Consumer Price Index: Update of Boskin Commission's Estimate of Bias
(Letter Report, 02/01/2000, GAO/GGD-00-50).

Pursuant to a congressional request, GAO provided information on the
Advisory Commission to Study the Consumer Price Index's (CPI) updated
estimate of CPI bias, focusing on the: (1) methodological changes the
Bureau of Labor Statistics (BLS) made to the CPI since December 1996,
when the Advisory Commission (also referred to as the Boskin Commission)
issued its final report; and (2) opinions of the five former Boskin
Commission members on how much of the bias in the CPI that the
Commission estimated in its December 1996 report remains after recent
methodological changes to the CPI.

GAO noted that: (1) between December 1996, when the Boskin Commission
issued its final report estimating that the CPI overstates the cost of
living by 1.1 percentage points annually, and June 1999, when GAO began
this review, BLS had made seven methodological changes that affected the
calculation of the CPI; (2) in addition, as of June 1999, BLS had
announced three methodological changes that had not yet been
implemented; (3) four former members of the Commission responded to
GAO's questions about the extent to which recent methodological changes
in the CPI have reduced its overstatement of the changes in the cost of
living--that is, bias--as defined in the Commission's December 1996
report; (4) although all four of these former Boskin Commission members
said that the seven methodological changes made to the CPI have reduced
some of the bias in the CPI, they had different responses regarding the
extent of the remaining bias; (5) their point estimates of the remaining
bias varied from 0.73 to 0.9 percentage points annually after taking
into account those seven changes; (6) the former Boskin Commission
members believe that most of the remaining bias is due to what the
Commission referred to as "new products/quality change bias;" (7)
according to the Commission, this type of bias occurs when new products
are not included in the CPI or when they are included after a long
delay, which results in the CPI not capturing price decreases that often
occur after a product is introduced in the marketplace; and (8)
according to the Commission, new products/quality change bias occurs
when the CPI does not adequately measure the portion of a price increase
that is due to an improvement in the quality of a product or service
instead of to an increase in the cost of living.

--------------------------- Indexing Terms -----------------------------

 REPORTNUM:  GGD-00-50
     TITLE:  Consumer Price Index: Update of Boskin Commission's
	     Estimate of Bias
      DATE:  02/01/2000
   SUBJECT:  Price indexes
	     Economic analysis
	     Economic policies
	     Cost of living
	     Inflation
	     Prices and pricing
	     Projections
IDENTIFIER:  Consumer Price Index

******************************************************************
** This file contains an ASCII representation of the text of a  **
** GAO report.  Delineations within the text indicating chapter **
** titles, headings, and bullets are preserved.                 **
**                                                              **
** No attempt has been made to display graphic images, although **
** figure captions are reproduced.  Tables are included, but    **
** may not resemble those in the printed version.               **
**                                                              **
** Please see the PDF (Portable Document Format) file, when     **
** available, for a complete electronic file of the printed     **
** document's contents.                                         **
**                                                              **
** A printed copy of this report may be obtained from the GAO   **
** Document Distribution Center.  For further details, please   **
** send an e-mail message to:                                   **
**                                                              **
**                                            **
**                                                              **
** with the message 'info' in the body.                         **
******************************************************************

United States General Accounting Office
GAO

Report to the Ranking Minority Member, Committee

on Finance

U.S. Senate

February 2000

GAO/GGD-00-50

CONSUMER PRICE INDEX
Update of Boskin Commission's Estimate of Bias

Ordering Information

The first copy of each GAO report and testimony
is free. Additional copies are $2 each. Orders
should be sent to the following address,
accompanied by a check or money order made out
to the Superintendent of Documents, when
necessary. VISA and MasterCard credit cards are
accepted, also. Orders for 100 or more copies to
be mailed to a single address are discounted 25
percent.

Order by mail:

U.S. General Accounting Office
P.O. Box 37050
Washington, DC 20013

or visit:

Room 1100
700 4th St. NW (corner of 4th and G Sts. NW)
U.S. General Accounting Office
Washington, DC

Orders may also be placed by calling (202) 512-
6000 or by using fax number (202) 512-6061, or
TDD (202) 512-2537.

Each day, GAO issues a list of newly available
reports and testimony. To receive facsimile
copies of the daily list or any list from the
past 30 days, please call (202) 512-6000 using a
touch-tone phone. A recorded menu will provide
information on how to obtain these lists.

For information on how to access GAO reports on
the INTERNET, send e-mail message with "info" in
the body to:

[email protected]

or visit GAO's World Wide Web Home Page at:

http://www.gao.gov

United States
General Accounting Office
Washington, D.C. 20548-0001
Official Business
Penalty for Private Use $300
Address Correction Requested

                    Bulk Rate
               Postage & Fees Paid
                       GAO
                 Permit No. G100

(410449)

Contents
Page 201                    GAO/GGD-00-50 CPI Bias
Letter                                                                      1
                                                                             
Appendix I                                                                 22
Background Information
on the Consumer Price
Index
                           Construction of the CPI                         23
                                                                             
Appendix II                                                                29
Questionnaire Sent to
the Former Members of
the Boskin Commission
                                                                             
Appendix III                                                               41
Comments From the Bureau
of Labor Statistics
                                                                             
Appendix IV                                                                43
Comments From Dr.
Michael J. Boskin
                                                                             
Related GAO Products                                                       46
                                                                             
Tables                     Table 1: The Boskin Commission's                 5
                           December 1996 Estimates of Bias in
                           the CPI
                           Table 2: Responses of Four Former               14
                           Boskin Commission Members Regarding
                           the Effect of 10 Methodological
                           Changes to the CPI by Source of Bias
                                                                             
Figures                    Figure 1:  Former Boskin Commission              3
                           Members' Point Estimates for the
                           Remaining Bias in the CPI,
                           Considering the Seven Methodological
                           Changes in Effect as of June 1999
                           Figure 2:  Former Boskin Commission             12
                           Members' Point and Plausible Range
                           Estimates of the Remaining Bias in
                           the CPI
                                                                             

Abbreviations

BLS       Bureau of Labor Statistics
CEX       Consumer Expenditure Survey
CPI       Consumer Price Index
CPI-U     Consumer Price Index for All Urban
Consumers
CPI-W     Consumer Price Index for Urban Wage
Earners and Clerical Workers

B-284299

Page 2                      GAO/GGD-00-50 CPI Bias
     B-284299

February 1, 2000

The Honorable Daniel Patrick Moynihan
Ranking Minority Member
Committee on Finance
United States Senate
 
 Dear Senator Moynihan:

The Consumer Price Index (CPI) is the principal
measure of trends in consumer prices and inflation
in the United States, according to the Bureau of
Labor Statistics (BLS), which publishes the CPI.1
The CPI significantly affects federal revenues and
spending because automatic cost-of-living
adjustments to federal tax brackets and many
federal benefit programs are calculated based on
changes in the CPI. In December 1996, the Advisory
Commission to Study the CPI, which was appointed
by the Senate Finance Committee, issued a report
estimating that the CPI overstates changes in the
cost of living by 1.1 percentage points annually.
(The Advisory Commission is commonly referred to
as "the Boskin Commission," after its chairman
Michael J. Boskin.) The Commission's report also
included the Congressional Budget Office's
estimate that such an overstatement-that is, bias-
in the CPI would cause a $1.07 trillion rise in
the national debt by the year 2008.

This report responds to your request that we
obtain an update from the five former Boskin
Commission members of the Commission's estimate of
the CPI bias. You asked for an updated estimate
because of recent reports of changes BLS has made
in the methodology for calculating the CPI.

As agreed with your office, our objectives were to

ï¿½    identify the methodological changes BLS made
to the CPI since December 1996, when the Boskin
Commission issued its final report; and
ï¿½    obtain the opinions of the five former Boskin
Commission members on how much of the bias in the
CPI that the Commission estimated in its December
1996 report remains after recent methodological
changes to the CPI.

As further agreed, we did not attempt to assess
the validity of the Boskin Commission's 1996
estimate or the validity of the estimates former
Commission members gave for this report.

Results in Brief
Between December 1996, when the Boskin Commission
issued its final report estimating that the CPI
overstates the cost-of-living by 1.1 percentage
points annually, and June 1999, when we began this
review,  BLS had made seven methodological changes
that affected the calculation of the CPI.  In
addition, as of June 1999, BLS had announced three
methodological changes that had not yet been
implemented.

Four former members of the Boskin
Commission-Michael J. Boskin, Ellen R. Dulberger,
Robert J. Gordon, and Dale Jorgenson-responded to
our questions about the extent to which recent
methodological changes in the CPI have reduced its
overstatement of the changes in the cost of
living-that is, bias-as defined in the
Commission's December 1996 report.2 Although all
four of these former Boskin Commission members
said that the seven methodological changes made to
the CPI have reduced some of the bias in the CPI,
they had different responses regarding the extent
of the remaining bias. As shown in figure 1, their
point estimates of the remaining bias varied from
0.73 to 0.9 percentage points annually after
taking into account those seven changes.

Figure 1:  Former Boskin Commission Members' Point
Estimates for the Remaining Bias in the CPI,
Considering the Seven Methodological Changes in
Effect as of June 1999

aDr. Boskin gave the range of 0.8 to 0.85 for his
point estimate of the remaining bias because he
believes the bias cannot be estimated precisely.
Source:  Former Boskin Commission members'
responses to GAO questions in 1999 and the Boskin
Commission's December 1996 report.

     The former Boskin Commission members believe
that most of the remaining bias is due to what the
Commission referred to as "new products/quality
change bias."  According to the Commission, this
type of bias occurs when new products are not
included in the CPI or when they are included
after a long delay, which results in the CPI not
capturing price decreases that often occur after a
product is introduced in the marketplace. Further,
according to the Commission, new products/quality
change bias occurs when the CPI does not
adequately measure the portion of a price increase
that is due to an improvement in the quality of a
product or service instead of to an increase in
the cost of living.

Background
The CPI is a widely used indicator of inflation.
BLS calculates the CPI each month based on a
complex methodology that uses prices collected on
a wide range of goods and services from outlets
throughout urban areas in the United States. The
CPI significantly affects the U.S. budget because
automatic cost-of-living adjustments in federal
income tax brackets, federal retirees' pensions,
and various federal programs, such as Social
Security and food stamps, are linked to changes in
the CPI.

Although the CPI is used to make cost-of-living
adjustments, it is not, technically, a cost-of-
living index. A true cost-of-living index would
measure the change in the cost of obtaining a
fixed level of economic well-being or
utility-something that is not clearly defined.
What the CPI measures is clearly defined; it
measures the change in prices of a market basket
of goods and services purchased directly by urban
consumers. These goods and services fall into
eight major groups, including food and beverages,
housing, apparel, transportation, medical care,
recreation, education and communication, and other
goods and services that people buy for day-to-day
living. (See app. I for more background
information on the CPI.)

In June 1995, out of concern that the CPI may have
been inaccurately measuring cost-of-living
increases, the U.S. Senate Finance Committee
created the Advisory Commission to Study the
Consumer Price Index. The Commission was tasked
with determining how effectively the CPI measured
the cost of living and inflation and recommending
ways to improve the CPI. Appointed to the
Commission were five economists, including Michael
J. Boskin, Ellen R. Dulberger, Robert J. Gordon,
Zvi Griliches, and Dale Jorgenson.

In December 1996, the Boskin Commission issued its
final report,3 which took the position that
different sources of bias in the CPI methodology
were causing the CPI to overestimate cost-of-
living increases by approximately 1.1 percentage
points per year.4 An underlying reason for the
bias, according to the Commission, was that the
CPI was based on prices of a fixed market basket
of goods and services of fixed quantities from a
fixed set of outlets. This resulted in the CPI not
taking into account how consumers substitute one
item or outlet for another, that is, respond to
price increases and opportunities to pay less for
goods and services by changing what they purchase
and where they shop.

In addition, the Commission reported that the CPI
was biased because BLS did not promptly include
new products in the CPI and, therefore, did not
measure price changes in many new products.
According to the Commission, because the price of
a product often decreases after its introduction
in the marketplace, delays in including new
products in the CPI prevent the CPI from measuring
these price decreases. Further, the Commission
said that the CPI often erroneously attributed
price increases to inflation instead of to quality
improvements-advances in products and services
that have affected peoples' lives. For example,
the Commission said that the CPI was biased
because it did not estimate how much of the
increase in medical care costs was due to more
successful surgeries and more effective drugs or
the replacement of surgery with drugs. Table 1
shows the sources and estimated amounts of bias
reported by the Boskin Commission.

Table 1: The Boskin Commission's December 1996
Estimates of Bias in the CPI
           Sources of bias             Commission'
                                        s December
                                              1996
                                       estimate of
                                              bias
                                       (percentage
                                            points
                                         annually)
Commission's     Description of bias:   
name for bias    Inability of the CPI
                methodology to
                adequately account for
                -
Upper level      Consumers purchasing          0.15
substitution     an item from one BLS
                category of goods
                (called an "item
                stratum") instead of a
                preferred higher
                priced item in a
                different item
                stratum-for example,
                renting a video
                instead of going out
                to see a movie.
Lower level      Consumers purchasing          0.25
substitution     an item different from
                a preferred higher
                priced item from the
                same item stratum-for
                example, purchasing
                lower priced granny
                smith apples instead
                of their normally
                purchased red
                delicious apples.
New              Price changes in new          0.60
products/quality products not included
change           in the CPI, price
                decreases in products
                before they are
                included in the CPI,
                and changes in the
                quality of products.
New outlets      Consumers changing            0.10
                where they make
                purchases in order to
                get a better price-for
                example, shopping at a
                warehouse store
                instead of a
                department store.
Total                                         1.10
Plausible range                        0.80 - 1.60
Source: Boskin Commission's December 1996 report
and descriptions by GAO based on that report.

The Commission gave 1.1 percentage points annually
as the point estimate of the bias in the CPI and
reported that the plausible range for the bias was
from 0.8 to 1.6 percentage points annually. The
implication of the Commission's estimate is that
the cost of living may have been increasing less
than the CPI, decreasing when the CPI showed
little or no increase, or decreasing more than the
CPI when it showed a decrease.

Although the Boskin Commission's estimate of bias
in the CPI was controversial, many economists
agreed that the CPI probably overstated the level
of inflation. However, many believed that there
was insufficient data to make an estimate of bias.5
BLS agreed with the Commission's estimate of upper
level substitution bias but stated that the
estimates for lower level substitution and new
outlets bias may be too high.  BLS expressed
skepticism about the Commission's estimate of new
products/quality change bias and noted that the
evidence the Commission used for this estimate was
sparse and that the Commission did not have a well-
defined methodology for deriving its estimate. In
response, the members of the Commission said that
future research might find that  some parts of the
Commission's estimate were too high, but it might
also find that other parts were too low.

Scope and Methodology
To identify all the methodological changes made to
the CPI since December 1996 when the Boskin
Commission issued its final report, we obtained a
list of methodological changes from BLS in June
1999. BLS' list of 10 changes included all
methodological changes that took effect for the
CPI calculation between December 1996 and June
1999 and all methodological changes announced
before June 1999 that were not yet effective. In
addition, BLS provided brief descriptions of the
methodological changes; reference materials; and,
where available, an estimate of the effect of the
changes on the CPI and/or subindexes of the CPI.

To obtain the opinions of the five former Boskin
Commission members on how much of the bias in the
CPI that the Commission estimated in its December
1996 report remains after recent methodological
changes to the CPI, we sent the Commission members
a questionnaire. The questionnaire included
information provided by BLS on the 10
methodological changes made since December 1996 or
announced to take effect in the future.  It asked
for estimates of the extent to which the
Commission's estimate of bias could be reduced or
increased as a result of each methodological
change. In addition, for each methodological
change that the former Commission members believed
may have reduced or increased the bias, the
questionnaire asked which source of bias may have
been affected. (See app. II for a copy of the
questionnaire sent to the Commission members.) We
also provided the former Commission members copies
of all the reference material provided by BLS.

Our questionnaire did not ask the former
Commission members about methodological changes
that took effect for the CPI calculation before
December 1996 because (1) the Commission issued
its final report in December 1996 and (2) its
final report took into account methodological
changes that had been made to the CPI since the
Commission's September 1995 Interim Report.

Four former members of the Boskin Commission
responded to our questionnaire. The fifth former
Commission member did not respond due to illness.

We conducted follow-up telephone interviews with
all the respondents to obtain clarification on
their questionnaire responses where needed.
Information obtained during follow-up interviews
indicated that in some cases, there was some
misunderstanding about the meaning of some of the
questions in the questionnaire. Therefore, in some
cases, our report is based on the responses to
questions asked in the follow-up interviews
instead of the responses to the questionnaire.
Because our objective was to obtain the expert
opinion of the former members of the Boskin
Commission, we did not attempt to validate their
responses or estimates.

We conducted our work in Washington, D.C., from
June 1999 through December 1999 in accordance with
generally accepted government auditing standards.

Methodological Changes to CPI Since Boskin
Commission Report
Between December 1996, when the Boskin Commission
issued its final report, and June 1999, when we
began this review,  BLS had made seven
methodological changes that affected the
calculation of the CPI.  In addition, as of June
1999, BLS had announced three methodological
changes that had not yet been implemented.6

In addition to the seven completed and three
planned methodological changes, BLS implemented a
major revision of the CPI in January 1998. BLS
implements major revisions roughly every 10 years
to keep the CPI current and accurate. In the 1998
major revision, as with major revisions in the
past, a new market basket of goods and services
was introduced into the CPI. The 1998 major
revision made other changes, including changes in
geographic areas from which prices are collected.
Since BLS considers major revisions to be part of
the established CPI methodology, it did not
include the 1998 major revision in its list of
methodological changes.

The seven methodological changes already completed
as of June 1999 are the following:

ï¿½    Changes in the Methods for Pricing of
Hospital Services (hospital services pricing):7 In
January 1997, BLS redefined its categories -
called "item strata" - of goods and services
related to hospital services for which it collects
information on price changes. For example, BLS
eliminated the item strata for hospital rooms and
created a new item strata for hospital services.
According to BLS, it did this in order to gather
data on price changes for treatments, even if
aspects of the treatments had changed. For
example, BLS wanted to be able to compare the
price of a treatment, even if the treatment
changed from being given on an inpatient basis to
an outpatient basis. BLS did not provide an
estimate of the annual effect of this change on
the CPI.

ï¿½    Adjustment of Personal Computer Prices
(personal computer pricing): In January 1998, BLS
implemented a new method for measuring changes in
personal computer prices. The new method
decomposes the price of personal computers into
implicit prices for each important feature and
component. According to BLS, the new method
enables BLS to adjust for quality improvements in
new models when it measures price increases. BLS
did not estimate the annual effect of this change
on the overall CPI, but it estimated that the
change lowered the rate of growth in the personal
computer index by approximately 6.5 percent in
1998.

ï¿½    Accounting for Consumer Substitution within
CPI Categories (implementation of the geometric
mean estimator): In January 1999, BLS began using
a new formula called the geometric mean estimator
to combine the prices collected on items each
month into indexes for the various item strata.
(For example, apples and chicken each have their
own item stratum, and BLS calculates indexes for
price changes in the item strata. 8) The geometric
mean estimator replaced a formula called the
arithmetic mean estimator in most item strata. The
geometric mean estimator takes into account that
from one period to the next consumers may change
the quantity that they purchase of a particular
item within an item stratum because of a change in
the relative price of the item.

The item strata for which the geometric mean
estimator is being used represent approximately 61
percent of the consumer spending covered by the
CPI. They include all item strata in the "food and
beverages," "apparel," and "other goods and
services" major groups. According to BLS, the
geometric mean is not being used for some item
strata in the other major groups because consumers
are less likely to respond to price increases in
those item strata by making item substitutions and
thereby changing the quantity that they purchase.
Therefore, BLS is not using the geometric mean
estimator in certain item strata, such as rent of
primary residence, electricity, cable television,
physicians' services, and eyeglasses and eye care.

BLS estimated that implementation of the geometric
mean estimator will reduce the rate of growth of
the CPI by approximately 0.2 percentage point per
year.

ï¿½    Adjustment of Television Prices (television
pricing): In January 1999, BLS implemented a new
method for measuring changes in television prices.
The new method decomposes the price of televisions
into implicit prices for each important feature
and component. According to BLS, the new method
enables BLS to adjust for quality improvements in
new models when it measures price increases. BLS
did not estimate the annual effect of this change
on the overall CPI, but it estimated that the
change would have lowered the rate of growth in
the television index by approximately 0.1 percent
per year during the period August 1993 to August
1997.

ï¿½    Changes to the Treatment of Utility Refunds:
In January 1999, BLS began disregarding earlier
periods' utility refunds that appear in a
consumer's current bill. Under the previous method
for gathering prices for utilities, the refund
amount was subtracted from the consumer's current
bill, and the price used in the CPI calculation
could have been as low as zero. BLS did not
provide an estimate of the annual effect of this
change on the CPI.  However, according to BLS, the
change will not affect the long-run rate of growth
of the CPI.

ï¿½    Treating Mandated Pollution Control Measures
as Price Increases: In January 1999, BLS began
treating price increases due to legislatively
mandated modifications to goods and services for
pollution control purposes as price increases.
Before this change, such price increases were
regarded as a change in quality and were excluded
from the CPI. BLS did not provide an estimate of
the annual effect of this change on the CPI.
However, according to BLS, any effect will be an
increase in the CPI, and the extent of the
increase will depend on what future requirements
for pollution controls are implemented.

ï¿½    Revision of the Housing Estimation System:9
In January 1999, BLS implemented a new housing
sample of rental units and an estimation method
for homeowners' housing costs, which uses only
data on rental units. Before this change, the
housing sample included owner-occupied and rental
units, and homeowners' housing costs were
estimated from the costs experienced by renters
whose units were comparable to owner-occupied
units in the housing sample. BLS did not provide
an estimate of the annual effect of this change on
the CPI.

The three planned methodological changes that were
not implemented for the CPI calculation as of June
1999 are the following:

ï¿½    Change from Area- to Item-Based Rotation
Procedures (sample rotation procedures):10 BLS has
changed the process for updating the samples of
outlets where it selects items for price
collection. The new samples are selected according
to expenditure category, such as televisions and
soups. The old process updated outlet and item
samples by geographic area. In part because of the
large number of outlets and items covered, the
process of sample rotation takes several years to
complete.  According to BLS, the first outlets and
items selected under the new sample rotation
procedures were used in the CPI for October 1999.

According to BLS, this change will enable it to
introduce new goods into the CPI in a more timely
fashion. BLS did not provide an estimate of the
annual effect of this change on the CPI.

ï¿½    Reduction of the Average Age of Expenditure
Weights (age of expenditure weights): According to
BLS, it has begun work to enable the future
implementation of a methodological change that
will result in more timely information being used
when the component of the CPI calculation referred
to as "expenditure weights" is updated (the
frequency of the updates is the subject of the
methodological change discussed below).
Expenditure weights are used to determine the
extent to which price changes on certain goods
will affect the overall CPI. Expenditure weights
represent, in essence, the proportion of the
typical consumer's expenditures that are spent on
an item or category of goods. Therefore, if ground
beef had an expenditure weight of one-third of 1
percent, that would mean that according to CPI
data, one-third of 1 percent of the typical
consumer's expenditures are made for ground beef.

BLS is increasing the size of the sample of
consumers surveyed to collect the data used to
develop expenditure weights.  According to BLS,
this methodological change, along with system
enhancements, will reduce the average age of the
data for developing expenditure weights from 3-ï¿½
years to 2 years beginning with the CPI for
January 2002. BLS did not provide an estimate of
the annual effect of this change on the CPI.

ï¿½    Increased Frequency of Expenditure Weights
Updates (expenditure weights updates): Beginning
with the release of the CPI for January 2002, BLS
plans to update expenditure weights every 2 years.
Expenditure weights have been updated roughly
every 10 years when BLS implemented a major
revision of the CPI. BLS did not provide an
estimate of the annual effect of this change on
the CPI.

Former Boskin Commission Members' Responses on
Remaining Bias in the CPI
All four former Boskin Commission members said
that the methodological changes had reduced some
of the bias in the CPI. However, they had
different responses on the extent of the remaining
bias. Their responses about how the methodological
changes affected the different sources of bias
identified by the Commission indicated that they
share the view that (1) lower level substitution
bias has been greatly reduced and (2) new
products/quality change bias is still the largest
source of bias in the CPI.

Total Remaining Bias
Of the estimates of remaining bias in the CPI,
taking into consideration the seven methodological
changes that were already effective as of June
1999, Dr. Jorgenson had the highest point
estimate-0.9 percentage point-and the widest
plausible range-from 0.6 to 1.4 percentage points.
Dr. Gordon had the lowest point estimate-0.73
percentage point-and a plausible range from 0.5 to
1.2 percentage points. The point and plausible
range estimates of Drs. Boskin and Dulberger fell
between those of Drs. Jorgenson and Gordon.

Drs. Boskin, Dulberger, and Jorgenson all provided
0.6 percentage point as the lowest point in the
plausible range for remaining bias, taking into
consideration the seven methodological changes
that were already effective as of June 1999. Drs.
Boskin, Dulberger, and Gordon all provided 1.2
percentage points as the highest point in the
plausible range. Figure 2 shows the former
members' point and plausible range estimates of
remaining CPI bias taking into consideration (1)
the seven methodological changes already effective
as of June 1999; and (2) the combination of the
seven methodological changes already effective as
of June 1999 and the three planned methodological
changes.

Figure 2:  Former Boskin Commission Members' Point
and Plausible Range Estimates of the Remaining
Bias in the CPI

Note: Dr. Boskin gave a range for his point
estimate for each grouping of changes because he
believes the bias cannot be estimated precisely.
He gave 0.8 to 0.85 as the range for his point
estimate accounting for the seven completed
changes and 0.7 to 0.75 as the range for his point
estimate accounting for the combination of the
completed and planned changes.
Source:  Former Boskin Commission members'
responses to GAO questions in 1999 and the Boskin
Commission's December 1996 report.

Dr. Boskin said his estimates reflect his view
that the remaining bias is about three-quarters of
the 1.1 percentage points bias originally
estimated by the Commission. Further, Dr. Boskin
and Dr. Dulberger stated that they could not
estimate how much reduction in bias could be
attributed to some of the methodological changes.
Dr. Boskin said that he did not provide a specific
estimate when methodological changes likely
reduced the bias by less than 0.1 percentage
point. In part, to compensate for the reductions
they could not provide specific estimates for, Dr.
Boskin's and Dr. Dulberger's point estimates for
the remaining bias in the CPI show a greater
amount of bias reduction than the total of the
specific estimates they made for individual
methodological changes.

Dr. Boskin said that when the Boskin Commission
developed its estimate of bias, it took into
account the changes BLS had announced it was
planning to make. Therefore, according to Dr.
Boskin, the 1.1 percentage points estimate should
not be reduced as a result of the changes BLS had
announced but not completed before December 1996.11

Dr. Gordon said that he used estimates published
by the President's Council of Economic Advisers as
the basis for some of his estimates of bias
reduction due to individual methodological
changes. 12 Dr. Gordon said that when the Boskin
Commission developed its estimate of bias, it did
not take into account the changes BLS had
announced it was planning to make. Therefore, he
believes the 1.1 percentage points estimate should
be reduced as a result of some of the changes BLS
had announced but not completed before December
1996. Also, unlike the estimates of the other
former Commission members, Dr. Gordon's point
estimate for remaining bias incorporated an amount
for bias reduction that he attributed to the 1998
major revision.

Dr. Jorgenson's estimates of remaining bias take
into account only the bias reduction-0.2
percentage point-due to the implementation of the
geometric mean estimator. Dr. Jorgenson said that
he was not willing to make estimates for
individual methodological changes when BLS had not
estimated how the methodological changes would
affect the CPI.

Reduction in Bias Categorized by Source of Bias
As discussed in the background section, the Boskin
Commission's December 1996 report attributed the
bias in the CPI to four different sources. For
each methodological change that the former Boskin
Commission members said had or would affect the
Commission's estimate of bias, we asked the former
Boskin Commission members to identify the source
of bias affected. We also asked them to estimate
how much of the bias would be reduced or increased
by the methodological changes.

Table 2 summarizes, by source of bias, the four
former Commission members' responses about the
effect of methodological changes on the bias in
the CPI.

Table 2: Responses of Four Former Boskin
Commission Members Regarding the Effect of 10
Methodological Changes to the CPI by Source of
Bias
Source of bias         Boskin  Responses of former commission members
                 Commission's  regarding amount of bias reduced by 10
                     December  methodological changes
                         1996
                  estimate of
                         bias
                  (percentage
                        point
                    annually)
Upper level              0.15  Different responses were provided. Dr. Gordon
substitution                   estimated a 0.12 reduction. The others said
                              they could not estimate the amount of
                              reduction. Dr. Boskin indicated that the
                              reduction would be less than 0.10, and Dr.
                              Dulberger said the reduction would be very
                              small.
Lower level              0.25  All estimated a 0.2 reduction due to
substitution                   implementation of the geometric mean
                              estimator.
New                      0.60  Different responses were provided. Dr. Gordon
products/quality               estimated a 0.12 reduction due to four
change                         different changes. The others said they could
                              not estimate the amount of reduction. Drs.
                              Boskin and Dulberger indicated the reduction
                              would be small. All said changes in personal
                              computer and television pricing may have
                              affected this bias.
New outlets              0.10  All said no reduction occurred.
Source: Former Boskin Commission members'
responses to GAO questions in 1999 and the Boskin
Commission's December 1996 report.

Upper Level Substitution Bias
The former Boskin Commission members provided
different responses about the effect of
methodological changes related to upper level
substitution bias. Unlike Drs. Boskin, Dulberger,
and Jorgenson, Dr. Gordon stated that the
Commission's 0.15 percentage point estimate for
upper level substitution bias can be reduced by
0.1 due to the 1998 major revision to the CPI.13
Dr. Gordon also said an additional 0.02 percentage
point would be eliminated due to the change in
expenditure weights updates planned to go into
effect for the CPI calculation in January 2002.
Therefore, according to Dr. Gordon, the remaining
upper level substitution bias will be
approximately 0.03 percentage point annually.

Drs. Boskin and Dulberger said that the planned
methodological change on expenditure weights
updates will reduce upper level substitution bias.
Both said they could not estimate the extent of
the reduction, but Dr. Boskin indicated the
reduction would be less than 0.10 percentage
point, and Dr. Dulberger said the reduction would
be very small. Dr. Jorgenson said he could not
estimate how the expenditure weights updates
change would affect upper level substitution bias
or whether the effect would be an increase or a
decrease in upper level substitution bias. The
former Commission members did not identify any
other methodological changes that would affect
upper level substitution bias.

Lower Level Substitution Bias
All four former Boskin Commission members agreed
that implementation of the geometric mean
estimator eliminated about 0.2 percentage point of
lower level substitution bias. This indicates that
only about 0.05 remains of the 0.25 percentage
point of annual bias attributed to lower level
substitution bias by the Boskin Commission's
December 1996 report.

New Products/Quality Change Bias
The four former Boskin Commission members gave
somewhat different responses regarding the details
of changes related to new products/quality change
bias. However, none believed that new
products/quality change bias has been
significantly reduced. Drs. Boskin, Dulberger, and
Gordon indicated that new products/quality change
bias has been slightly reduced. Dr. Jorgenson said
he could not estimate the effect of any of the
methodological changes related to new
products/quality change bias and did not know if
the changes increased or decreased new
products/quality change bias.

All four former Boskin Commission members
responded that methodological changes in personal
computer and television pricing may have had an
effect on new products/quality change bias. Dr.
Gordon estimated that changes in personal computer
and television pricing reduced new
products/quality change bias by 0.04 and 0.02
percentage points annually, respectively. (Dr.
Gordon said he based his estimate for changes in
personal computer prices on an estimate made by
the Council of Economic Advisers.) Drs. Boskin and
Dulberger said they could not estimate the effect
of these changes individually, but they indicated
that the changes had slightly reduced new
products/quality change bias. Dr. Jorgenson said
he could not estimate the effect of the changes.

Drs. Boskin, Dulberger, and Gordon noted that BLS'
estimate of the effect of the methodological
change in personal computer pricing appears very
low in light of other estimates of the extent to
which personal computer prices have been
declining. Dr. Dulberger questioned how BLS was
using the results from the new methodology to
adjust personal computer prices. Similarly, for
the methodological change in television pricing,
Drs. Boskin, Dulberger, and Gordon noted that BLS'
estimate of the effect of this methodological
change on the television subindex was too low.
Drs. Boskin and Gordon cited a BLS research paper
that showed the change would have a much greater
effect, and Dr. Dulberger stated that there looks
like there may be a problem or error in BLS'
estimate.  In response to the former Commission
members' comments, BLS stated that it made the
estimates for personal computers and televisions
that were included in our questionnaire by
comparing the subindexes calculated without the
respective methodological changes with the
subindexes calculated with the methodological
changes.  BLS said that for these estimates, it
used data from specific time periods-1998 for
personal computers and August 1993 to August 1997
for televisions-and the effect of the changes may
be different in other time periods.

Dr. Gordon attributed a 0.01 percentage point
reduction in new products/quality change bias to
the change in hospital services pricing. He based
his estimate of 0.01 percentage point on estimates
by the Council of Economic Advisers. Drs. Boskin,
Dulberger, and Jorgenson did not attribute any
reduction in bias to hospital services pricing.
Drs. Boskin and Dulberger said that the Commission
knew that the methodological change in hospital
services pricing was about to be made and took any
potential reduction of bias into account when it
developed its December 1996 estimate.

Dr. Gordon said that the planned change in sample
rotation procedures will reduce new
products/quality change bias by about 0.05
percentage point. He based this estimate on an
estimate by the Council of Economic Advisers. The
other three former members did not believe this
change reduced the 1.1 percentage points estimated
bias. Dr. Boskin said that the Commission knew BLS
planned to change its sample rotation procedures.
Therefore, he said, the Commission took into
account any reduction in bias that might occur due
to this change when it developed its estimate of
1.1 percentage points bias.

New Outlets Substitution Bias
None of the recent methodological changes
addressed outlet substitution bias, according to
the former Boskin Commission members.

Agency Comments
     We provided copies of a draft of this report
to BLS, the Office of Management and Budget, and
each of the former Boskin Commission members for
their review and comment.

     In a letter dated January 18, 2000, the
Commissioner of BLS stated that BLS does not
believe it is currently possible to produce
reliable estimates of bias in the CPI. She stated
that the measurement issues considered by the
Boskin Commission are complex and that there is
considerable uncertainty attached to the Boskin
Commission's estimates, especially those relating
to new products/quality change bias.  The BLS
Commissioner said that BLS has been at the
forefront of price measurement research and
operational innovation and has introduced many
important improvements in CPI methods over time.
She said that BLS plans to produce an additional
index beginning in 2002 that will more completely
account for consumer response to relative price
change.  She also said that BLS will continue to
develop and evaluate potential improvements to CPI
methods and to implement any that can further
improve the accuracy of the CPI.  The BLS
Commissioner's letter is reprinted in appendix
III. The Office of Management and Budget did not
provide any comments on this report.

     In a letter dated January 13, 2000, Dr.
Boskin commented about the difficulty of measuring
price changes, the importance of some of the
methodological changes, and the good job BLS has
been doing in the last few years in improving the
CPI.  He said that a sizable bias remains in the
CPI, which he hopes will be reduced by future
improvements, but that some bias is inevitable.
Further, Dr. Boskin said that making the necessary
improvements in the CPI will be difficult, will
require resources, but will produce benefits far
exceeding their cost.  Dr. Boskin's letter is
reprinted in appendix IV.  Dr. Dulberger provided
technical comments; changes were made in the
report, where appropriate.

     As agreed with your office, unless you
announce the contents of this report earlier, we
plan no further distribution until 30 days after
the date of the report.  At that time, we will
send copies of this report to Senator William
Roth, Chairman of the Senate Committee on Finance;
and Representative Bill Archer, Chairman, and
Representative Charles B. Rangel, Ranking Minority
Member, House Committee on Ways and Means.  We
will also send copies to the Honorable Jacob J.
Lew, Director of the Office of Management and
Budget; the Honorable Alexis Herman, Secretary of
Labor; the Honorable Katharine G. Abraham,
Commissioner of BLS; and other interested parties.
We will also make this report available to others
on request.

     If you have any questions regarding this
report, please contact me or Kathy Peyman at 202-
512-8676.  Key contributors to this assignment
were Kathleen Scholl and Martin de Alteriis.

     Sincerely yours,

     Nancy R. Kingsbury
Acting Assistant Comptroller General

_______________________________
1BLS is a part of the U.S. Department of Labor.
2One former Commission member, Zvi Griliches, was
not able to respond to our questions due to the
illness that led to his death in November 1999. We
greatly appreciate the assistance Dr. Griliches
gave us on various GAO reviews in the past.
3Final Report of the Advisory Commission to Study
the Consumer Price Index, Committee on Finance,
United States Senate, December 4, 1996.
4The Boskin Commission issued an interim report on
September 15, 1995, which estimated the CPI's
overstatement of the cost of living to be 1.5
percentage points per year. However, the
Commission's final report revised this estimate,
in part, to take into account the effect of
methodological changes that had been made to the
CPI since issuance of the interim report.
5Most of the disagreement with the Boskin
Commission's estimate focused on the Commission's
estimate of new product/quality change bias.
6 Our review did not cover an additional
methodological change involving the adjustment of
prices for audio and video products that BLS
implemented in January 2000.  This change had not
been announced as of June 1999 when we obtained
from BLS the list of methodological changes used
for this review.
7BLS announced before issuance of the Boskin
Commission's December 1996 report that it planned
to make this change to hospital services pricing.
8See appendix I for a description of how the
indexes for the item strata are used to develop
the CPI.
9BLS announced before issuance of the Boskin
Commission's December 1996 report that it planned
to  make this revision of the housing estimation
system.
10 BLS announced before issuance of the Boskin
Commission's December 1996 report that it planned
to make this change in sample rotation procedures.
11These changes were changes in hospital services
pricing, revision of the housing estimation
system, and changes in sample rotation procedures.
12Economic Report of the President, Transmitted to
the Congress February 1999 Together with the
Annual Report of the Council of Economic Advisers,
U.S. Government Printing Office, 1999, p. 94.
13Page 44 of the Boskin Commission's December 1996
report states: "The BLS is preparing for a
benchmark revision in January 1998, when the CPI
will incorporate new expenditure weights .
However, BLS will retain the modified Laspeyres
formula, so that our estimates of bias will carry
over to the revised CPI."

Appendix I
Background Information on the Consumer Price Index
Page 28                     GAO/GGD-00-50 CPI Bias
The Bureau of Labor Statistics (BLS) produces the
Consumer Price Index (CPI) by measuring the
average change over time in the prices paid by
urban consumers for a market basket of consumer
goods and services. The selection of items for the
market basket is determined from detailed records
of purchases made by thousands of individuals and
families, as reported on periodic surveys. The
items selected for the market basket, such as
potatoes, are to be priced each month at specific
retail outlets, such as grocery stores and
supermarkets, in urban areas throughout the
country. According to BLS, in 1999, price takers
collected the prices of about 96,000 items (goods
and services) in 87 urban areas of the country.
These prices were collected from about 30,000
retail and service establishments and from about
27,000 landlords and tenants, who provided data on
housing units.

The CPI is used as a measure of price changes to
make economic decisions in the private and public
sectors. According to BLS, the CPI has three major
uses:

Economic indicator of inflation: The
administration, Congress, and the Federal Reserve
use trends in the CPI as an aid to formulating
fiscal and monetary policies. Business and labor
leaders as well as private citizens use the CPI as
a guide to making economic decisions.

Escalator for wages, benefit payments, and tax
brackets: Collective bargaining units use the CPI
to adjust the wages of workers. Also, it is the
basis for automatic changes in some federal
benefit payments. For example, in December 1998,
as a result of changes in the CPI, payments to 44
million Social Security beneficiaries and 6.6
million Supplemental Security Income recipients
were adjusted for inflation. The benefits of
approximately 18 million food stamp recipients in
1999 were affected by changes in the CPI. Payments
to millions of railroad, military, and federal
civilian retirees and survivors are also affected
by changes in the CPI. The CPI is also used to
adjust key elements of the individual income tax
to limit the extent to which individuals must pay
higher taxes solely because of inflation. For
example, the amount allowed for personal
exemption, the amount of the standard deduction,
and tax brackets are adjusted annually according
to changes in the CPI.

Deflator of selected economic statistical data
series: The CPI is used to adjust selected
economic statistical series for price changes and
to translate these series into inflation-free
dollars. Examples of data series that are adjusted
by the CPI include retail sales, hourly and weekly
earnings, and components of the National Income
and Product Accounts.

The CPI was initiated during World War I, when
rapid increases in the prices of goods and
services, particularly in shipbuilding centers
where workers were demanding wage adjustments,
made such an index essential for calculating
cost-of-living adjustments. In 1921, BLS began
regular publication of an index representing the
expenditures of urban wage and clerical workers,
which was then called the Cost-of-Living Index.
The name of the index was changed to the CPI
following controversy during World War II over the
index's validity as a measure of the cost of
living. According to BLS, the CPI has always been
a measure of the changes in prices for goods and
services purchased for family living.1

Major revisions have been made to the CPI about
once each decade to update the market basket, with
the most recent revision occurring in January
1998. Because consumers' buying habits change, new
studies were made of what goods and services
consumers were purchasing; and major revisions to
the CPI were made in 1940, 1953, 1964, 1978, and
1987 as well as 1998. In the 1978 major revision,
BLS began publication of a new index for all urban
consumers-the CPI-U. According to BLS, the CPI-U,
which represents the expenditures of about 87
percent of the population, takes into account the
buying patterns of professional employees,
part-time workers, the self-employed, the
unemployed, and retired people as well as those
previously covered in the CPI. BLS has continued
publication of the older index, the CPI-W, which
represents the expenditures of urban wage and
clerical workers or about 32 percent of the
population.

Construction of the CPI
BLS begins construction of the CPI by selecting a
collection of goods and services usually bought by
urban consumers. The collection of goods and
services, called items, is known as the market
basket. The CPI market basket is developed from
detailed expenditure information that is provided
by thousands of families and individuals who
participate in the Consumer Expenditure Survey
(CEX), which is conducted for BLS by the Bureau of
the Census over several years. For example, the
1998 CPI revision was based on CEX data collected
from 1993 through 1995, from about 36,000
individuals and families. BLS uses expenditure
data from the CEX to select the categories of
items from which it selects specific, unique
commodity and service items to be priced for the
CPI.

BLS measures price changes each month by checking
the prices of the items in the market basket and
then comparing the aggregate costs of the market
basket with those for the previous month. BLS
price takers obtain prices for most of the items
by visiting or contacting thousands of retailers,
service providers, and landlords and tenants each
month.

Classification of Market Basket Items
BLS classified all CEX expenditure items into 211
item strata, which are arranged into eight major
groups: (1) food and beverages; (2) housing; (3)
apparel; (4) transportation; (5) medical care; (6)
recreation; (7) education and communication; and
(8) other goods and services, such as haircuts and
tobacco smoking products. Taxes that are directly
associated with the prices of specific goods and
services, such as sales and excise taxes, are also
included.2

Expenditure Weights of Market Basket Items
Expenditure weights are used to give proportionate
emphasis for price changes of one item in relation
to other items in the CPI. Expenditure weights
allow the CPI to distinguish between items that
have a major impact on consumers and to provide
appropriate emphases to price changes associated
with these items. For example, if ground beef were
assigned a weight representing about one-third of
1 percent of the expenditures of the typical urban
consumer; and if beef steaks, such as sirloin and
rib steaks, were assigned a smaller weight
representing about one-tenth of 1 percent, then
the price changes of ground beef would have about
3 times as much impact on the overall CPI as
similar price changes for beef steaks.

Weights derived from consumers' expenditures, as
reported in the CEX, are assigned to the 211 item
strata. To compute the weights, BLS first totals
the amount spent on an item stratum, such as white
bread, during the base weighting period by CEX
respondents, whom BLS refers to as consumer units.3
BLS then divides that total by the number of
consumer units, which results in an average
expenditure per unit. Next, the average
expenditures per unit are weighted with data from
the Decennial Census to represent the U.S. urban
population. To do so, the average expenditures per
unit are multiplied by certain factors to
represent the geographic dispersion of the urban
population. Finally, these nationwide urban
expenditures on the market basket items are
totaled into an aggregate amount. The 211
expenditure weights are the percentages of this
aggregate amount that are spent on each of the 211
item strata (e.g., white bread).

Expenditure weights remain fixed until the next
major revision of the CPI and serve as a benchmark
from which price comparisons are calculated. The
weights of the components for the 1998 major
revision are those derived from the 1993 through
1995 CEX.

Collecting Prices of Market Basket Items
Each month, BLS price takers visit or call
thousands of retail stores, service
establishments, rental units, and doctors' offices
all over the United States. Each month, they
record the prices of about 80,000 items.4

To determine which retail outlets its price takers
should visit to obtain monthly price quotations
for nonrent items, BLS sponsors the continuing
Point-of-Purchase Survey (POPS), which is
conducted by the Bureau of the Census. The survey
respondents are asked if they purchased any items
in an expenditure category of items, such as
televisions and soups, and, if so, the names of
all the outlets where they made their purchases.
BLS uses the results from the survey to select
outlets from which to collect prices on items to
be included in the CPI market basket.

BLS price takers visit each selected retail outlet
to initially select items that will be priced
either monthly or bimonthly. For each outlet,
categories of items are selected for pricing.
Using probability selection methods that are based
on revenues and volume information that is
provided by the retail outlet, BLS price takers
use a table of random numbers to select for
pricing a unique item within the specified
categories.

BLS collects rent prices for rental units in a
different manner from that used to identify and
price other items in the market basket. BLS uses
monthly price changes of rental units in the CPI
housing survey for the residential rent and
homeowners' equivalent rent items in the CPI
housing component.5 Residential rent and
homeowners' equivalent rent are estimated from
approximately 27,000 rented units. Each month, BLS
price takers obtain information from rental units
on the rent for the current month, the previous
month, and the services that the landlord
provides. These data are used to measure changes
in rent prices for residential rent as well as
homeowners' equivalent rent.

Replacement of Market Basket Items No Longer
Available for Pricing
BLS price takers attempt to collect price
information for the same item (e.g., 1 dozen pink
carnations, with greenery, wrapped in paper, and
not delivered) as during their previous visit to
the retail outlet or rental unit. However, in many
instances, an identical item is not available for
purchase in each subsequent visit. In these
situations, price takers are to follow certain
procedures to make a substitution-selection of a
new version (replacement) that is similar to the
old version of the item that is no longer
available.6

In selecting a substitution the price takers are
to follow specific guidance for choosing the new
version. In general, the price taker is to select
the item with specifications most consistent with
the old version.7 After the price taker selects a
new version and records the information about the
item, the information is sent to BLS headquarters
in Washington, D.C., where it is coded, entered
into computer systems, verified, and examined by
commodity analysts.

Adjustments by Commodity Analysts
Commodity analysts review the information
collected by the price takers to determine if the
original item and its substitution are comparable.
Generally, when the two items are considered
comparable, the difference between the prices of
the items is used in the calculation of the CPI.
However, if the commodity analysts find
dissimilarities between the original item and its
substitution, they may make a price adjustment to
account for the dissimilarities. The adjustments
are made to avoid counting in the CPI (as
inflation or deflation) any price changes due to a
difference in the quality, size, or quantity of
the original item and its substitute.

Commodity analysts make two types of
adjustments-direct and indirect. Analysts make
direct adjustments when they have data on the ways
the original and substitution items differ and
have information with which to assess the value of
those differences. To make these adjustments, the
analysts use the specific cost of a quality change
that can be estimated either by the manufacturers
of the items or by the use of statistical models,
which BLS calls hedonic regressions. Direct
adjustments are also made when an item's size or
quantity changes.

Indirect adjustments are made when the commodity
analysts do not have sufficient data to make
direct adjustments. Indirect adjustments impute
the pure price change from the original item to
its substitution by averaging the rates of price
changes experienced by the same type of items in
the CPI.8

Calculating the index
Following the review and adjustment of individual
prices by the commodity analysts, the prices are
aggregated into lower level indexes, such as
"funeral expenses." Then the lower level indexes
are aggregated into higher level indexes, such as
the overall "all-items" indexes for the CPI-U and
CPI-W. To aggregate the prices, BLS uses the
geometric mean and the Laspeyres arithmetic mean
formulas. Basically, a number of lower level
indexes are calculated from price data, and these
indexes are then combined to form higher level
indexes.

Lower level indexes
Each month BLS calculates approximately 7,000
monthly lower level indexes that BLS refers to as
item stratum indexes.9 These indexes measure the
price change for a single item stratum in a
geographic area, such as watches in St. Louis, or
dental services in Pittsburgh. Most of the lower
level indexes are calculated with the geometric
means formula. The rest of the lower level indexes
are constructed with the Laspeyres arithmetic mean
formula. The weights used in the lower level index
calculations come from the continuing POPS.

BLS uses the geometric mean formula in lower level
indexes that represent approximately 61 percent of
total consumer spending. In contrast to the
arithmetic mean formula that uses fixed (constant
over time) quantity weights, the geometric mean
formula employs a set of fixed expenditure
proportions as weights in the calculation. This
difference recognizes that consumers alter the
quantities of goods or services within the lower-
level index. For example, the geometric mean is
used to calculate the "ice cream and related
products" lower level index to allow for the
changes in purchases that consumers make when the
price of ice cream increases (or decreases) in
relation to the price of frozen yogurt. The
geometric mean formula also treats price increases
and decreases symmetrically so that if a price
increases one month and then decreases by the same
amount the next month, the index will essentially
be at the same level as it was before the price
initially increased.

Generally, the Laspeyres arithmetic mean formula
is used in lower level indexes in the categories
of shelter services, utilities and government
charges, and medical services. These are the
categories in which, according to BLS, consumers
have a limited ability to make substitutions in
their purchases. For example, for utilities and
government charges, consumers would have to move
to different localities to change their purchases
because these services are provided by regulated
monopolies or local governments.

Higher level indexes
The Laspeyres formula uses lower level index
numbers and fixed-quantity weights to calculate
various higher level indexes. These higher level
indexes are weighted averages of various subsets
of the lower level indexes, or, in the case of the
U.S. city average all-items index, a weighted
average of all 8,018 lower level indexes. Examples
of higher level indexes include those for regions,
such as the south or midwest; and indexes for
expenditure categories, such as footwear or bakery
products. The weights in calculations of the
higher level indexes are the quantities of the
goods and services that were purchased at the time
the CPI market basket was established, which are
the quantities purchased by urban consumers from
1993 through 1995. The resulting index numbers can
be compared over time to indicate how much more
(or less) it costs consumers to purchase the items
in that index.

_______________________________
1 According to BLS, it has used a cost-of-living
conceptual framework in making decisions about
constructing the CPI. A cost-of-living index would
measure changes over time in the amount that
consumers need to spend to reach a certain utility
level or standard of living. According to BLS, a
complete cost-of-living index would go beyond the
items included in the CPI's market basket and take
into account other governmental or environmental
factors, such as crime and water quality, that
affect consumers' well-being.
2The CPI includes various governmental-charged
user fees, such as water and sewerage charges,
auto registration fees, and vehicle tolls. Taxes
not directly associated with the purchase of
consumer goods and services, such as income and
Social Security taxes, are excluded. In addition,
the CPI does not include investment items, such as
stocks, bonds, real estate, and life insurance,
because they relate to savings, not daily living
expenses.
3The CEX collects data from "consumer units,"
which are defined by BLS as either financially
independent, unrelated individuals or groups of
individuals who pool their resources to make joint
consumption decisions.
4Prices are not collected monthly on all items in
the CPI. Some are collected bimonthly, and rent
information is collected every 6 months for
housing units.
5BLS determines the value of owner-occupied
housing by using a rental equivalent method, which
estimates the amount of rent that would be paid if
the owner-occupied housing were rented.
6Price takers in the CPI housing survey return to
the same address in each collection period and
record information about the residential unit at
that address. Substitutions do not take place
between residential units as they do elsewhere in
the CPI. However, adjustments are made to make the
current unit similar to what it was at the prior
price collection.
7BLS has different procedures for the price takers
to follow to bring into the CPI new products or
services from the POPS that are not substitutions
for items that are in the market basket.
8 For additional information about how commodity
analysts decide whether to make adjustments and
the adjustment methods they use, see our report,
Consumer Price Index: Impact of Commodity
Analysts' Decicionmaking Needs to Be Assessed
(GAO/GGD-99-84, June 15, 1999).
9 Lower level indexes are calculated for each of
BLS' 186 priced item strata in each of 38
geographic regions. Including unpriced item
strata, there are a total of 8,018 lower level
indexes.

Appendix II
Questionnaire Sent to the Former Members of the
Boskin Commission
Page 40                     GAO/GGD-00-50 CPI Bias

Appendix III
Comments From the Bureau of Labor Statistics
Page 42                     GAO/GGD-00-50 CPI Bias

Appendix IV
Comments From Dr. Michael J. Boskin
Page 44                     GAO/GGD-00-50 CPI Bias

Related GAO Products
Page 47                     GAO/GGD-00-50 CPI Bias
Consumer Price Index:  Impact of Commodity
Analysts' Decisionmaking Needs to Be Assessed
(GAO/GGD-99-84, June 15, 1999).

Bureau of Labor Statistics:  Making the CPI More
Reflective of Current Consumer Spending (GAO/T-GGD-
98-115, Apr. 29, 1998).

Consumer Price Index:  More Frequent Updating of
Market Basket Expenditure Weights Is Needed
(GAO/GGD/OCE-98-2, Oct. 9, 1997).

Consumer Price Index:  Cost-of-Living Concepts and
the Housing and Medical Care Components (GAO/GGD-
96-166, Aug. 26, 1996).

Economic Statistics:  Status Report on the
Initiative to Improve Economic Statistics (GAO/GGD-
95-98, July 7, 1995).

Economic Statistics:  Measurement Problems Can
Affect the Budget and Economic Policymaking
(GAO/GGD-95-99, May 2, 1995).

Prescription Drug Prices:  Official Index
Overstates Producer Price Inflation (GAO/HEHS-95-
90, Apr. 28, 1995).

Developing a Consumer Price Index for the Elderly
(GAO/T-GGD-87-22, June 29, 1987).

Stabilizing Social Security-Which Wage Measure
Would Best Align Benefit Increases With Revenue
Increases? (GAO/IMTEC-85-13, Aug. 27, 1985).

Funds Needed to Develop CPI Quality Control System
(GAO/GGD-83-32, Apr. 1, 1983).

A CPI for Retirees Is Not Needed Now but Could Be
in the Future (GAO/GGD-82-41, June 1, 1982).

A Consumer Price Index for Retirees and
Alternatives for Controlling Indexing (Testimony,
Apr. 20, 1982).

Measurement of Homeownership Costs in the Consumer
Price Index Should Be Changed (GAO/PAD-81-12, Apr.
16, 1981).

*** End of Document ***