CONSUMER EXPENDITURE
SURVEYS
“Overview and Recent Developments in the Consumer
Expenditure Surveys”
NBER
July 22
22-23,
23, 2002
1
Outline
 Overview
O
i
 Recent
Research
 Data Comparisons
 Recent Developments
 Current/Future Research
2
The survey is designed to
represent a
 National
Probability Sample
– using the most recent decennial census,
augmented by new construction permits
– consisting of primary sampling units (PSU)
– based
b d on probabilities
b biliti proportional
ti l to
t
population size
– consists of counties,
counties group of counties
counties, or
independent cities
3
Housing Units are selected
selected...
 Within
each PSU using such information
as:
–
–
–
–
vacancy status
number of persons residing in housing unit
value of the housing unit
rent paid for the housing unit
4
5
Definitions
 CE
iis interested
i
d in
i determining
d
i i consumer
units - based on the financial relationship of
th members
the
b off th
the hhousehold.
h ld
6
How do we define
Consumer Units?

Members
b off a hhousehold
h ld related
l d by
b blood,
bl d marriage,
i
adoption,
d i
or other legal arrangement

Single person living alone or sharing a household with others
but who is financially independent

Two or more persons living together who are financially
dependent
7
How does this compare to others?
 SIPP
andd CPS use a household
h
h ld
definition… All persons who occupy a
h i unit
housing
it
8
 Although
we have slightly different
definitions - the CE based on the financial
relationship between members -- the
difference between consumer unit and
housing unit is small.
small About 3% of our
housing units contain more than one
consumer unit.
unit
9
Respondent

CE - One person responds for the entire consumer unit.
unit
We ask for the person most knowledgeable of
expenditures for the entire consumer unit

CPS - One person, preferably most knowledgeable about
the labor force activities of the others.
others

SIPP - A separate
p
qquestionnaire is administered to each
member over age 15
10
Reference Person
 All
three
h surveys hhave a similar
i il definition
d fi i i the person or one of the persons who owns
or rents
t the
th unit.
it
11
Survey Instruments
 Diary
– 2 consecutive 1-week
– includes
» detailed expenditures
for food, personal care,
household supplies
expenditures
– excludes
» expenditures
di
for
f out-off
town trips
 Interview
– 5 quarterly, only
inventoryy and basic
sample data from 1st
– excludes expenditures
for:
» housekeeping supplies
(e.g., postage stamps)
» personal care products
» non-prescription drugs
12
Socio demographic Variables
Socio-demographic
 Collected
during each interview
– Member level
– Consumer
C
unit
i level
l l
» reference person
» CU
 Collected
2nd and/or 5th interview
– Income, work experience, contributions
13
Publication Tables

are “integrated”
integrated
– Neither survey collects the entire universe of
expenditures.
expenditures
– Some data are only collected in one instrument
– Some
S
ddata
t are collected
ll t d in
i both
b th instruments.
i t
t
– For these areas we determine which is the best
source and use that in our publications
14
 Both
the Interview and Diary samples are
purposely non-clustered
 Past
P research
h has
h shown
h
that
h clustered
l
d
samples for expenditure data would not
yield
i ld expenditure
di
patterns that
h are
representative of the entire area
 Clustered samples tend to cut down on
travel expenses and hours per schedule.
15
Let’s look at sample size
 Census
uses the term “cases”
cases when
referring to its collection workload and
costs.
costs
 A Case is defined as one interview or one
di
diary
visit.
ii
 For households with more than one
consumer unit, each consumer unit would
be counted separately
16
Let’s look at some Census figures
 In
order to collect the equivalent of 7800
households completing 5 interviews or 2
diaries we must field about
diaries,
 60,000
cases for Interview
and
 25,000
cases for Diary
17
How does this translate to
households?
 60,000
represents the size of the sample
prior to any
p
y refinement.
 About 1/5 of these cases are Interview 1 -the bounding interview
 The field representatives refine the sample
to determine those housing units that are
out of scope of our survey
18
 Once
eligible households are determined
p
attempt
p to collect
the field representatives
the data.
 As in every survey they encounter refusals
refusals.
 The result is the number of completed cases
that can be used to determine expenditures
19
Let’s do the math
math...
 Total
Cases
 Bounding
B
di Interview
I t i
– equals
 Out
O off S
Scope
 Refusals
 Net Sample Yield
60,000
12 000
12,000
48,000
8 800
8,800
8,000
31,200
Thi translates
This
l
to about
b
7800 consumer units
i .
20
In Comparison
 CE
fields about 60,000 Interview cases and
25 000 Di
25,000
Diary cases eachh year
 SIPP
fields about 129,000 cases each year
 CPS
fields about 875,000 cases each year
21
Average Time per Case
 CE
– Diary
– Interview
 SIPP
 CPS
 All
3.1 hours
4.0 hours
3.65 hours
1 3 hours
1.3
hours include travel time
22
Census Budgets
 CE
$24 653 000
$24,653,000
 SIPP
$
$31,400,000
 CPS
$55,000,000
23
Cost per Case
 CE
$290 00
$290.00
 SIPP
$193 00
$193.00
 CPS
$65 00
$65.00
24
Response Rates
82%
80%
78%
76%
74%
72%
1997
1998
1999
Interview
2000
2001
Diary
25
 How
many households completed all 5
interviews?
 about
b
75%
 How
many households completed both
diaries?
 about 92%
26
Outline
 Research
using Interview, “four
quarter”, complete income reporters
 Importance of attrition
 Importance of income
– New bracketing procedure
 Quality
of global questions
– Interview vs. Diary
27
Attrition in the Interview survey
 Structure
of Interview Survey
– up to five interviews; four expenditure quarters
– quarterly
l data
d independent,
i d
d
weights
i h adjusted
dj
d
 Many
researchers use “four quarter” CUs
– adjust for age and tenure
 Recent
study
y examined 1997-2000
– With a 2nd through 5th interview
– Accounted for type
yp of non-interview
28
Percent of CUs who complete all four
inter ie s 1984-2000
interviews,
1984 2000
00
20
99
19
98
19
97
19
96
19
95
19
94
19
93
19
92
19
91
19
90
19
89
19
88
19
87
19
86
19
85
19
19
84
90%
80%
70%
60%
50%
40%
30%
20%
10%
0%
29
Probability of completing next
i t i for
interview
f those
th
completing
l ti 22nd
d
94
92
90
88
86
84
82
Complete 3rd interview
Complete 3rd and 4th
Complete all four
30
Fo r qquarter
Four
arter CUs vs.
s non-interviews
non inter ie s
Four quarter At least one
CUs
non-interview
50.6
40.9
83.2%
41.0%
2.6
2.3
Average quarterly
expenditures
$8,981
$7,504
Average per-capita
expenditures
$3,442
$3,212
Average age
Percent home-owner
Average size
31
Family type composition
40%
Four quarter CU's
35%
At least one non-interview
30%
25%
20%
15%
10%
5%
0%
Married with Other married Single
g p
parents
children
families
Singles
g
Other families
32
CUs who
ho ret
return
rn after an initial refusal
ref sal
60%
50%
40%
30%
20%
10%
0%
1st refusal in 2nd interview
1st refusal in 3rd interview
1st refusal in 4th interview
Total
33
Next steps
 Continue
to examine “intermittent”
intermittent
responders
 Examine
E
i non-responders
d ffor 11st iinterview
i
34
The use of brackets in collecting
income data
 Structure
of income component
– 5 member level income sources; 14 CU level sources
– 2nd
2 d and
d 55th
h iinterview,
i
annuall recall
ll
– many missing
 Many
researchers use complete income reporters
 Conducted study
y of using
g brackets
– cognitive results suggested use of ranges
– Data ggenerated ranges
g
– Initiated brackets in April 2001
35
Percent of CUs with missing CU level
income sources,
sources 2001 (April - Dec)
90
80
70
60
50
40
30
No missing sources
one missing
2 missing
3 or more missing
20
10
0
36
Reporters and Use of Brackets, selected
CU le
level
el income sources,
so rces 2001
100%
80%
60%
40%
20%
0%
Interest
Pensions
k ’ Comp
C
Worker’s
Percent who report receipt of source
Percent of reporters with actual amounts
P
Percent
t off reporters
t using
i brackets
b k t
P
Percent
t missing
i i
37
Distribution of Interest income,
Actual vs.
vs bracket,
bracket 2001
80
70
60
50
40
30
20
10
0
-2
00
0
2
-3
00
0
3
-4
00
0
4
-5
00
0
5
-1
00
00
10
-1
50
00
15
-2
00
00
20
-3
00
00
30
-4
00
00
40
-5
00
00
>
50
00
0
1
<
10
00
Actual
Bracket
38
Reporters and use of brackets for
salar income,
salary
income 2001
100%
80%
60%
40%
20%
0%
Reference person
Spouse
Other
Percent who report receipt of salary
Percent of reporters with actual amounts
Percent of reporters using brackets
Percent missing
39
50
00
-1
0,
00
0
10
-1
5,
00
0
15
-2
0,
00
0
20
-3
0,
00
0
30
-4
0,
00
0
40
-5
0,
00
0
50
-7
0,
00
0
70
-9
0,
00
90
0
-1
20
,0
00
>
12
0,
00
0
5
<
Distribution of Salary Income,
Actual vs.
vs bracket,
bracket 2001
20
15
Actual
Bracket
10
5
0
40
Percent complete reporters and
average income,
income by quarter
09
0.9
50000
0.85
45000
0.8
40000
0.75
35000
0.7
30000
I
2000
II
2000
III
2000
IV
2000
I
2001
Percent complete reporters
II
2001
III
2001
IV
2001
FINCBTAX
41
Next steps
 Examine
effect on complete reporters and
total income
 Examine
E
i the
h effect
ff on response
– Do responders use brackets more often
 Consider
using brackets for Assets and
Liabilities
 Income imputation
42
Global vs.
s Detail Food Expenditures
E pendit res
 Interview
vs
vs. Diary
 Use Diary in published tables
 Global questions on both
– what has been your usual weekly expense at the
grocery store or supermarket?
– About how much of this amount was for non-food
items?
43
Trend in ratios
ratios, Global vs
vs. Detail
13
1.3
1.2
1.1
1
0.9
1993
1994
1995
1996
1997
1998
Battistin (Interview Global/Diary Detail)
Interview Global/Diary Global
Diary Global/Diary Detail
Diary Global/Diary Detail (no recalls)
1999
2000
44
Comparison of Food-at-home global
q estions 1999
questions,
8,000
6,000
4,000
2,000
0
All
One person
CE - Interview
Two persons
Three persons
CE -diary
Four persons
Five or more
persons
PSID
45
Equivalent Interview Food-at-home expenditures
(
(global);
) CE
C vs. PSID
S by age cohort, 1999
3500
3000
2500
2000
1500
CE - Food at home
PSID - Food at home
CE Food at home - with non-Groceries
1000
500
0
0 5 6-10
0-5
6 10 11
1115
1616
20
2121
25
2626
30
3131
35
3636
40
4141
45
4646
50
5151
55
5656
60
6161
65
6666
70
7171
75
7676
80
Age of member
NOTE: U
NOTE
Use “f
“four quarter”
t ” CU
CUs and
d adjust
dj t weights
i ht bby age and
d ttenure.
Expenditures are adjusted using equivalence scale of square root of family size
Allocate equivalent expenditures to all members of CU.
46
Next steps
 Examine
the diary global and detail
questions
 Recall
R ll vs. C
Complete
l
 Continue comparison of CE and PSID
 New Diaries
47
Comparing
C
i CE Expenditure
E
dit
Estimates with Data from Other
Sources
T monitor
To
i consistency
i
off results
l
and
To help identify areas where CE data collection
g be improved
p
and methods might
48
Issues to Consider
 Account
for differences in content or
concept (focus on components)
– can be
b reconciled
il d
– cannot be reconciled
 Source
–
–
–
–
of data
Household survey
Census
Administrative
Trade association publications
49
CE Interview/American Housing
Survey Ratios based
b d on M
Medians
di
1.2
1.0
08
0.8
1985
1987
1989
1991
0.6
04
0.4
0.2
0.0
Market
Mortgage Property
value of
principal
taxes
owned and interest
home
Property
insurance
Rent
50
CE Diary/Progressive Grocer Ratios
based on Aggregates
4.0
3.5
3.0
2.5
1994
1995
1996
1997
2.0
1.5
1.0
0.5
0.0
Total food
Fresh fruits
& vegs
vegs.
Non-alc
beverages
Fish &
seafood
51
CE Interview/MEPS Interview Health
Insurance Premiums based
b d on A
Aggregates
t
1.2
1.0
0.8
0.6
04
0.4
0.2
0.0
1996
Private
Employer hosp/phys
hosp/phys-family
family
Non-employer
Employer hosp/phys-single
hosp/phys single
52
CE Integrated/Personal Consumption
E pendit res Ratios based
Expenditures
b d on A
Aggregates
1.20
1.00
0.80
1996
1997
1998
1999
0.60
0.40
0.20
0.00
Apparel
Food
Gasoline
Rented
Vehicle
dwellings purchases
53
Attention on CE and PCE
 Definitions
of populations and expenditures
 Data sources and periodicity
 Trends over time in ratios
– CE/PCE aggregates
 Example
for total expenditures with
j
for select differences
adjustments
 BLS
BLS-CE
CE
group
Project: focus of one commodity
54
Populations
 CE:
consumer units
and persons
– Civilian non-institutional
population and some
institutional
– Continental U.S., Alaska,
and Hawaii
– Urban and Rural
 PCE: persons
– Individuals
resident
» Persons resident in U.S. and
th
those
who
h physically
h i ll llocated
t d iin
U.S. and have resided, or expect
to reside in U.S. for 1 year or
more
» Employees of U.S. businesses
abroad for 1 year or less
» U.S. government civilian and
military personnel stationed
abroad regardless of time
– Nonprofit institutions serving
i di id l
individuals
55
Definitions
 PCE
 CE
– Value of goods and services
– What consumers spend:
purchased by the personal
transaction costs including
g
sector
t ((excludes
l d iintra-sector
t
t
excise and sales taxes of
transactions) includes excise
goods and services acquired
and sales taxes
during reference period
» Spent
p by
y individuals
» Operating expenses of
– Primarily out of pocket
nonprofit institutions serving
expenditures (OOP)
individuals
reported by consumers plus
» Value of food,
food fuel
fuel, clothing
clothing,
in kind food and rent as
rent of dwellings, and
pay, and value of food
financial services received in
kind by individuals; and net
stamps
purchases of used goods
56
Data Sources
So rces and Periodicity
Periodicit
 PCE
 CE
– Household Surveys
» Interview
» Diary
– Periodicity
»
»
»
»
Annual
Quarterly
Monthly
y
Weekly
– Sources
» Statistical reports primarily from Census
Bureau (e.g.,
(e g Censuses of Manufactures,
Manufactures
Wholesale Trade, Retail Trade, Service
Industries)
» Government administrative and
regulatory agency reports
» Reports from private organizations
(trade sources)
– Periodicityy
»
»
»
»
Benchmark (detailed)
Annual
Quarterly
Monthly
57
Comparison of the CE/PCE Ratio
Over Time (1984-97) for
Selected Categories
Using PCE data based on the 1987
and 1992 Benchmark Estimates
58
Trends in CE/PCE Ratios
 Linear
regressions using the 1987
benchmark PCE estimates are depicted in
red ( )
 Linear regressions using the 1992
b h
benchmark
k PCE estimates
i
are ddepicted
i d iin
yellow ( )
59
Expenditure Categories Grouped
Using
i 1992 Benchmark
B h
k Trends
T d
 Stable
 Stable: Additional Groups
– slope coefficient is
close to zero ((+ or -))
– High
g
 Decreasing
– slope coefficient is
negative
 Increasing
» CE/PCE >= 1.0
– CE Moderately Lower
» 0.75
0 75 <= CE/PCE < 1.0
10
– CE Even Lower
» CE/PCE < 0.75
– slope coefficient is
positive
60
I A
I.
A. Stable
Stable, High Ratio
Vehicle Purchases
1.2
1
0.8
0.6
0.4
0.2
84
85
86
87
88
89
90
91
92
93
94
95
96
97
61
I B
I.
B. Stable
Stable, Mid Ratio
Rent, Utilities, and Public Services
Total Food
1.2
1.2
1
1
0.8
0.8
0.6
0.6
0.4
0.4
0.2
0.2
84 86 88 90 92 94 96
84
86
88
90
92
94
96
62
I B
I.
B. Stable
Stable, Mid Ratio
Televisions, Radios, Sound Equipment
1.2
1
0.8
0.6
0.4
0.2
84
85
86
87
88
89
90
91
92
93
94
95
96
97
63
I C
I.
C. Stable
Stable, Low Ratio
Tobacco Prods. & Supplies
Public Transportation
1.2
1.1
1
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
1.2
1.1
1
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
84 86 88 90 92 94 96
84 86 88 90 92 94 96
64
II Decreasing
II.
Alcoholic Beverages
Food Away From Home
1.2
1.2
1
1
0.8
0.8
06
0.6
0.6
0.4
0.4
0.2
0.2
84 86 88 90 92 94 96
84 86 88 90 92 94 96
65
II Decreasing
II.
Apparel
pp
and Services
Other Apparel P &S
1.2
1.2
1
1
0.8
0.8
06
0.6
0.6
0.4
0.4
0.2
0.2
84 86 88 90 92 94 96
84 86 88 90 92 94 96
66
III Increasing
III.
Footwear
1.2
1
0.8
0.6
0.4
0.2
84
85
86
87
88
89
90
91
92
93
94
95
96
97
67
Sample Comparison of Aggregate
C vs. PCE
CE
C
 Populations
 In
PCE but out of scope for CE
 In CE but out of scope for PCE
 Partly
P l out off scope for
f CE (e.g.,
(
non-profit
fi
institutions serving households and employer
payments))
 Components operationally defined differently
68
In PCE but Out of Scope for CE


Population - Persons resident in U.S.
– Employees of U.S. businesses working abroad and U.S.
government and military personnel stationed abroad
– Military living on-base
on base in the U.S.
US
– All persons in institutions and the homeless for whom
expenditures are made
– Non-profit institutions serving households
Expenditures
– Value of home production for own consumption on farms
– Standard clothing issued to military
– Services
S i
ffurnished
i h d without
ith t paymentt bby fi
financial
i l
intermediaries except life insurance carriers
69
In CE but Out of Scope for PCE
 Expenditures
– Transaction from household to household (includes
used vehicles purchased from another CU)
– Social Security contributions
70
Partly Out Of Scope For CE and
Partl Defined Quite
Partly
Q ite Differently
Differentl
 Health
Care Expenditures
– CE medical care OOP; PCE medical care expenditures made
by households, insurance companies, employers, and non
nonprofits (=current expenditures of non-profits+payments by
patients to profit and government facilities)
 Religious
and Welfare
– CE cash contributions; PCE religious and welfare not
i l di child
including
hild care (for
(f nonprofits=current
fit
t expenditures
dit
nett
of receipts for commodities; for profit and gov’t=receipts
from users)
71
Defined Quite Differently
 Education Expenditures
– CE education
ed cation OOP
– PCE education and research expenditures made by households and for
profit and non-profit institutions serving households (for private=current
expenditures net of receipts for meals
meals, etc
etc. and for gov’t
gov’t=student
student tuition
payments; fees paid to other schools plus current expenditures)
 Life Insurance and Pension Plans
– CE premiums paid and consumer contributions OOP
– PCE expense of handling: operating and administrative expenses,
premiums paid net of benefits and dividends,
dividends profits of some companies
 Owner-Occupied
Housing Expenditures
– CE expenditures for owner occupied housing (interest and charges
charges,
property taxes, maintenance and repairs, and other expenses)
– PCE imputed space rent
72
Aggregate
gg g and Ratio Comparison
p
Source
Original
($billions)
Expenditure
Flow
“Comparable” “Comparable”
($billions)
($billions)
$2,069
$2,568
(0.69)
(0.86)
Consumer
Expenditures
$2,985
Personal
Consumption
Expenditures
$4,211
$2,553
(0.61)
$3,001
(0.71)
Ratio CE/PCE
0.71
0.810
0.856
73
Reasons for Differences
 Overall
– Populations differ
– Expenditures defined differently
– Items in scope versus out of scope
– Sources of data differ (household survey vs. business and tax
reports)
– Expert judgement
– Proxy response

Component
– Content

Trends
– Benchmarks and adjustments each year between benchmarks
74
Analytical Example
1992 E
Expenditures
dit
for
f
Footwear in the CE and PCE
75
Rationale for Year and Item Category
Categor
 Why
1992?
– Data are available at finer level of item detail
– Data represent latest benchmark year
 Why
footwear?
– Apparel of which footwear is a component is a
g y where the difference in CE and PCE estimates
category
is large and increasing
– Footwear has a small number of component items in
both CE and PCE and is thus
th s easier to handle
handle.
76
Expenditures for footwear, total and by
major
j item
it
category,
t
1992 CE and
d PCE
Item category
CE Total
Men’s footwear
Boys’ footwear
Women’s footwear
Girls’ footwear
PCE Total
Men’s footwear, except athletic
Women’s footwear, except athletic
House slippers
Rubber and plastic
Footwear, except rubber, n.e.c.
Used footwear
Boot and shoe cut stock and findings
Annual expenditures
(millions of dollars)
$23,124
$7,182
3,050
10,828
2,064
$32,903
$6,267
9,452
555
8,376
8,210
41
2
77
Derivation of CE Estimates
 Diary
survey is the source for all footwear data
though both Interview and Diary collect
expenditures
dit
for
f footwear.
f t
 Individual expenditure reports originate in three
ways.
– Directly reported by respondent
– Allocation of expenditures where respondent reports
expenditure for a combination of items
– Imputation of expenditures where respondent
acknowledges purchase, but does not provide value
78
Derivation of PCE Estimates
 Process
uses data created for preparation of inputoutput accounts for U. S.
 The
benchmark purchase value of goods and
services is calculated to determine allocable
output.
 Total
purchase value is allocated among
intermediate and end users.
79
Amount of value added to total footwear
estimate
ti t by
b factor,
f t 1992 PCE
Factor
Total
Basic value
Wholesale margin
Transportation cost
Wholesale taxes
Retail margin
Retail taxes
Value added
(millions of dollars)
$32,903
$14,926
3 371
3,371
121
11
13,153
1,321
80
Production and allocation of footwear, 1992
I/O A
Accounts (Millions of dollars)
Basic
value
l
Production
Shipments
Adjustments:
Secondary production
Nonemployer receipts
Filer misreporting
receipts
Nonemployer
misreporting receipts
Trans.
cost
Comm &
whole.
h l taxes
Retail
margin
i
Retail
taxes
Purchase
value
l
4,748
3,589
143
12
13,153
1,321
22,966
201
7
8
0
0
0
216
4,547
3,582
135
12
13,153
1,321
22,750
4,727
1
9
9
2
21
Adjusted shipments
Less: Balancing Record
Allocable shipments
Imports
Adjustments:
Census re-exports
NIPA territorial adj. to imports
Whole.
margin
i
11 638
11,638
-87
204
Allocable imports
11,755
Allocable shipments & imports
16,302
11,755
3,582
135
12
13,153
1,321
34,505
Production and allocation of footwear, 1992
I/O A
Accounts – cont. (Millions of dollars)
Allocation of Production
Exports
I
Intermediate
di production
d i
Government purchases – Federal
Government purchases – State & local
Other nondurables purchases
Change
g in intermediate ggoods
inventories
Unspecified costs
Change in wholesale inventories
Change in retail inventories
PCE Military Clothing
PCE Footwear
Basic
value
Whole.
margin
Trans.
cost
Comm &
whole. taxes
Retail
margin
Retail
taxes
Purchase
value
603
159
11
147
1
109
6
1
8
0
6
4
0
2
0
0
0
0
0
0
718
169
12
157
1
31
30
41
314
39
14,926
2
1
13
68
3
3,371
0
0
0
2
0
121
0
0
0
1
0
11
0
13,153
0
1,321
33
31
54
385
42
32,903
16,302
3,582
135
12
13,153
1,321
34,505
Evaluation of CE and PCE Estimates
Standard
errors and confidence
intervals
Expert
judgment
Content
difference in component
categories
83
Standard error and 95% confidence interval
f total
for
t t l footwear
f t
expenditures,
dit
1992 CE
It
Item
Value
( illi
(millions
off dollars)
d ll )
Total footwear
$23,124
Standard error
1 145
1,145
95% confidence intervals
U
Upper
li
limit
it
$25 368
$25,368
Lower limit
$20,880
84
Expert Judgment
Data
adjustment in CE
– Allocation procedures
– Imputation procedures
Trade
margin calculation in PCE
– Wholesale
– Retail
85
Impact of allocation procedures on
foot ear estimates
footwear
estimates, 1992 CE
Item
Total footwear
Total
expenditure
Allocated
expenditure
% Allocated
$23 124
$23,124
$2 617
$2,617
11 32%
11.32%
Men’s
7,182
692
9.63%
Boys’
3,050
674
22.09%
10,828
938
8.67%
2,064
313
15.19%
Women’s
Girls’
86
Calculation of Wholesale Trade Margin in
PCE Estimates
E ti t
 Initial
margin estimate is computed for
wholesalers whose primary business is footwear.
– Total sales receipts and the cost of purchases of
footwear are obtained from Census of Wholesale Trade
(CWT).
(CWT)
– Misreporting adjustments are made to sales receipts
and the cost of purchases from IRS and other data.
– Changes in the value of inventories held at beginning
and end of year are added from CWT and Annual
T d Survey.
Trade
S
87
Calculation of Wholesale Trade Margin in
PCE Estimates
E ti t – cont.
t
 Margin
estimate for wholesalers whose primary
business is footwear is adjusted to account for two
f t
factors.
– These wholesalers may also trade in other businesses.
– Wholesalers whose primary business is not footwear
may also have footwear operations.
 CWT
data
d do
d not di
distinguish
i i h bbetween ffootwear
and non-footwear operations in either case.
88
Calculation of Wholesale Trade Margin in
PCE Estimates
E ti t – cont.
t
A
harmonization procedure is used based on sales
receipt data which is available for all businesses
engaged in footwear trade.
– Trade margin rate (wholesale trade margin / sales receipts) is
calculated for wholesalers whose primary business is footwear.
– The trade margin rate is applied to footwear sales receipts of all
wholesalers of footwear.
– Trade margin rates for other commodities handled by footwear
wholesalers are computed.
computed
– These rates are applied to non-footwear sales receipts of footwear
wholesalers.
89
Calculation of Wholesale Trade Margin in
PCE Estimates
E ti t – cont.
t
 Ideally,
the trade margin generated by applying
these rates to commodity lines handled by
footwear wholesalers = the trade margin derived
by evaluating purchases, costs, and inventory
adjustments.
 In
practice, the trade margins are not equal, so
adjustments
j
are made to the margin
g rates for
commodity lines and kinds of businesses until the
margins are harmonized.
90
Calculation of Wholesale Trade Margin in
PCE Estimates
E ti t – cont.
t
 Effect
of this harmonization procedure
– Margin
g based on purchases,
p
, costs and inventory
y
adjustments is $3,635 million
– Harmonized margin calculated by harmonization
procedure
d
iis $3,589
$3 589 million
illi
– Procedure results in 1.3% drop in footwear margin.
91
Calculation of Retail Trade Margin in PCE
E ti t
Estimates
 Procedures
are similar to those used in calculating
wholesale trade margin due to data limitations.
 Overall
retail trade margins can be computed by
type
yp of outlet
– Department store
– Hardware store
– Grocery store
but not by merchandise line, like footwear.
92
Calculation of Retail Trade Margin in PCE
E ti t – cont.
Estimates
t
 Ideally,
the trade margin generated by applying
these rates to merchandise lines handled by each
t
type
off outlet
tl t = the
th trade
t d margin
i derived
d i d by
b
evaluating purchases, costs, and inventory
adjustments for those outlets.
outlets
 In
practice, the trade margins are not equal, so the
h
harmonizing
i i procedure
d
iis applied.
li d
93
Calculation of Retail Trade Margin in PCE
E ti t – cont.
Estimates
t
 Effect
of this harmonization procedure
– Margin
g based on purchases,
p
, costs and inventory
y
adjustments of shoe stores applied to all footwear sales
is $14,463 million.
– Harmonized
H
i d margin
i calculated
l l t d ffrom hharmonization
i ti
procedure is $13,153 million.
– Procedure results in 9.1%
% drop
p in footwear margin.
g
94
Examples of content differences in
component categories of footwear
foot ear
 PCE
includes athletic footwear for participant
sports in footwear, CE includes such footwear in
recreation
ti expenditures.
dit
(Currently we cannot isolate these
expenditures in either PCE or CE to make an adjustment to the
aggregates.)
 PCE
includes boot and shoe cut stock and
findings in footwear
footwear. CE does not include such a
category in footwear. It is likely such
expenditures
p
are included in shoe repair
p services.
95
R
Research
hS
Summary
96
Issues we need to consider:
 Respondent
Burden
 Household Underreporting
 Nonresponse
p
 Uses and Users
 Cost
97
Recent Developments
Introduction of the Computer Assisted Personal
Interview instrument – April
p 2003
9 month field test underway – due to be completed in
September 2002
Redesign of the Diary Instrument
4 month field test due to start in
September 2002
i l
implementation
t ti scheduled
h d l d for
f January
J
2004
98
Ongoing Work
 Global
vs
vs. detail
 CE/PCE comparisons
 ACNielson research
 Diary research
 Income
I
imputation
i
t ti
 Study of teenage spending
99
Possible Areas for Research

Individual Diaries

Bracketing Assets and Liabilities

Randomized Interview applications

Targeted Interview applications

Interview supplements

Augmenting collected data with secondary source
data
100