS)         government social insurance,
  data.                               federal government defined
                                      benefit and defined
                                      contribution pension plan
                                      contributions, and federal
                                      government taxes on production
                                      and imports--federal agency
                                      data.

PCE for electricity and             Employer contributions to
  natural gas services--annual        private pension plans and
  Energy Information                  imputed interest on plans'
  Administration (EIA) data.          claims on employers for
                                      private pension plans--plan-
                                      level tabulations of Internal
                                      Revenue Service (IRS) form
                                      5500 data from the Pension
                                      Benefit Guaranty Corporation
                                      (PBGC) (third estimates).

PCE for insurance                   Employer contributions to
  services--A.M. Best data            state and local government
  (second and third estimates).       pension plans and imputed
                                      interest on plans' claims on
                                      employers for state and local
                                      government pension
                                      plans--complete sample of state
                                      and local governments'
                                      financial and actuarial report
                                      data (third estimates).

Private fixed investment and
  state and local government
  gross investment in
  structures--Census Bureau
  annual construction spending
  data.

Private fixed investment in         Employer contributions for
  equipment--Census Bureau annual     group health insurance--medical
  survey of manufactures (ASM)        expenditure panel survey data
  data (second and third              (second and third estimates).
  estimates).

Private fixed investment in         State and local government
  research and                        taxes on production and
  development--National Science       imports, subsidies, and
  Foundation survey data (second      current surplus of government
  and third estimates).               enterprises--Census Bureau
                                      census of governments/
                                      government finances (COG/GF)
                                      data (second and third
                                      estimates). (2)

Private fixed investment in         State government taxes on
  software--Census Bureau SAS         production and imports--Census
  data.                               Bureau COG/GF data (second and
                                      third estimates) and state
                                      government tax collections
                                      data. (3)

Exports of services and             Imputed interest paid by
  imports of services--annual         depository institutions other
  data from BEA surveys.              than commercial banks-- Federal
                                      Deposit Insurance Corporation
                                      and Credit Union National
                                      Association data.

General government wages and        Imputed interest paid by life
  salaries--BLS QCEW data.            insurance carriers--American
                                      Council of Life Insurers data
                                      (second and third estimates).

Federal government                  Imputed interest paid by
  spending--Department of Defense     property and casualty
  contract awards data, Federal       insurance companies and net
  Procurement Data System data,       insurance settlements--A.M.
  U.S. Budget data, and "budget       Best data (second and third
  exhibit" data.                      estimates).

State and local government          Farm proprietors' income--U.S.
  spending--Census Bureau COG/GF      Department of Agriculture
  data (second and third              data.
  estimates). (2)

                                    Federal government subsidies
                                      and current surplus of federal
                                      enterprises-- U.S. Budget data.

Adjusted comprehensive data

Change in private inventories       Monetary interest, net,
  for all industries except           nonfarm proprietors' income,
  farm-Census Bureau annual           corporate profits, and capital
  wholesale trade survey data,        consumption allowance-IRS SOI
  ARTS data, and ASM data and         data (second and third
  IRS Statistics of Income (SOI)      estimates).
  data (second and third
  estimates).

Private fixed investment in
  own-account software-BLS
  occupational employment
  statistics survey data on
  production costs (second and
  third estimates).

Direct indicator data

PCE for nondiesel                   Employer contributions to
  gasoline--monthly EIA data on       private pension plans and
  quantities of gasoline              imputed interest on plans'
  supplied and BLS data on            claims on employers for
  average prices.                     private pension plans--PBGC
                                      aggregate tabulations of IRS
                                      form 5500 data (second
                                      estimates).

Rental of tenant-occupied           Employer contributions to
  nonfarm housing and imputed         state and local government
  rental of owner-occupied            pension plans and imputed
  nonfarm housing--Census Bureau      interest on plans' claims on
  American housing survey (AHS)       employers for state and local
  data or current population          government pension
  survey (CPS) data on housing        plans--incomplete sample of
  stocks and BLS consumer price       state and local governments'
  index (CPI) data on average         financial and actuarial report
  rent.                               data (second estimates).

PCE for insurance                   State government taxes on
  services--A.M. Best                 production and imports--Census
  Review/Preview data (first          Bureau quarterly tax survey
  estimates).                         data (first estimates). (3)

Private fixed investment in         Rental of tenant-occupied
  equipment--Census Bureau            nonfarm housing and imputed
  monthly survey of                   rental of owner-occupied
  manufacturers' shipments,           nonfarm housing within rental
  inventories, and orders (MSM)       income of persons--Census
  data.                               Bureau AHS
                                      data or CPS data on housing
Change in private inventories         stocks and BLS CPI data on
  for nondurable-goods                average rent.
  manufacturing industries-- MSM
  data (first estimates).           Imputed interest paid by
                                      property and casualty
                                      insurance companies and net
                                      insurance settlements//A.M.
                                      Best Review/Preview data
                                      (first estimates).

Change in private inventories
  for merchant wholesale trade      Corporate profits--Compustat
  industries--Census Bureau           data and Census Bureau
  monthly wholesale trade survey      Quarterly Financial Report
  data (first estimates).             data (first estimates).

Indirect indicator data

Private fixed investment in         Monetary interest, net--NIPA
  entertainment, literary, and        estimates of personal interest
  artistic originals--Census          income, net interest paid by
  Bureau SAS data.                    government, and personal
                                      interest payments (first
                                      estimates).

Government consumption of           Nonfarm proprietors' income--
  fixed capital--NIPA estimates       various activity indicators
  of government gross investment      and NIPA estimates of wages
  and prices.                         and salaries (first
                                      estimates).

Trend-based data

State and local government          State and local government
  spending other than for wages       taxes on production and
  and salaries, employer              imports, subsidies, and
  contributions for government        current surplus of government
  social insurance, and               enterprises (first and second
  consumption of fixed capital        estimates). (2)
  (first and second
  estimates). (2)

(1.) Unless otherwise noted, the comprehensive data
incorporated for the current quarterly estimates are also
incorporated for the annual revision estimates. See table 3
for more information.

(2.) Fiscal year COG/GF data incorporated for the second
annual revision estimates only cover about half of the
calendar year; the remaining half of the calendar year is
estimated with judgmental trends.

(3.) Fiscal year state government tax collections data
incorporated for the first annual revision estimates only
cover about half of the calendar year; the remaining half of
the calendar year is estimated with quarterly tax survey
data.

Note. In this table "first" "second" and "third" refer to
the first, second, and third annual revision estimates,
respectively. The references to the annual revision
estimates in parentheses indicate for which of the estimates
the data are incorporated. Where no reference is included in
parentheses, the data are incorporated for each of the
annual revision estimates.

Chart 2. Shares of Source Data for the Annual
Revision and Comprehensive Revision GDP and
GDI Estimates

GDP Estimates

First annual revision

Comprehensives data            73.5%
Adjusted comprehensive data     0.5%
Direct indicator data          15.3%
Indirect indicator data         4.0%
Trend-based data                6.7%

Second annual revision

Comprehensive data              84.3%
Adjusted comprehensive data      0.9%
Direct indicator data            9.0%
Indirect indicator data          2.7%
Trend-based data                 3.0%

Third annual revision

Comprehensive data              87.6%
Adjusted comprehensive data      0.9%
Direct indicator data            9.0%
Indirect indicator data          2.5%

Comprehensive revision

Comprehensive data              96.7%
Adjusted comprehensive data      0.9%
Indirect indicator data          2.3%

GDI Estimates

First annual revision
Comprehensive data              53.0%
Direct indicator data           15.2%
Indirect indicator data         13.4%
Trend-based data                18.3%

Second annual revision

Comprehensive data              58.3%
Adjusted comprehensive data     31.6%
Direct indicator data            7.8%
Indirect indicator data          0.5%
Trend-based data                 1.8%

Third annual revision

Comprehensive data              64.8%
Adjusted comprehensive data     31.6%
Direct indicator data            3.2%
Indirect indicator data          0.3%

Comprehensive revision

Comprehensive data              68.0%
Adjusted comprehensive data     31.6%
Indirect indicator data         0.3%

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

U.S. Bureau of Economic Analysis

Note: Table made from pie chart.