Attachment S - BLS Handbook of Methods

Chapter 16
Consumer Expenditures and Income

C

onsumer expenditure surveys are specialized studies
in which the emphasis is on data related to family
expenditures for goods and services used in day-today living. In addition to data on family expenditures, the
Consumer Expenditure Survey (CE) of the Bureau of Labor
Statistics (BLS, the Bureau) collects information on the
amount and sources of family income, changes in assets and
liabilities, and demographic and economic characteristics of
family members.
Background

IN THIS CHAPTER
Background................................................................. 1
Current Survey: 1980 to the present........................... 1
Interview Survey.................................................... 2
Diary Survey.......................................................... 3
Integrated survey data............................................ 3
Data collection and processing.............................. 3
Sample Design............................................................ 5
Selection of households ........................................ 5
Cooperation levels.................................................. 5
Estimation methodology........................................ 6
Precision of the estimates....................................... 6
Presentation ................................................................ 7
Evaluation Research.................................................... 7
Survey Uses................................................................ 9
Survey Limitations...................................................... 9

The Bureau’s studies of family living conditions rank among
its oldest data-collecting functions. The first nationwide
expenditure survey was conducted during 1888–1891, to
study workers’ spending patterns as elements of production
costs. With special reference to competition in foreign trade,
the survey emphasized the worker’s role as a producer,
rather than as a consumer. In response to rapid price changes
prior to the turn of the 20th century, a second survey was
administered in 1901. The resulting data provided the weights
for an index of prices of food purchased by workers that was
used until World War I, as a deflator for workers’ incomes
and expenditures. A third survey, conducted during 1917–
1919, provided weights for computing a cost-of-living index,
now known as the Consumer Price Index (CPI). The Bureau
conducted its next major survey, covering only urban wage
earners and clerical workers, during 1934–1936, primarily to
revise CPI weights.
During the Great Depression of the 1930s, the use of
consumer expenditure surveys extended from the study of
the welfare of selected groups to more general economic
analysis. Concurrent with its 1934–1936 investigation, the
Bureau cooperated with four other Federal agencies in a fifth
survey, the 1935–1936 study of consumer `purchases, which
presented consumption estimates for both urban and rural
segments of the population. A sixth survey, in 1950, covered
only urban consumers; this survey was an abbreviated version
of the 1935–1936 study. A seventh survey, the 1960–1961
Survey of Consumer Expenditures, once again included both
urban and rural families and provided the basis for revising
the CPI weights, while supplying material for broader
economic, social, and market analyses.

The next major survey to collect information on
expenditures of householders in the United States was
conducted in 1972–1973. That survey, while providing
continuity with the content of the Bureau’s previous surveys,
departed from the past in its collection techniques. Unlike
earlier surveys, the U.S. Census Bureau, under contract to
BLS, conducted all sample selection and field work. Another
significant change was the use of two independent surveys
to collect the information—a diary survey and an interview
panel survey. A third major change was the switch from an
annual recall to a quarterly recall (in the Interview Survey)
and daily recordkeeping of expenditures (in the Diary
Survey). Again, the resulting data were used to revise CPI
weights.
Current Survey: 1980 to the present
The need for more timely data than could be supplied by
surveys conducted every 10 to 12 years—intensified by the
rapidly changing economic conditions of the 1970s—led to
the initiation of the current continuing survey in late 1979.
Since then, data have been available annually. The objectives
1

of the CE remain the same: to provide the basis for revising
weights and associated pricing samples for the CPI and to
meet the need for timely and detailed information on the
spending patterns of different types of families.
Like the 1972–1973 survey, the current survey consists of
two separate surveys, each with a different data collection
technique and sample. In the Interview Survey, each family
in the sample is interviewed every 3 months over five
calendar quarters. The sample for each quarter is divided into
three panels, with consumer units being interviewed every
3 months in the same panel of every quarter. The Diary (or
recordkeeping) Survey is completed by the respondent family
for two consecutive 1-week periods.
The sample housing unit is notified in advance by a letter
informing the occupants about the purpose of the survey and the
upcoming visit by the interviewer. Both the Interview Survey
and the Diary Survey are conducted primarily by personal
visits with some telephone usage. The interviewer uses a
structured questionnaire to collect both the demographic and
expenditure data in the Interview Survey. The demographic
data for the Diary Survey are collected by the interviewer,
whereas the expenditure data are entered on the diary form
by the respondent. Any eligible household member who is at
least 16 years old can serve as the respondent in either survey.
The unit for which expenditure reports are collected is the
set of eligible individuals constituting a consumer unit that
is defined as 1) all members of a particular housing unit who
are related by blood, marriage, adoption, or some other legal
arrangement, such as foster children; 2) a person living alone
or sharing a household with others, or living as a roomer in a
private home, lodging house, or in permanent living quarters
in a hotel or motel, but who is financially independent; or 3)
two or more unrelated persons living together who pool their
income to make joint expenditure decisions. Students living
in university-sponsored housing are also included in the
sample as separate consumer units. Information on members
living in the consumer unit is identified by their relationship
to the reference person, who is defined as the first member
mentioned by the respondent when asked to “Start with the
name of the person or one of the persons who owns or rents
the home.”
Survey participants report dollar amounts for goods and
services purchased by any member of the consumer unit
during the reporting period, regardless of whether payment
was made at the time of purchase. Expenditure amounts
for items purchased by the consumer unit include all
applicable sales and excise taxes. Excluded from expenditure
total amounts are any business-related expenditures and
expenditures for which the family is reimbursed.
The Interview Survey collects detailed data on an estimated
60 to 70 percent of total family expenditures. In addition,
global estimates are obtained for food and other selected
items. These global estimates account for an additional 20
to 25 percent of total expenditures. On average, it takes 60
minutes to complete the interview.
In the Diary Survey, detailed data are collected on all
expenditures made by consumer units during their 2-week

participation in the survey. It takes approximately 25 minutes
over three visits for the interviewer to collect the demographic
data and to instruct the respondent on how to keep the diary.
It is estimated that it takes the respondent 15 minutes each
day to complete the diary.
Quality control is provided by a re-interview program,
which evaluates the performance of the individual
interviewer, to determine how well the procedures are being
carried out in the field. The re-interview is conducted by a
Census Bureau supervisor. Subsamples of approximately 9
percent of households in both the Interview and Diary Survey
are re-interviewed on an ongoing basis.
All data collected in both surveys are subject to Census
Bureau and BLS confidentiality requirements that prevent the
disclosure of the respondents’ identities. The information that
respondents provide is used solely for statistical purposes.
All Census Bureau and BLS employees who work with the
CE data take an oath of confidentiality and are subject to fines
and imprisonment for improperly disclosing information
provided by respondents. Names and addresses are removed
from all forms and datasets prior to transmission from the
Census Bureau to BLS and are not included in any statistical
releases. At BLS, the data are processed and stored on secure
servers, with access limited to employees having security
clearances. As a further precaution, BLS applies certain
restrictions to the microdata available on the public-use
files. These include geographical and value restrictions that
prevent identification of respondents.
Interview Survey
The Interview Survey is designed to collect data on the types
of expenditures that respondents can be expected to recall
for a period of 3 months or longer. In general, expenditures
reported in the Interview Survey are either relatively large,
such as those for property, automobiles, and major appliances,
and/or that occur on a fairly regular basis, such as for rent,
utilities, or insurance.
Each occupied sample unit is interviewed once per quarter
for five consecutive quarters. After the fifth interview, the
sample unit is dropped from the survey and replaced by
a new sample unit. For the survey as a whole, 20 percent
of the sample in each quarter are new families introduced
into the sample, to replace families that have completed
their participation. Data collected in each quarter are treated
independently, so that estimates are not dependent upon a
particular family participating in the survey for a full five
quarters.
For the initial interview, information is collected on
demographic and family characteristics and on the inventory
of major durable goods of each consumer unit. Expenditure
information is also collected in this interview, using a 1-month
recall, but is used—along with the inventory information—
solely for bounding purposes, that is, to classify the unit for
analysis and to prevent duplicate reporting of expenditures in
subsequent interviews.
The second through fifth interviews use uniform
questionnaires to collect expenditure information for the
2

previous 3 months. Data collected in these questionnaires,
which are arranged by major expenditure component (for
example, housing, transportation, medical, and education),
form the basis of the expenditure estimates derived from the
Interview Survey. Also, wage, salary, and other information
on the employment of each member of a consumer unit are
collected in the second interview and updated in the fifth
interview. Expenditure data are collected via two major types
of questions. The first set of questions asks the respondent
for the month of purchase of each reported expenditure. The
second asks for a quarterly amount of expenditures. The use
of these two questions varies, depending on the types of
expenditures collected. Most of the data are collected using
the direct monthly method. A portion of the data is collected
by asking for quarterly expenses, but this also includes asking
for the amount that was spent in the current month so as to
only have the expenses that occur in the 3-month reference
period. In the fifth and final interview, an annual supplement
is used to obtain a financial profile of the consumer unit. This
profile consists of information on the income of the consumer
unit as a whole, including unemployment compensation;
income from royalties, dividends, and estates; alimony
and child support. A 12-month recall period is used in the
collection of income- and asset-type data.

by four classifications of goods and services—food away
from home, food at home, clothing, and all other goods and
services—a breakdown designed to aid the respondent in
recording the entire consumer unit’s daily purchases. The
items reported are subsequently coded by the Census Bureau,
so BLS can aggregate individual purchases for representation
in the CPI and for presentation in statistical tables.
Integrated survey data
Integrated data from the BLS Diary and Interview Surveys
provide a complete accounting of consumer expenditures and
income, which neither survey component alone is designed to
do. Some expenditure items are collected only by the Diary
or Interview Survey. For example, the Diary collects data
on detailed food expenditures and items, such as postage
and nonprescription drugs, which are not collected in the
Interview. The Interview collects data on expenditures for
overnight travel and information on insurance reimbursements
for medical-care costs and automobile repairs, which are not
collected in the Diary. Data on average annual expenditures
that come exclusively from the Interview Survey, including
global estimates, such as those for food and alcoholic
beverages, average about 95 percent of the total estimated
spending, based on integrated Diary and Interview data. For
items unique to one or the other survey, the choice of which
survey to use as the source of data is obvious. However,
there is considerable overlap in coverage between the
surveys. Because of the overlap, the integration of the data
presents the problem of determining the appropriate survey
component from which to select the expenditure items. When
data are available from both survey sources, the more reliable
of the two is selected, as determined by statistical methods.
The selection of the survey source is evaluated every 2 years.

Diary Survey
The primary objective of the Diary Survey is to obtain
expenditures data on small, frequently purchased items, which
can be difficult to recall even a few weeks later. These items
include food and beverage expenditures at home and in eating
places; housekeeping supplies and services; nonprescription
drugs; and personal care products and services. The Diary
Survey is not limited to these types of expenditures but, rather,
includes all expenses that the consumer unit incurs during the
survey week. Expenses incurred by family members while
away from home overnight and for credit and installment
plan payments are excluded.
Two separate questionnaires are used to collect Diary
data: a Household Characteristics Questionnaire and a
Record of Daily Expenses. In the Household Characteristics
Questionnaire, the interviewer records information pertaining
to age, sex, race, marital status, and family composition, as
well as information on the work experience and earnings
of each member of the consumer unit. This socioeconomic
information is used by the Bureau to classify the consumer
unit for publication of statistical tables, as well as for
economic analysis. Data on household characteristics also
provide the link in the integration of Diary expenditure data
with Interview expenditure data that permit the publication
of a full profile of consumer expenditures by demographic
characteristics.
The daily expense record is designed as a self-reporting,
product-oriented diary, in which respondents record a detailed
description of all expenses for two consecutive 1-week
periods. Diary recording keeping can start on any day of the
week. Data collected each week are treated as statistically
independent. The diary is divided by day of purchase and

Data collection and processing
Due to differences in format and design, the Interview
Survey and the Diary Survey are collected and processed
separately. The U.S. Census Bureau, under contract with
BLS, carries out data collection for both surveys. In addition
to its collection duties, the Census Bureau does field editing
and coding, checks consistency, ensures quality control, and
transmits the data to BLS. In preparing the data for analysis
and publication, BLS performs additional review and editing
procedures.
Interview Survey. Beginning April 2003, Census field
representatives (FR) started collecting the Interview data
using a Computer Assisted Personal Interview (CAPI)
instrument. This was a major improvement from the paperand-pencil data collection that had been in place since 1980.
The CAPI instrument enforces question skip patterns, allows
for data confirmation of high expenditure values, and reduces
processing time. The FR performs some coding of expenses—
by selecting from a predetermined list—for example, vehicle
make and model; trip destination; and types of services for
alterations, maintenance, and repair on owned and rented
properties.
3

Data are electronically transferred from a FR’s laptop at
completion of the interview to the Census Master Control
System. The Census Bureau’s Demographics Surveys
Division then reformats the data into datasets and performs
special processing, such as converting missing values to
special characters and merging data records into the required
BLS output structure. Some data, like vehicle and mortgage
records, are copied into an input file that is loaded on the
laptops for subsequent interviews the next quarter. This way,
a few fields are updated each quarter, rather than an entire
data record. As mentioned earlier, names and addresses of
respondents are not transmitted.
At BLS, a series of automated edits are applied to monthly
data. These edits check for inconsistencies, identify missing
expenditure amounts for later imputation, impute values for
missing demographic variables, calculate weights, and adjust
data to include sales tax, exclude business expenses, and net
out reimbursed expenditures.
Monthly data files then are combined into quarterly
databases, and a more extensive data review is carried
out. During this data review, BLS conducts the following
steps: verifies counts and means by region, checks family
relationship coding inconsistencies, and inspects selected
extreme values for expenditure and income categories. Other
adjustments convert mortgage and vehicle loan payments into
principal and interest (using associated data on the interest
rate and term of the loan). In addition, BLS verifies the various
data transformations it performs. Cases of questionable
data values or relationships are investigated, and errors are
corrected prior to release of the data for public use.
Three major types of data adjustment routines—
imputation, allocation, and time adjustment—improve
estimates derived from the Interview Survey. Data imputation
routines account for missing or invalid entries and address
all fields in the database, except assets. Allocation routines
are applied, when respondents provide insufficient detail
to meet tabulation requirements. For example, combined
expenditures for the fuels and utilities group are allocated
among the components of that group, such as natural gas
and electricity. Time adjustment routines are used to classify
expenditures by month, prior to aggregation of the data to
calendar-year expenditures. Tabulations are made before and
after data adjustment routines, to analyze the results.
The CE implemented multiple imputations of income data,
starting with the publication of 2004 data. Prior to that, only
income data collected from complete income reporters were
published. However, even complete income reporters may
not have provided information on all sources of income for
which they reported receipt. With the collection of bracketed
income data starting in 2001, this problem was reduced
but not eliminated. One limitation was that bracketed data
only provided a range in which income falls, rather than a
precise value for that income. In contrast, imputation allows
income values to be estimated when they are not reported.
In multiple imputations, several estimates are made for the
same consumer unit, and the average of these estimates is
published.

The CE Interview questionnaire is revised every 2
years to incorporate new products and services, to clarify
instructions, to improve navigation through the instrument,
to incorporate changes required for the CPI, and to streamline
the interview by deleting outdated items. Whereas changes to
the questionnaire are made biennially, CE staff continuously
monitors the emergence of new goods and services available
in the marketplace, as well as changes in the relative
importance of existing items in consumers’ budgets. New
items are incorporated in a product index that enables
Census field representatives to classify these new items by
the appropriate item codes. Also, updated information on
how to report new goods and services is provided to the field
representatives on a regular basis.
Diary Survey. At the beginning of the 2-week collection
period, the Census Bureau interviewer, using the Household
Characteristics Questionnaire (a CAPI instrument), records
demographic information on members of each sampled
consumer unit. At this time, the interviewer also leaves the
Diary questionnaire—or daily expenditure record—with the
consumer unit, to record expenditures for the week.
At the end of the first week, the interviewer collects the
diary, reviews the entries, answers any questions that the
respondent may have, and leaves a second diary. At the end
of the second week, the interviewer picks up the second diary
and reviews the entries. During this time, the interviewer
again uses the Household Characteristics Questionnaire to
collect previous-year information on work experience and
income. Each week of a consumer unit’s participation in the
survey is treated as a separate entity.
The Census Bureau performs preliminary processing
activities, including a number of data edits and adjustments.
Data in the diaries are reviewed during a field edit for
completeness and consistency. After the diaries are sent to
the Census Bureau’s National Processing Center, expenditure
data captured in the diaries are key-entered into electronic
formats; and a computer file of the database containing
these data is produced and transmitted monthly to Census
headquarters, along with image files of the diaries. Census
headquarters merges the expenditure data with the data
collected in the Household Characteristics Questionnaire,
removes personal identifying information, and transmits
the merged file monthly to BLS. At BLS, data are processed
by computer to calculate population weights, to impute
demographic characteristics for missing or inconsistent
demographic data, to impute values for weeks worked when
nonresponse is encountered, and to apply appropriate sales
taxes to the expenditure items.
Like the Interview Survey, three monthly diary data files
are combined into quarterly databases; and BLS screens the
data for invalid coding and inconsistent relationships, as well
as for extreme values recorded or keyed erroneously. BLS
then corrects any coding and extreme-value errors found.
Two types of data adjustment routines—allocation and
imputation—improve the Diary Survey estimates. Allocation
routines transform reports of nonspecific items into specific
4

ones. For example, when respondents report expenditures
for meat, rather than beef or pork, allocations are made,
using proportions derived from item-specific reports in other
completed diaries. BLS imputes missing attributes, such as
age and gender, or a product’s type of packaging needed for
mapping Diary expenditures. Income data from the Diary
Survey are processed in the same way as in the Interview
Survey.

local governments. The area frame is a list of housing units
in places that do not issue construction permits or where a
significant number of addresses are incomplete (e.g., they
have PO boxes or rural route numbers instead of street
addresses). This list is generated by Census Bureau staff
canvassing those areas and making maps of the housing units.
Finally, the group quarters frame is a list of housing units that
are owned or managed by organizations for residents who
live in group arrangements, such as college dormitories and
retirement communities.
The Census Bureau selects a sample of approximately
12,500 addresses per year to participate in the Diary Survey.
Usable diaries are obtained from approximately 7,100
households at those addresses. Diaries are not obtained from
the other addresses, due to refusals, vacancies, ineligibility,
or the nonexistence of a housing unit at the selected address.
The placement of diaries is spread equally over all 52 weeks
of the year.
The Interview Survey is a rotating panel survey, in which
approximately 15,000 addresses are contacted in each calendar
quarter of the year. One-fifth of the addresses contacted
each quarter are new to the survey and provide “bounding”
interviews that provide baseline data, which are not used
to compute the survey’s published expenditure estimates.
Excluding these bounding interviews and interviews not
completed (due to refusals, vacancies, ineligibility, and
the nonexistence of a housing unit at the selected address),
usable interviews are obtained from approximately 7,100
households each quarter. After a housing unit has been in the
sample for five consecutive quarters, it is dropped from the
survey, and a new address is selected to replace it.

Sample Design
Selection of households
The CE is a nationwide household survey, representing the
entire U.S. civilian noninstitutional population. It includes
people living in houses, condominiums, apartments, and
group quarters, such as college dormitories. It excludes
military personnel living on base, nursing home residents, and
people in prisons. The civilian noninstitutional population
represents more than 98 percent of the total U.S. population.
The selection of households for the survey begins with the
definition and selection of primary sampling units (PSUs).
PSUs are small clusters of counties grouped together into
geographic entities called “core-based statistical areas”
(CBSAs), which are defined by the Office of Management
and Budget (OMB) for use by federal statistical agencies in
collecting, tabulating, and publishing federal statistics. The
CE currently uses the OMB definitions from 2000. There
are two types of CBSAs: metropolitan and micropolitan.
Metropolitan CBSAs are areas that have an urban “core”
of 50,000 or more people, plus adjacent counties that have
a high degree of social and economic integration with the
core as measured by commuting ties. Micropolitan CBSAs
are similar to metropolitan CBSAs but are areas that have an
urban core of 10,000 to 50,000 people.
The current geographic sample used in the survey consists
of 91 PSUs that are classified into four categories:

Cooperation levels
Response data for the 2011 CE are shown in table 1. For the
Interview Survey, each unique housing unit provides up to
four usable interviews per year. For the Diary Survey, totals
refer to housing units in weeks 1 and 2 of the survey, with
each unique housing unit providing up to two usable diaries.
Most Diary respondents participate in both weeks.

•

21 “A” PSUs, which are metropolitan CBSAs with a
population over 2.7 million people

•

38 “X” PSUs, which are metropolitan CBSAs with a
population under 2.7 million people

•

16 “Y” PSUs, which are micropolitan CBSAs

•

•

16 “Z” PSUs, which are non-CBSA areas that are often
referred to as “rural” PSUs

Type A nonresponses are refusals, temporary absences,
and noncontacts.

•

Type B nonresponses are vacant housing units, housing
units with temporary residents, and housing units under
construction.

•

Type C nonresponses are nonresidential addresses, such
as destroyed or abandoned housing units, and housing
units converted to nonresidential use.

There are three general categories of nonresponse:

The 75 “A”, “X”, and “Y” PSUs also are used by the
Consumer Price Index program.
Within these 91 PSUs, the list of addresses from which the
sample is drawn comes from four sources called “sampling
frames.” The largest sampling frame is the “unit” frame,
which comes from the 2000 Census 100-Percent Detail File.
The unit frame has about 80 percent of the addresses selected
for the survey. It is supplemented by three smaller frames—
the permit, area, and group quarters frames. The permit
frame is a list of new housing units constructed after the 2000
census and is generated from construction permits filed with

Response rates are defined to be the percent of eligible
housing units (that is, the designated sample less Type B
and Type C nonresponses) from which usable interviews are
collected. In the 2011 Interview Survey, there were 47,979
eligible housing units, from which 34,302 usable interviews
were collected, resulting in a response rate of 71.5 percent.
5

In the 2011 Diary Survey, there were 18,537 eligible housing
units, from which 13,986 usable interviews were collected,
resulting in a response rate of 75.4 percent. (See table 1.)

Several factors are involved in computing the weight of
each consumer unit from which a usable interview is received.
Each consumer unit is initially assigned a base weight that
is equal to the inverse of the consumer unit’s probability of
being selected for the sample. Base weights in the CE are
typically around 10,000, which means that a consumer unit
in the sample represents 10,000 consumer units in the U.S.
civilian noninstitutional population―itself plus 9,999 other
consumer units that were not selected for the sample. The
base weight is then adjusted by the following factors to
correct for certain nonsampling errors:

Estimation methodology
The estimation of population quantities of interest, such as
the average expenditure per consumer unit on a particular
item category, is achieved through the use of weights. Each
consumer unit in the survey is assigned a weight that is
the number of similar consumer units in the U.S. civilian
noninstitutional population the sampled consumer unit
represents. Using these weights, the average expenditure per
consumer unit on a particular item category is estimated with
the formula:

∑w y
y=
∑w
i∈S

i∈S

i

Weighting control factor. This adjusts for subsampling in
the field. Subsampling occurs when a data collector visits
a particular address and discovers multiple housing units,
where only one housing unit was expected.

i

Noninterview adjustment factor. This adjusts for interviews
that cannot be conducted in occupied housing units, due to
a consumer unit’s refusal to participate in the survey or the
inability to contact anyone at the housing unit, in spite of
repeated attempts. This adjustment is based on region of the
country, household tenure (owner or renter), consumer unit
size, and race of the reference person.

i

where,
y = the average expenditure per consumer unit on the item
category,
yi = the expenditure made by the ith consumer unit on the
item category,
wi = the weight of the ith consumer unit in the sample, and
S = the sample of consumer units that participated in the
survey.

Calibration factor. This adjusts the weights to 24 “known”
population counts to account for frame undercoverage.
These known population counts are for age, race, household
tenure (owner or renter), region of the country, and urban or
rural. The population counts are updated quarterly using the
Current Population Survey estimates. Each consumer unit is
given its own unique calibration factor. There are infinitely
many sets of calibration factors that make the weights add up
to the 24 known population counts, and the CE uses nonlinear
programming to select the set that minimizes the amount of
change made to the “initial weights” (initial weight = base
weight x weighting control factor x noninterview adjustment
factor).

For example, if yi is the expenditure on eggs made by the
i th consumer unit in the sample during a given time period,
then y is an estimate of the average expenditure on eggs made
by all consumer units in the U.S. civilian noninstitutional
population during that time period.
If one wants to estimate the proportion of consumer units
that purchased eggs during a given time period, then the
same formula is applied, where yi is set equal to 1, if the
i th consumer unit purchased eggs during the time period, and
0 if it did not. When this binary definition of yi is used, y
is an estimate of the proportion of all consumer units in the
U.S. civilian noninstitutional population that purchased eggs
during the given time period.

Precision of the estimates
The precision of the estimator y is measured by its standard
error. Standard errors measure the sampling variability
of the CE estimates. That is, standard errors measure the
uncertainty in the survey estimates caused by the fact that
a random sample of consumer units from across the United
States is used, instead of collecting data from every consumer
unit in the nation.
The CE’s standard errors are estimated using the method
of “balanced repeated replication.” In this method the
sampled PSUs are divided into 43 groups (called strata), and
the consumer units within each stratum are randomly divided
into two half samples. Half of the consumer units are assigned
to one half sample, and the other half are assigned to the other
half sample. Then, 44 different estimates of y are created,
using data from only one half sample per stratum. There are
many combinations of half samples that can be used to create
these replicate estimates, and the CE uses 44 of them that are

Table 1. Analysis of response in the Consumer
Expenditure Survey, 2011
Sample unit
Housing units designated for survey
Less: Type B and type C nonresponses

Interview
Survey
60,079

Diary
Survey
1

25,336

12,100

6,799

47,979

18,537

Less: Type A nonresponses

13,677

4,551

Equals: Interviewed units

34,302

13,986

71.5

75.4

Equals: Eligible units

Percent of eligible units interviewed

The number of Diary Survey addresses (12,668) multiplied by two
weekly diaries.
1

6

created in a “balanced” way with a 44´44 Hadamard matrix.
The standard error of is then estimated by:
SE
SE ( y ) =

1
44
4

4
44

∑(y

r

1972–1973 and 1984 onward are available on the BLS website.
The Diary and Interview Survey public use microdata
contain files of expenditure and income reports of each
consumer unit. To protect the identities of respondents,
selected geographic detail is eliminated, and selected income
and expenditure variables may be topcoded. The Interview
files contain expenditure data in two formats: MTAB files
that present monthly values in an item-coding framework,
based on the CPI pricing scheme, and EXPN files that
organize expenditures by the section of the Interview survey
instrument, in which they are collected. Expenditure values
on the EXPN files cover different time periods, depending
on specific questions asked; these files also contain relevant
nonexpenditure information not found on the MTAB files.
The public-use microdata files include quarterly expenditure
summary variables at the consumer unit level. The annual
Interview and Diary microdata files are available, beginning
with 1990, as well as for selected earlier years.
Articles that include analyses of CE data are published
online in the Monthly Labor Review (MLR), in the publication
Beyond the Numbers, and in CE data comparisons and research
reports. Other survey information is available on the Internet,
including answers to frequently asked questions, copies of
the Interview and Diary Survey instruments, a glossary of
terms, and order forms for survey products. Starting with the
2000 data, estimates of standard errors for integrated Diary
and Interview Survey data are available on the BLS website.
More detailed expenditure tables are available upon
request. These tables are sorted by the same demographic
variables as the standard tables but have more expenditure
subcategories. Estimates for these subcategories have higher
variance than the standard published tables.

− y )2

r =1

where,
y r is the r th replicate estimate of y.

The coefficient of variation is a related measure of sampling
variability that measures the variability of the survey estimate
relative to the mean. It is defined by the equation:

C
VCV ( y ) =

SESE( y )
× 100%
y

Presentation
Information from the CE is available in press releases,
reports, and analytical papers. Newer public use microdata
from the survey are available on the CE section of the
BLS website (http://www.bls.gov/cex), and older data are
available on USB flash drive. Tabular data also are available
at the same location on the BLS website and by contacting
the BLS Consumer Expenditure Survey Division directly.
(See website.)
Publications from the CE generally include tabulations of
average expenditures and income arrayed by consumer unit
characteristics, such as consumer unit size, age of reference
person, and income. Tabulations by two variables (crosstabulations) are available for selected characteristics, such as
age by income and consumer unit size by income. Integrated
Diary and Interview Survey data covering 12 months of data
are published on twice a year basis, and tabulations starting with

Evaluation Research
Consumer expenditure surveys undergo continuous
evaluation, by comparing results with other sources and by
performing internal statistical, qualitative, and cognitive
analyses to address current methodological concerns. To
improve expenditure estimates, research efforts related to
the data collection instruments, field procedures, and sources
of potential survey error (examples, nonresponse bias and
measurement error) began in the mid-1980s, and have since
become standard practice. A separate Branch of Research
and Program Development (BRPD) was established within
the Division of Consumer Expenditure Survey of BLS,
in 1999, with the mission of developing and conducting
methodological studies to improve survey instruments, field
procedures, and overall survey data quality. In recent years,
BRPD has focused on redesigning the survey, analyzing the
data for quality assessment, and developing improved data
collection tools.

Table 2. Precision of the Consumer Expenditure
Survey expenditure estimates, integrated Diary and
Interview Survey data, 2011
Item
category
Total expenditures
Food

Average
annual
expenditure
per consumer
unit,( y )

Standard
error, SE
(y)

Coefficient
of
variation,
CV ( y )
(in percent)

$49,705

$436

0.88

6,458

66

1.02

Housing

16,803

180

1.07

Apparel

1,740

36

2.06

Transportation

8,293

147

1.77

Healthcare

3,313

61

1.84

Entertainment

2,572

43

1.69

Personal care

634

12

1.90

Cash contributions

1,721

71

4.11

Personal insurance
and pensions

5,424

85

1.57

Other

2,747

80

2.90

The Gemini Project. Since 2009, research efforts have placed
an emphasis on updating the design of the CE. In 2009, the
CE program initiated the Gemini Project, with the goal of

7

promoting improved expenditure estimates in the CE—
through an updated survey design—with a focus on reducing
measurement error. The multi-year survey redesign project
includes research, pilot testing, and transition to the new
survey design. The scope of the project is broad and open
to a wide range of design alternatives, such as new modes,
modular surveys, and innovative technologies. Throughout
the Gemini Project, program staff will conduct ongoing work
to develop and test potential survey design changes, aimed at
improving data quality, increasing the analytic value of the
data for users, and supporting greater operational flexibility,
to respond to changes in the data-collection environment. For
further information on the Gemini Project, see http://www.
bls.gov/cex/geminiproject.htm.

the comparisons that are made are those for a wide range
of expenditure categories, as well as comparisons of data
on income and on assets and liabilities. For information
on published articles and presentations comparing CE data
with those from other sources, see http://www.bls.gov/cex/
cecomparison.htm.
Developing improved data collection methods. A number
of studies have been conducted to develop and to evaluate
new data collection methods—identifying methods that are
feasible, as well as approaches for implementation. A records
feasibility study in 2011 found that a majority of CE data
needs could be collected solely through respondent financial
records; however, categorizing the information obtained
from records into CE expenditure categories require a great
deal of time. A balance edit study in 2011 assessed the effects
of balancing a household’s cash flow with the household’s
outlays on overall data quality and respondent experience,
finding that conducting the edit generated additional
expenditure and income reporting, but remained unsuccessful
in balancing the cash flow of the household. The study also
identified a number of feasibility issues that would make
implementation in the survey problematic.
Several small-scale lab studies conducted in 2012
assessed how respondents react to varying reference periods
(the time period for which an expenditure is reported) and
how respondents report for other household members
(proxy reporting). The reference period study showed that
respondents have minimal trouble switching from one
reference period to another, however varying the reference
period had no meaningful effect on the level of expenditures
reported. The lab study investigating proxy reporting found
that asking questions upfront about expenditure categories
relevant to other household members—and then reminding
the respondents of the other household members when
collecting expenditures for those categories—appears to
effectively capture additional expenditures without adding
much time to the survey.
Another 2012 study evaluated different capabilities and
features available on various expense tracking applications
that are downloadable to mobile devices (e.g., smart phones,
tablets). Feedback from users asked to report their daily
expenses, using an expense tracking application provided
insight for integrating that type of technology into the current
data collection methodology. Other research in 2012 that
focused on the effectiveness of using a bounding interview
to prevent “telescoping” of expenditures (reporting expenditures outside the reference period) led to a decision to
eliminate the bounding interview beginning in 2015. This
step will reduce overall respondent burden and survey costs,
with minimal expected impact on data quality. In 2013, a
field test of a Web-based diary was conducted, testing the
feasibility and impact of using a Web diary to collect CE
diary expenditures.

Research overview. BRPD has continued research—both in
support of the redesign effort—and as an ongoing effort to
improve data quality, while balancing survey costs. Current
evaluation research focuses on two main areas: (1) assessing
data quality, and (2) developing improved data collection
methods to produce higher quality estimates. The first area
is useful for reviewing the current survey and for evaluating
any impacts of potential future designs. The second area
provides valuable insight for informing decisions on design
options.
Assessing data quality. Several recent studies have addressed
a general overview of methods for measuring data quality.
In 2012, a study reviewed literature and previous work on
how to measure survey error, resulting in a proposal that the
CE use a multi-method-indicators (MMI) approach to track
error through (1) internal indicators from the survey data and
paradata, (2) external indicators involving comparisons of CE
data to other sources, and (3) periodic record check studies.
A 2011 small-scale record check study analyzed the accuracy
of respondent reports, showing that respondents accurately
reported expenditure amounts (within 5–10 percent of the
actual expenditure amount) just over half of the time and very
rarely missed reporting an item (under-reporting) or reported
additional items that were not purchased, according to the
records (over-reporting). In 2012, a data-quality profile team
developed a framework for the collection and organization
of information that could be used for tracking survey
quality. The framework provides an approach to routinely
produce metrics for monitoring quality over time, allowing
for evaluation of survey improvements and highlighting
areas that need improving. Other research has focused on
quantifying specific aspects of survey quality, such as the
development of a summary index for respondent burden that
can be used to compare the effect of alternative survey design
options on respondent burden, which is believed to be linked
to data quality.
Finally, as an important part of the CE quality assurance
program, estimates from CE data are regularly compared
with corresponding estimates from other data sources to
evaluate the soundness of CE estimates at any point in time,
as well as the consistency of estimates over time. Among

Future research. Upcoming research projects will address
the issue of measurement error, along with new modes and
8

methods for collecting data. Some of those studies will
include the evaluation of data collection at the individual
household member level, rather than at the household level,
as well as alternative collection modes, such as online diary
data collection via mobile devices.

over time. The chained index is designed to better measure
the change in the cost of living, as compared with the CPI-U
and CPI-W, which measure the change in a fixed market
basket of goods and services in retail outlets. The C-CPI-U
uses expenditure data from different time periods, to reflect
the effect of substitution that consumers make across item
categories, in response to changes in the relative prices of
goods and services.

Survey Uses
The importance of the CE is its ability to allow data users
to examine the association of expenditures and income of
consumers to consumer characteristics. CE survey data
are of value to government and private agencies interested
in studying the welfare of particular segments of the
population, such as the elderly, low-income families,
urban families, and those receiving food stamps. Data
also are used by economic policymakers interested in the
effects of policy changes on levels of living among diverse
socioeconomic groups, and econometricians find the data
useful in constructing economic models. Market researchers
find consumer expenditure data valuable in analyzing the
demand for groups of goods and services. In addition, the
Department of Commerce uses the survey data as a source
of information for revising its benchmark estimates of
selected items in the expenditure and income components of
the national accounts and in preparing supplemental poverty
thresholds, the Internal Revenue Service uses expenditures
to calculate alternate sales tax standard deductions, and the
Department of Defense uses the data in determining cost-ofliving allowances for military personnel living off military
bases.
As in the past, the revision of the Consumer Price Index
remains a primary reason for undertaking the Bureau’s
extensive Consumer Expenditure Survey. Results of the
CE are used to select new “market baskets” of goods and
services for the index, to determine the relative importance
of components, and to derive cost weights for the baskets. In
August 2002, the Bureau of Labor Statistics began publishing
another index, the Chained Consumer Price Index for All
Consumers (C-CPI-U), which supplements the CPI for All
Urban Consumers (CPI-U) index and the CPI for Urban
Wage Earners and Clerical Workers (CPI-W) index. The use
of expenditure data from different time periods distinguishes
the C-CPI-U from the other two CPI measures, which use
a single expenditure base period to compute price change

Survey Limitations
Sample surveys are subject to two types of errors—sampling
and nonsampling. Sampling error is the uncertainty caused
by the fact that observations are taken from a random sample
of population members and not from the entire population.
Nonsampling error is the rest of the error and arises, regardless
of whether data are collected from a sample or from the entire
population. Nonsampling errors can be attributed to many
sources, such as differences in the interpretation of questions,
inability or unwillingness of respondents to provide correct
information, data processing errors, etc.
Another way of analyzing error is to divide it into variance
and bias. The variance is a measure of how close different
estimates would be to each other, if it were possible to repeat
the survey over and over, using different samples. However,
it is not feasible to repeat the survey over and over, therefore,
statistical theory allows the variance to be estimated. A small
variance indicates that multiple independent samples would
produce values that are consistently very close to each other.
Bias is the difference between the “expected” value of an
estimate and its “true” value. A statistic may have a small
variance but a large bias, or it may have a large variance but
a small bias. For an estimate to be considered accurate, both
its variance and its bias must be small.
The Bureau of Labor Statistics is constantly trying to
reduce the error in the CE estimates. Variance and sampling
error are reduced by using a sample of respondents that is
as large as possible, given resource constraints. Improving
the accuracy of the estimates was the primary reason for the
significant expansion in the sample for both the Interview and
Diary Surveys in 1999. Going forward, the Bureau will strive
to reduce nonsampling error through a series of computerized
and professional data reviews, as well as through continuous
survey process improvements, theoretical research, and the
Gemini Project redesign.

9