View original document

The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.

and firms make mistakes on, or deliberately falsify, their tax returns, then
the estimates of GNP components will
be incorrect. The BEA used information based on a 1977 IRS study of
underreported
and unreported income
to adjust for possible "underground
economy" activities. The revision, back
to 1950, extrapolates
data under the
assumption
that the ratio of underreported to reported income remains constant. This revision results in substantial increases to several NIP A
categories. For example, using estimates for 1977, compensation
of
employees rises $11.3 billion, and proprietors' income rises $46.5 billion. Personal consumption
expenditures
rise
$21.6 billion. The adjustment
to proprietors' income represents an addition
of more than 50 percent to the previously reported series. The total
adjustment
added a little more than 1.0
percent to the level of GNP.
The current benchmark revisions
also affect measurement
of the personal
saving rate. A curious phenomenon of
recent years has been the large decline
in the personal saving rate. The revisions do not alter the recent trend of
historically low saving rates, but do
raise the level of personal income,
mostly due to adjustments
for the
"underground
economy." Consequently, the saving rate does shift up a
bit, replacing the record low rate set in
1985 third quarter of 2.7 percent with a
new low of 3.7 percent.

Federal Reserve Bank of Cleveland
Research Department
P.O. Box 6387
Cleveland, OH 44101

Material may be reprinted
provided that the
source is credited. Please send copies of reprinted
materials
to the editor.

Impact on Business Cycle Timing
and Magnitudes
With respect to business cycles since
1960, the revisions do not alter the peak
and trough quarters (that is, the duration of upswings and downturns),
but
they do affect the amplitudes in each
cycle. For almost every cycle, the
change in amplitude results from the
base-period shift. For the five contractions, the revised estimates show three
periods with steeper declines than previously reported. As for the five expansion periods, all of them show less
vigorous Increases.
Since the base-period shift is a major
source of revision for all cycles, caution
must be used when making comparisons
involving several years difference. As
previously noted, the 1982 base prices
give a more accurate picture of recent
events, but are not fully suited for comparisons to the previous decade. What
is interesting,
though, is that the 198182 contraction
turns out to be more
severe, and the current expansion has
less strength than previously thought.

and for the misreporting
on tax
returns, provide good examples that the
revisions supply a necessary update of
the NIPA.
Although the 1985 benchmark revision does not drastically alter our view
of the economy, it does provide some
new and useful information.
For example, gross investment
as a percent of
GNP is now higher than previously estimated. Although definitional changes
do account for a portion of the increase,
the new estimates might become a part
of the current debate over tax incentives for business investment.
Further,
past comprehensive
revisions, such as
the 1976 revision, have provided a better picture of the service sector's growing importance.
The effect of the current revision on
government
policies and business decisions will probably be small, but such
an outcome could not be fully-known beforehand. The comprehensive
revisions,
however, update the NIPA statistics to
incorporate changing patterns in the
economy, and are well worth the effort.

Conclusion

Further Reading

The new data demonstrate
a greater
accuracy by reducing the statistical
discrepancy,
that is, the difference
between GNP measured on the product
side and GNP measured on the income
side. On average, the yearly statistical
discrepancy falls from 0.15 percent to
0.11 percent. Furthermore,
adjustments for the timing problem associated with merchandise
trade data,

"An Advanced Overview of the Comprehensive
Revision of the National Income and Product
Accounts,"
Survey of Current Business vol. 65
(October 1985): 19·28.
"Revised Estimates
of the National Income and
Product Accounts of the United States, 1929·85:
An Introduction,"
Survey of Current Business vol.
65 (December 1985): 1·19.

BULK RATE
U.S. Postage Paid
Cleveland, OH
Permit No. 385

Address Correction Requested: Please send
corrected mailing label to the Federal Reserve
Bank of Cleveland, Research Department,
P.O. Box 6387, Cleveland, OH 441Ol.

Federal Reserve Bank of Cleveland

April 15, 1986
ISSN 0428·1276

ECONOMIC
COMMENTARY
The Bureau of Economic Analysis
(BEA), a part of the U.S. Commerce
Department,
produces the National
Income and Product Accounts (NIPA)
statistics. These statistics summarize
the nation's total economic activity,
and provide statistical views of our
gross national product (GNP).
NIPA statistics are important economic tools. They provide information
that helps government
policymakers
and business leaders to understand
past and present economic activity, and
to make decisions that have farreaching effects in the economy. Aside
from regularly scheduled revisions, the
BEA also releases benchmark revisions
approximately
every five years, as new
census data become available.' The
most recent benchmark revision took
place in December 1985, the eighth
such revision of its kind.
The NIPA statistics measure economic activity in terms of current
(nominal) dollars and in constant (real)
dollars. Current-dollar
series reflect
data before a correction for price
changes, while constant-dollar
series
are adjusted to remove the effect of
price changes over time.
In the BEA's 1985 revision, all
current-dollar
series in the NIPA statistics were revised back to 1972, while
many were revised for earlier periods
as well. All constant-dollar
series and
price indexes were revised back to 1929
to reflect a base-period shift from 1972
to 1982.
Two types of major changes are evident in the recent revision: definitional
and classificational
changes, and statistical changes. Although changes in
definition and classification
are important, the statistical changes have a
much greater impact on the NIPA staTheodore G. Bernard is a research assistant at the
Federal Reserve Bank of Cleveland. The author
would like to thank KI Kowalewski for his helpful
comments.
The views stated herein are those of the author
and not necessarily those of the Federal Reserve
Bank of Cleveland or of the Board of Governors of
the Federal Reserve System.

A Revised Picture:
Has Our View of the
Economy Changed?
by Theodore G. Bernard

tistics, both in terms of percent
changes and conceptual revisions. This
Economic Commentary analyzes the
major statistical changes in the
December 1985 benchmark revisions of
the NIPA.2
Figure 1 Real GNP growth rates
Percent

change,

year over year

6
5
4
3
2

1
0

-I

~IJ

U""

-2
-3

1974

1976

c:::::J Previous

1978

1980

I
l
1982

1984

_Revised

SOURCE: U.S. Department
of Commerce,
Bureau of Economic Analysis.

Rebasing Constant-dollar Series
The statistical changes are the BEA's
self-described "outstanding
feature" of
the revisions. Shifting the base period
for the price indexes to 1982 from 1972
in the calculation of constant-dollar
series is the most important statistical
change. The year 1982 was chosen for
the new base period because it is the
latest year for which data will not be
revised until the next comprehensive
revision.
The base-period shift restates
constant-dollar
output into different
units of measurement.
As a result,
there is an approximate
doubling of real
GNP levels that reflects the difference
in measurement
units. For example,
l. The NIPA statistics are revised on a regular
basis as new and more comprehensive
data
become available. Quarterly data are routinely
revised twice after the preliminary
figures are
issued, then further revisions occur annually in
the following three years.

with 1972 as the base period, constantdollar output in 1984 had been estimated at $1.6 trillion, but with 1982 as
the base period, constant-dollar
output
in 1984 is now estimated at about $3.4
trillion. This does not imply a doubling
of actual output-only
a doubling of the
units used to measure output.
The rebasing, or restating, of
constant-dollar
series does more than
just change the units of measurement.
Since the economy does not remain
static, patterns of consumption
and
investment change over time; rebasing
updates the NIPA statistics to reflect
these changes.
For example, price inflation in the
economy since 1972, the previous base
period, has caused changes in purchasing patterns. Generally, purchasers
have tended to shift away from goods
with large price increases and toward
goods with smaller price increases, or
even with price decreases. Shifting the
base period, thus incorporating
new
purchasing
patterns and relative price
schemes, systematically
affects the
growth rates of constant-dollar
series.
These growth rates are a ratio of a given
period's total output to the output of
the base period. Within the ratio, GNP
component quantities are "weighted"
by their respective prices for some chosen year. For the sake of comparison,
only one set of prices can be used to
"weight" the output for both periods,
When compared to the 1972 base
period, restating real GNP levels in
1982 prices will give greater "weight"
to slow-growing quantities associated
with fast-growing prices (energy items,
for example), while less "weight" goes
to fast-growing quantities associated
with slow-growing prices (such as

2. Ten definitional
and classificational
changes
attempt to modernize the accounts, so that they
more accurately
reflect our evolving economy.
The net effect of these changes accounted for
about one-third of the $26.9 billion increase in
current-dollar
GNP in 1972 and for about onefourth of the $11l.9 billion increase in 1984. The

computing equipment). Historical
trends tend to verify this inverse relationship, although some exceptions
occur, thus meaning that, under such
circumstances, shifting to a more
recent base period will lower measured
growth rates for real GNP.
The choice of the base period will
therefore affect our estimate of real
GNP growth rates unless there is little
relative price change among the various goods and services. In spite of this
phenomenon, the BEA shifts the base
period because rebasing produces measures of output more relevant to current prices.
Although the net result of all the
benchmark revisions is to raise the
level of real GNP, the new estimates
show lower growth rates than
previously published data for almost all
major components of real GNP. Revised
estimates for current-dollar GNP show
an unchanged 9.9 percent average
annual growth rate for the years 197284, but revised constant-dollar GNP
growth drops 0.2 percentage point to
2.5 percent. (See figure 1.) Downward
revisions occur in nine out of 12 yearly
growth rates, the largest one being a
l.1 percentage point decline in 1973.
How much of this decline comes solely from the rebasing? The BEA figures
show that, while holding other things
constant, rebasing reduces real GNP
growth rates an average of 0.4 percentage point per year. (See table 1.) In general, any measure of real GNP will tend
to understate growth in years preceding the base period and correspondingly overstate growth in years subsequent to the base period. This happens
because the price "weights" remain constant, even though spending patterns
might be altered due to subsequent
changes in the relative prices of goods.
Some caution should be used, therefore, when comparing NIPA statistics
over an extended period of time. The
1982 base might be more appropriate
for measuring economic activity in the
1980's, but the 1972 base might be a
more reliable measurement of growth
in the early 1970's. However, using
several different indexes for different
time periods is impractical and inconvenient. Using the most current index
provides a more up-to-date measure of
recent economic activity.

details of these changes are documented in the
October and December 1985 issues of the Survey
0/ Current Business.

Table 1 Revisions in Average Annual Rates of Change Over the
Period 1972-84 for GNP and its Major Components
[Percent]
Source of constantdollar revision

Constant
dollars

GNP
Personal consumption
expenditures
Durable goods
Nondurable goods
Services
Gross private domestic
investment
Fixed investment
Nonresidential
Structures
Producers' durable
equipment
Residential
Change in business
inventories
Net exports of goods
and services
Exports
Imports
Government purchases of
goods and services
Federal
National defense
Nondefense
State and local
SOURCE: U.S. Department

Revision

Current
dollar
revision

Base
period
shift

Other

-0.2

0.0

-0.4

0.2

-0.4
-0.1
-0.5
-0.4

0.1
0.3
0.1
0.0

-0.4
-0.4
-0.5
-0.3

-0.1
0.0
-0.1
-0.1

-0.4
-0.6
-0.8
0.5

0.2
0.1
-0.2
-0.3

-0.9
-0.8
-l.0
0.5

0.3
0.1
0.4
0.3

-1.2
0.6

0.0
0.7

-l.6
-0.1

0.4
0.0

0.1
-l.1

0.0
0.2

-0.2
-1.2

0.3
-0.1

-0.1
-0.1
-0.2
0.3
0.0

0.0
0.2
0.1
0.2
-0.1

0.0
-0.1
-0.3
-0.1
0.0

-0.1
-0.2
0.0
0.2
0.1

of Commerce, Bureau of Economic Analysis.

A look at information reflected in
previous indexes illustrates the alterations that can occur when revisions are
made. In the 1976 comprehensive revision, when price indexes were rebased
from 1958 prices to 1972 prices, real
growth rates also declined. For the
period 1958-74, average annual growth
of real GNP decreased 0.2 percentage
point. Offsetting changes to the components of GNP kept the aggregate
revision relatively small. The sharp
rise in food and energy prices, relative
to other goods, since 1972 would have
created a significant impact on real
GNP if these elements were incorporated into the 1976 revision. According
to the January 1976 Survey of Current
Business, the BEA found it "regrettable
that it was not possible to take the
energy price increase into account.">

3. "The National Income and Product Accounts
of the United States: Revised Estimates, 192974," Survey 0/ Current Business vol. 56 (January
1976): 1-34.

Since energy prices influence the
prices of many other goods, a rebasing
that adjusts for higher energy prices
should be expected to lower the real
growth rates of several GNP components. Because the 1972 base period did
not reflect the relative rise in energy
prices, real imports of petroleum, in recent years, were overstated relative to
their value in 1982 dollars. When the
1982 base adjusts for higher energy
prices, real petroleum imports receive a
greater "weight" relative to the 1972based accounts. Since imports are a negative item in the GNP accounts, the
greater "weight" attached to petroleum
imports will reduce GNP growth rates in
the 1982 basis relative to the 1972 basis.

The current benchmark revision of
the NIP A statistics offers some interesting information. For example, productivity, as measured by output per
man-hour worked, is usually systematically affected by the GNP revisions.
Since the current base-period shift
tended to lower output growth rates,
and output is a part of productivity, it
is not too surprising to find that productivity growth rates have also fallen.
To make the decline even worse, manhours worked have been revised
upward by the Bureau of Labor Statistics. The pattern of productivity
growth remains fairly similar to previously published data, but the revised
growth rate for nonfarm productivity
shows a 0.3 percentage point decline,
on average, for 1972-84. Thus, the
incorporation of the revised data does
not change the well-publicized productivity slowdown of the last decade.
The GNP fixed-weight price index
shows the same effects from a baseperiod shift as do real GNP growth
rates. Prices that increased the least
from 1972 to 1982 generally will have a
greater effect on the change in the
fixed-weight index than before the rebasing. Therefore, the measured price
increase will also be less. Over the
period 1972-84, the GNP fixed-weight
price index is now estimated at a 6.6
percent annual growth rate, down from
7.3 percent. (See figure 2.) About half of
the decrease is attributable to the baseperiod shift. The price deflator for producers' durable equipment, a component
of nonresidential fixed investment, displayed the largest downward revision,
5.2 percentage points. The combination
of the rebasing and the incorporation of
a new computer price index are responsible for almost the entire change.
Other Statistical Changes
The current benchmark revisions include other major statistical changes,
including source-data changes and
changes in methodology. These changes
incorporate new source-data not previously available and reflect new estimating techniques, as the two tend to
go hand-in-hand. Together they account
for about 30 changes. The most important include: 1) Improvement in the
price index for computing equipment;

2) correction of a timing problem associated with the monthly U.S. merchandise trade data; and 3) adjustments for
misreporting on tax returns.
The first change deals with developing a price deflator for computing
equipment. The BEA, with the advice
and assistance of IBM Corporation,
developed a new price index that more
accurately reflects the prices of computing equipment. The previous index
assumed computer prices did not change
between 1972 and 1984. The new index
shows that computer prices have
declined by an average annual rate of
14.0 percent from 1972-84. The incorporation of the new computer price
index raises the average annual growth
rate of constant-dollar computer purchases, a component of producers'
durable equipment (PDE), by almost
15.0 percentage points! This increase
translates into a substantial rise in
PDE expenditures in 1972 dollars.
Figure 2 GNP Fixed-weight
Percent change, year over year

Price Index

9
8
7
6
5
4

3
2

o~~~~~~~~~~~~~~
1974 1976 1978 1980
c::::J

Previous

1982

1984

_Revised

SOURCE: U.S. Department of Commerce,
Bureau of Economic Analysis.

When measured in 1972 dollars,
expenditures for PDE, over the period
1972-84, are now $107.9 billion more
than previously reported. This
increase, almost solely due to the revision of computer purchases, pushed the
annual growth rate of real PDE up 4.8
percentage points. Despite the strong
upward revision due to the new computer price index, rebasing to 1982 dollars, as is evident in table 1, becomes
the dominant influence in lowering real
PDE growth.
Further effects of this statistical
change are carried over into the revision of data on exports. The upward

revision to exports more than offsets
the negative impact of rebasing, primarily due to the incorporation of the
computer price index. Thus, real exports
have grown 5.5 percent yearly, which is
0.1 percentage point higher than in earlier estimates based on 1972 prices.
A second major statistical change
became necessary due to the large
growth in the net export sector. From
1983 second quarter to 1985 third quarter, real imports rose an astounding
32.4 percent, causing problems in the
processing of import data. The U.S.
Customs Service reports merchandise
trade data to the Census Bureau within
15 days of each month's end. The enormous increase in imports has created a
substantial increase in volume and
variability of "carry-over" data-that
is, data received too late for inclusion
into the proper monthly report. This
"carry-over" effect results in a serious
timing problem and diminishes the
reliability of quarterly changes in net
export data. Revised data for August
1985, for example, showed that "carryover" documents accounted for 17 percent of the value of imports and 11 percent of the value of exports.
The BEA used the comprehensive
revisions to adjust the period 1983
second quarter to 1985 second quarter
for this "carry-over" data. The most
dramatic revision occurred in 1984
fourth quarter and 1985 first quarter.
Real net exports of merchandise were
originally reported to have risen $15.7
billion in the last quarter of 1984, then
to have fallen $14.3 billion at the start
of 1985.
Current revisions now reverse this
pattern to indicate that net exports fell
in 1984 fourth quarter and then rose in
1985 first quarter! This revision caused
real GNP growth estimates to swing
from the originally reported 4.3 percent
in 1984 fourth quarter and 0.3 percent
in 1985 first quarter to 0.6 percent and
3.7 percent respectively. Although the
data does not alter the overall view of
the net export sector, it does give us a
better picture of quarterly changes.
Finally, the third main statistical
change in NIPA statistics attempts to
improve the estimates for misreporting
on tax returns. Federal tax return
information is used for estimating several components of GNP. If consumers

computing equipment). Historical
trends tend to verify this inverse relationship, although some exceptions
occur, thus meaning that, under such
circumstances, shifting to a more
recent base period will lower measured
growth rates for real GNP.
The choice of the base period will
therefore affect our estimate of real
GNP growth rates unless there is little
relative price change among the various goods and services. In spite of this
phenomenon, the BEA shifts the base
period because rebasing produces measures of output more relevant to current prices.
Although the net result of all the
benchmark revisions is to raise the
level of real GNP, the new estimates
show lower growth rates than
previously published data for almost all
major components of real GNP. Revised
estimates for current-dollar GNP show
an unchanged 9.9 percent average
annual growth rate for the years 197284, but revised constant-dollar GNP
growth drops 0.2 percentage point to
2.5 percent. (See figure 1.) Downward
revisions occur in nine out of 12 yearly
growth rates, the largest one being a
l.1 percentage point decline in 1973.
How much of this decline comes solely from the rebasing? The BEA figures
show that, while holding other things
constant, rebasing reduces real GNP
growth rates an average of 0.4 percentage point per year. (See table 1.) In general, any measure of real GNP will tend
to understate growth in years preceding the base period and correspondingly overstate growth in years subsequent to the base period. This happens
because the price "weights" remain constant, even though spending patterns
might be altered due to subsequent
changes in the relative prices of goods.
Some caution should be used, therefore, when comparing NIPA statistics
over an extended period of time. The
1982 base might be more appropriate
for measuring economic activity in the
1980's, but the 1972 base might be a
more reliable measurement of growth
in the early 1970's. However, using
several different indexes for different
time periods is impractical and inconvenient. Using the most current index
provides a more up-to-date measure of
recent economic activity.

details of these changes are documented in the
October and December 1985 issues of the Survey
0/ Current Business.

Table 1 Revisions in Average Annual Rates of Change Over the
Period 1972-84 for GNP and its Major Components
[Percent]
Source of constantdollar revision

Constant
dollars

GNP
Personal consumption
expenditures
Durable goods
Nondurable goods
Services
Gross private domestic
investment
Fixed investment
Nonresidential
Structures
Producers' durable
equipment
Residential
Change in business
inventories
Net exports of goods
and services
Exports
Imports
Government purchases of
goods and services
Federal
National defense
Nondefense
State and local
SOURCE: U.S. Department

Revision

Current
dollar
revision

Base
period
shift

Other

-0.2

0.0

-0.4

0.2

-0.4
-0.1
-0.5
-0.4

0.1
0.3
0.1
0.0

-0.4
-0.4
-0.5
-0.3

-0.1
0.0
-0.1
-0.1

-0.4
-0.6
-0.8
0.5

0.2
0.1
-0.2
-0.3

-0.9
-0.8
-l.0
0.5

0.3
0.1
0.4
0.3

-1.2
0.6

0.0
0.7

-l.6
-0.1

0.4
0.0

0.1
-l.1

0.0
0.2

-0.2
-1.2

0.3
-0.1

-0.1
-0.1
-0.2
0.3
0.0

0.0
0.2
0.1
0.2
-0.1

0.0
-0.1
-0.3
-0.1
0.0

-0.1
-0.2
0.0
0.2
0.1

of Commerce, Bureau of Economic Analysis.

A look at information reflected in
previous indexes illustrates the alterations that can occur when revisions are
made. In the 1976 comprehensive revision, when price indexes were rebased
from 1958 prices to 1972 prices, real
growth rates also declined. For the
period 1958-74, average annual growth
of real GNP decreased 0.2 percentage
point. Offsetting changes to the components of GNP kept the aggregate
revision relatively small. The sharp
rise in food and energy prices, relative
to other goods, since 1972 would have
created a significant impact on real
GNP if these elements were incorporated into the 1976 revision. According
to the January 1976 Survey of Current
Business, the BEA found it "regrettable
that it was not possible to take the
energy price increase into account.">

3. "The National Income and Product Accounts
of the United States: Revised Estimates, 192974," Survey 0/ Current Business vol. 56 (January
1976): 1-34.

Since energy prices influence the
prices of many other goods, a rebasing
that adjusts for higher energy prices
should be expected to lower the real
growth rates of several GNP components. Because the 1972 base period did
not reflect the relative rise in energy
prices, real imports of petroleum, in recent years, were overstated relative to
their value in 1982 dollars. When the
1982 base adjusts for higher energy
prices, real petroleum imports receive a
greater "weight" relative to the 1972based accounts. Since imports are a negative item in the GNP accounts, the
greater "weight" attached to petroleum
imports will reduce GNP growth rates in
the 1982 basis relative to the 1972 basis.

The current benchmark revision of
the NIP A statistics offers some interesting information. For example, productivity, as measured by output per
man-hour worked, is usually systematically affected by the GNP revisions.
Since the current base-period shift
tended to lower output growth rates,
and output is a part of productivity, it
is not too surprising to find that productivity growth rates have also fallen.
To make the decline even worse, manhours worked have been revised
upward by the Bureau of Labor Statistics. The pattern of productivity
growth remains fairly similar to previously published data, but the revised
growth rate for nonfarm productivity
shows a 0.3 percentage point decline,
on average, for 1972-84. Thus, the
incorporation of the revised data does
not change the well-publicized productivity slowdown of the last decade.
The GNP fixed-weight price index
shows the same effects from a baseperiod shift as do real GNP growth
rates. Prices that increased the least
from 1972 to 1982 generally will have a
greater effect on the change in the
fixed-weight index than before the rebasing. Therefore, the measured price
increase will also be less. Over the
period 1972-84, the GNP fixed-weight
price index is now estimated at a 6.6
percent annual growth rate, down from
7.3 percent. (See figure 2.) About half of
the decrease is attributable to the baseperiod shift. The price deflator for producers' durable equipment, a component
of nonresidential fixed investment, displayed the largest downward revision,
5.2 percentage points. The combination
of the rebasing and the incorporation of
a new computer price index are responsible for almost the entire change.
Other Statistical Changes
The current benchmark revisions include other major statistical changes,
including source-data changes and
changes in methodology. These changes
incorporate new source-data not previously available and reflect new estimating techniques, as the two tend to
go hand-in-hand. Together they account
for about 30 changes. The most important include: 1) Improvement in the
price index for computing equipment;

2) correction of a timing problem associated with the monthly U.S. merchandise trade data; and 3) adjustments for
misreporting on tax returns.
The first change deals with developing a price deflator for computing
equipment. The BEA, with the advice
and assistance of IBM Corporation,
developed a new price index that more
accurately reflects the prices of computing equipment. The previous index
assumed computer prices did not change
between 1972 and 1984. The new index
shows that computer prices have
declined by an average annual rate of
14.0 percent from 1972-84. The incorporation of the new computer price
index raises the average annual growth
rate of constant-dollar computer purchases, a component of producers'
durable equipment (PDE), by almost
15.0 percentage points! This increase
translates into a substantial rise in
PDE expenditures in 1972 dollars.
Figure 2 GNP Fixed-weight
Percent change, year over year

Price Index

9
8
7
6
5
4

3
2

o~~~~~~~~~~~~~~
1974 1976 1978 1980
c::::J

Previous

1982

1984

_Revised

SOURCE: U.S. Department of Commerce,
Bureau of Economic Analysis.

When measured in 1972 dollars,
expenditures for PDE, over the period
1972-84, are now $107.9 billion more
than previously reported. This
increase, almost solely due to the revision of computer purchases, pushed the
annual growth rate of real PDE up 4.8
percentage points. Despite the strong
upward revision due to the new computer price index, rebasing to 1982 dollars, as is evident in table 1, becomes
the dominant influence in lowering real
PDE growth.
Further effects of this statistical
change are carried over into the revision of data on exports. The upward

revision to exports more than offsets
the negative impact of rebasing, primarily due to the incorporation of the
computer price index. Thus, real exports
have grown 5.5 percent yearly, which is
0.1 percentage point higher than in earlier estimates based on 1972 prices.
A second major statistical change
became necessary due to the large
growth in the net export sector. From
1983 second quarter to 1985 third quarter, real imports rose an astounding
32.4 percent, causing problems in the
processing of import data. The U.S.
Customs Service reports merchandise
trade data to the Census Bureau within
15 days of each month's end. The enormous increase in imports has created a
substantial increase in volume and
variability of "carry-over" data-that
is, data received too late for inclusion
into the proper monthly report. This
"carry-over" effect results in a serious
timing problem and diminishes the
reliability of quarterly changes in net
export data. Revised data for August
1985, for example, showed that "carryover" documents accounted for 17 percent of the value of imports and 11 percent of the value of exports.
The BEA used the comprehensive
revisions to adjust the period 1983
second quarter to 1985 second quarter
for this "carry-over" data. The most
dramatic revision occurred in 1984
fourth quarter and 1985 first quarter.
Real net exports of merchandise were
originally reported to have risen $15.7
billion in the last quarter of 1984, then
to have fallen $14.3 billion at the start
of 1985.
Current revisions now reverse this
pattern to indicate that net exports fell
in 1984 fourth quarter and then rose in
1985 first quarter! This revision caused
real GNP growth estimates to swing
from the originally reported 4.3 percent
in 1984 fourth quarter and 0.3 percent
in 1985 first quarter to 0.6 percent and
3.7 percent respectively. Although the
data does not alter the overall view of
the net export sector, it does give us a
better picture of quarterly changes.
Finally, the third main statistical
change in NIPA statistics attempts to
improve the estimates for misreporting
on tax returns. Federal tax return
information is used for estimating several components of GNP. If consumers

and firms make mistakes on, or deliberately falsify, their tax returns, then
the estimates of GNP components will
be incorrect. The BEA used information based on a 1977 IRS study of
underreported
and unreported income
to adjust for possible "underground
economy" activities. The revision, back
to 1950, extrapolates
data under the
assumption
that the ratio of underreported to reported income remains constant. This revision results in substantial increases to several NIP A
categories. For example, using estimates for 1977, compensation
of
employees rises $11.3 billion, and proprietors' income rises $46.5 billion. Personal consumption
expenditures
rise
$21.6 billion. The adjustment
to proprietors' income represents an addition
of more than 50 percent to the previously reported series. The total
adjustment
added a little more than 1.0
percent to the level of GNP.
The current benchmark revisions
also affect measurement
of the personal
saving rate. A curious phenomenon of
recent years has been the large decline
in the personal saving rate. The revisions do not alter the recent trend of
historically low saving rates, but do
raise the level of personal income,
mostly due to adjustments
for the
"underground
economy." Consequently, the saving rate does shift up a
bit, replacing the record low rate set in
1985 third quarter of 2.7 percent with a
new low of 3.7 percent.

Federal Reserve Bank of Cleveland
Research Department
P.O. Box 6387
Cleveland, OH 44101

Material may be reprinted
provided that the
source is credited. Please send copies of reprinted
materials
to the editor.

Impact on Business Cycle Timing
and Magnitudes
With respect to business cycles since
1960, the revisions do not alter the peak
and trough quarters (that is, the duration of upswings and downturns),
but
they do affect the amplitudes in each
cycle. For almost every cycle, the
change in amplitude results from the
base-period shift. For the five contractions, the revised estimates show three
periods with steeper declines than previously reported. As for the five expansion periods, all of them show less
vigorous Increases.
Since the base-period shift is a major
source of revision for all cycles, caution
must be used when making comparisons
involving several years difference. As
previously noted, the 1982 base prices
give a more accurate picture of recent
events, but are not fully suited for comparisons to the previous decade. What
is interesting,
though, is that the 198182 contraction
turns out to be more
severe, and the current expansion has
less strength than previously thought.

and for the misreporting
on tax
returns, provide good examples that the
revisions supply a necessary update of
the NIPA.
Although the 1985 benchmark revision does not drastically alter our view
of the economy, it does provide some
new and useful information.
For example, gross investment
as a percent of
GNP is now higher than previously estimated. Although definitional changes
do account for a portion of the increase,
the new estimates might become a part
of the current debate over tax incentives for business investment.
Further,
past comprehensive
revisions, such as
the 1976 revision, have provided a better picture of the service sector's growing importance.
The effect of the current revision on
government
policies and business decisions will probably be small, but such
an outcome could not be fully-known beforehand. The comprehensive
revisions,
however, update the NIPA statistics to
incorporate changing patterns in the
economy, and are well worth the effort.

Conclusion

Further Reading

The new data demonstrate
a greater
accuracy by reducing the statistical
discrepancy,
that is, the difference
between GNP measured on the product
side and GNP measured on the income
side. On average, the yearly statistical
discrepancy falls from 0.15 percent to
0.11 percent. Furthermore,
adjustments for the timing problem associated with merchandise
trade data,

"An Advanced Overview of the Comprehensive
Revision of the National Income and Product
Accounts,"
Survey of Current Business vol. 65
(October 1985): 19·28.
"Revised Estimates
of the National Income and
Product Accounts of the United States, 1929·85:
An Introduction,"
Survey of Current Business vol.
65 (December 1985): 1·19.

BULK RATE
U.S. Postage Paid
Cleveland, OH
Permit No. 385

Address Correction Requested: Please send
corrected mailing label to the Federal Reserve
Bank of Cleveland, Research Department,
P.O. Box 6387, Cleveland, OH 441Ol.

Federal Reserve Bank of Cleveland

April 15, 1986
ISSN 0428·1276

ECONOMIC
COMMENTARY
The Bureau of Economic Analysis
(BEA), a part of the U.S. Commerce
Department,
produces the National
Income and Product Accounts (NIPA)
statistics. These statistics summarize
the nation's total economic activity,
and provide statistical views of our
gross national product (GNP).
NIPA statistics are important economic tools. They provide information
that helps government
policymakers
and business leaders to understand
past and present economic activity, and
to make decisions that have farreaching effects in the economy. Aside
from regularly scheduled revisions, the
BEA also releases benchmark revisions
approximately
every five years, as new
census data become available.' The
most recent benchmark revision took
place in December 1985, the eighth
such revision of its kind.
The NIPA statistics measure economic activity in terms of current
(nominal) dollars and in constant (real)
dollars. Current-dollar
series reflect
data before a correction for price
changes, while constant-dollar
series
are adjusted to remove the effect of
price changes over time.
In the BEA's 1985 revision, all
current-dollar
series in the NIPA statistics were revised back to 1972, while
many were revised for earlier periods
as well. All constant-dollar
series and
price indexes were revised back to 1929
to reflect a base-period shift from 1972
to 1982.
Two types of major changes are evident in the recent revision: definitional
and classificational
changes, and statistical changes. Although changes in
definition and classification
are important, the statistical changes have a
much greater impact on the NIPA staTheodore G. Bernard is a research assistant at the
Federal Reserve Bank of Cleveland. The author
would like to thank KI Kowalewski for his helpful
comments.
The views stated herein are those of the author
and not necessarily those of the Federal Reserve
Bank of Cleveland or of the Board of Governors of
the Federal Reserve System.

A Revised Picture:
Has Our View of the
Economy Changed?
by Theodore G. Bernard

tistics, both in terms of percent
changes and conceptual revisions. This
Economic Commentary analyzes the
major statistical changes in the
December 1985 benchmark revisions of
the NIPA.2
Figure 1 Real GNP growth rates
Percent

change,

year over year

6
5
4
3
2

1
0

-I

~IJ

U""

-2
-3

1974

1976

c:::::J Previous

1978

1980

I
l
1982

1984

_Revised

SOURCE: U.S. Department
of Commerce,
Bureau of Economic Analysis.

Rebasing Constant-dollar Series
The statistical changes are the BEA's
self-described "outstanding
feature" of
the revisions. Shifting the base period
for the price indexes to 1982 from 1972
in the calculation of constant-dollar
series is the most important statistical
change. The year 1982 was chosen for
the new base period because it is the
latest year for which data will not be
revised until the next comprehensive
revision.
The base-period shift restates
constant-dollar
output into different
units of measurement.
As a result,
there is an approximate
doubling of real
GNP levels that reflects the difference
in measurement
units. For example,
l. The NIPA statistics are revised on a regular
basis as new and more comprehensive
data
become available. Quarterly data are routinely
revised twice after the preliminary
figures are
issued, then further revisions occur annually in
the following three years.

with 1972 as the base period, constantdollar output in 1984 had been estimated at $1.6 trillion, but with 1982 as
the base period, constant-dollar
output
in 1984 is now estimated at about $3.4
trillion. This does not imply a doubling
of actual output-only
a doubling of the
units used to measure output.
The rebasing, or restating, of
constant-dollar
series does more than
just change the units of measurement.
Since the economy does not remain
static, patterns of consumption
and
investment change over time; rebasing
updates the NIPA statistics to reflect
these changes.
For example, price inflation in the
economy since 1972, the previous base
period, has caused changes in purchasing patterns. Generally, purchasers
have tended to shift away from goods
with large price increases and toward
goods with smaller price increases, or
even with price decreases. Shifting the
base period, thus incorporating
new
purchasing
patterns and relative price
schemes, systematically
affects the
growth rates of constant-dollar
series.
These growth rates are a ratio of a given
period's total output to the output of
the base period. Within the ratio, GNP
component quantities are "weighted"
by their respective prices for some chosen year. For the sake of comparison,
only one set of prices can be used to
"weight" the output for both periods,
When compared to the 1972 base
period, restating real GNP levels in
1982 prices will give greater "weight"
to slow-growing quantities associated
with fast-growing prices (energy items,
for example), while less "weight" goes
to fast-growing quantities associated
with slow-growing prices (such as

2. Ten definitional
and classificational
changes
attempt to modernize the accounts, so that they
more accurately
reflect our evolving economy.
The net effect of these changes accounted for
about one-third of the $26.9 billion increase in
current-dollar
GNP in 1972 and for about onefourth of the $11l.9 billion increase in 1984. The