The revisions to GDP, GDI, and their major components.
Fixler, Dennis J. ; Greenaway-McGrevy, Ryan ; Grimm, Bruce T. 等
THE NATIONAL income and product accounts (NIPAs) provide a timely,
comprehensive, and accurate picture of the condition of the U.S.
economy. The two featured measures, gross domestic product (GDP) and
gross domestic income (GDI), are measures of the same concept of total
activity in the U.S. economy. GDP measures activity as the sum of all
final expenditures in the economy; it is detailed on the product side of
the domestic income and product account. GDI measures activity as the
sum of all incomes generated in production; it is detailed on the income
side of the account. Thus, in concept GDP and GDI measure the same
economic activity, but in practice, they differ because each is
constructed using different source data. (1)
Measuring the accuracy of the national accounts estimates is a
long-standing challenge for several reasons:
* The early GDP and GDI estimates are based on partial and
preliminary source data as well as trend projections when data are not
available. They are intended to provide an "early read" on the
general picture of economic activity for decisionmakers. These early
estimates are subsequently revised as more complete information become
available.
* The source data used to estimate the national economic accounts
come from a mixture of survey, tax, and other business and
administrative data. They are subject to a mix of sampling and
nonsampling errors and biases that cannot be measured in terms of
standard errors.
* The NIPAs are regularly revised to reflect changes in the
economic concepts and methods necessary for the accounts to provide a
relevant and accurate picture of the evolving U.S. economy. These
updates range from expanding the definition of investment to include
research and development activity as well as the production of
entertainment, literary, and artistic originals to updating seasonal
adjustment factors to reflect the most recent seasonal patterns.
As a result, accuracy cannot be assessed by conventional
statistical measures, such as standard errors. Instead, we assess
accuracy by examining magnitudes and patterns of NIPA revisions (see the
box "Accuracy, Reliability, and Uncertainty"). The NIPAs are
revised for a variety of reasons. Some revisions are due to the
replacement of early extrapolations for missing source data or
preliminary survey data with more complete and accurate annual and
benchmark data, such as economic census data. (2) Other revisions are
the result of updates to the concepts on which the accounts are based.
These revisions to concepts and definitions can be substantial. In six
comprehensive revisions--in 2009, 2003, 1999, 1995, 1991, and 1985--the
average percentage change in the levels of current-dollar GDP for
selected periods was 2 percent, and of that change, about one-third was
from concepts and definitions and two-thirds from statistical revisions.
In the 2013 comprehensive revision, the dollar level of GDP was revised
up an average of 3.1 percent in 1993-2012, mainly because of the
recognition of research and development and creation of artistic and
literary originals as investment. (3) Another major cause of revisions
is updated seasonal adjustment factors, which Fixler and Grimm (2002)
reported accounted for an average revision to GDP (without regard to
sign) of 1.0 percentage point.
BEA's principal standard of reliability is based on a
comparison of its early estimates to the "latest" estimates,
which are revised to incorporate the most up-to-date concepts,
statistical methods, and the most complete and accurate source data
available. These comparisons indicate that revisions do not
substantively change BEA's measures of long-term growth, the
picture of business cycles, and trends in major components of GDP.
Economic policy decisions should not need to be reconsidered in the
light of revisions to GDP estimates, and policymakers should be able to
rely on the early estimates as correctly indicating the state of the
economy. More specifically, BEA judges the qualitative reliability of
its early estimates by whether they present the same general picture of
economic activity as the latest estimates in terms of the following:
* Long-term growth rates
* Trends in saving, investment, government spending, corporate
profits, and other key components of GDP and GDI
* Broad features of the business cycle, including the timing and
depth of recessions, the strength of recoveries, and the major
components contributing to growth and contractions
* The patterns of quarterly growth, including whether growth in any
particular period is high or low relative to trend, is accelerating or
decelerating, or is positive or negative
Quantitatively, revisions are measured as the changes from an
earlier vintage of a given estimate to a later vintage of that estimate,
for example, from the third current quarterly estimate to the first
annual revision estimate (see the box "Vintages and Timing of
Revisions"). Because the latest vintage incorporates the most
recent comprehensive revision, they incorporate all the available source
data that are believed to be the most reliable. Over the long run, this
study finds the following:
* Revisions to long-term growth rates are small, averaging less
than 0.1 percentage point for average growth rates over the 1993-2012
period.
* With the exception of conceptual revisions like the addition of
R&D, there are no substantial revisions, as measured by shares of
GDP or GDI for key components such as investment and government
expenditures or for the national saving rate.
* The overall pattern of change in GDP over time is little changed
by the revisions (chart 1).
In the short run, there are three vintages of "current
quarterly" estimates of GDP--the advance, the second, and the third
estimates. Each estimate is produced using a wide mix of source
data--preliminary survey results, such as the Census Bureau's
surveys of monthly retail trade and quarterly services as well as
manufacturers' shipments and inventories, various indirect
indicators, trade industry data, and some trend projections--that are
later revised to reflect more complete information. (4)
The three vintages of GDP estimates successfully indicate the
following:
* The direction of change in real GDP 96 percent of the time
* The acceleration or deceleration of growth about 75 percent of
the time
* The relative magnitude of growth--whether it was above, near, or
below trend (near trend is within one standard deviation from the mean)
about 83 percent of the time
* The cyclical peaks before five of the seven recessions in
1969-2012
* The cyclical troughs of five of the seven recessions (5) Early
quarterly estimates are replaced successively
with three vintages of "current annual" estimates that
are primarily based on increasingly complete annual source data.
Approximately every 5 years, the annual estimates are replaced with
benchmark estimates when BEA conducts its comprehensive revisions of the
NIPAs.
The measures of reliability used by BEA and discussed in this
article are mean absolute revisions (MARs) and mean revisions (MRs). The
MARs to the annual rates of change--that is, taking the average of the
revisions without regard to sign--from the current quarterly estimates
to the latest estimates for 19932012 of both current-dollar and real GDP
have averaged somewhat more than 1 percentage point. (See the box
"Mean Revisions, Mean Absolute Revisions, and Standard
Deviations.") The MRs--that is taking the averages and allowing
both positive and negative values for the revisions--have averaged about
-0.1 percentage point.
The MARs between the current quarterly estimates are smaller. For
example, the MAR from the advance estimates to the second estimates of
GDP is 0.5 percentage point, and the MAR from the advance estimates to
the third estimates is 0.6 percentage point. The MAR from the second
estimates to the third estimates is 0.2 percentage point.
MRs indicate whether the revisions are generally positive or
negative. Because revisions may be offsetting, the MRs are much smaller
than the MARs. The MR for real GDP from the advance estimates to the
latest estimates is between 0 and -0.1 percentage point, much of which
reflects the result of NIPA comprehensive revisions. The MRs from the
second and third estimates to the latest estimates are both between 0
and 0.2 percentage point. To put these MRs into context for 1993-2012,
the mean growth of real GDP was 2.5 percent, and the growth rates ranged
from -8.3 percent to 7.8 percent.
[GRAPHIC 1 OMITTED]
The remainder of this article discusses (1) revisions to quarterly
estimates of GDP and its components, (2) revisions to annual estimates
of GDP and its components, (3) revisions to quarterly estimates of GDI
and its components, (4) revisions to annual estimates of GDI and its
components, and (5) the relationship of quarterly estimates of GDP and
GDI and weighted averages of the two measures. These are followed by a
brief summary and conclusions.
Revisions to Quarterly Estimates of GDP
The featured reliability measures featured MRs and MARs from the
three vintages of current quarterly estimates to the latest estimates.
Standard deviations are also shown in selected tables, providing
information about the distribution of the revisions.
Mean absolute revisions
The measures of revisions for real and current-dollar GDP and their
components are shown in table 1. In 1993-2012, the MARs for both
current-dollar GDP and real GDP are somewhat larger than 1.2 percentage
points for both the second and third current quarterly estimates, with
those for real GDP slightly smaller than those for current-dollar GDP.
For both measures, the MARs decline slightly but steadily from the
advance estimates to the third estimates. (6)
In general, previous BEA revision studies have found little or no
tendency for MARs to decline with successive vintages of estimates, in
either real or current dollars. However, this study--which drops the
1983 to 1992 time period included in earlier studies--finds that the
MARs for GDP and for many of its components decline with successive
vintages and with more complete and revised source data. (7)
The MARs for current-dollar personal consumption expenditures (PCE)
are about 0.3 percentage point smaller than those for the corresponding
vintages of current-dollar GDP; they are 1.0 percentage point for the
advance estimates and 0.9 percentage point for both the second and third
estimates. Those for real PCE are slightly smaller, about 0.8 percent
for the second and third vintages. Both measures tend to decline
slightly with successive vintages. Within PCE, the MARs for durable
goods are noticeably larger, about 4.3 percent for current-dollars and
4.4 percent for real durable goods; the MARs do not decline with
successive vintages.
The MARs for nondurable goods are also larger than those for all
PCE, and show some tendency to decline with successive vintages of
estimates. The MAR for the advance estimate is 2.0 percentage points.
The MARs for the second and third estimates are 1.8 percentage points.
The MARs for the real estimates are about 0.2 percentage point smaller
than those for current-dollar estimates.
The MARs for PCE services are the smallest of those of any of the
major PCE components shown in the table. The MARs for current-dollar PCE
are 0.9 percentage point for the advance estimate, and 0.8 percentage
point for both the second and third estimates. The MARs for the
corresponding real estimates are slightly smaller, but the differences
are less than 0.1 percentage point. Again, there is a modest tendency
for the MARs to decline with successive vintages. Fixler and Grimm 2011
presented average revisions for price indexes for GDP and major
components. They were found to be small and generally not statistically
significant. BEA also produces monthly estimates of PCE and prices.
These are discussed for the first time in the box "Revisions to
Monthly Estimates of Price Indexes for Personal Consumption
Expenditures."
The MARs for the other components of GDP are all considerably
larger than those for GDP and PCE The MARs for gross private domestic
investment are relatively large; they range from 5.2 percentage points
to 5.9 percentage points for both current-dollar and real investment.
Both increase 0.6 percentage point from the advance to the second
estimates, then decrease 0.2 percentage point to the third estimates.
The MARs for fixed investment are smaller; in current dollars, they
cluster near 2.7 percentage points and near 2.8 percentage points in
constant dollars. The smaller sizes are the result of the exclusion of
inventory investment, which in a previous study were found, using an
alternative methodology, to have large revisions; see Fixler and others
(2011).
Within private fixed investment, MARs for current-dollar
nonresidential structures decrease from 8.3 percentage points for the
advance estimates to 7.7 percentage points for the second estimates and
7.1 percentage points for the third estimates. Real structures have a
similar pattern, but at values roughly 1 percentage point lower.
The expansion of the category "equipment and software" to
include other intellectual property products" influenced the sizes
of the MARs for the category. (8) They are near 5 percentage points for
both current-dollar and real estimates for all the current quarterly
vintages. In the new estimates, the expansion increased the size of this
category by a little more than one fourth, with a generally increasing
trend ratio. To the extent that the quarterly pattern of change for
R&D plus literary and artistic originals is different from that for
equipment and software, the MARs are increased due to the expansion. The
MARs for the three current quarterly estimates are roughly 1 percent
point higher than those found in the previous study. The MARs for the
real estimates are also higher, but by roughly 0.3 percentage point. The
change also tends to increase MARs for broader investment categories,
but because the shares of the revision in totals is not large, the
effects are not easily observed.
The MARs for the various current quarterly vintages of residential
structures investment are only modestly smaller than those for equipment
and intellectual property products investment. They decline across
successive vintages in both real and current dollars. The MARs for real
investment are about 0.4 percentage point smaller than those for current
dollars.
The MARs for exports and imports, in both current and real dollars,
and for the current quarterly vintages range from about 3 percentage
points to somewhat more than 4 percentage points. (9) Each of the four
series has MARs that decline steadily across the successive vintages of
current quarterly estimates.
The MARs for total government consumption expenditures and gross
investment are all roughly 1.5 percentage points in both current and
real dollars. They show little tendency to decline with successive
vintages. MARs for federal expenditures are about 2.75 percentage points
and show no particular tendency to decline with successive vintages.
Within federal, defense expenditures' MARs decline from 3.5
percentage points for the advance estimates to 3.3 percentage points for
the third estimates. The MARs for real estimates are somewhat smaller,
ranging from 3.4 percentage points for the advance estimates to 3.0
percentage points for the third estimates. Nondefense expenditures'
MARs increase with successive vintages, ranging from 4.1 percentage
points for the advance estimates, to 4.6 percentage points for the third
estimates, in both real and current dollars.
The MARs for state and local government consumption expenditures
and gross investment decline from 2.1 percentage points for the
current-dollar advance estimates to 1.9 percentage points for both the
second and third estimates. MARs for the real estimates are about 1.6
percentage points and show no tendency to decrease across vintages.
Mean revisions
The MRs for real and current-dollar GDP are rather small, generally
much smaller, than the MARs for GDP because the MARs are functions of
both the MRs and the variances of the estimates, which by definition are
positive. This occurs because individual revisions are both positive and
negative and tend to offset one another. For most of the measures shown
here, there is little or no tendency for MRs to become smaller with
successive vintages of estimates.
The MRs for both current-dollar and real GDP are rather small and
generally negative. The MRs for advance estimates of current-dollar GDP
and real GDP are 0.1 percentage point and -0.1 percentage point,
respectively. The MRs for the second and third current-dollar estimates
of GDP are negative. For real GDP, the MRs are -0.2 percentage point.
The MRs for the components and subcomponents of real and current-dollar
GDP are generally small and both positive and negative, and the signs of
the MRs sometimes vary with successive vintages. The MRs for
current-dollar structures investment and exports are generally much
larger than for other components and subcomponents. The MRs for GDP and
its components reflect definition changes that are part of comprehensive
revisions.
The standard deviations of revisions are somewhat larger than the
MARs for GDP and its components, although the standard deviations for
vintages have similar patterns and relationships to those of the MARs.
The standard deviations can be used to test whether the MRs are
statistically significant. For 1993-2012, the MRs for GDP or its
components and subcomponents generally are not statistically different
from zero and, therefore, are not statistically significant and do not
indicate bias. (10)
Charts 2 and 3 provide supplemental pictures of the revisions.
Chart 2 is a histogram that shows the numbers of times that revisions
from the advance to the latest estimates are within various size
categories. The chart reflects average revisions (both median and mean)
of -0.06 percent. The smallest and largest revisions are -4.53 percent
and 4.01 percent, respectively.
Chart 3 is a scatter diagram showing advance estimates of real GDP
and the corresponding size of the revisions from those advance estimates
to the latest estimates. The scatter shows no apparent relationship
between the sizes of the advance estimates and the ultimate extent of
revisions to the latest estimate. The chart indicates that there is
nothing systematic about revisions available at the time of earlier
estimates.
Table 2 presents revisions triangles that show the MARs for the
various vintages of current-dollar GDP estimates to later estimates. All
tend to increase with later vintages. For example, the MARs for the
second estimates increase rapidly to the first annual estimates, and
more slowly thereafter. The revisions to each successive vintage
decrease steadily; that is, revisions decline with later vintages.
Because comprehensive revisions of the NIPAs are made about every 5
years, the first annual revision estimates contain major redefinitions
and reclassifications about one-fifth of the time; the second annual
revision estimates, two-fifths; and the third annual revision estimates,
three-fifths. Thus, the MARs for the successive annual revision
estimates increasingly reflect the effects of these changes as well as
the incorporation of increasing amounts of annual data that are
available with 1-to-3-year lags.
These same patterns generally hold for the five major components of
GDP. The MARs for the various vintages of estimates of PCE to the
various later vintages of estimates are modestly lower than those for
GDP, but follow the same general patterns. The patterns of MARs for the
three components of PCE also generally follow the same patterns. The
MARs for PCE durable goods are several times those of total PCE. MARs
for PCE nondurable goods are less than half those for all PCE but also
show the same patterns. The MARs for PCE services also show similar
patterns but are roughly half the size of the corresponding MARs for
nondurable goods, and roughly one-third to one-fourth those of durable
goods.
[GRAPHIC 2 OMITTED]
[GRAPHIC 3 OMITTED]
The MARs for the various vintages of gross private domestic
investment have a generally similar pattern to those of GDP and PCE, but
they are four to five times as large as those of GDP. The MARs for the
various vintages of fixed investment follow roughly the same patterns
but are about half the size of those of the corresponding vintage pairs
of gross private domestic investment. The difference is the exclusion of
change in private inventories, for which percent changes are not
meaningful because the series of estimates switches sign from one
quarter to the next. The MARs for the various vintage pairs of fixed
nonresidential investment and its three components also show similar
patterns but with somewhat larger sizes. The patterns of MARs for the
various estimates of residential investment are similar to those for the
various estimates of fixed nonresidential investment, and the MARs are
of roughly similar sizes.
The pattern of MARs for the various vintages of both imports and
exports to later vintages are roughly the same and are similar to the
patterns for the other components of GDP. The MARs of the advance
estimates to later vintages of imports estimates are considerably larger
than those for the corresponding exports estimates. The MARs for later
vintages of imports estimates to later intermediate vintages are,
however, smaller than the corresponding MARs for exports. The MARs for
both exports and imports estimates to the latest estimates, however, are
similar in size for each vintage, including the advance estimates.
The MARs for the various vintages of estimates of government
consumption expenditures and gross investment also show patterns similar
to those for GDP, but they are typically roughly 0.5 percentage point
larger. The patterns of MARs for the components and subcomponents of
government are roughly similar to the aggregate but with somewhat larger
values. The MARs for state and local government are somewhat smaller
than those for the other components. This probably does not indicate
greater reliability; relatively little new quarterly information comes
in at the times of the later estimates, and the sizes of the revisions
reflect this.
There are several characteristics of the pattern of the revisions
of the various vintages to successive revisions that hold generally but
not universally. The revisions from the second to the third vintages of
estimates are typically the smallest for any adjacent pair of vintages.
For any given vintage, the MARs tend to increase as later vintages
are used as the standards of revisions. For any vintage of later
estimates, the MARs tend to decrease in size for all vintages (when
later vintages of estimates are used as the earlier estimates). That is,
for any row (vintage of earlier estimates), moving across columns of
later estimates results in increasingly large MARs. Also, for any given
column (vintage of later estimates), going down rows generally results
in decreasing MARs.
Table 3 shows MARs for real GDP and its components. The sizes of
the MARs are quite close to those of MARs for current-dollar GDP. The
patterns are also very similar to the current-dollar patterns. These
occur because the revisions to deflators used to construct the real
estimates are quite small in comparison with the current-dollar
revisions. Most estimates of the components of GDP are made in current
dollars and then deflated.
Table 4 (page 12) provides additional information about the
revisions from earlier vintages of estimates to the latest estimates.
Columns 1 and 2 show the standard deviations of the latest estimates of
real and current-dollar GDP and its components and subcomponents at the
same level of detail as tables 1 and 2. These are measures of the
volatility of the estimates. They range from somewhat less than 3
percentage points for real and current-dollar GDP to about 15 percentage
points for current-dollar residential investment.
Columns 2 and 3 expand on columns 3 and 4 of table 1; they show
standard deviations of the revisions of the same set of measures,
including the three annual revision vintages of estimates. They show a
wide variation in size, ranging from 0.9 percentage point for
current-dollar GDP and 0.8 percentage point for real GDP to 12.1
percentage points for current-dollar exports and 7.0 percentage points
for gross private domestic investment. The revision standard deviations
may be scaled by the standard deviations of the estimates in order to
facilitate comparisons among components. When the revision standard
deviations are scaled by the standard deviations of the latest
corresponding estimates, the results are much closer together. For the
revisions, the standardized standard deviations have relatively small
ranges, 3.1 scaled units for the current-dollar estimates and 1.8 scaled
units for the real estimates. Thus, the more volatile series for GDP and
components that have larger standard deviations also tend to have larger
revision standard deviations.
Revisions around cyclical turning points
Comprehensive revisions have generally preserved the patterns of
change in and around recessions. The 2013 revision did so, despite the
large upward revisions in levels yielded by the inclusion of research
and development and artistic originals in investment. Panel 1 of chart 4
shows the rates of change in real GDP before, during, and after the
1990-91 recession. Only minor revisions to the rates of change were
introduced, in comparison with the overall pattern of change in GDP.
Panels 2 and 3 show the revisions to real GDP before, during, and after
the 2000 and 2007-2009 recessions. The revisions in and around the 2000
recession are very minor in comparison with the quarter-to-quarter
variations in the rates of change in real GDP. The revisions in the
2007-2009 recession are somewhat larger, but the largest is only 1.7
percentage points in third quarter of 2008; this compares with a range
of rates of change from 4 percentage points to more than -8 percentage
points.
[GRAPHIC 4 OMITTED]
The 2013 comprehensive revision preserved the pattern of volatility
of GDP, including the effects of business cycles. Chart 5 shows the
paths of eight-quarter moving variances of GDP over the entire period
since the ebbing, in the mid-1980s, of a period of higher volatility
that has been commented on by a number of observers. (11) Both the
prerevision and postrevision estimates closely resemble one another. The
large movements, indicating increased volatility during and after the
three recessions since the mid-1980s are quite similar. In addition, the
smaller fluctuations around the larger movements are largely the same.
The most notable change is that the comprehensive revision estimates
have modestly lower volatility during and immediately after the sharp
2007-2009 recession. (The lags in the volatility patterns following
recessions reflect that a given quarter's percent change stays in
the variance calculations for the seven following quarters.) On average,
the variances were revised little; the average value was 4.51 percentage
points before the comprehensive revision and 4.47 percentage points
after the comprehensive revision, for an average revision of -0.04
percentage point. Without regard to sign, the average revision in
variances was 0.57 percentage point. These compare with variances that
range from 1 percentage point to 22 percentage points.
[GRAPHIC 5 OMITTED]
Revisions to Annual Estimates of GDP
Summary statistics for revisions of annual frequency estimates of
real and current-dollar GDP and its major components to the latest
estimates are shown in table 5. As with the quarterly frequency
estimates, the mean revisions are small and are both positive and
negative.
The early annual estimates of GDP are those available in March of
the following year and are mostly composed of third current quarterly
estimates, and their mean revisions are therefore similar to those of
those estimates.
The mean absolute revisions are fractions of the sizes of those for
the current quarterly estimates for both real and current-dollar GDP.
One reason for this is that annual frequency estimates are not subject
to revisions to seasonal adjustment factors. The MARs decline with the
successive annual estimates of both current-dollar and real GDP. MARs
also decline steadily for most vintages of most components of GDP.
Exceptions are increases from the previous vintage for the second annual
vintage of real personal consumption expenditures, the second annual
vintage of both current-dollar and real imports, and the third annual
vintage of real exports. As with total GDP, the MARs of components are
much smaller than those for quarterly frequency estimates.
Definition changes as part of comprehensive NIPA revisions may
change both the levels and growth rates of GDP; for example, the
introduction of software as investment in the 1999 comprehensive
revision raised both the levels and the growth rates of GDP. The
introduction of R&D and artistic originals as investment in the 2013
comprehensive revision raised the levels of GDP but did little to change
the growth rates. The ratio of private fixed investment to GDP increased
by 0.016 to 0.019, but the pattern of movements of the ratio of private
fixed investment to GDP was essentially unchanged (chart 6).
[GRAPHIC 6 OMITTED]
Revisions to Quarterly Estimates of GDI
Advance estimates of GDI are not prepared, and since 1995, second
current quarterly estimates of GDI in the fourth quarter of each year
have not been prepared. These estimates of GDI are not prepared mainly
because of a lack of source data for estimating corporate profits.
Estimates are prepared for the other major components of GDI, such as
compensation of employees and proprietors' income. Starting with
the first quarter of 2002, a "fourth" vintage of estimates of
the previous quarter has been prepared using information from the Bureau
of Labor Statistics' quarterly census of employment and wages
(QCEW). It is used to revise the estimates of compensation of employees,
national income, and GDI. It is released at the time of the second
quarterly estimate of the following quarter.
The MAR to the latest estimates from the third estimates of GDI is
the same as that for the third estimate to the latest estimate of GDP
(table 6), at 1.2 percentage points. The MAR increases in the first
annual estimate then declines with the second and third annual
estimates. Only the MAR for the third annual estimates is lower than
that for the third current quarterly estimate. The MARs for the various
vintages of national income are all moderately and increasingly larger
than those for GDI with successive vintages, but they also decline
steadily with successive vintages. The MARs for compensation of
employees increase slightly with successive vintages of current
quarterly estimates, reaching 2.9 percentage points for the third
estimates, then decrease steadily to 1.3 percentage points for the third
annual vintage. MARs for the other components of GDI are larger to much
larger, depending on the component. They show little or no tendency to
decrease with successive vintages of estimates, but all decline with
successive vintages of annual estimates.
The MRs of GDI and its major components are much smaller than the
corresponding MARs and have both positive and negative signs. An
exception is corporate profits; the MRs are negative for all vintages,
and show no tendency to decline with successive vintages of estimates.
Standard deviations for the various components and vintages are all
larger to much larger than those for GDI and generally tend to diminish
in size with successive vintages (table 7). As with the components of
GDP, these MRs are not statistically significant.
Table 8 shows the MARs for GDI and most components from each
vintage to each subsequent vintage. Entries also include the fourth
current quarterly estimates for compensation, national income, and GDI;
these incorporate information from the QCEW. Fourth estimates of most
other components of GDI are not made, and the missing vintage pairs for
the fourth estimates are left blank. One must view the fourth estimates
with caution because there were no fourth estimates prior to the first
quarter of 2002, and the statistics are not fully comparable with the
rest of the entries, which are for 1993-2012. Thus, the fourth
estimates' MARs are not further discussed.
The patterns of MARs for successive vintages of GDI estimates and
vintages of standards are, at most, roughly similar to those found for
GDP. For example, there is a decline in the MARs for GDI from the third
estimates to the third annual estimates, but the MAR for the first
annual estimate to the latest is larger than for any other vintage of
estimates. The increases in MARs for the various vintages of GDI
estimates are not monotonic with increasingly later vintages of
estimates used as standards. The whole set of MARs for GDI tends to be
modestly larger than the comparable ones for GDP.
The various MARs for compensation of employees are the smallest for
any of the other components of GDI but larger than those of several
components of GDP. There are sharp drops in MARs for the fourth and
later vintages of estimates, relative to those of the earlier vintages
of estimates; the MAR to the latest estimates drops by 40 percent from
the third estimates to the fourth estimates. There are, however, the
same patterns of increasing MARs with increasingly later vintages of
target estimates. The patterns of MARs for national income are roughly
similar to those for GDI, but are larger for each vintage/standard pair
in the table.
The MARs for other components of GDI are generally larger than
those for compensation, and sometimes very much larger. At the extreme,
the MARs for the various estimates of corporate profits are several to
many times as large as the corresponding ones for compensation (table
6). The MARs for some GDI components are not shown, because definition
changes have caused excessive changes in the time series for the
components.
The patterns of increases and decreases described for
current-dollar GDP hold generally for GDI and components, but there are
deviations from them; that is, MARs sometimes decrease from vintage to
vintage used as the standards, or increase from vintage to vintage of
the estimates for any given standard. Nevertheless, there is a tendency
for increases across rows of table 8 and decreases down columns. In
general, the MARs for the various components and vintages tend to be
larger than those for the various GDP components.
Revisions to Annual Estimates of GDI
Summary statistics for revisions of annual frequency estimates of
real and current-dollar GDI and its major components are shown in table
9.12 The estimates of revisions are expressed in percent changes of the
annual estimates. As with the quarterly frequency estimates, the mean
revisions are small and both positive and negative. The early annual
estimates are mostly composed of third current quarterly estimates, and
their mean revisions are therefore similar to those of those estimates.
The mean absolute revisions for annual GDI and its components are
fractions of the sizes of those for the current quarterly estimates for
both real and current-dollar GDI. One reason for this is that annual
frequency estimates are not subject to revisions to seasonal adjustment
factors. The MARs decline for the successive annual estimates of GDI.
The declines also reflect the incorporation of later annual frequency
source data that is more complete or revised from earlier data. In
particular, the second annual estimates are the first vintage to
incorporate data from the Statistics of Income from the Internal Revenue
Service. MARs also decline steadily for most vintages of most components
of GDI.
The MARs for GDI decline 40 percent from the first to second annual
estimates, then increase modestly to the third annual estimate. In
contrast, the MARs for national income increase 41 percent from the
first to the second annual estimates before declining by more than half
to the third annual estimate. The MARs for the vintages of the
components show little tendency to decline with successive vintages and,
with the exception of proprietors' income are lower for the third
annual estimates than for the early annual estimates.
The overall size of GDI has changed noticeably due to the
recognition of intellectual property products as capital investment;
these changes were introduce to the accounts in two phases. Software was
introduced to the accounts in 1999, and research and development and
artistic originals were introduced in 2013. This may be seen in chart 7,
which shows the third current quarterly estimates of GDI, the estimates
in place immediately before the 2013 revision, and the latest estimates.
The dotted line links the third estimates (but is not a time series). It
is consistently lower than the later estimates until the second half of
2009. The solid line shows the latest estimates after the 2013
comprehensive revision; it is consistently above the dashed line of the
immediately previous estimates, by somewhat less than 3 percent in 1993
and somewhat less than 4 percent in 2012. Although other definition
revisions also occurred, the intellectual property additions to GDI
account for the bulk of the revisions to GDI (and GDP).
The effects of capitalizing intellectual property products in the
accounts have tended to somewhat lower the ratios of the components of
GDI to aggregate GDI simply by increasing the size of the denominator.
Among the components, there has been considerable smoothing of the
patterns of the ratios over time, but the basic patterns of increases
and decreases over time have been largely maintained. The three panels
of chart 8 show the ratios for compensation of employees, corporate
profits, and consumption of fixed capital. All three vintages of
estimates of the ratios for compensation show a general decline over
1993-2012, with temporary increases around the times of the 2001 and
2007-2009 recessions (panel 1). The shorter-term fluctuations, however,
are considerably more pronounced for the third estimates than for the
later two sets of estimates. The estimates for the most recent two
vintages--immediately before and after the comprehensive revision--have
very similar patterns over time, with differences in levels that reflect
the increased size of GDI.
The three vintages of estimates of corporate profits likewise show
very similar general movements, and again there is a good deal less
short-term volatility in the later two vintages (panel 2). The ratios of
consumption of fixed capital have two distinct revisions in patterns
(panel 3). Until about 2009, the third estimates were much more volatile
than either of the later sets of estimates. The smoother pattern
thereafter is similar to the later vintages. This reflects a revised
treatment of capital losses associated with natural disasters that was
introduced in the 2009 comprehensive revision; this included the
elimination of a spike in the third quarter of 2005 that was due to the
destruction associated with hurricane Katrina. The general upward
revision in level in the post-2013 comprehensive estimates is largely
due to the introduction of new types of intellectual property, as
discussed above. This general upward revision is the income-side twin to
the product- side revisions due to the new types of intellectual
property.
[GRAPHIC 7 OMITTED]
[GRAPHIC 8 OMITTED]
Comparing GDP and GDI revisions
At the time of the advance NIPA estimates, data are available to
allow BEA to calculate estimates of GDP and its major components. Data
are also available to allow BEA to calculate estimates of most GDI
components. For corporate profits, however, the data are from a too
small and unrepresentative
sample to permit the estimation. As a result, advance estimates of
aggregate GDI are not prepared.
Second estimates of GDP are published for each quarter, and for GDI
in the first three quarters of each calendar year. In the fourth quarter
of each year, profits reports are delayed sufficiently that no second
estimates of profits or GDI are made.
Third estimates of both GDP and GDI are published for each quarter.
At the time of the third estimates, about 67 percent of the source data
for GDP components are based on comprehensive data or direct indicators,
21 percent on indirect indicators, and 12 percent on trend-based
estimates. In comparison, about 14 percent of the source data for the
third estimates of GDI components is from direct sources, 56 percent
from indirect sources, and 30 percent from trend-based estimates. (13)
As a result, considerably more judgment goes into the construction of
the third estimates of GDI than into the third estimates of GDP.
Annual revision estimates are made each year for both GDP and GDI.
These use annual frequency data that are superior to the quarterly
frequency data that are available to support the current quarterly
estimates. Of particular importance, at the time of the second annual
revision estimates, data are available from the Internal Revenue Service
Statistics of Income to support the estimates of GDI. As a result, the
second annual estimates of GDI no longer contain trend-based estimates,
very little is based on indirect sources, and most direct indicators
have also been replaced by comprehensive data, which make up 94 percent
of the estimates.
As mentioned earlier, about every 5 years there is a comprehensive
revision to GDP. However, no corresponding benchmark estimates of GDI
are made because of a lack of information to produce them. As a result,
there are statistical discrepancies between GDP and GDI in the benchmark
years.
The reliability of the various vintages of quarterly estimates of
GDP and GDI is examined in table 10, which shows the MARs of each
vintage to the latest-available estimates of GDP and GDI. Smaller MARs
indicate greater reliability. Column 1 shows the MARs to the latest GDP
estimates for the successive vintages of GDP estimates. They get
steadily smaller with the successive vintages and are half as large for
the third annual vintage as they are for the advance vintage. Column 4
shows the MARs of the latest GDI estimates to the successive vintages of
GDI estimates. They have no tendency to get smaller with successive
vintages, although the third current quarterly vintage's MAR is
smaller than those of the three annual vintages. The MARs for third
current quarterly estimates of both GDP and GDI are the same size. The
MARs for the three annual vintages of estimates of GDP, however, are
noticeably smaller than those for the annual vintages of estimates of
GDI.
Column 2 of table 10 shows the MARs of the vintages of GDI
estimates to the latest estimates of GDP, and column 3 shows the MARs
for the vintages of GDP estimates to GDI. The MARs for successive
vintages of GDI to GDP increase. The MARs for successive vintages of GDP
to the latest estimates of GDI show no particular tendency to increase
or decrease. The MARs for GDP and GDI to their latest vintages are, by
definition, the same.
In addition to examining revisions to GDP and GDI in isolation, one
may also use ordinary least squares regressions to explain the latest
estimates. Table 11 shows this using the third current quarterly
estimates of both measures as explanatory variables. Although constant
terms are used in half of the regressions, they are never statistically
significant and have very little impact on the coefficients of other
variables. The first panel shows equations explaining the latest
estimates of GDP. The first equation explains the latest estimates of
GDP by the third current quarterly estimates of GDP. The regression
coefficient of the third estimates is just below one, it is highly
statistically significant (t-test statistics are in parentheses below
the estimated coefficients), and the R-square indicates that it explains
just under seven-tenths of the variance of the latest estimates.
Equation 3 uses the third estimate of GDI as the explanatory variable
for GDP and finds only marginally different results, with an R-square
just 0.017 lower than that of equation 1. This suggests that the third
estimate of GDI is only slightly less good as a forecast of the latest
estimate of GDP as the third estimate of GDP. Equation 5 uses the third
estimates of both GDP and GDI as explanatory variables. Both are
statistically significant, with the coefficient of GDP being about
one-quarter larger than that of GDI. The R-square of the equation is
modestly larger than those of the equations with only one of the
measures as an explanatory variable. This finding is consistent with the
hypothesis that the third estimates of GDI contain explanatory power
that is in addition to that of the contemporaneous estimates of GDP.
The second panel shows equations explaining the latest estimates of
GDI. Equation 7 estimates the latest estimate of GDI as a function of
the third estimate of GDI. As with GDP in equation 1, the coefficient of
GDI is close to one and highly significant. The R-square is only
slightly lower than that for equation 1. Thus, the third estimates of
GDI are about as accurate as the latest estimates of GDI, as the third
estimates of GDP are about as accurate as the latest estimates of GDP.
Equation 9 estimates of GDI as a function of the third estimates of
GDP. GDP is statistically highly significant, but the equation's
R-square is about one-sixth lower than those of either equation 7 or
equation 1. Thus, although GDP is a good estimator of the latest
estimates of GDI, it is less successful than GDI as an estimator of GDP.
Equation 11 estimates GDI as a function of the third estimates of
both GDP and GDI. GDP is not statistically significant, but the R-square
is only slightly below that for equation 7. The results of equations 9
and 11 together suggest that the third estimates of GDP do not contain
useful information about GDI once estimates of GDI are available.
A further evaluation of the usefulness of early estimates of GDP
and GDI as estimators of GDP can be made by calculating the MARs of the
latest estimates of GDP and GDI relative to estimators composed of the
weighted sums of the third estimates GDP and GDI.
Table 12 shows MARs to the latest estimates of GDP and GDI of the
third estimates of GDP, GDI, and three weighted sums. (14) The weights
are chosen for illustrative purposes, and are loosely based on the
relative coefficient sizes of the two measures reported as equations 5
and 6 of table 10. The first line shows MARs for the latest estimates of
GDP from the third estimates of GDP, GDI, two weighted sums of the
measures, and an unweighted average of the two measures. The lower MARs
of the weighted sums of GDP and GDI are reflective of the additional
information about GDP that is contained in GDI. The MARs reinforce the
regression finding that the third estimates of GDI have predictive power
for the latest estimates of GDP that is in addition to that of those of
the corresponding vintage of GDP. This, in turn, may reflect the notion
that the estimates of GDP incorporate judgments that are based on more
than just the available source data and that the estimates of GDP and
GDI are not made in isolation from one another. (15)
The second line shows MARs to the latest estimates of GDI from the
third estimates of GDP, GDI, and the same three weighted sums of the two
measures. The larger the weight of GDI, the smaller the resulting MARs
are. The weighted sums have smaller MARs than those of GDP. These
results, combined with the lower R-squares of equations for GDI that
include GDP, suggest that there is little or no useful information about
the latest estimates of GDI in the third estimates of GDP; that is, the
third estimate of GDP does not add to the information provided by the
third estimate of GDI.
Weighted averages of GDP and GDI have smaller variances than those
for either measure alone. This occurs because the unknown measurement
errors in the vintages are unlikely to be perfectly correlated. Table 13
shows the variances of the third and the latest vintages of estimates of
GDP, GDI, and selected weighted sums of the two measures. For both
vintages, the estimates of GDI have the largest variances (bold), and
weighted sums of the two, with weights favoring GDP, have the smallest
variances (shaded). Line 3 shows the MARs of the measures and of
weighted sums of the third to the latest estimates. The smallest MARs
(shaded) are for weighted sums, with weights ranging from half to
two-thirds for GDP, and the rest for GDI. All of the weighted sums have
smaller MARs than that for GDP, but GDI (bold) has the largest MAR of
all those shown. If both GDP and GDI are each interpreted as the sums of
the unobserved true economic activity and of measurement errors, it is
possible to infer that the weighted averages are more accurate measures
of activity than either GDP or GDI alone. This occurs because some of
the measurement errors are averaged out, reducing subsequent revisions
in the weighted averages.
Another way of comparing GDP and GDI is to look at their
performance near turning points. This is done in table 14, which shows
MARs from the third to the latest estimates of GDP, GDI, and selected
weighted averages of GDP and GDI in the most recent six recessions. The
first line shows the MARs for all of the quarters from those immediately
before to those immediately after the turning points of the recessions.
All of the weighted averages have MARs of 0.2 percentage point or more
below those of GDP and GDI alone, and the 50-50 weighted average has the
lowest MAR, 0.3 percentage point less than that of GDP and 0.4
percentage point less than that of GDI. This is consistent with the
suggestion that weighted averages of GDP and GDI are superior estimators
of the path of economic activity around turning points at the time of
the third estimates.
The remaining lines of the table disaggregate the MARs by the
quarters of the turning points and those immediately before and after
peaks and troughs. The weighted averages generally have lower MARs, with
those before and those at turning points generally favoring greater
weights for GDP. Weights favoring GDI do best in quarters immediately
following troughs. The largest MARs before and at turning points are for
GDI, but the largest MARs after turning points are for GDP.
Based on the foregoing, measures of GDP as well as those of GDI
should yield improved understanding of the recent course of the economy.
The better reliability observed using weighted averages of GDP and GDI
will not, however, improve understanding of the detailed workings of the
economy because there is no obvious way of distributing the averaging
among the major components of GDP and GDI. Thus, averages can only
provide supplemental summary information about the recent course of the
aggregate economy.
Summary and Conclusions
The results of this review are fully consistent with those of
previous BEA studies. The finding that GDI has additional information
about GDP is stronger in this study than was previously reported. In
summary:
* The estimates of GDP and GDI are accurate; the MARs for early
vintages of both measures are somewhat above 1 percentage point.
* The MRs for both GDP and GDI are near zero and reflect the
improvements in measures of economic activity and the expansions of the
definition of what is included in economic activity that have been
introduced in the comprehensive NIPA revisions that have adapted the
NIPAs to an evolving economy. Nonzero values are not indications of
bias.
* The quarterly estimates are accurate indicators of whether the
economy is growing at rates above, near, or below the long-term trend.
* The MARs for the annual estimates of GDP, GDI, and their major
components are less than half of those for the current quarterly
estimates; they decline steadily in size from the early annual estimates
to the third annual estimates.
* The fourth estimates of GDI, national income, and compensation of
employees, which are published 5 months after the end of each quarter,
reduce the MAR for compensation, but not the MARs for GDI or national
income.
* GDI provides additional and reliable information about the course
of true economic activity, which is never observed.
Although not discussed here, the finding of the previous study that
"revisions in the major components of GDP and GDI have preserved
the trends found in the early estimates" still holds.
Accuracy, Reliability, and Uncertainty
How accurate are the gross domestic product (GDP) estimates? This
is a fairly common question, often inspired by the release of revised
estimates with the passage of time. From a statistical perspective, when
one speaks of the accuracy of an estimate, one is usually referring to
the difference between the estimate and some "true" value. For
example, the difference between a sample mean and a population mean that
is as taken as the true value.
The fact that BEA provides a sequence over time of estimates for a
given quarter--which are referred to as vintage estimates--implies that
users of the data should understand that there is some uncertainty
surrounding the estimates, especially the early estimates. Because the
true value of GDP can never be observed, its accuracy cannot be
assessed.
However, over time, BEA acquires more and better information about
GDP and schedules several revisions to the advance estimate of quarterly
GDP. It is therefore able to revise its estimate of GDP. These revisions
are believed to be a better estimates of the true value of GDP. The
revised estimates can be viewed as repeated estimates of the aggregate
economic activity for that quarter. By assessing the performance of
these revised estimates, BEA can assess the reliability of estimates.
The reliability of the GDP estimates, as assessed by studying the
revision patterns, seeks to answer these questions: how similar are the
repeated estimates of GDP for a given quarter? And do they tell the same
story? Quantitatively, reliability is assessed by measuring the revision
magnitudes and the corresponding means and standard deviations.
Qualitatively, it is assessed by looking at such measures as the
frequency of directional changes in the estimates. These measures are
explained more fully in the text.
By using the means and standard deviations for revisions between
different vintages of the GDP estimates given in tables 15 and 16 (page
22), a user can construct confidence intervals for an estimate. For
example, consider the difference between the advance and second
quarterly estimate of real GDP for 1993-2012. Table 1 shows that the
mean difference between these two estimates is 0.10 percentage point.
(Note, however, that this mean revision is statistically
indistinguishable from zero). Using the standard deviation of the
revision between the advance and second estimates, 0.654 percentage
point, and assuming a normal distribution, one can construct a 90
percent confidence interval for the revisions.
[bar.x] [+ or -] [z.sub.a/2] s = 0.10 [+ or -] 1.65 x 0.654 = 0.10
[+ or -] 1.08,
where x bar is the mean, [z.sub.a/2] defines a 90 percent
confidence interval (1.65), and s is the standard deviation. The same
procedure could be used for all revisions between vintages. So what can
be said about the uncertainty of the advance estimates? Using a 90
percent confidence criterion, the revision between the advance and
second estimates is in the interval (-0.98 to 1.18). One could use this
information to estimate the second estimate given the advance estimate.
For example, the advance estimate for the first quarter of 2013 was 2.5
percent at an annual rate. Using the above equation, one could say with
90 percent confidence that the second estimate would be between 1.52
percent and 3.68 percent. In fact, the second estimate was 2.4 percent.
Given that BEA routinely revises its estimates during the course of
a year, one might ask why BEA produces point estimates of GDP instead of
interval estimates. In his review of the then new estimates of national
income, Kuznets (1948) remarked, "The very fact that the estimates
are cast in the form of unique series and not of ranges, is itself an
invitation to treat them as firm results and tend to discourage
questioning whether a total of x billion might just as well read x + a
or x - a." Although interval estimates would inform users of the
uncertainty surrounding the estimates, most users prefer point
estimates, and so they are featured. However, BEA provides the
information that enables an interested user to construct their own
interval estimate.
Vintages and Timing of Revisions
The Bureau of Economic Analysis prepares quarterly and annual
estimates of gross domestic product (GDP) and gross domestic income
(GDI). It prepares three current quarterly vintages of GDP
estimates--advance, second, and third estimates. The advance estimates
for a quarter are released about a month after the quarter ends. The
second estimates for the quarter are released about 2 months after the
quarter ends. And the third estimates are released about 3 months after
the quarter ends. In addition, as part of the annual revision of the
national income and product accounts (NIPAs) that are released in late
July of each year, the quarterly estimates for the 3 preceding
years--and possibly more--are revised.
For GDI, BEA prepares a fourth vintage of quarterly estimates.
These revised estimates, which incorporate data from the quarterly
census of employment and wages, are released with the second estimates
of GDP for a quarter. These revised estimates are available beginning
with the estimates for the first quarter of 2002.
BEA initially prepares four vintages of annual estimates--early
annual, first annual, second annual, and third annual estimates. For GDP
and GDI, the early estimates are the sum of the third quarterly
estimates for that year. The estimates of GDP are released in March with
the third estimates for the fourth quarter of that year. The estimates
of GDI are released with the release of the fourth estimates for the
fourth quarter of the year in late May. In most years, the quarterly
estimates for the first quarter of the previous year are from the first
current annual estimate released the previous summer. The current annual
estimates for the 3 preceding years are revised as part of the annual
NIPA revision. After the third annual revision of the estimates for a
year is released, these estimates are generally not revised or released
again until the next comprehensive revision. Occasionally, however,
revisions extend further back in time.
Annual NIPA revision estimates are superseded by comprehensive NIPA
revisions, which occur about every 5 years. These revisions incorporate
changes in definitions, in classifications, and in statistical
methodology. The most recent comprehensive revision was released in late
July 2013. It presented revised annual estimates for 1929-2012 and
revised quarterly estimates for 1947-2012. The latest available
estimates for 1990-2012 in this study are the comprehensive revision
estimates of 2013.
Mean Revisions, Mean Absolute Revisions, and Standard Deviations
By convention, revisions are calculated as the later vintage
estimates less the earlier vintage estimates; that is, for any time t,
the revision is
[R.sub.t] = [L.sub.t] - [E.sub.t]
where L is the percent change in the later vintage quarterly or
annual estimates, and E is the percent change in the earlier vintage
estimates. Percent changes in quarterly estimates are at annual rates,
this corresponds to the convention generally used for the estimates.
The mean revision is the average of the revisions in the sample
period.
MR = [[SIGMA].sub.t] [R.sub.t] /n, t = 1, ..., n
The revisions can be positive or negative, so they may be
offsetting. As a result, it is also useful to look at the mean absolute
revision, which is the average of the absolute revisions in the sample
period.
MAR = [[SIGMA].sub.t] [absolute value of [R.sub.t]]/n, t = 1, ...,
n
For some purposes, it is also useful to calculate the standard
deviation of the revisions. The standard deviation is the square root of
the variance of the revisions. In turn, the variance is the average of
the square of the deviation of the revisions about their mean.
SD(R) = [Var(R).sup.1/2]
and
Var (R)= [[SIGMA].sub.t] ([[R.sub.t] - MR).sup.2]/n, t = 1, ... , n
The mean absolute revision and the standard deviations are
complementary measure of the distribution of the revision around their
mean.
Revisions to Monthly Estimates of Price Indexes for Personal
Consumption Expenditures
The reliability measures used to analyze prices are the mean
revisions (MRs) and mean absolute revisions (MARs) for the current
monthly estimates to the later vintage estimates of the percent changes
in two indexes: (1) the price index for personal consumption
expenditures (PCE) and (2) the index for personal consumption
expenditures excluding food and energy (PCEX). The MRs and MARs are
calculated using monthly data for 1996-2013 for PCE and monthly data for
2000-2013 for PCEX (see the table). Both MRs and MARs increase for any
vintage of estimates as progressively later vintages of estimates are
used as standards. The revisions are measured in terms of annualized
percent changes; for example, the first vintage of estimates to the
latest revision estimates of PCE has a MAR of 0.92 percent, which
corresponds to a month-to-month rate MAR of 0.07 percent.
The MRs for PCE and PCEX follow similar patterns. All have positive
values. (1) Both have progressively larger MRs for the earlier vintages
as progressively later vintages are used as standards. The MRs for PCE
range from 0.04 for the third estimate to the latest estimate to 0.18
for the first estimate to the latest estimate. A difference in patterns
is that the smallest MR for PCE is for the third estimate to the latest
estimate, and the smallest MR for PCEX is for the second estimate to the
third estimate. In addition, for PCE, the MR for first estimate to the
third estimate is larger than the MR for the second estimate to the
latest estimate, with values of 0.14 and 0.10, respectively. For PCEX,
the MR for the first estimate to third estimate is smaller than the MR
for the second estimate to the latest estimate, with values of 0.10 and
0.12, respectively.
The MARs for PCE and PCEX also follow similar patterns. Both have
progressively larger MARs for the earlier vintages as progressively
later vintages are used as standards. The MARs for PCE range from 0.36
for the second estimate to third estimate to 0.92 for the first estimate
to the latest estimate. The MARs for PCEX range from 0.23 for the second
estimate to third estimate to 0.66 for the first estimate to the latest
estimate.
(1.) The current quarterly estimates of PCE prices were also found
to be positive and not statistically significant; see Fixler,
Greenaway-McGrevy, and Grimm (2011), 27.
Average Revisions to Successive Vintages of Monthly Estimates of
Annualized Percent Changes in Price Indexes for Personal Consumption
Expenditures (PCE) and PCE Excluding Food and Energy
Vintage of revision
used as standard (1)
Second Third Latest
estimate estimate estimate
Mean revision
PCE
First monthly estimate 0.08 0.14 0.18
Second monthly estimate ... 0.06 0.10
Third monthly estimate ... ... 0.04
PCE excluding food and energy
First monthly estimate 0.07 0.10 0.19
Second monthly estimate ... 0.03 0.12
Third monthly estimate ... ... 0.09
Mean absolute revision
PCE
First monthly estimate 0.40 0.50 0.92
Second monthly estimate ... 0.36 0.86
Third monthly estimate ... ... 0.80
PCE excluding food and energy
First monthly estimate 0.25 0.35 0.66
Second monthly estimate ... 0.23 0.57
Third monthly estimate ... ... 0.53
(1.) First is the first monthly estimate, second is the
second monthly estimate, and third is the third monthly
estimate. Latest is the latest estimate.
References
Fixler, Dennis J. 2004. "Revisions to GDP Estimates in the
United States." Paper presented at the OECD Workshop on Revisions,
Paris. October 7.
Fixler, Dennis J., and Bruce T. Grimm. 2002. "Reliability of
GDP and Related NIPA Estimates." Survey of Current Business 82
(February): 9-27.
Fixler, Dennis J., and Bruce T. Grimm. 2003. "Revisions,
Rationality, and Turning Points in GDP." Paper presented at the
American Economic Associations meeting, Washington, DC, January 3-5. BEA
Working Paper WP-2003-01.
Fixler, Dennis J., and Bruce T. Grimm. 2005. "Reliability of
the NIPA Estimates of U.S. Economic Activity." Survey of Current
Business 85 (February): 9-19.
Fixler, Dennis J., and Bruce T. Grimm. 2006. "GDP Estimates:
Rationality Tests and Turning Point Performance." Journal of
Productivity 25 (June): 213-229.
Fixler, Dennis J., and Bruce T. Grimm. 2008. "The Reliability
of the GDP and the GDI Estimates." SURVEY of Current Business 88
(February): 16-32.
Fixler, Dennis J., Ryan Greenaway-McGrevy, and Bruce T. Grimm.
2011. "Revisions to GDP, GDI, and Their Major Components."
Survey of Current Business 91 (July): 9-31.
Grimm, Bruce T. 2005. "Alternative Measures of U.S. Economic
Activity in Business Cycles and Business Cycle Dating" BEA Working
Paper WP-2005-05.
Grimm, Bruce T., and Robert P. Parker. 1998 "Reliability of
the Quarterly and Annual Estimates of GDP and Gross Domestic
income." SURVEY OF CURRENT BUSINESS 78 (December): 12-21.
Grimm, Bruce T., and Teresa L. Weadock. 2006. "Gross Domestic
Product: Revisions and Source Data." Survey of Current Business 86
(February): 11-15.
Holdren, Alyssa E. 2014. "Gross Domestic Product and Gross
Domestic Income: Revisions and Source Data." Survey of Current
Business 94 (June): 1-11.
Holdren, Alyssa E., and Bruce T. Grimm. 2008. "Gross Domestic
Income: Revisions and Source Data." Survey of Current Business 88
(December): 14-20.
Kuznets, Simon. 1948. National Income: A New Version." Review
of Economics and Statistics 30 (August): 151-179.
Yorgason, Daniel R., and Sarah P. Scott. 2012. "An Analysis of
Revisions to BEA's International Economic Accounts." Survey of
Current Business 92 (November): 76-102.
Young, Alan H., "Reliability and Accuracy of Quarterly GDP
Estimates: A Review." In The New System of Economic Accounts,
edited by John W. Kendrick, 423-449. Norwell, MA: Kluwer Academic
Publishers.
(1.) The difference between GDP and GDI is allocated to the income
side of the account, and is known as the statistical discrepancy.
(2.) Economic censuses are conducted by the Census Bureau every 5
years and provide detailed product information on the goods and services
produced by each industry in the United States. These data are the basis
for BEA's benchmark input-output accounts, which are used to set
the benchmarked level of GDP.
(3.) The end point of this study is 2012 because at the time this
study was conducted, there were no annual revision estimates available
for 2013. The 1983-92 period, which was included in several earlier
studies, has been dropped in order to emphasize more recent issues
affecting reliability. In addition, this study was conducted with data
that does not reflect the 2014 annual revision of the NIPAs.
(4.) For more information on the source data underlying GDP and GDI
estimates, see Holdren (2014).
(5.) No major measure of economic activity captures all of the
cyclical peaks and troughs of all of the postwar recessions. This
applies to GDP and GDI as well as the four monthly frequency measures
emphasized by the Business Cycle Dating Committee of the National Bureau
of Economic Research in determining peaks and troughs. See Grimm (2005).
(6.) This is somewhat different from the results of previous BEA
studies, which found no particular tendency of MARs to increase or
decrease with successive estimates.
(7.) This finding of declines with successive vintages is the
expected finding: the absence of this finding in previous studies was
puzzling because the later vintages, with more and better source data,
would be expected to be closer to the latest estimates.
(8.) For this study, we combine the NIPA components
"equipment" and "intellectual property products." In
previous studies, we used the (now outdated) NIPA component
"equipment and software."
(9.) BEA has published an extensive analysis of the international
transactions accounts (Yorgason and Scott 2012).
(10.) A Jarque-Bera statistic of 1.53 for the group of revisions
indicates that the hypothesis of normality cannot be rejected, with a
p-value of 0.47.
(11.) The eight-quarter moving average is an adaptation of the
method used by Blanchard and Simon (2001) in their study of the
volatility of GDP. The value for 1984:IV is the variance of GDP for
1983:I-1984:IV, the value for 1985:I is the variance for 1983:II-1985:I,
and so on.
(12.) Early annual estimates of GDI are available in May of the
following year.
(13.) For a discussion of the source data available to estimate GDP
and GDI, see Grimm and Weadock (2006), Holdren and Grimm (2008), and
Holdren (2014).
(14.) The weighted sums are of growth rates of GDP and GDI. The
weighted sums of levels would yield somewhat different growth rates.
(15.) Allan Young reported that estimates of GDP made a month
earlier than the advance estimates had a MAR of similar size to those of
the advance estimates (Young 1996). This finding is also reflective of
the quality of judgmental inputs into GDP estimates in the presence of
progressively less accurate source data.
Table 1. Revisions, Current Quarterly Estimates to the Latest
Estimates, Changes In Gross Domestic Product and Its
Major Components, 1993-2012
[Percentage points]
Mean revision Standard deviation
Current Current
Vintage dollar Real dollar Real
Gross domestic product
Advance 0.12 -0.06 1.69 1.61
Second -0.01 -0.17 1.56 1.53
Third -0.10 -0.19 1.53 1.49
Personal consumption
expenditures
Advance -0.06 -0.07 1.27 1.19
Second -0.09 -0.10 1.13 1.05
Third -0.10 -0.10 1.15 1.05
Durable goods
Advance -0.12 -0.23 5.50 5.53
Second -0.06 -0.15 5.35 5.38
Third -0.15 0.21 5.28 6.00
Nondurable goods
Advance -0.21 0.01 2.59 2.31
Second -0.33 -0.02 2.39 2.32
Third -0.33 -0.10 2.45 2.25
Services
Advance -0.05 -0.12 1.14 0.99
Second -0.04 -0.18 1.09 1.00
Third -0.05 -0.14 1.05 1.01
Gross private domestic
investment
Advance 0.59 -0.66 6.63 6.97
Second -0.11 -0.99 7.15 7.20
Third -0.24 -1.03 6.88 6.90
Fixed investment
Advance 0.70 -0.34 3.47 3.54
Second 0.06 -0.84 3.44 3.53
Third -0.01 -0.92 3.53 3.50
Nonresidential
Advance 0.08 -0.60 4.36 4.39
Second -0.61 -1.18 4.48 4.46
Third -0.76 -1.22 4.47 4.43
Structures
Advance 2.85 1.39 10.81 9.17
Second 1.47 0.16 10.62 9.46
Third 0.86 -0.11 10.17 7.92
Equipment and
intellectual
property products
Advance -0.55 -1.12 6.10 5.72
Second -1.10 -1.58 6.44 5.84
Third -1.36 -1.40 6.62 6.08
Residential
Advance 0.67 0.06 6.02 4.88
Second 0.38 -0.22 5.29 4.52
Third 0.09 -0.44 5.34 4.47
Change in private
inventories (1)
Net exports of goods and
services (1)
Exports
Advance 4.98 1.34 12.09 4.64
Second 1.29 0.31 5.23 4.00
Third 0.19 0.24 4.28 3.81
Imports
Advance 0.71 0.77 5.53 5.09
Second 0.14 -0.18 4.62 3.87
Third 0.11 -0.06 4.13 3.78
Government consumption
expenditures and gross
investment
Advance 0.30 0.03 2.02 1.88
Second 0.08 -0.09 2.06 1.90
Third 0.07 -0.02 2.07 1.85
Federal
Advance 0.10 -0.06 3.40 3.29
Second 0.00 0.01 3.58 3.47
Third 0.03 0.12 3.51 3.39
Defense
Advance -0.14 -0.28 4.61 4.56
Second -0.12 -0.10 4.46 4.33
Third -0.05 -0.01 4.39 4.24
Nondefense
Advance 0.45 0.20 5.79 5.48
Second 0.00 -0.05 6.14 5.81
Third -0.05 -0.11 6.17 5.89
State and local
Advance 0.45 0.11 2.67 2.08
Second 0.12 -0.11 2.52 2.00
Third 0.13 -0.18 2.55 2.14
Mean absolute revision
Current
Vintage dollar Real
Gross domestic product
Advance 1.35 1.29
Second 1.28 1.25
Third 1.21 1.20
Personal consumption
expenditures
Advance 1.00 0.95
Second 0.89 0.82
Third 0.87 0.82
Durable goods
Advance 4.31 4.40
Second 4.33 4.42
Third 4.24 4.65
Nondurable goods
Advance 1.96 1.68
Second 1.82 1.69
Third 1.83 1.61
Services
Advance 0.86 0.81
Second 0.84 0.79
Third 0.80 0.76
Gross private domestic
investment
Advance 5.16 5.27
Second 5.82 5.87
Third 5.61 5.63
Fixed investment
Advance 2.71 2.79
Second 2.68 2.85
Third 2.74 2.92
Nonresidential
Advance 3.39 3.51
Second 3.61 3.72
Third 3.49 3.75
Structures
Advance 8.32 7.10
Second 7.67 6.96
Third 7.13 5.89
Equipment and
intellectual
property products
Advance 4.94 4.63
Second 5.23 4.83
Third 5.44 4.98
Residential
Advance 4.49 3.81
Second 3.89 3.48
Third 3.61 3.45
Change in private
inventories (1)
Net exports of goods and
services (1)
Exports
Advance 4.14 3.69
Second 3.29 3.11
Third 2.69 3.00
Imports
Advance 4.37 4.16
Second 3.28 3.00
Third 3.19 2.91
Government consumption
expenditures and gross
investment
Advance 1.65 1.52
Second 1.58 1.48
Third 1.59 1.45
Federal
Advance 2.76 2.60
Second 2.75 2.76
Third 2.72 2.71
Defense
Advance 3.55 3.39
Second 3.43 3.18
Third 3.33 3.04
Nondefense
Advance 4.22 4.16
Second 4.43 4.38
Third 4.56 4.61
State and local
Advance 2.10 1.59
Second 1.85 1.61
Third 1.89 1.64
(1.) Percent changes cannot be calculated because
of the presence of both positive and negative
values.
Table 2. Mean Absolute Revisions, Changes In Current-
Dollar GDP and Its Components, 1993-2012
[Percentage points]
Vintage of revision
used as standard
First
Vintage Second Third annual
Gross domestic product
Advance 0.56 0.63 0.93
Second ... 0.28 0.79
Third ... ... 0.72
First annual ... ... ...
Second annual ... ... ...
Third annual ... ... ...
Personal consumption
expenditures
Advance 0.32 0.38 0.72
Second ... 0.21 0.64
Third ... ... 0.60
First annual ... ... ...
Second annual ... ... ...
Third annual ... ... ...
Durable goods
Advance 1.10 1.50 2.86
Second ... 0.37 2.68
Third ... ... 2.96
First annual ... ... ...
Second annual ... ... ...
Third annual ... ... ...
Nondurable goods
Advance 0.67 0.65 1.14
Second ... 0.22 0.92
Third ... ... 0.87
First annual ... ... ...
Second annual ... ... ...
Third annual ... ... ...
Services
Advance 0.28 0.42 0.68
Second ... 0.32 0.72
Third ... ... 0.64
First annual ... ... ...
Second annual ... ... ...
Third annual ... ... ...
Gross private
domestic investment
Advance 2.90 3.11 4.89
Second ... 1.04 4.45
Third ... ... 4.22
First annual ... ... ...
Second annual ... ... ...
Third annual ... ... ...
Fixed investment
Advance 1.29 1.68 2.55
Second ... 0.82 2.18
Third ... ... 2.21
First annual ... ... ...
Second annual ... ... ...
Third annual ... ... ...
Fixed nonresidential
investment
Advance 1.68 1.90 3.11
Second ... 0.85 2.71
Third ... ... 2.68
First annual ... ... ...
Second annual ... ... ...
Third annual ... ... ...
Nonresidential structures
investment
Advance 3.75 4.35 7.33
Second ... 1.99 6.91
Third ... ... 6.00
First annual ... ... ...
Second annual ... ... ...
Third annual ... ... ...
Equipment and intellectual
property investment
Advance 1.68 2.02 3.78
Second ... 0.88 3.44
Third ... ... 3.28
First annual ... ... ...
Second annual ... ... ...
Third annual ... ... ...
Residential investment
Advance 1.56 1.94 3.67
Second ... 0.73 2.85
Third ... ... 2.64
First annual ... ... ...
Second annual ... ... ...
Third annual ... ... ...
Exports
Advance 1.06 2.67 2.42
Second ... 2.56 3.64
Third ... ... 2.71
First annual ... ... ...
Second annual ... ... ...
Third annual ... ... ...
Imports
Advance 2.72 2.74 3.59
Second ... 1.19 2.57
Third ... ... 2.28
First annual ... ... ...
Second annual ... ... ...
Third annual ... ... ...
Government consumption
expenditures and
gross investment
Advance 0.72 0.79 1.33
Second ... 0.27 1.20
Third ... ... 1.19
First annual ... ... ...
Second annual ... ... ...
Third annual ... ... ...
Federal government
Advance 1.02 0.95 2.66
Second ... 0.31 2.54
Third ... ... 2.60
First annual ... ... ...
Second annual ... ... ...
Third annual ... ... ...
Federal defense
Advance 0.96 1.03 2.77
Second ... 0.34 2.70
Third ... ... 2.80
First annual ... ... ...
Second annual ... ... ...
Third annual ... ... ...
Federal nondefense
Advance 1.95 1.97 4.80
Second ... 0.46 4.78
Third ... ... 4.75
First annual ... ... ...
Second annual ... ... ...
Third annual ... ... ...
State and local government
Advance 0.76 0.95 1.32
Second ... 0.34 1.04
Third ... ... 1.04
First annual ... ... ...
Second annual ... ... ...
Third annual ... ... ...
Vintage of revision
used as standard
Second Third
Vintage annual annual Latest
Gross domestic product
Advance 1.11 1.21 1.35
Second 0.98 1.13 1.28
Third 0.94 1.08 1.21
First annual 0.64 0.88 0.99
Second annual ... 0.55 0.89
Third annual ... ... 0.68
Personal consumption
expenditures
Advance 0.94 1.03 1.00
Second 0.80 0.92 0.89
Third 0.81 0.94 0.87
First annual 0.59 0.74 0.77
Second annual ... 0.49 0.72
Third annual ... ... 0.66
Durable goods
Advance 3.37 3.75 4.31
Second 3.16 3.83 4.33
Third 3.22 3.71 4.24
First annual 2.32 2.88 3.23
Second annual ... 2.05 2.66
Third annual ... ... 1.87
Nondurable goods
Advance 1.59 1.86 1.96
Second 1.45 1.76 1.82
Third 1.45 1.76 1.82
First annual 1.08 1.42 1.73
Second annual ... 0.86 1.29
Third annual ... ... 1.14
Services
Advance 0.88 1.00 0.86
Second 0.83 0.96 0.84
Third 0.82 0.89 0.80
First annual 0.52 0.74 0.68
Second annual ... 0.53 0.67
Third annual ... ... 0.61
Gross private
domestic investment
Advance 5.50 5.43 5.15
Second 5.36 5.74 5.82
Third 5.07 5.51 5.61
First annual 3.10 3.60 4.26
Second annual ... 2.92 3.71
Third annual ... ... 2.91
Fixed investment
Advance 2.74 2.85 2.71
Second 2.52 2.73 2.68
Third 2.60 2.78 2.74
First annual 1.93 2.38 2.42
Second annual ... 1.49 1.89
Third annual ... ... 1.80
Fixed nonresidential
investment
Advance 3.68 4.08 3.39
Second 3.53 3.97 3.61
Third 3.51 3.95 3.49
First annual 3.09 3.66 2.94
Second annual ... 2.50 2.37
Third annual ... ... 2.32
Nonresidential structures
investment
Advance 8.28 8.30 8.32
Second 7.80 8.00 7.67
Third 7.16 7.73 7.13
First annual 3.95 4.70 4.81
Second annual ... 2.82 3.67
Third annual ... ... 3.38
Equipment and intellectual
property investment
Advance 4.98 4.07 4.94
Second 5.24 4.42 5.23
Third 5.17 4.70 5.44
First annual 4.05 3.65 4.41
Second annual ... 2.90 3.90
Third annual ... ... 2.53
Residential investment
Advance 4.02 4.14 4.49
Second 3.38 3.50 3.89
Third 3.11 3.23 3.81
First annual 2.04 2.47 3.35
Second annual ... 1.29 2.52
Third annual ... ... 1.99
Exports
Advance 1.97 1.16 4.14
Second 4.27 4.55 3.29
Third 3.02 3.50 3.20
First annual 3.06 3.47 2.69
Second annual ... 2.91 1.93
Third annual ... ... 1.16
Imports
Advance 3.97 4.03 4.37
Second 2.86 2.74 3.28
Third 2.62 2.47 3.19
First annual 1.82 2.01 2.58
Second annual ... 1.12 1.90
Third annual ... ... 1.39
Government consumption
expenditures and
gross investment
Advance 1.36 1.57 1.65
Second 1.30 1.62 1.58
Third 1.27 1.58 1.59
First annual 0.72 1.26 1.37
Second annual ... 0.98 1.22
Third annual ... ... 1.10
Federal government
Advance 2.79 3.32 2.76
Second 2.74 3.37 2.75
Third 2.84 3.43 2.72
First annual 1.49 2.61 2.38
Second annual ... 1.67 2.02
Third annual ... ... 2.20
Federal defense
Advance 3.45 3.69 3.55
Second 3.37 3.69 3.43
Third 3.40 3.72 3.33
First annual 1.79 2.41 2.89
Second annual ... 1.68 2.77
Third annual ... ... 2.44
Federal nondefense
Advance 4.81 5.36 4.22
Second 4.76 5.19 4.43
Third 4.70 5.27 4.56
First annual 3.11 4.40 4.05
Second annual ... 3.08 3.65
Third annual ... ... 3.95
State and local government
Advance 1.50 1.79 2.10
Second 1.33 1.55 1.85
Third 1.29 1.58 1.89
First annual 0.81 1.31 1.58
Second annual ... 1.04 1.40
Third annual ... ... 1.24
Table 3. Mean Absolute Revisions, Changes in Real GDP
and Its Major Components, 1993-2012
[Percentage points]
Vintage of revision
used as standard
First
Vintage Second Third annual
Gross domestic product
Advance 0.52 0.56 0.94
Second ... 0.21 0.81
Third ... ... 0.79
First annual ... ... ...
Second annual ... ... ...
Third annual ... ... ...
Personal consumption
expenditures
Advance 0.31 0.38 0.78
Second ... 0.17 0.71
Third ... ... 0.67
First annual ... ... ...
Second annual ... ... ...
Third annual ... ... ...
Durable goods
Advance 1.14 1.66 3.10
Second ... 0.88 2.72
Third ... ... 3.16
First annual ... ... ...
Second annual ... ... ...
Third annual ... ... ...
Nondurable goods
Advance 0.72 0.64 1.10
Second ... 0.28 1.06
Third ... ... 0.93
First annual ... ... ...
Second annual ... ... ...
Third annual ... ... ...
Services
Advance 0.29 0.43 0.71
Second ... 0.31 0.75
Third ... ... 0.70
First annual ... ... ...
Second annual ... ... ...
Third annual ... ... ...
Gross private
domestic investment
Advance 2.79 2.80 3.90
Second ... 1.06 2.39
Third ... ... 4.45
First annual ... ... ...
Second annual ... ... ...
Third annual ... ... ...
Fixed investment
Advance 1.19 1.54 2.48
Second ... 0.72 2.14
Third ... ... 2.07
First annual ... ... ...
Second annual ... ... ...
Third annual ... ... ...
Fixed nonresidential
investment
Advance 1.61 1.95 3.16
Second ... 0.87 2.69
Third ... ... 2.64
First annual ... ... ...
Second annual ... ... ...
Third annual ... ... ...
Nonresidential structures
investment
Advance 3.75 4.28 6.57
Second ... 2.38 6.69
Third ... ... 5.52
First annual ... ... ...
Second annual ... ... ...
Third annual ... ... ...
Equipment and intellectual
property investment
Advance 1.76 2.27 3.64
Second ... 1.11 3.10
Third ... ... 3.17
First annual ... ... ...
Second annual ... ... ...
Third annual ... ... ...
Residential investment
Advance 1.43 1.70 3.07
Second ... 0.79 2.60
Third ... ... 2.44
First annual ... ... ...
Second annual ... ... ...
Third annual ... ... ...
Exports
Advance 2.11 2.51 2.97
Second ... 1.01 2.22
Third ... ... 2.27
First annual ... ... ...
Second annual ... ... ...
Third annual ... ... ...
Imports
Advance 2.68 2.90 4.04
Second ... 0.96 2.60
Third ... ... 2.48
First annual ... ... ...
Second annual ... ... ...
Third annual ... ... ...
Government consumption
expenditures and
gross investment
Advance 0.68 0.73 1.09
Second ... 0.25 0.92
Third ... ... 0.87
First annual ... ... ...
Second annual ... ... ...
Third annual ... .... ...
Federal government
Advance 1.11 1.02 2.27
Second ... 0.38 2.10
Third ... ... 2.15
First annual ... ... ...
Second annual ... ... ...
Third annual ... ... ...
Federal defense
Advance 1.01 1.07 2.59
Second ... 0.39 2.25
Third ... ... 2.21
First annual ... ... ...
Second annual ... ... ...
Third annual ... ... ...
Federal nondefense ... ... ...
Advance 1.90 1.98 4.48
Second ... 0.62 4.36
Third ... ... 4.46
First annual ... ... ...
Second annual ... ... ...
Third annual ... ... ...
State and local government
Advance 0.68 0.95 1.08
Second ... 0.45 0.89
Third ... ... 0.87
First annual ... ... ...
Second annual ... ... ...
Third annual ... ... ...
Vintage of revision
used as standard
Second Third
Vintage annual annual Latest
Gross domestic product
Advance 1.09 1.22 1.29
Second 1.09 1.19 1.25
Third 1.05 1.12 1.20
First annual 0.73 0.91 1.00
Second annual ... 0.69 1.01
Third annual ... ... 0.62
Personal consumption
expenditures
Advance 0.85 0.94 0.95
Second 0.78 0.82 0.82
Third 0.78 0.85 0.82
First annual 0.52 0.62 0.74
Second annual ... 0.40 0.61
Third annual ... ... 0.51
Durable goods
Advance 3.29 3.82 4.40
Second 3.47 3.76 4.42
Third 3.83 3.65 4.65
First annual 2.80 2.84 3.34
Second annual ... 2.17 3.45
Third annual ... ... 2.18
Nondurable goods
Advance 1.42 1.56 1.68
Second 1.38 1.53 1.69
Third 1.36 1.52 1.61
First annual 1.03 1.23 1.55
Second annual ... 0.83 1.30
Third annual ... ... 1.08
Services
Advance 0.90 0.89 0.81
Second 0.89 0.90 0.79
Third 0.84 0.83 0.76
First annual 0.50 0.66 0.67
Second annual ... 0.52 0.67
Third annual ... ... 0.56
Gross private
domestic investment
Advance 5.69 6.20 5.27
Second 5.27 6.23 5.87
Third 5.07 6.01 5.63
First annual 3.30 4.86 4.76
Second annual ... 3.29 3.70
Third annual ... ... 3.41
Fixed investment
Advance 2.74 2.79 2.79
Second 2.58 2.88 2.85
Third 2.59 2.89 2.92
First annual 2.00 2.42 2.48
Second annual ... 1.51 1.94
Third annual ... ... 1.68
Fixed nonresidential
investment
Advance 3.55 3.47 3.51
Second 3.38 3.63 3.72
Third 3.29 3.65 3.75
First annual 2.72 3.25 3.16
Second annual ... 1.81 2.13
Third annual ... ... 1.97
Nonresidential structures
investment
Advance 6.84 7.03 7.10
Second 6.97 6.62 6.96
Third 5.74 6.01 5.89
First annual 3.52 4.13 4.56
Second annual ... 2.50 3.34
Third annual ... ... 3.03
Equipment and intellectual
property investment
Advance 3.86 3.95 4.63
Second 4.30 4.27 4.83
Third 4.22 4.37 4.98
First annual 3.36 3.98 4.54
Second annual ... 2.83 3.82
Third annual ... ... 2.60
Residential investment
Advance 4.05 3.58 3.81
Second 3.82 3.73 3.48
Third 3.68 3.53 3.45
First annual 2.94 3.30 3.49
Second annual ... 2.37 3.10
Third annual ... ... 2.21
Exports
Advance 3.41 3.59 3.69
Second 2.63 2.96 3.11
Third 2.60 2.91 3.00
First annual 1.70 2.06 2.35
Second annual ... 1.25 1.88
Third annual ... ... 1.32
Imports
Advance 4.19 4.40 4.16
Second 2.84 2.98 3.00
Third 2.82 2.91 2.91
First annual 1.86 2.31 2.37
Second annual ... 1.48 2.37
Third annual ... ... 1.67
Government consumption
expenditures and
gross investment
Advance 1.32 1.53 1.52
Second 1.24 1.52 1.48
Third 1.22 1.52 1.45
First annual 0.75 1.25 1.23
Second annual ... 0.95 1.00
Third annual .... ....... 0.95
Federal government
Advance 2.48 2.73 2.60
Second 2.48 2.83 2.76
Third 2.53 2.89 2.71
First annual 1.52 2.13 2.05
Second annual ... 1.42 1.85
Third annual ... ... 1.74
Federal defense
Advance 3.51 3.10 3.39
Second 3.31 2.97 3.18
Third 3.33 2.94 3.04
First annual 2.46 2.60 2.86
Second annual ... 1.75 2.51
Third annual ... ... 2.18
Federal nondefense ... ... ...
Advance 4.84 4.82 4.16
Second 4.77 4.82 4.38
Third 4.71 4.94 4.61
First annual 2.59 3.41 3.92
Second annual ... 2.35 3.59
Third annual ... ... 3.87
State and local government
Advance 1.41 1.67 1.59
Second 1.33 1.52 1.61
Third 1.23 1.52 1.64
First annual 0.78 1.34 1.34
Second annual ... 1.04 1.09
Third annual ... ... 0.94
Table 4. Standard Deviations of Revisions, Early Vintages to the
Latest Estimates, Changes in Current-Dollar GDP and
Its Components, 1993-2012
[Percentage points]
Standard Standard
deviation of deviation of
estimates revision
Current Real Current Real
Vintage dollar dollar
Gross domestic product 2.79 2.59 ... ...
Advance ... ... 1.685 1.611
Second ... ... 1.562 1.534
Third ... ... 1.534 1.485
First annual ... ... 1.259 1.270
Second annual ... ... 1.089 1.206
Third annual ... ... 0.907 0.765
Personal consumption
expenditures 2.64 2.07 ... ...
Advance ... ... 1.269 1.187
Second ... ... 1.131 1.047
Third ... ... 1.150 1.050
First annual ... ... 0.985 0.991
Second annual ... ... 0.981 0.710
Third annual ... ... 0.941 0.581
Durable goods 8.98 9.23 ... ...
Advance ... ... 5.504 5.525
Second ... ... 5.353 5.383
Third ... ... 5.284 6.001
First annual ... ... 4.212 4.268
Second annual ... ... 3.627 5.394
Third annual ... ... 2.739 3.201
Nondurable goods 5.38 2.66 ... ...
Advance ... ... 2.592 2.313
Second ... ... 2.389 2.315
Third ... ... 2.450 2.254
First annual ... ... 2.317 2.121
Second annual ... ... 1.738 1.698
Third annual ... ... 1.501 1.435
Services 1.94 1.53 ... ...
Advance ... ... 1.138 0.992
Second ... ... 1.087 ...
Third ... ... 1.048 1.008
First annual ... ... 0.861 0.830
Second annual ... ... 0.855 0.825
Third annual ... ... 0.841 0.693
Gross private
domestic investment 12.94 12.82 ... ...
Advance ... ... 6.627 6.973
Second ... ... 7.153 7.202
Third ... ... 6.879 8.899
First annual ... ... 5.620 6.225
Second annual ... ... 4.851 4.870
Third annual ... ... 3.697 5.845
Fixed investment 8.45 8.01 ... ...
Advance ... ... 3.465 3.543
Second ... ... 3.435 3.525
Third ... ... 3.530 3.495
First annual ... ... 3.151 3.220
Second annual ... ... 2.416 2.459
Third annual ... ... 2.307 2.111
Fixed nonresidential
investment 8.61 8.25 ... ...
Advance ... ... 4.355 4.393
Second ... ... 4.476 4.461
Third ... ... 4.457 4.428
First annual ... ... 3.893 4.041
Second annual ... ... 2.993 2.615
Third annual ... ... 3.171 2.490
Structures investment 15.44 13.73 ... ...
Advance ... ... 10.807 9.171
Second ... ... 10.620 9.456
Third ... ... 10.171 7.923
First annual ... ... 6.046 5.701
Second annual ... ... 4.600 4.345
Third annual ... ... 4.259 3.837
Equipment and intellectual ... ... ... ...
property investment 11.65 12.24 ... ...
Advance ... ... 6.104 5.724
Second ... ... 6.444 5.837
Third ... ... 6.621 6.081
First annual ... ... 5.489 5.761
Second annual ... ... 5.428 5.178
Third annual ... ... 3.296 3.428
Residential investment 14.94 13.67 ... ...
Advance ... ... 6.022 4.876
Second ... ... 5.286 4.515
Third ... ... 5.341 4.468
First annual ... ... 4.338 5.036
Second annual ... ... 3.374 4.333
Third annual ... ... 3.641 3.138
Net exports of goods
and services (1) ... ... ... ...
Exports 11.91 9.08 ... ...
Advance ... ... 12.067 4.638
Second ... ... 5.225 3.999
Third ... ... 4.277 3.810
First annual ... ... 3.938 3.138
Second annual ... ... 3.810 2.629
Third annual ... ... 2.652 2.315
Imports 13.31 9.02 ... ...
Advance ... ... 5.529 5.088
Second ... ... 4.623 3.868
Third ... ... 4.134 3.777
First annual ... ... 3.482 3.131
Second annual ... ... 2.523 2.698
Third annual ... ... 2.129 2.116
Government consumption ... ... ... ...
expenditures and
gross investment 3.62 3.27 ... ...
Advance ... ... 2.018 1.878
Second ... ... 2.055 1.903
Third ... ... 2.066 1.851
First annual ... ... 1.700 1.556
Second annual ... ... 1.525 1.301
Third annual ... ... 1.382 1.181
Federal government 7.96 7.10 ... ...
Advance ... ... 3.399 3.292
Second ... ... 3.581 3.470
Third ... ... 3.513 3.391
First annual ... ... 3.032 2.949
Second annual ... ... 2.537 2.461
Third annual ... ... 4.107 2.202
Defense 10.22 9.99 ... ...
Advance ... ... 4.605 4.556
Second ... ... 4.458 4.333
Third ... ... 4.387 4.241
First annual ... ... 3.738 4.103
Second annual ... ... 3.994 3.436
Third annual ... ... 3.530 2.795
Nondefense 6.64 6.36 ... ...
Advance ... ... 5.793 5.484
Second ... ... 6.137 5.810
Third ... ... 6.168 5.892
First annual ... ... 5.189 4.922
Second annual ... ... 4.730 4.582
Third annual ... ... 4.920 4.908
State and local 3.38 2.81 ... ...
Advance ... ... 2.670 2.077
Second ... ... 2.521 2.003
Third ... ... 2.545 2.142
First annual ... ... 2.004 1.720
Second annual ... ... 1.763 1.406
Third annual ... ... 1.542 1.313
Scaled
standard
deviation
Current Real
Vintage dollar
Gross domestic product ... ...
Advance 0.603 0.622
Second 0.559 0.593
Third 0.549 0.574
First annual 0.451 0.491
Second annual 0.390 0.466
Third annual 0.325 0.295
Personal consumption
expenditures ... ...
Advance 0.480 0.572
Second 0.428 0.505
Third 0.435 0.506
First annual 0.373 0.478
Second annual 0.371 0.342
Third annual 0.356 0.280
Durable goods ... ...
Advance 0.613 0.598
Second 0.596 0.583
Third 0.588 0.650
First annual 0.469 0.462
Second annual 0.404 0.584
Third annual 0.305 0.347
Nondurable goods ... ...
Advance 0.482 0.870
Second 0.444 0.871
Third 0.456 0.848
First annual 0.431 0.798
Second annual 0.323 0.639
Third annual 0.279 0.540
Services ... ...
Advance 0.586 0.648
Second 0.560 0.655
Third 0.540 0.658
First annual 0.444 0.542
Second annual 0.441 0.539
Third annual 0.433 0.453
Gross private
domestic investment ... ...
Advance 0.512 0.544
Second 0.553 0.562
Third 0.532 0.538
First annual 0.434 0.486
Second annual 0.375 0.380
Third annual 0.286 0.456
Fixed investment ... ...
Advance 0.410 0.442
Second 0.406 0.440
Third 0.418 0.436
First annual 0.373 0.402
Second annual 0.286 0.307
Third annual 0.273 0.264
Fixed nonresidential
investment ... ...
Advance 0.508 0.532
Second 0.520 0.541
Third 0.519 0.537
First annual 0.452 0.490
Second annual 0.347 0.317
Third annual 0.368 0.302
Structures investment ... ...
Advance 0.700 0.668
Second 0.688 0.689
Third 0.659 0.577
First annual 0.392 0.415
Second annual 0.298 0.316
Third annual 0.276 0.279
Equipment and intellectual ... ...
property investment ... ...
Advance 0.524 0.468
Second 0.553 0.473
Third 0.568 0.497
First annual 0.471 0.471
Second annual 0.466 0.423
Third annual 0.283 0.280
Residential investment ... ...
Advance 0.403 0.357
Second 0.354 0.330
Third 0.357 0.327
First annual 0.290 0.368
Second annual 0.226 0.317
Third annual 0.177 0.230
Net exports of goods
and services (1) ... ...
Exports ... ...
Advance 1.013 0.511
Second 0.439 0.440
Third 0.359 0.420
First annual 0.331 0.346
Second annual 0.320 0.290
Third annual 0.223 0.255
Imports ... ...
Advance 0.415 0.564
Second 0.347 0.429
Third 0.310 0.419
First annual 0.262 0.347
Second annual 0.190 0.299
Third annual 0.160 0.235
Government consumption ... ...
expenditures and
gross investment ... ...
Advance 0.558 0.575
Second 0.568 0.582
Third 0.572 0.566
First annual 0.470 0.476
Second annual 0.422 0.398
Third annual 0.382 0.361
Federal government ... ...
Advance 0.427 0.464
Second 0.450 0.489
Third 0.442 0.478
First annual 0.381 0.416
Second annual 0.319 0.347
Third annual 0.516 0.310
Defense ... ...
Advance 0.451 0.456
Second 0.436 0.434
Third 0.429 0.425
First annual 0.366 0.411
Second annual 0.391 0.344
Third annual 0.345 0.280
Nondefense ... ...
Advance 0.872 0.862
Second 0.924 0.913
Third 0.929 0.926
First annual 0.781 0.773
Second annual 0.712 0.720
Third annual 0.741 0.771
State and local ... ...
Advance 0.790 0.738
Second 0.746 0.712
Third 0.753 0.761
First annual 0.593 0.611
Second annual 0.521 0.500
Third annual 0.456 0.467
Table 5. Average Revisions of Annual Estimates,
Changes in Current-Dollar GDP and Its Components,
Annual Vintages to Latest Estimates, 1993-2012
[Percentage points]
Mean revision Mean absolute
revision
Current Real Current Real
Vintage dollar dollar
Gross domestic product
Early annual (1) -0.04 -0.08 0.46 0.56
First annual -0.01 0.06 0.35 0,41
Second annual 0.10 0.23 0.31 0.34
Third annual 0.13 0.27 0.23 0.29
Personal consumption
expenditures
Early annual (1) -0.03 0.02 0.41 0.41
First annual -0.06 0.09 0.30 0.32
Second annual 0.07 0.24 0.30 0.36
Third annual 0.15 0.30 0.26 0.34
Gross private
domestic investment
Early annual (1) -0.27 -0.80 1.83 2.05
First annual -0.11 -0.47 1.74 1.77
Second annual 0.34 0.29 1.25 1.15
Third annual 0.35 0.45 1.19 1.19
Fixed investment
Early annual (1) -0.35 -0.80 1.26 1.47
First annual -0.25 -0.54 1.11 1.19
Second annual 0.38 0.38 1.02 0.89
Third annual 0.26 0.31 0.81 0.79
Change in private
inventories (2)
Net exports of goods
and services(2)
Exports
Early annual (1) 0.21 0.30 0.57 0.80
First annual 0.28 0.18 0.45 0.63
Second annual 0.05 -0.07 0.39 0.47
Third annual 0.02 -0.15 0.36 0.52
Imports
Early annual (1) 0.16 0.04 0.47 0.64
First annual 0.11 -0.06 0.24 0.33
Second annual 0.03 0.19 0.26 0.66
Third annual 0.07 -0.47 0.17 0.57
Federal government
Early annual (1) 0.01 0.13 0.48 0.75
First annual -0.29 -0.03 0.51 0.47
Second annual -0.30 -0.09 0.44 0.36
Third annual -0.26 0.02 0.43 0.32
State and local
government
Early annual (1) 0.13 -0.16 0.93 0.85
First annual 0.19 0.07 0.77 0.69
Second annual 0.13 0.00 0.66 0.63
Third annual -0.03 -0.06 0.49 0.31
(1.) Early annual estimates are available in late April of
the following year.
(2.) Percent changes cannot be calculated because of the presence
of both positive and negative values.
Table 6. Revisions, Earlier Vintages to the Latest Estimates,
Changes in GDI and Its Major Components, 1993-2012
[Percentage points]
Mean
Mean absolute Standard
Vintage revision revision deviation
Gross domestic income
Advance ... ... ...
Second ... ... ...
Third -0.10 1.21 1.85
First annual -0.01 1.57 1.99
Second annual 0.22 1.24 1.65
Third annual 0.21 0.94 1.29
Private consumption
of fixed capital
Advance -0.10 8.05 20.20
Second -0.15 7.91 20.22
Third -0.89 8.46 24.73
First annual -1.87 9.05 26.40
Second annual -2.11 9.23 27.30
Third annual -1.86 8.95 26.92
Taxes on production
and imports
Advance -0.08 2.67 3.68
Second 0.00 2.62 3.53
Third 0.09 2.46 3.40
First annual 0.30 2.03 2.99
Second annual 0.25 1.78 2.79
Third annual 0.18 1.58 3.03
National Income
Advance ... ... ...
Second ... ... ...
Third -0.32 2.33 3.12
First annual 0.06 2.17 2.96
Second annual 0.22 1.89 2.87
Third annual 0.15 1.61 2.71
Compensation of
employees
Advance 0.15 2.55 3.63
Second -0.02 2.80 3.92
Third 0.09 2.89 4.14
First annual -0.12 1.68 2.26
Second annual 0.05 1.34 2.29
Third annual 0.07 1.32 1.83
Proprietors' income
Advance 0.45 7.42 9.14
Second 0.42 7.66 9.57
Third 0.48 7.54 9.55
First annual 0.86 7.79 9.83
Second annual 0.93 6.42 8.52
Third annual 0.69 4.91 6.06
Nonfarm proprietors'
income
Advance 0.45 6.12 7.93
Second 0.41 6.01 7.84
Third 0.45 6.00 7.86
First annual 1.16 6.36 7.91
Second annual 0.47 6.88 10.97
Third annual 0.18 5.05 7.83
Corporate profits
with IVA and CCAdj
Advance ... ... ...
Second ... ... ...
Third -2.28 18.35 26.09
First annual -1.99 15.05 20.77
Second annual -2.23 11.75 19.36
Third annual -3.05 10.36 17.65
Net interest and
miscellaneous payments
Advance -0.06 9.42 26.57
Second 0.43 13.21 20.74
Third -0.32 13.35 20.09
First annual 1.00 9.57 13.06
Second annual 0.85 7.27 9.08
Third annual -1.35 6.44 8.91
CCAdj Capital consumption adjustment IVA Inventory valuation adjustment
Table 7. Standard Deviations, Revisions to Latest Estimates,
Changes in GDI and Its Components, 1993-2012
[Percentage points]
Standard Standard
deviation deviation Scaled
of of standard
Vintage estimates revisions deviation
Gross domestic income 3.18 ... ...
Advance ... ... ...
Second ... ... ...
Third ... 1.853 0.583
First annual ... 1.989 0.626
Second annual ... 1.651 0.520
Third annual 1.291 0.406
Private consumption 2.83 ... ...
of fixed capital
Advance ... 20.201 7.127
Second ... 20.218 7.133
Third ... 24.734 8.726
First annual ... 26.395 9.132
Second annual ... 27.298 9.631
Third annual ... 26.920 9.497
Taxes on production 3.40 ... ...
and imports
Advance ... 3.675 1.082
Second ... 3.532 1.040
Third ... 3.396 1.000
First annual ... 2.988 0.880
Second annual ... 2.786 0.820
Third annual ... 3.027 0.891
National Income 3.64 ... ...
Advance ... ... ...
Second ... ... ...
Third ... 3.115 0.857
First annual ... 2.953 0.815
Second annual ... 2.870 0.789
Third annual ... 2.711 0.746
Compensation of employees 3.81 ... ...
Advance ... 3.631 0.952
Second ... 3.920 1.026
Third ... 4.138 1.086
First annual ... 2.289 0.601
Second annual ... 1.830 0.480
Third annual ... 1.828 0.480
Proprietors' income 8.30 ... ...
Advance ... 9.144 1.101
Second ... 9.566 1.152
Third ... 9.550 1.150
First annual ... 9.827 1.184
Second annual ... 8.515 1.025
Third annual ... 6.061 0.730
Nonfarm proprietors' income 7.55 ... ...
Advance ... 7.934 1.051
Second ... 7.843 1.039
Third ... 7.864 1.042
First annual ... 7.912 1.048
Second annual ... 10.972 1.454
Third annual ... 7.825 1.037
Corporate profits with 24.33 ... ...
IVA and CCAdj
Advance ... ... ...
Second ... ... ...
Third ... 26.093 1.072
First annual ... 20.773 0.854
Second annual ... 19.360 0.796
Third annual ... 17.649 0.725
Net interest and 16.29 ... ...
miscellaneous payments
Advance ... 26.574 1.632
Second ... 20.737 1.273
Third ... 20.094 1.234
First annual ... 13.057 0.802
Second annual ... 9.083 0.558
Third annual ... 8.908 0.547
CCAdj Capital consumption adjustment
IVA Inventory valuation adjustment
Table 8. Mean Absolute Revisions, Changes in GDI
and Its Components, 1993-2012
[Percentage points]
Vintage of revision
used as standard
Fourth First
Vintage Second Third (1) annual
Gross domestic income
Third ... ... 1.31 0.88
Fourth ... ... ..... 1.05
First annual ... ... ..... ...
Second annual ... ... ..... ...
Third annual
Private consumption
of fixed capital
Advance 0.43 1.29 ... 4.36
Second ... 0.96 ... 4.28
Third ... ... ... 3.73
First annual ... ... ... ...
Second annual ... ... ... ...
Third annual ... ... ... ...
Taxes on production
and imports
Advance 0.64 1.07 ... 1.93
Second ... 0.78 ... 1.74
Third ... ... ... 1.58
First annual ... ... ... ...
Second annual ... ... ... ...
Third annual ... ... ... ...
National income
Third ... ... 1.86 1.25
Fourth ... ... ... 1.86
First annual ... ... ... ...
Second annual ... ... ... ...
Third annual ... ... ... ...
Compensation of
employees
Advance 0.77 0.97 2.48 1.94
Second ... 0.28 2.98 1.22
Third ... ... 3.15 2.09
Fourth ... ... ... 0.85
First annual ... ... ... ...
Second annual ... ... ... ...
Third annual ... ... ... ...
Proprietors' income
Advance 0.90 1.13 ... 4.67
Second ... 0.60 ... 4.63
Third ... ... ... 5.15
First annual ... ... ... ...
Second annual ... ... ... ...
Third annual ... ... ... ...
Nonfarm proprietors'
income
Advance 0.54 0.71 ... 3.82
Second ... 0.37 ... 3.72
Third ... ... ... 3.77
First annual ... ... ... ...
Second annual ... ... ... ...
Third annual ... ... ... ...
Corporate profits
with IVA and
CCAdj
Third ... ... ... 12.43
First annual ... ... ... ...
Second annual ... ... ... ...
Third annual ... ... ... ...
Net interest
and miscellaneous
payments
Advance 0.88 2.16 ... 4.82
Second ... 2.74 ... 7.39
Third ... ... ... 7.60
First annual ... ... ... ...
Second annual ... ... ... ...
Third annual ... ... ... ...
Vintage of revision
used as standard
Second Third
Vintage annual annual Latest
Gross domestic income
Third 1.08 1.34 1.21
Fourth 1.35 1.56 1.27
First annual 0.99 1.31 1.57
Second annual ..... 0.79 1.24
Third annual ..... ... 0.94
Private consumption
of fixed capital
Advance 4.57 4.49 8.05
Second 4.49 4.42 7.91
Third 3.91 3.77 8.46
First annual 2.89 3.79 9.05
Second annual ... 3.01 9.23
Third annual ... ... 9.95
Taxes on production
and imports
Advance 1.95 1.88 2.67
Second 1.83 1.88 2.46
Third 1.68 1.80 2.62
First annual 1.17 1.48 2.03
Second annual ... 0.96 1.78
Third annual ... ... 1.58
National income
Third 1.53 1.72 2.33
Fourth 2.22 2.41 3.41
First annual 1.14 1.79 2.17
Second annual ... 1.17 1.89
Third annual ... ... 1.61
Compensation of
employees
Advance 1.81 1.94 2.55
Second 1.85 1.94 2.89
Third 1.95 1.50 2.89
Fourth 0.78 1.50 1.72
First annual 1.05 1.22 1.68
Second annual ..... 0.77 1.34
Third annual ..... ... 1.32
Proprietors' income
Advance 6.68 5.89 7.42
Second 7.01 6.23 7.66
Third 6.88 6.21 7.54
First annual 5.57 6.41 7.79
Second annual ... 4.43 6.42
Third annual ... ... 4.91
Nonfarm proprietors'
income
Advance 5.70 5.45 6.12
Second 5.67 5.32 6.01
Third 5.61 5.32 5.00
First annual 4.94 5.80 6.36
Second annual ... 5.67 6.88
Third annual ... ... 5.05
Corporate profits
with IVA and
CCAdj
Third 13.37 16.29 18.35
First annual 7.58 13.17 15.05
Second annual ... 9.56 11.75
Third annual ... ... 10.36
Net interest
and miscellaneous
payments
Advance 7.70 8.73 9.42
Second 11.44 12.70 13.21
Third 11.50 12.87 13.33
First annual 7.17 10.23 9.57
Second annual ... 6.63 7.27
Third annual ... ... 6.44
CCAdj Capital consumption adjustment IVA Inventory valuation adjustment
(1.) Fourth estimates begin in 2002 and apply only to GDI, national
income, and compensation of employees.
Table 9. Average Revisions, Annual Vintages to
Latest Estimates, Change in GDI and Its Major
Components, 1993-2012
[Percentage points]
Mean
Vintage Mean absolute
revision revision
Gross domestic income
Early annual 0.03 0.56
First annual 0.05 0.55
Second annual 0.14 0.33
Third annual 0.20 0.40
Private consumption
of fixed capital
Early annual 0.56 2.07
First annual 0.62 1.97
Second annual 0.80 1.69
Third annual -0.37 1.53
Taxes on production
and imports
Early annual 0.34 1.27
First annual 0.22 0.87
Second annual 0.13 0.77
Third annual 0.02 0.54
National income
Early annual -0.05 0.60
First annual 0.07 0.65
Second annual 0.66 0.92
Third annual 0.23 0.41
Compensation of employees
Early annual -0.16 0.73
First annual -0.01 0.35
Second annual 0.04 0.35
Third annual 0.04 0.40
Proprietors' income
Early annual 1.15 2.69
First annual 0.95 3.95
Second annual 1.07 3.63
Third annual 1.47 3.19
Nonfarm proprietors'
income
Early annual 0.80 3.12
First annual 0.98 3.81
Second annual 1.02 3.76
Third annual 1.50 3.27
Corporate profits
with IVA and CCAdj
Early annual -0.25 5.98
First annual -0.48 4.78
Second annual -0.49 2.70
Third annual 0.57 3.04
Net interest and
miscellaneous payments
Early annual -0.44 7.79
First annual 0.42 7.16
Second annual 0.12 3.53
Third annual -1.74 3.97
CCAdj Capital consumption adjustment
IVA Inventory valuation adjustment
Table 10. Mean Absolute Revisions, Earlier Vintages to
Latest Estimates, Changes in GDP and GDI, 1993-2012
[Percentage points]
Latest GDP Latest GDI
Vintage Earlier Earlier Earlier Earlier
GDP GDI GDP GDI
Advance 1.35 1.26 1.75 1.74
Second
Third 1.21 1.07 1.71 1.21
First annual 0.99 1.22 1.69 1.57
Second annual 0.89 1.31 1.73 1.24
Third annual 0.68 1.38 1.95 1.29
Latest 1.75 1.75
Note. For example, the mean absolute revision of the
third estimates of GDI to the latest estimates of
GDP is 1.07
Table 11. Equations Explaining the Latest Estimates of Changes
in GDP Using the Third Estimates of GDP and GDI in 1993-2012
GDP GDI
(third (third
Constant estimate) estimate) R square
GDP
Equation 1 0.9729 0.6998
(30.00)
Equation 2 0.1466 0.9485 0.7004
(0.39) (13.50)
Equation 3 0.9353 0.6828
(29.20)
Equation 4 0.2866 0.8896 0.6851
(0.76) (13.03)
Equation 5 0.5245 0.4248 0.7295
(3.83) (3.11)
Equation 6 -0.0413 0.5467 0.4276 0.7330
(0.11) (-3.72) (3.06)
GDI
Equation 7 0.9827 0.6595
(26.03)
Equation 8 -0.0634 0.9928 0.6596
(0.89) (12.29)
Equation 9 1.0016 0.5569
(22.42)
Equation 10 0.2317 0.9630 0.5580
(0.45) (9.92)
Equation 11 0.1306 0.8598 0.6617
(0.72) (4.92)
Equation 12 -0.1507 0.1456 0.8697 0.6622
(0.33) (0.77) (4.87)
Note. The t-test statistics are in parentheses.
Table 12. Mean Absolute Revisions Third to the Latest Estimates,
Changes in GDP and GDI, 1993-2012
[Percentage points]
Latest Third .67GDP+ .6GDP .5GDP
GDP 33GDI +.4GDI +.5GDI GDI
GDP 1.21 1.16 1.16 1.17 1.39#
GDI 1.71# 1.58 1.55 1.52 1.45
Note. Shaded backgrounds indicate the weighted sums
that yield the smallest MARs.
Table 13. Variances and Mean Absolute Revisions,
Changes in Current-Dollar GDP and GDI, 1993-2012
[Percentage points]
.75P+ .67P+ .5P+
Variances GDP .25I .33I .5I
Third 2.46 2.44 2.44 2.46
Latest 2.79 2.73 2.73 2.78
Mean absolute
revisions to the
latest estimates
Third 1.21 1.06 1.04 1.04
.33P+ .25P+ GDI
Variances .67I .75I
Third 2.49 2.51 2.59#
Latest 2.87 2.93 3.18#
Mean absolute
revisions to the
latest estimates
Third 1.12 1.20 1.45#
NOTES. The variances and mean absolute revisions are for the
third estimates to the latest estimates. P is GDP and I is
GDI. Shaded backgrounds indicate the weighted sums that
yield the smallest MARs.
Note: The weighted sums that yield the smallest MARs are
indicated with #.
Table 14. Mean Absolute Revisions Around Cyclical
Turning Points, Changes in Current-Dollar GDP and GDI
[Percentage points]
GDP .75P+.25I .67P+.33I .5P+.5I
All quarters 1.48 1.32 1.26 1.17
Prior quarter 1.03 0.87 0.85 0.88
Peak quarter 1.25 1.04 0.98 1.05
After quarter 1.87 1.37 1.21 0.94
Prior quarter 1.12 1.12 1.12 1.19
Trough quarter 1.31 1.28 1.22 1.08
After quarter 1.89 1.78 1.76 1.73
.33P+.67I .25P+.75I GDI
All quarters 1.20 1.29 1.59
Prior quarter 0.95 1.03 1.29
Peak quarter 1.40 1.57 2.09
After quarter 0.73 0.74 0.80
Prior quarter 1.31 1.37 1.56
Trough quarter 1.25 1.48 2.20
After quarter 1.69 1.71 1.82
NOTES. The data provide mean absolute revisions for the
third estimates to the latest estimates for the six most
recent recessions. P is GDP and I is GDI. Shaded backgrounds
indicate the weighted sums that yield the smallest MARs.
Table 15. Standard Deviations of Revisions to Changes
in Real GDP and Its Major Components, 1993-2012
[Percentage points]
Vintage used as standard
First
Vintage Second Third annual
Gross domestic product
Advance 0.654 0.688 1.193
Second ... 0.277 1.031
Third 1.003
First annual ... ... ...
Second annual
Third annual ... ... ...
Personal consumption
expenditures
Advance 0.276 0.832 0.812
Second 0.464 0.981
Third ... ... 0.858
First annual ... ... ...
Second annual ... ... ...
Third annual ... ... ...
Gross private domestic
investment
Advance 3.710 3.815 6.253
Second 1.292 5.560
Third ... ... 5.029
First annual ... ... ...
Second annual ... ... ...
Third annual ... ... ...
Fixed investment
Advance 1.566 2.251 3.328
Second ... 1.775 2.687
Third ... ... 2.931
First annual ... ... ...
Second annual
Third annual ... ... ...
Exports
Advance 2.670 2.960 4.647
Second ... 1.394 3.676
Third ... ... 3.677
First annual ... ... ...
Second annual ... ... ...
Third annual ... ... ...
Imports
Advance 3.681 3.509 4.560
Second ... 2.281 3.894
Third ... ... 3.016
First annual ... ... ...
Second annual ... ... ...
Third annual ... ... ...
Federal government
Advance 1.889 1.764 3.352
Second ... 0.568 3.268
Third ... ... 3.247
First annual ... ... ...
Second annual ... ... ...
Third annual ... ... ...
State and local government
Advance 1.037 1.218 1.554
Second ... 0.422 1.207
Third ... ... 1.182
First annual ... ... ...
Second annual ... ... ...
Third annual ... ... ...
Vintage used as standard
Second Third
Vintage annual annual Latest
Gross domestic product
Advance 1.376 1.562 1.611
Second 1.281 1.446 1.534
Third 1.229 1.393 1.485
First annual 0.929 1.159 1.270
Second annual 0.937 1.206
Third annual ... ... 0.765
Personal consumption
expenditures
Advance 1.206 1.307 1.269
Second 1.033 1.165 1.131
Third 1.065 1.174 1.150
First annual ... 0.972 0.985
Second annual ... 0.633 0.981
Third annual ... ... 0.941
Gross private domestic
investment
Advance 6.528 7.067 6.627
Second 6.757 7.154 7.153
Third 6.321 6.740 6.879
First annual 4.219 4.658 5.620
Second annual ... 3.905 4.851
Third annual ... ... 3.783
Fixed investment
Advance 3.310 3.676 3.465
Second 3.119 3.652 3.436
Third 3.313 3.777 3.530
First annual 2.214 2.894 3.151
Second annual ... 2.137 2.416
Third annual ... ... 2.307
Exports
Advance 5.598 6.072 5.225
Second 4.537 5.039 4.277
Third 4.339 5.009 3.938
First annual 4.230 5.371 3.810
Second annual ... 1.758 2.652
Third annual ... ... 2.225
Imports
Advance 4.943 5.073 5.529
Second 4.147 4.042 4.623
Third 3.608 3.485 4.134
First annual 2.818 2.970 3.482
Second annual ... 1.471 2.523
Third annual ... ... 2.129
Federal government
Advance 3.340 4.745 3.399
Second 3.444 4.972 3.581
Third 3.456 4.969 3.513
First annual 2.043 4.435 3.032
Second annual ... 3.881 2.537
Third annual ... ... 4.107
State and local government
Advance 1.762 2.157 2.670
Second 1.650 1.928 2.521
Third 1.568 1.932 2.545
First annual 1.060 1.661 2.004
Second annual ... 1.424 1.763
Third annual ... ... 1.542
Table 16. Mean Revisions to the Latest Estimates, Changes
in Real GDP and Its Components, 1993-2012
[Percentage points]
Vintage of revision
used as standard
Vintage Second Third First
annual
Gross domestic product
Advance 0.10 0.13 -0.04
Second ... 0.03 -0.14
Third ... ... -0.17
First annual ... ... ...
Second annual ... ... ...
Third annual ... ... ...
Personal consumption
expenditures
Advance 0.03 0.05 -0.03
Second ... 0.02 -0.06
Third ... ... -0.08
First annual ... ... ...
Second annual ... ... ...
Third annual ... ... ...
Gross private domestic
investment
Advance 0.71 0.84 0.87
Second ... 0.13 0.16
Third ... ... 0.03
First annual ... ... ...
Second annual ... ... ...
Third annual ... ... ...
Fixed investment
Advance 0.64 0.71 1.26
Second ... 0.07 0.62
Third ... ... 0.55
First annual ... ... ...
Second annual ... ...
Third annual ... ... ...
Exports
Advance 0.19 -0.38 -0.13
Second ... 1.28 0.90
Third ... ... -0.19
First annual ... ... ...
Second annual ... ... ...
Third annual ... ... ...
Imports
Advance 0.58 0.60 0.51
Second ... 0.02 -0.06
Third ... ... -0.09
First annual ... ... ...
Second annual ... ... ...
Third annual ... ... ...
Federal government
Advance 0.10 0.07 0.38
Second ... -0.03 0.38
Third ... ... 0.32
First annual ... ... ...
Second annual ... ... ...
Third annual ... ... ...
State and local government
Advance 0.32 0.31 0.23
Second ... -0.01 -0.09
Third ... ... -0.08
First annual ... ... ...
Second annual ... ... ...
Third annual ... ... ...
Vintage of revision
used as standard
Vintage Second Third Latest
annual annual
Gross domestic product
Advance -0.32 -0.35 -0.06
Second -0.46 -0.46 -0.17
Third -0.25 -0.48 -0.19
First annual -0.25 -0.28 -0.03
Second annual ... -0.02 0.23
Third annual ... ... 0.26
Personal consumption
expenditures
Advance -0.11 -0.20 -0.06
Second -0.15 -0.24 -0.09
Third -0.17 -0.26 -0.10
First annual -0.08 -0.15 -0.01
Second annual ... -0.07 -0.07
Third annual ... ... 0.14
Gross private domestic
investment
Advance -0.01 0.06 0.59
Second -0.56 -0.53 -0.11
Third -0.69 -0.60 -0.24
First annual -0.69 -0.47 -0.29
Second annual ... 0.17 0.42
Third annual ... ... 0.15
Fixed investment
Advance 0.15 -0.01 0.70
Second -0.47 -0.61 0.06
Third -0.50 -0.60 -0.01
First annual -0.99 -0.94 -0.59
Second annual 1... 0.03 0.42
Third annual ... ... 0.42
Exports
Advance -0.69 -0.05 -0.05
Second 0.93 0.82 1.29
Third -0.11 -0.26 0.19
First annual -0.32 -0.51 0.22
Second annual ... -0.38 0.42
Third annual ... ... 0.05
Imports
Advance 1.03 1.05 0.71
Second 0.40 0.40 0.14
Third 0.30 0.19 0.11
First annual 0.27 0.17 0.21
Second annual ... -0.08 -0.02
Third annual ... ... 0.06
Federal government
Advance 0.42 -0.01 0.10
Second 0.32 -0.12 0.00
Third 0.36 -0.08 0.03
First annual 0.06 -0.43 -0.30
Second annual ... -0.50 -0.39
Third annual ... ... 0.13
State and local government
Advance 0.25 0.73 0.45
Second -0.11 0.36 0.12
Third -0.09 0.37 0.13
First annual -0.11 0.37 0.23
Second annual ... 0.43 0.35
Third annual ... ... -0.08