Feasibility of food price monitoring in rural areas final report

              »!»•»
ys^ Feasiblity of Food Price
Monitoring in Rural Areas
rood and Final Report
Office*
Andy* and October 1996
■Hi
USDA
Feasiblity of Food Price
Monitoring in Rural Areas
Final Report
October 1996
Authors:
Joel Popkin
Jack Rutner
Kathryn Kobe
Rakesh Kochhar
Eva Jacobs
Helen Jensen
Submitted by:
Joel Popkin and Company
1101 Vermont Avenue, NW
SuHe 201
Washington, D.C 20005
Project Director: Joel Popkin
Principal Investigator: Jack L Rutner
Submitted to:
Office of Analysis and Evaluation
USDA Food and Consumer Service
3101 Park Center Drive, Rm. 214
Alexandria, VA 22302
Project Officer Margaret Andrews
This study was conducted under Contact No. 53-3198-5-063 wft to rood and Coniumw Service,
United States Department of Aancuhuro, under i» cMhoriNof AoFoodSlompActof 1977,ot
amended, PDmtsot vtew or opinions stated minis report do not nooonony wproMnt me official
nmitinn etlttm Food and <**««—TIM Sarvica.
/
FINAL FEASIBILITY REPORT
FEASIBILITY OF FOOD PRICE MONITORING IN RURAL AREAS
Table of Contents
Page
ACKNOWLEDGMENTS i
GLOSSARY OF ABBREVIATIONS ii
EXECUTIVE SUMMARY iii
I. INTRODUCTION J-l
B. BACKGROUND D-l
HI. THE HYPOTHETICAL DATA COLLECTION SYSTEMS: AN OVERVIEW ffl-1
m.A. Conceptual Factors m-1
III.B. Methodological Issues III-4
m.C. Organizational Structure 01-13
III.D. Components of Cost 111-16
IV. THE BLS ALTERNATIVE IV-1
IV.A. Summary Description of BLS Alternative IV-1
IV.B. Organizational and Operational Factors IV-3
V. THE BLS/NASS AND ERS ALTERNATIVE V-l
V.A. Summary Description of the BLS/NASS and ERS Alternative V-l
V.B. Organizational and Operational Factors V-3
V.C. Budgetary Factors V-4
VI. THE COMBINATION BLS/PRTVATE CONTRACTOR ALTERNATIVE VI-1
VIA. Summary Description of the BLS/Private Contractor Alternative VI-1
VLB. Organizational and Operational Factors VI-2
VLC. Budgetary Factors VI-5
VD. THE SCANNER ALTERNATIVE VH-1
VII.A. Summary Description of the Scanner Alternative VII-1
VII.B. Organizational and Operational Factors VII-2
VII.C. Budgetary Factors VII-3
VD!. PROS AND CONS OF THE ALTERNATIVES VHI-1
VHLA Introduction VHI-1
Vm.B. The BLS-type Alternatives VHI-2
Vm.C. A Scanner-type System VIII-9
APPENDLX 1: MATRIX
APPENDLX 2: DATA COLLECTION FIELD STUDY
APPENDIX 3: THE POPS
APPENDLX 4: COST ELEMENTS FOR BLS AND OTHER ALTERNATIVES
APPENDIX 5: THE BLS ALTERNATIVE: DERIVATION OF COSTS
APPENDLX 6: THE BLS/NASS ALTERNATIVE: DERIVATION OF COSTS
APPENDIX 7: THE BLS/PC ALTERNATIVE: DERIVATION OF COSTS
APPENDLX t: STATISTICIAN'S FINAL TECHNICAL MEMO
APPENDLX 9: THE EXPERT PANEL
^
AdrnflHtodgnaatt
Joel Popkin and Company wishes to acknowledge and thank the following individuals and
organizations that contributed to this study.
• At USDA's Food and Consumer Service, Margaret Andrews, the COR, provided
valuable guidance in conducting the research.
• The members of the Expert Panel-Bill Hawkes, Eva Jacobs, Helen Jensen, Richard
Mantovani and Linda Neuhauser—provided critical comments and important suggestions
to Joel Popkin and Company during every phase of the project.
• At the Bureau of Labor Statistics, many people provided generous support and
information throughout this study, especially Robert Cage, Bill Cook, Charles Fortuna,
John Greenlees, Walter Lane, Jane Martinez, Charles Mason and Janet Williams.
• Many people at the Bureau of the Census also provided valuable assistance and
information, including Chester Bowie, Audrey Brinkley, Ronald Dopkowski and David
Hubble.
• Individuals at private organizations provided important information about alternative
sources for and methods of collecting food price data. Those individuals are Joe Garrett
of A.C. Nielsen, Richard Mantovani of Macro International, Kirk Wolter of NORC,
Michael Moritz of International Resources, Inc., and Amy Starer of Genesys.
• At USDA's National Agricultural Statistical Service, Robin Roark, Bob Milton and Doug
Kleweno provided valuable information about NASS' operations.
• At USDA's Economic Research Service, Jim MacDonald and Annette Clausen provided
valuable information about ERS' operations.
3
GLOSSARY OF ABBREVIATIONS
BLS Bureau of Labor Statistics / U.S. Department of Labor
CATI Computer Aided Telephone Interview
CES Consumer Expenditure Survey
CPI Consumer Price Index
CPS Current Population Survey
ERS Economic Research Service / U.S. Department of Agriculture
FCS Food and Consumer Service / U.S. Department of Agriculture
JPC Joel Popkin and Company
NASS National Agricultural Statistical Service / U.S. Department of Agriculture
PC Private Contractor
POPS Point-of-Purchase Survey
PSU Primary Sampling Unit
TPOPS POPS conducted by telephone interview
USDA U.S. Department of Agriculture
f
EXECUTIVE SUMMARY
The Food and Consumer Service (FCS) of the U.S. Department of Agriculture (USDA)
desires to monitor how the movement in prices of food items typically bought by rural
households compares with the movement of those bought by urban households. Information
on the movement of prices paid by urban consumers for food items consumed at home has
been available since 1978 when the Bureau of Labor Statistics (BLS) defined the scope of the
Consumer Price Index (CPI) to be all urban households. The most logical and direct answer
to the FCS question is to develop a food price index for rural households that can be compared
with the CPI component for food at home. But, there are various ways that can be
accomplished. This report identifies, describes and evaluates them.
The first step in any of them is to determine the areas of the country that comprise an
adequate representative sample ofrural households and then conduct a point-of-purchase survey
(POPS) for rural households. The Census Bureau (Census) now conducts a POPS for BLS
which determines where urban households represented in the CPI shop. That survey for rural
households may show that they shop largely in urban areas. If so, the FCS could construct a
rural price index by reweighting BLS price indexes for food item categories by the percentage
shares of spending by rural households on those categories, if those shares differ from urban
ones. Data on those shares are collected in the annual Survey of Consumer Expenditures (CES)
that BLS now uses for urban spending patterns.
If the POPS shows mat rural households shop for an important share of their food in rural
areas, it will be necessary to construct a food-at-home price index for rural households that is
comparable to the CPI. The most direct way to do mat is for BLS to construct a rural food
price index using the same methods it uses for the CPI. Doing that is the best way to insure
that die rural and urban food price indexes are comparable, so that differences in movement
between them can be ascribed to market behavior differences, not to methodological ones.
iii
BLS is interested in conducting this work for FCS. The main consideration in deciding
whether to contract with BLS or another organization would be differences in costs that were
large enough to outweigh the advantages associated with the degree of comparability
achievable. It turns out that, based on the cost estimates developed by Joel Popkin and
Company (JPC) for this report, BLS is cost competitive. The competitiveness of BLS extends
to both the start-up period and to the subsequent annual costs of the ongoing program, vis-a-vis
all approaches that can now be costed, save one. In that one, price collection is turned over
to the National Agricultural Statistical Service (NASS). This approach envisions that NASS
staff would be trained by BLS, use BLS methods and specifications for selecting items to be
priced and for pricing them, and that BLS would compile the index, using CPI methods and
programs, once NASS sent them the price data. Some cumbersomeness and the need for more
complex USDA coordination of the process, probably by FCS, must be weighed against the
saving which, at this stage of cost estimation is probably not significant.
The BLS cost is below that of another approach explored in this report. It is premised
on contracting for price collection with a private firm. The approach has two variants. In one,
the contractor is assumed to have a small staff, dedicated to food price collection, that will
travel from one rural place to another. In the other, the contractor is assumed to use a part-time
price-collection staff dispersed over the country to minimize travel costs including time. The
cost of the dispersed staff is below that of the small-staff model. But neither is cost
competitive with the BLS nor NASS alternatives. That is because BLS can utilize a large,
dispersed part-time staff-its own-and because its staff is already trained for the task of price
collection.
Besides the BLS, NASS and private contractor approaches, a fourth alternative was
considered It is to use optically scanned data, augmented by some staff field collection to
obtain price quotes. While the use of such data for price indexes may eventuate in the future,
its use now as a source for rural prices would significantly diminish comparability with the
urban CPI. To compare the movements of urban and rural prices would require mat a totally
iv
new scanner price index be calculated for urban households. This would make the use of
scanner data too costly under present circumstances.
As noted earlier, the first step in this process is to conduct a POPS. Indications are that
private contractors may be able to conduct that survey at lower cost than Census. If so, BLS
might choose not to publish the index as its own official index or to publish it with a technical
footnote that might include a BLS assessment of the potential impact of the non-Census POPS
on the comparability of the index. The FCS will have to consider the type of BLS imprimatur
it desires in deciding who should conduct the POPS.
Once the price collection is begun, it will probably be necessary to calculate the rural
food-at-home index for several years to compare its movement with the CPI for food at home.
A sound way to make the comparison is to calculate and publish the index every other month,
which would coincide with BLS pricing cycles for the CPI. Percentage changes in the
bimonthly index and its components from their value a year earlier would facilitate
comparisons with the urban CPI without having to obtain a time series long enough to facilitate
seasonal adjustment. An understanding of the results of the comparison would be easier if the
rural index can be published for each of the five major categories of food at home. By
analyzing those categories, and the more detailed indexes likely to be available in tabulations
that may not be officially publishable by BLS, it should be possible to determine if rural prices
rise faster or slower than urban prices, and, if so, why. Is the difference in aggregate behavior
across die board, or is it due to certain items purchased in certain types of stores? Obviously,
if there were no differences of concern to FCS, it could continue to track food prices faced by
rural households by using BLS indexes for detailed food expenditures in the urban CPI and,
if necessary, weighting them by the rural spending patterns measured in the CES.
This report will form the basis for the development by JPC of a recommended action
agenda for USDA That agenda, when presented to USDA officials at a briefing, is the final
deliverable of tins contract
?
L INTRODUCTION
This Final Feasibility Report is one of the final deliverables under USDA contract #53-
3198-5-083, "Feasibility of Food Price Monitoring in Rural Areas," awarded to JPC. The
remaining deliverables are the Action Agenda and Briefing Materials that reflect a weighing
of all the issues identified and evaluated in this report bearing on the question of how to
measure the movement of food prices faced by rural households.
The purpose of this section is to inform the reader of the organization of this report. This
report contains seven major sections (II through VIII) and nine appendixes. Section II provides
background about the reasons for the study. It also explains the phases in which the study was
conducted and indicates the individuals and groups associated with each phase.
Section III provides essential discussion of the data collection systems that hypothetically
could be used to achieve FCS's objectives. That section, and the rest of the study, are
structured to delineate four kinds of considerations in devising the data collection systems:
conceptual; methodological; organizational; and budgetary.
Sections IV through VII of the report are each devoted to the description and evaluation
of one of the four approaches considered to be most likely to achieve the analytical objectives
of FCS. The organizational, operational and budgetary aspects of each is discussed as well.
Due to the complexity of deriving costs for the alternative approaches, detailed information on
the derivation of costs has been placed in several appendixes. Appendix 4 describes the
general costing model used. Appendixes 5,6 and 7 present detailed cost estimates for the three
alternatives for which costs could be estimated.
Section VIII, the final section of the main report, presents the pros and cons of each of
the four alternative approaches. These pros and cons are a basic input to the development of
the Action Agenda mentioned earlier.
1-1
/
There are five other appendixes in addition to the four previously described. Appendix
1 summarizes, in a matrix format, the range of research questions that FCS could have interest
in and how each might be answered. The matrix was used in the early stages of the project
to help in defining precisely the main question the FCS wished to focus on in this study.
Appendix 2 reports the results of a field test conducted to determine the time
requirements that might be encountered in collecting food prices in rural areas. Appendix 3
describes the point-of-purchase survey, a key instrument in determining where rural households
shop, and, to some extent, the items they buy. A number of statistical sampling questions
arose in mis study; key questions are addressed in the report of a consulting statistician found
in Appendix 8. A final appendix identifies the members of the Expert Panel and their
respective areas of expertise.
1-2
L
D. BACKGROUND
The affordability of and access to a nutritious and safe food supply is an important objective
for the USDA's FCS. The primary goal of USDA food-assistance pre grams is the provision
of nutrition security. Questions have been raised about whether persons living in rural areas
of the U.S. generally have the same access to sources of nutritious food the urban population
has. Past studies have shown, albeit on a limited basis, that the cost of acquiring a given
market basket of food also can be higher in underserved areas and that varieties of nutritious
foods are more limited.
A related question, and one about which nothing is presently known, is how do food prices
behave in rural areas? The CPI compiled by BLS measures price changes only in the urban
areas of the U.S. For that reason FCS is interested in determining the feasibility of monitoring
food price changes in rural areas so they can be compared with those experienced by urban
households. The purpose of this project has been to examine methodologies to monitor prices
in rural areas and to assess their feasibility.
Work on the project was divided into four distinct phases. The first was meeting with FCS,
forming an Expert Panel and then meeting with the Expert Panel and FCS staff. The second
was the Conceptual Assessment Report. The third was the Feasibility Study and the final one
is die Action Agenda that will provide recommendations.
In the first phase, members of the research team met with FCS staff to focus on defining the
specific research objectives of rite study. The need to define specific objectives arises because
the affordability of food in rural areas of the U.S. can be assessed in a number of ways. Also,
price collection methodologies are not independent ofthe question being asked. What emerged
from the first meeting of the research team with FCS led to a Memo of Understanding (MOU).
It contained a "matrix" listing six possible research questions and the price-collection
methodologies that would satisfy each question. The matrix is found in Appendix 1.
n-i
/o
Another purpose of meeting with FCS was to form an Expert Panel that would advise the
research team. The members selected for that Panel were expected to contribute independent
ideas based on their expertise in the areas of price collection and food stamp and nutrition
programs. The Expert Panel was also expected to review the design of the study and interim
findings from different stages of the research.1
Part of the first phase of the project was to meet with the Expert Panel and with FCS. An
important part of the discussion during this meeting was the matrix of research questions and
price-collection methodologies. The decisions to emerge from that meeting were critical to the
writing of the Revised Study Plan. The Revised Study Plan outlined the final research
objectives for this project and formed the basis for all subsequent work.
A key decision of the first Expert Panel meeting was that the principal focus of this project
would be the following question:
How does the movement in prices offood items typically bought by rural households
compare with the movement in prices of food items typically bought by urban households?2
One way of answering that question is to construct a price index for food at home faced by
households living in rural areas of the U.S. using techniques similar to those used by BLS.
An index produced by such methods could then be compared to the CPI for food at home
which is based on prices faced by urban households. The rural food price index would be
national in scope, it would cover the households in rural areas not included in the CPI, and it
would not be limited to food-stamp households living in rural areas or to the market basket
purchased by those households. It was also decided that the research would examine the
'See Appendix 9 for a list of the members of the Expert Panel and the areas of their
cupeilise.
2This is a variant of question 4 in the matrix found in Appendix 1.
n-2
J/
feasibility of constructing a price index for food at home using scanner technologies to collect
price data.
The decision to focus generally on comparing price changes, essentially to replicate a CPI
for food at home in rural areas, leaves unexplored the possibility that at some point interest
may surface in a method for comparing the level of prices in rural areas with the level in urban
areas. Price-collection methodologies currently in use, especially the BLS methodology, are
specifically designed to measure price changes over time. A comparison of price levels across
different areas may serve some USDA goals but would require a significant departure from
those methodologies.
The second phase of this study involved the preparation of a Conceptual Assessment Report.
The purpose of the report was to review various possible methodologies for price collection
in rural areas. An important goal was to develop a full understanding of the BLS methodology
for compiling the CPI with particular reference to its applicability to monitoring food prices
in rural areas.
The methodological review for the Conceptual Assessment Report encompassed issues such
as sampling procedures, staff training, price collection methods, data editing, and index
construction. That work was accomplished through a combination of literature review and
interviews with staff members from BLS, Census, and private data-collection companies.
Alternative price-collection procedures such as the use of electronic scanners were also consid-ered.
The Expert Panel convened twice more during the second phase to provide input to the
research team. In the first of those two meetings, the Conceptual Assessment Report was the
principal subject of discussion. The review I alternative methodologies during that meeting
was key to developing an initial list of hypothetical data collection systems. Most of the data
collection systems hewed closely to the BLS methodology differing mainly in their organiza-
II-3
/-
tional and operational aspects. The most significant methodological departure was a
hypothetical system relying upon the collection of price data via electronic scanners.
In the second of the two meetings during phase two, the Expert Panel convened to review
the preliminary list of hypothetical data collection systems. That meeting and a subsequent
meeting with FCS were instrumental in shaping the final list of hypothetical data collection
systems presented in this report.
The third phase of the study was to discuss conceptual, operational and cost factors
associated with several hypothetical price collection scenarios. The present report is the
culmination of the third phase.
The final phase of this project will conclude with the delivery of an Action Agenda that will
provide a set of recommendations regarding the best methodology for monitoring food price
changes in rural areas.
0-4
/3
ID. THE HYPOTHETICAL DATA COLLECTION SYSTEMS: AN OVERVIEW
The purpose of this section is to provide an overview of the hypothetical data collection
systems that will allow FCS to monitor changes in prices of food items typically bought by
rural households. These systems are intended to lead to the construction of a price index for
food purchased by rural households. That index could then be monitored relative to an index
of food prices faced by urban households.
As previously indicated, there are two principal kinds of collection systems being considered
to arrive at the price index needed to compare food-price movements confronting rural and
urban households. One system relies upon the BLS methodology and would lead to rural price
indexes that can be compared directly to the CPI for urban households. The second system
uses scanner technologies to collect prices in outlets frequented by rural households. Because
this system is a significant departure from BLS methodology, the index produced by it cannot
be compared to the CPI for urban households. Thus, this system would require producing both
an urban food price index and a rural food price index.
The discussion in this section first reviews the concepts that are germane to consumer-price-collection
systems as well as those that are of particular relevance for price collection in areas
where rural households shop. The section then touches upon key methodological issues that
impact upon the organization and final cost of a price-collection system. Details on the
different hypothetical data collection systems are provided in subsequent sections ofmis report
HLA. Conceptual Factors
nLA.1. The Scorn of a Rural Food Price Index
The objective is to determine how changes in food prices faced by rural households compare
with changes in food prices faced by urban households. A readily available measure of the
latter is the food-at-home component of die CPI. Because the CPI would serve as the principal
comparator for the rural food price index, it was decided that the rural index would be
nationally representative covering the geographic areas not presently included in the CPI. In
ra-i
/S
other words, rural households covered by this index will be those residing in areas outside of
metropolitan areas and other urban places with at least 2,500 people. Rural households, so
defined, account for about 14 percent of all U.S. households.
Another key aspect of the rural food price index is that it would cover all food-at-home
items typically purchased by rural households. In other words, the index will not be
constrained to reflect specific market baskets, such as the Thrifty Food Plan market basket, or
to reflect the expenditure patterns of specific types of rural households, such as food-stamp
households. Price data for the index will be collected mainly from a sample of stores located
in rural areas. If rural households are found to shop in urban stores, price data for those stores
will be collected either from BLS or from scanner data. The rural food price index will be
designed to measure the change over time in food pr~.es faced by rural households. The
principal policy question addressed by the index is whether the CPI yields an adequate
approximation to changes in rural households' food costs.
m.A-2. Price Inforcg and Their Confrtnrction
The purpose of a "price index" is to provide a way for combining many separate prices into
a single number. This makes it possible to compare the percentage change in price from one
time to another for a group of commodities and/or services. The determination of which prices
to combine will depend upon the use to which the price index will be put Thus, for example,
a price index that is meant to measure the course of consumer food prices would obviously
combine only the prices of food items typically purchased by consumers.
The construction of a price index requires a method for collecting the data needed for the
index and a method for aggregating the data into an index. With respect to data, there are two
key components of an index: expenditures and prices. Data on expenditures are used to
combine the prices of the various hems mat are part of an index. For indexes attempting to
mearjre changes in consumer prices, the expenditure component of an index is based on
household expenditure patterns. The expenditure component is generally drawn from a specific
m-2
is
time period and is not changed frequently. For example, the expenditure component of the CPI
is revised by BLS every ten years.3
The construction ofa consumer price index can be decomposed into the following five steps:
1. Survey households to determine expenditure patterns;
2. Survey households to determine the outlets (i.e., stores) from which purchases are
made and use the results to select a sample of outlets from which prices will be
collected;
3. Conduct an initial survey of the sample of outlets to select the items whose prices will
be representative of each expenditure category to be included in the price index;
4. Perform routine surveys of the same outlets to collect the prices of the representative
items at different points in time;
5. Verify price data and compute the price index for each time period.
Each of the five steps outlined above can be performed in a number of different ways
depending upon the specific objective and choice of methods. In this report, die word
"methodology" will be used to refer to all aspects of price index construction as enumerated
above.4 The various hypothetical data collection systems described below in this paper share
3Some changes in purchasing habits are captured every year based on the results of annual
expenditure surveys. This involves adjusting the probabilities with which items are selected
for the CPI pricing sample. BLS has also introduced a new method to sample outlets and
aggregate price quotations for each Hem. However, die expenditure weights mat are used to
combine prices of different items, technically called strata, are revised infrequently which is
one reason that the CPI may not be a true representation of changes in the cost of living faced
by consumers. Households—urban and rural—are likely to switch away from commodities that
have become more expensive and towards commodities mat have become relatively less
expensive, but such cost-moderation behavior by consumers is not immediately reflected in the
CPI.
4There are, of course, other aspects of index number construction but they are not critical
for the purposes of this report An example of a methodological issue not discussed here is
the choke of an index number formula. Several other issues were covered previously in the
Conceptual Assessment Report.
ra-3
l<-
a common objective and confront similar methodological choices. They differ with regard to
some details but also share in several others, The common aspects of list methodologies used
by the different hypothetical data collection systems are discussed below.
m.R Methodological Issues
TTT.B.I. A PCTfthmrt Methodology
In view of the research objective, a key part of mis research has been a thorough review of
BLS methodology and its applicability to the construction of a rural food price index. Because
the BLS methodology serves as the point of departure for all but one of the hypothetical data
collection systems, it is described here as the "benchmark" methodology. The purpose of this
section is to describe mis methodology in the framework of the five steps (listed earlier)
needed to construct a price index.3
The first step mat must be taken for the construction of a price index is to determine the
expenditure patterns of households mat will be covered by the index. Because of the national
scope of the CPI, these data are collected by means of surveying a random sample of
households. The design of the sample and the actual expenditure survey are done on an
ongoing basis by Census for BLS. This survey is called the Consumer Expenditure Survey
(CES). The objective of the CES is to provide the basis for revising the expenditure
components and pricing samples for the CPI and to meet the need for timely and detailed
information on consumption patterns of different types of families in different areas of the
country.* An important aspect of the CES is mat it covers both rural and urban households.
'Further details on the BLS methodology can be found in the Conceptual Assessment
Report or in the BLS Handbook of Methods, U.S. Department of Labor, Bureau of Labor
Statistics, Bulletin 2414, September 1991
*Tbe CES consists of two parts, a quarterly personal Interview and a Diary Survey which
is completed at home by the respondent for two consecutive 1-week periods. The Diary
Survey was designed to collect detailed expenditures for frequently purchased items which
would not be remembered over a three month period. One such set of hems includes food.
The daily expense record is divided by day of purchase and by broad classifications of goods
and service! a breakdown mat aids the respondent when recording dairy purchases. Thediary
m-4
n
From the point of view of a rural food price index, this means that the first step is already
fulfilled by Census and the USDA does not nted to arrange for the collection of data on the
food expenditure patterns of rural households.
The sampling design mat Census uses for the CES has implications for the CPI as well as
the rural food price index. Census first maps out the U.S. into small areas called Primary
Sampling Units (PSU's) and then randomly selects a subset ofthose areas where it will conduct
its surveys. For a forthcoming revision of the CPI, Census has selected a sample of 107 PSU's
for the survey of consumer expenditure patterns. A map showing the locations of those PSU's
is on the next page. The map also shows that there are four principal types of PSU's: A, B,
C, and D. The A PSU's are the large metropolitan areas that are drawn automatically into the
CES. The B PSU's are a set of smaller metropolitan areas of the U.S. that are selected
probabilistically. The C and D PSU's, of which there are 28, are non-metropolitan areas also
selected probabilistically. Together, these four types of PSU's are intended to represent all of
the U.S., rural and urban.7
expenditure data provide the basis for establishing the classification and expenditure weights
for classes of items included in the CPI. While the basic weights for the CPI are changed only
about every ten years, the expenditure survey data are examined every year to adjust the
selection probabilities of items in the CPI and to determine the number of price quotes to be
assigned for collection, depending on the amounts ofexpenditures reported for different classes
of items. That procedure allows BLS to account somewhat for change in purchasing habits
in the decade mat usually passes between the major revisions in expenditure weights.
The C and D PSU's are similar in mat bom are a mixture of rural and (non-metropolitan)
urban areas. Indeed, BLS intended to make no distinction between the C and D PSU's. Those
two types of PSU's originally comprised a larger group of PSU's selected for inclusion in a
CPI mat would cover the entire country, not just the urban areas. When budgetary
considerations led to the decision not to expand the CPI to include rural areas, BLS retained
a smaller set of PSU's-the C PSU's-for inclusion in the CPI-U. Excluded from the CPI-U
are some PSU's mat are purely rural and some mat are a mix of rural and urban areas. These
are the D PSU's. It should be noted that the C PSU's do not include any purely rural areas.
ra-5
LT
PSUs in the Consumer Expenditure Survey
i
Once the upcoming revision of the CPI has taken place, the food-at-home component of the
CPI will include prices from the A, B and C PSU's. Households that will be covered by the
rural food price index reside in the C and D PSU's. Consequently, the C and D PSU's are
prime candidates in which to collect prices for a rural food-at-home price index. In all but one
of the data collection systems proposed below, it will be possible to use price data collected
by BLS in the urban portions of the C PSU's. Consequently, in the C PSU's, primary price
collection will take place mostly in the rural areas whereas in the D PSU's, primary price data
will be collected wherever rural households shop, in rural as well as urban outlets.
After BLS has determined the areas in which it is going to collect expenditure and price data
for the CPI, it needs to ascertain which outlets within those areas should be visited by its price-collection
staff. Thus, on behalf of BLS, Census conducts one more survey called a Point-of-
Purchase Survey (POPS). The purpose of this survey (see Appendix 3 for details) is to ask
households about the location of outlets (stores) in which they purchase specific groups of
items. BLS then samples from this list of outlets to obtain a set of stores from which prices
will be collected for the CPI. In contrast to the CES, it is important to note that the POPS
does not encompass many rural households, such as those in D PSU's, that are intended to be
covered by the rural food price index. Therefore, this type of survey will have to be conducted
of rural households in any system for constructing a rural price index.
The third step in the construction of a price index is to conduct initial visits to the outlets
selected from the POPS. This process, known as outlet initiation, serves two important
purposes. First, it helps secure the cooperation of the store for future visits by field
representatives. Second, the items in the store for which prices will be collected on a regular
basis are chosen during mis visit1 Prices of these items are also recorded at this time.
"These items are chosen on a probabilistic basis with items purchased more frequently
receiving a higher probability of selection. In outlets visited by bom rural and urban
households, it is possible mat the consumption patterns of urban households may reduce the
probability of iHecfmg hems favored by rural households for inclusion in the rural food price
index. See the Conceptual Assessment Report for more detail on the hem selection process.
ffl-7
Jc
After outlet initiation has been completed, step fear in the construction of an index begins
and prices are collected routinely from those outlets on a schedule dictated by the frequency
of the index. During each visit, the objective is to collect prices for the same set of items
selected during initiation. The gathering of reliable price data depends in a large way upon the
training of field representatives and the proper initiation of outlets. For that reason, the BLS
methodology is notable for the extensive two-stage training imparted to field representatives.
Another notable aspect of BLS methodology is the use of specification sheets. The
specification sheet is essentially a checklist covering a wide range of characteristics that could
be used to describe an item being priced. Specification sheets are supplied by BLS to its field
representatives for each of the most detailed categories of expenditures included in the CPI.
These checklists are completed in detail during store initiation and help to ensure that the same
item is priced during subsequent visits. BLS-type price collection systems proposed here call
upon the use of BLS training and price collection procedures.
In the final stage, price data are sent to BLS headquarters in Washington where they are
verified and aggregated into a price index. In this report, the term verification refers to a
process whereby the price data that BLS receives at headquarters in Washington are analyzed
to identify those that do not appear sensible and need further examination. For example, there
could be large differences among price quotes for the same item or the prices for some items
could be very different from those reported the previous time. For a price that does not appear
sensible, BLS verifies whether or not an error has been made in recording it Some of the
verification procedures simply determine whether the price is subject to sharp seasonal
fluctuations. Other procedures include asking BLS field offices to recheck the price. In the
BLS-type collection systems described later in this report, field checks would be the
responsibility of the organization mat is doing the price collection.
m.B.2. Methodological Aspects of BLS-Tvpe Price Collection Systems
All but one of the proposed hypothetical data collection systems are based on the BLS
methodology. BLS-type systems will all lead to a rural food price index that can be compared
directly to the urban food-at-home component of the CPI. As previously mentioned, the
DM
£\
consumer expenditure survey as currently conducted by Census is an adequate source of data
on the expenditure patterns of rural households. Therefore, the hypothetical data collection
systems need not replicate this step in the construction of the rural price index. But the
remaining steps in index computation—the point-of-purchase survey, outlet initiation, price
collection, and index compilation—need to be analyzed because the systematic collection of
food prices in outlets where rural households shop has no precedent.
Collecting food prices for the rural index using BLS methodology raises several issues that
affect the cost and analytical content of the resulting index. These issues are as follows:
1. Household survey of shopping patterns (POPS):
A. In what number of nonmetropolitan areas currently on the Census' PSU list will
price data be collected?
B. Should nonmetropolitan areas be added to the current list of PSU's sampled by
Census?
2. The number of categories of food needed to produce the rural price index and level of
detail that needs to be publishabL;
3. Number of outlet and price quotes needed per category of food expenditures;
4. Frequency of price collection and index calculation.
The issues raised above are common to all BLS-type data collection systems and the
resolution of these issues will also be shared by all systems. Each issue is elaborated on in
turn below.
I The POPS
The first issue is the number and location of rural areas in which a survey of household
shopping patterns should be conducted Where the survey occurs will determine where prices
will be collected for the rural food price index. In one scenario, a POPS could be conducted
in all 28 non-metropolitan PSU's now in the CES, and food price data collected in all 28. In
another scenario, the POPS could be conducted in a sub-sample of those 28 PSU's and prices
ra-9
>*—
collected in that subset In a third alternative, the rural POPS could, in addition to a subset of
the 28 PSU's, include other PSU's not now in the CES.9 Cost estimates will be provided
below for a range of possibilities.
Once die POPS is conducted in rural areas, its importance extends beyond just being used
to select the outlets at which food prices are collected for the rural food price index. The
POPS may reveal mat the bulk of shopping for food by rural households is actually at outlets
in urban places. In mat case, there may be no need to collect prices in rural areas and a rural
food price index could be constructed using existing BLS price data for the C PSU's. The
rural food price index so compiled may still differ from the food-at-home CPI for the C PSU's
if, in comparison to urban households, rural households buy food items in relatively different
quantities. If rural and urban households arc alike with respect to the items they buy and
where they buy mem, the two indexes will behave identically and there may be no need to
compile the rural food price index.10
'Since (he number of non-metropolitan PSU's is well below the number of states in the
U.S., it is inevitable mat several states will be left out of the pricing exercise. This does not
compromise the representativeness of the rural food price index but, for policy reasons, it is
possible mat it is desired some states not currently playing host to a C or a D PSU be included
in the collection of prices in rural areas. That objective could be achieved by taking another
"draw" of C and D PSU's by removing some from the current list and including some new
ones.
'"Suppose mat consumption patterns are found to differ across rural and urban households.
That possibility adds an interesting dimension to the comparison between the rural and urban
food price indexes. In footnote 3, it was noted mat the CPI is an approximation to the "true"
change in the cost of living faced by households. Economists have long been concerned with
the issue of how well changes in the CPI compare with changes in the true cost of living.
Once a rural food price index has been constructed, one could pose the following question:
How well do changes in the rural food price index approximate the true changes in food costs
fiwed by rural households? The answer to this question may not be the same as mat for a
corresponding question posed with respect to the urban food price index.
ffl-10
PJ
2. Number ofCategories ofFood to be Included in the Index
The CPI has five major expenditure categories for food at home. They are as follows:
Cereal and Bakery Products
Meats, Poultry, Fish and Eggs
Dairy Products
Fruits and Vegetables
Other food at home
Those categories are sub-divided further into approximately fifty expenditure categories which
are all represented in the published CPI at the present time. One of the issues for the rural
food price index is the level of detail on food categories that will be presented when the index
is published. The level of detail depends on USDA requirements. But factors entering into
the decision include providing users enough detail to understand the sources of differences
between the CPI and the rural food price index and to establish a necessary level of credibility
in the rural price collection methodology. The level of detail mat is eventually published will
also depend on choices made with respect to other methodological issues discussed in this
section as well as the statistical properties of the data collected for the rural food price index.
3. Number ofPrice Quotes Needed Per Category ofFood Expenditures
The number of price quotes per category of food expenditures needed for publication status
is uncertain at mis time. The answer depends upon factors such as the number of PSU's mat
are selected for pricing, the number of outlets located in those PSU's, and the observed
variance of food prices. Assuming mat one-half of the 28 C and D PSU's are selected for
pricing, BLS is confident mat its normal methodological procedures will yield a sufficient
number of price quotes for publication of the rural food-at-home index at the aggregate
level.11 If the variance of prices is low, publication at a more detailed level, say for the five
major categories listed above, may be possible. Publication at the detailed level may also be
"For the upcoming CPI revision, BLS plans to collect price quotes from four outlets for
each POPS expenditure category in each PSU. The four outlets are randomly selected for each
category. POPS currently has 49 categories of food expenditures. For some seasonal products
more than one price quote will be collected per outlet
ffl-11
V
made possible by increasing the number of PSU's in the pricing sample, by increasing the
number of outlets or by gathering more price quotes per category of food expenditures than
are normally obtained by BLS for the CPI. Because of the limited number of outlets in rural
areas, only the first and third options may be feasible, while cost considerations suggest the
third option as being the most feasible.
4. Frequency ofPrice Collection and Index Calculation
The final element to consider for the hypothetical data collection systems is the number of
times in a year price data will be collected and an index calculated. There are three
possibilities under consideration. One is monthly price collection and index calculation, a
second is bi-monthly, and the third is quarterly. An advantage of monthly and bi-monthly
price collection is that those procedures can be easily incorporated by BLS into its standard
operating practices in the event that BLS is involved in the construction of the rural food price
index. Price collection and index calculation on a quarterly basis may be a less costly
alternative, but it will make it more difficult to integrate the work with BLS operations for the
CPI.
m.B.3. Price Qgmtm MM *"*"" T«*«*W*
The scanner data collection system will have to undertake many of the steps required of the
BLS-type systems. Like the BLS-type systems, the implementation of a scanner system will
begin with a POPS. In mis case, though, the POPS will have to be conducted of urban and
rural households in the scanner-company data base ofhouseholds. Currently, one company has
about 46,000 useable urban households and about 2,000 rural households. The reason for
conducting the POPS from this list ofhouseholds, as opposed to Census' sample ofhouseholds,
is to maximise the probability mat the outlets in which households are observed to shop are
also found in the scanner-company's data base of retail outlets.
An impoitaut concern regarding the scanner alternative is that not all outlets in which rural
households shop will be in the scanner-company's outlet data base. That is because either the
stores were not originally sampled by the scanner company or the stores do not have scanner
ra-i2
A3-
equipment. For outlets with scanner equipment, the scanner company could make
arrangements to gamer the requisite price data. However, the other stores will require personal
visits for price collection and this may be fairly expensive.
A runner limitation of scanner data is that about 30 percent of food-item purchases are not
adequately scanned for price index purposes. These are items such as fresh fruits, vegetables,
meats, chicken and fish. The lacV of adequate price data on these items in scanner data bases
can only be overcome by personal visits to the stores.
Finally, the scanner alternative requires that an urban food price index be produced as well
as a rural food price index. That is because there may be some systematic (nonrandom)
reasons why the universe of households sampled by scanning companies differs from the BLS
universe of households. It is also possible that indexes based on scanned prices produce results
systematically different from those based on prices collected by BLS field representatives for
reasons not yet understood or researched. For these reasons, a rural food price index compiled
from scanner data may not be directly comparable to the food-at-home component of the CPI.
In mat case, a judgement about whether changes in food prices faced by rural and urban
households are the same can only be made if an urban index is also computed from scanner
data.
m.C. Organizational Structure
This section presents possible organizational structures for several hypothetical data collec-tion
systems for collecting prices for food items typically bought by rural households.
Differences across systems arise because the four broadly defined operational elements in
producing the rural price index can be accomplished by various configurations of organizations.
The operational elements and organizational configurations are summarized in Table III-l
below.
111-13
>6
m.C.l. BLS-Tvoe Price Collection Systems
In each BLS-type system, BLS will conduct the basic training of price collection agents.
That will ensure each system is using price collection methods comparable to those BLS uses
for the CPI. The major difference between BLS-type alternatives is in the identity of the
organization which collects prices. One principal alternative is for BLS to do the actual price
collection. The second is for other organizations to do the price collection. Under the second
alternative, an interface will have to be established between the organization doing the price
collection and BLS. Also, for that second alternative, a number of underlying assumptions
have been made. One is that, as noted, the field representatives who collect prices will be
trained by BLS in its collection techniques. The second is that BLS will do the necessary data
review and index computation using the price data collected. The final one is that BLS and
the other organization will confer about the scheduling of the price surveys and when the data
should be delivered to BLS.
TABLE m-1
Comparison of Hypothetical Data Collection Systems
Qptrittojil Ekaem
Alternatives
Food Price
Index
Coverage Methodology
Polat-of-
Parekate
survey Traiaiag
laitiation
and Price
Collection
Dita Verification
and Index
Calculation
BLS Rural HHDs BLS Census/PC BLS BLS BLS
BLS/NASS Rural HHDs BLS Census/PC BLS NASS BLS
BLS/PC (small staff) Rural HHDs BLS Census/PC BLS PC Small
Staff
BLS
BLS/PC (large staff) Rural HHDs BLS Census/PC BLS PC Large
Staff
BLS
SCANNER Rural and
Urban HHDs*
To be
developed
Scanner
oompany
Scanner
company
Electronically
and staff
*May be limited to scannable food items only.
BLS: BONN cfLator BMW
MASS: Naloatl Afrieaknl Statical Service
PC: ntrato ooofesdor
HHDt: ilomcooka
111-14
JO
The operational tasks that have to be completed for a BLS-type price collection system can
be decomposed as follows into three primary tasks, with the third task comprising several sub-tasks
that may be performed by different agencies.
1. Conducting the POPS;
2. Training of field representatives and review of data and index computation;
3. Doing the price survey:
a. Recruiting/supervision of field representatives;
b. Sending field representatives to BLS training;
c. Provision for travel to PSU's;
d. Initiation of outlets;
e. Routine price collection.
The organization mat will perform Task 1, which is the POPS, can be Census or a private
contractor. Whichever agency conducts the POPS will be responsible for all of the tasks
associated with that survey. Task 2 will be the responsibility of BLS, while Task 3 may be
performed by BLS or some other organizatioa Below, several different organizations that can
perform Task 3 will be discussed. The organization performing Task 3 would have the
responsibility of recruiting and supervising the field personnel, of directing outlet initiation, of
(ktermining travel arrangements, and scheduling for the routine price collection.
The simplest organizational structure is the one in which BLS performs Task 3. In that case,
USDA would simply contract with BLS to perform the entire process.
A second organizational structure would have another government statistical agency collect
prices in rural areas. A candidate to do mat is the NASS. The advantage of NASS is that it
has an extensive field operation with the result that it has field representatives sear the rural
PSU's. The disadvantage is that there would be extensive framing costs.
01-15
A third organizational structure would have a private company collect prices. In that
scenario, there are two possible ways the private contractor could organize price collection.
One way would be for a small number of field representatives to go from PSU to PSU to
collect all of the price data. The second possible structure is for the private contractor to have
a large number of field representatives situated across the country in or near the PSU's in
which data are to be collected. The trade-off between the two methods is a training/travel
trade-off. With a large number of field representatives, training becomes an appreciable cost
element On the other hand, with a small number of field representatives, travel becomes the
appreciable cost element
lH.C-2, Scanner JaMlMM
If scanner methods are used for collecting prices, the organizational structure would utilize
personnel of the scanner company. Those personnel would carry out the POPS to choose the
outlets at which prices will be collected. Then scanner personnel would collect the prices using
scanner technology where possible or conventional methods for outlets without scanner
equipment
The computation of the index would most likely have to be contracted out to a private
contractor, however. That is because most scanner companies are not set up to compute price
indexes. A private contractor would have to be used because the government agency that is
most suitable to do it, BLS, would find it difficult to fit the scanner data into its normal price
index computation procedures. In addition, it should be recalled mat since an urban price
index would have to be computed, BLS would have to produce duplicative indexes, its own
CPI and an urban index using the scanner data.
OLD. Components of Coat
The components of costs will be analyzed in the sections below by considering the
operational elements described in Table III-1. To a large extent those elements determine the
size of the personnel staff required for price collection under any given organizational structure.
The organizational structure determines how the personnel are chosen and trained, and then
m-16
2?
bow they are scheduled. Sections IV-VII describe various options for organizing die staff.
Budgetary factors are summarized as well in those sections. Appendixes 3-7 provide details
on how the budgetary factors were derived.
m-17
2>*
IV. THE BLS ALTERNATIVE
IV.A. Summary Description of BLS Alternative
One way to compare the behavior of food prices paid by rural households to the behavior
of prices paid by urban households is to follow exactly BLS methods when constructing the
rural food price index. That allows the rural price index to be directly comparable to the CPI
for food at home without the methods for producing the index being a factor in the
comparison. The most straightforward way of following BLS' methods would be to contract
the work to BLS. That would be feasible because BLS is interested in doing the work, and
if BLS were to produce the rural food-at-home index, its methods, procedures and trained
personnel would be used.
The use of BLS methods is the concept behind the hypothetical "benchmark" system
referred to as the BLS alternative. Nonetheless, as good as that benchmark will be, the rural
food price index produced in this alternative can still have more variance than the CPI for food
at home. That is because the rural food price index will have fewer quotes insofar as it will
represent only about 14 percent of U.S. households as compared with the 86 percent included
in the scope of the present CPI.
The way the BLS alternative is envisioned is for the tasks associated with the rural food
price index calculation to become an adjunct to the CPI price collection program. Hence, BLS
will perform all of the major duties listed in Section III.A., with the exception of the POPS
survey requesting information about the outlets in which rural households shop. That survey
will be conducted by Census or by a private contractor.
BLS' responsibilities will include recruiting and supervising the staff necessary for
collecting the prices used in a rural food index. BLS would use its own field representatives
to collect prices for the rural index. Thus, they would be the same field representatives who
are engaged in collecting prices for the CPI. Because of that, training of those individuals will
rv-i
37
be a part of BLS' regular CPI program and will not be a factor in the cost estimates. For this
alternative, training costs are assumed to be zero.
Initiation and routine price collection for the rural index would follow BLS methods and
procedures. Its methods would also be used to handle situations where the items being priced
disappear or the outlets themselves disappear. The price collection tasks would be part of a
broader job requiring the same set of skills BLS field representatives use in collecting prices
for the CPI price collection. Moreover, the BLS field representatives who collect prices for
the rural index will use the same item specifications and forms (and/or computer programs)
currently used for the CPI price collection. Hence, when the rural prices that BLS collects
arrive at BLS' Washington office, they will have been collected in the same way and be in the
same format as standard CPI food price data.
By being in the same format as CPI prices, the prices collected in rural areas can be merged
easily into the price verification and index calculation system BLS has in place for calculating
the CPI. That will ensure that the verification rules used for the CPI food-at-home index will
also be used for the rural food price index. Then, after the data have been checked and
verified, the same computer programs BLS uses to calculate the CPI food index can be used
to calculate the final rural food price index. That method will result in the rural and urban
food price indexes being methodologically comparable. What differences remain—if any—will
be due to substantive causes, and not to the methodology of price collection and index
computation.
In contracting out the price collection to BLS, a factor mat has come up in discussions with
BLS is the upcoming revision of the CPI. The discussions have centered around whether the
staff resources available to BLS are stretched thin because of the revision. The plausible time-line
for the development of the rural index is, however, likely to be such that there would be
almost no call on BLS resources until most of the revision has been completed. The most
optimistic scenario for beginning the rural food price index assumes mat a rural POPS would
be conducted no earlier mat 1997. In mat case, the earliest a price survey could be conducted
IV-2
1*.
is 1998. But, by that time, most of the revision activity should have been completed. That
would allow BLS to allocate resources to collecting prices in rural areas. More than likely,
though, the rural POPS would not be conducted until 1998. That means that the earliest a
price survey could be conducted is 1999. By then BLS should certainly be able to allocate the
resources needed for price collection in rural areas.
IV.B. Organizational and Operational Factors
Under the BLS alternative, organizational control of producing the index would be totally
in tiie hands of BLS. BLS would undertake all of the coordination and responsibility for: 1)
the recruiting, supervision and training of field representatives; 2) the initiation of stores for
price collection and the periodic visits by field representatives to collect prices for the rural
food price index; and 3) die review of the data and the index compilation.
The feasibility of die alternative is based on die use of BLS personnel who would otherwise
be collecting prices in the A, B and C PSU's. One reason it is considered feasible to use BLS
field representatives to perform die price collection for die rural food price index is that the
BL3 field representatives are not employed full-time. Currently, BLS field representatives
work approximately 30-40 percent of a full-time schedule for routine CPI price collection.
Assuming those representatives want additional work, they should be able to devote die
approximate 2-3 additional days needed for die collection of rural food prices, especially if the
collection frequency of those prices is less than monthly. Moreover, the price collection for
the rural food price index could be scheduled so as to take advantage of the less busy parts of
the CPI price collection schedule.
An alternative would be for BLS to hire and train entirely new personnel to do the rural
food pricing. But for die few days a month required to do die price collection for die rural
index, it is probably more coat effective to use the same field representatives who collect prices
for the CPI.
IV-3
33
Another reason it is considered feasible to use BLS personnel is based on the assumption
that the PSU's selected for the rural food price index are only from the C and D non-metropolitan
PSU's described in Section III. For the C PSU's in the USDA's sample, BLS
already has personnel mere and no travel will be involved to get to those PSU's. BLS field
representatives will be in those PSU's because they will be collecting prices from the area for
use in the CPI. For those field representatives, it should be a relatively minor extension of
their duties to collect any additional prices needed for the rural index in the same area. On the
other hand, for the D PSU's in the sample, field representatives will have to travel from the
nearest A, B or C PSU to collect prices. As can be seen from the map in Section III, though,
most of the D PSU's can be easily reached from a PSU where BLS field representatives
already reside. In general, most PSU's are no more than one and one-half hours by auto from
a nearby A, B or C PSU. Only four D PSU's will require air travel to reach. It should be
noted that even if rural areas besides the C and D PSU's on the map are selected for pricing,
it should still be feasible to use BLS personnel to collect prices, although a plane trip would
probably be required.
The absence of travel costs for the C PSU's reduces the cost of collecting prices in those
PSU's. It will be reduced further because some of the same outlets at which prices will be
collected for the CPI will also be the same outlets at which prices will be collected for the rural
index. Because of that overlap, there will be no need to collect prices in those outlets for the
rural index. The prices collected for the CPI would be used for the rural index albeit with
different weights.
IV.C. Budgetary Factors
An estimate of the cost elements for a budget of a BLS-only collection system is presented
at the end of mis section. JPC has developed cost estimates for a budget of a BLS-only
collection system which are presented in mis report The budget is for varying numbers of
PSU's for different price collection frequencies. These estimates were derived from data
supplied by BLS. Nevertheless, the estimates are rough, and intended only to help USDA
IV-4
3^
evaluate the alternatives presented in this study. They cannot be treated as an accurate measure
of final costs which can only be made with a complete design specification. Accordingly, they
are not binding on BLS and should not be used for budget submissions. The methodology for
estimating the budgetary factors is described in Appendixes 4 and 5. A high and low estimate
of travel costs is presented in Appendix 5 as well. The estimates presented here do not take
into account the fact that CPI price data are available in the C PSU's.
mPOPS Cost
The budgetary factors include three estimates of the POPS cost for three different PSU
configurations. The PSU configurations range from a maximum of 28 to a minimum of 10.
The 28 PSU configuration corresponds to the total number of C and D PSU's (see Section III).
The 10 PSU configuration would produce at best an experimental non-publishable index. The
14 PSU configuration was arrived at in consultation with BLS. BLS recommended that
number as one that would be adequate to yield an index publishable at the food-at-home level
and possibly at a more disaggregated level.12
For each PSU configuration, three POPS cost estimates were completed. One was based on
the cost of a Census computer aided telephone interview (CATI). The other two were based
on CATI costs of private contractors. For the Census costs, a range is given because of
uncertainty about the final costs for a food-only POPS. For one private contractor, the POPS
cost includes oversampling to be able to separate out urban and rural households within PSU's
that include both.
As more experience is gained by the organization doing the POPS, costs for the survey may
go down somewhat Nonetheless, the ongoing POPS cost will continue to be substantial. That
it because of two factors. One is the upcoming change in BLS sampling methods; and the
other is the near invariance of sampling costs to the number of questions asked. The upcoming
"For the 14 PSU configuration BLS has suggested choosing four PSU's from the west,
four for the norm-central, four from the south and two from the east
IV-5
str
change in BLS methods leads to a requirement that all PSU's be sampled each year so as to
resample one-fourth of all POPS categories. Although the schedule for resampling those
categories has yet to be determined, undoubtedly some of the categories to be resampled will
be food categories. So the assumption in this report is that at least some food categories will
be resampled every year. Consequently, all PSU's will have to be sampled every year.
Sampling all PSU's means, though, that the same number of households will have to be inter-viewed
each year for the POPS as will be interviewed initially. Separately, there is the second
factor of the near invariance in sampling costs regardless of the number of questions asked.
Hence, the cost for interviewing a household will be almost unchanged by the reduction in the
number of POPS categories in subsequent years. Hence, with the same number of households
and the same cost per household, continuing POPS costs will be substantial.
IY.C.2 Initwtiop Costs
The initiation costs are the costs for the first visit to the outlet when price collection begins.
Two estimates of initiation costs have been provided. One is first year cost for a full initiation.
The other is an ongoing cost It assumes that 25 percent of items will be newly initiated each
year, the same procedure BLS uses for the CPI. The initiation costs are invariant to the
number of times prices are collected per year. On the other hand they are sensitive to the
number of PSU's, and they are affected by the number of outlets in a PSU and the number of
items selected per outlet The initiation costs include travel costs to the PSU for initiation and
the longer time spent on initiation. The note on the table shows travel costs separately and
indicates what is included in travel costs. The estimate of travel costs takes advantage of the
fact mat reaching the C PSU's requires no travel.
IY.fr.1. Rfftfinc Price, Citation
Routine price collection costs are the actual costs of collecting the prices from the outlets
sampled from the POPS. The routine price collection costs include travel costs to the PSU.
For each configuration of PSU's, three estimates ofroutine price collection costs are presented,
depending on the frequency of price collection. BLS has indicated mat it could easily fit either
IV-6
3^
moodily or bimonthly frequencies into its production schedule. Fitting in a quarterly frequency
would be much more difficult
rv,r,4. Training and Management Bail
For the BLS alternative, training costs have not been estimated because they are expected
to be minimal. That is because most of the training will be a part of BLS' standard CPI
training program for its field representatives. Nonetheless, some training costs may be
encountered if there are an insufficient number of field representatives who want additional
work. One other cost not included is the management cost that BLS will need for including
the rural food-at-home index in its operations.
IV.C.5. The Effect of the Number of Quotes on Costs
For the 14 PSU configuration, two cost estimates have been presented, both for the BLS
alternative and for the other alternatives as well. One estimate assumes the standard number
of quotes will be collected from each outlet while the other configuration assumes twice the
standard number of quotes. The reason for estimating the cost for twice the number of quotes
was based partly on recent work by BLS mat showed that the variance of the CPI food-at-home
index can be reduced by increasing the number of quotes regardless of how mat number
is increased. It was also based partly on the observation mat the POPS cost is the most
substantial cost element ofprice collection. Thus, by doubling the quotes per PSU and reducing
the number of PSU's, substantial cost savings can be achieved with the expectation mat the
variance will be unaffected.
IV-7
37
TABLE IV-1
COOT ESTIMATES FOR BLS-ONLY PRICE COLLECTION SYSTEM
Census
Private Contractor
Private Contractor
Number of Quotes (est)
First Yc
Additional Year
Monthly
Bimonthly
Quarterly
BLS(e«L)
Verification
computation
NUMBER OF PSU'i
28 14 10
(thousand of dollars)
POPS COST
706-470 353-235 252-168
314 156 112
439 220 157
5900 2950 5900 2100
A B
INITIATION
50 25 42 18
27 14 22 10
ANNUAL COST OF
ROUTINE PRICE COLLECTION
244 120 140 88
122 60 70 44
82 40 47 30
VERIFICATION AND
COMPUTATION
109
6
109
6
109
6
109
6
owe in«•*§*• ami COM dwttiftm
ADDENDUM:
ANNUAL COST OP 1IAVIL
a 14 IS
70 35 26
35 11 13
23 12 9
The trwtl oorts ire it
md rouuiG price collection
•fPSlTi:
Travd figures include auto, plane, per diem A hotel, and labor costs (wages +
or l»eadouariari overheads are included in travel
)for
IV-I
sr
V. THE BLS/NASS AND ERS ALTERNATIVE
V.A. Summary Description of the BLS/NASS and ERS Alternative
NASS conducts surveys and collects data related to agricultural and rural statistics. It
maintains forty-five field offices distributed in every state except in New England, where it has
one field office in New Hampshire covering the six New England states. Within each field
office, NASS maintaias several supervisors .w are responsible for various areas of work.
Each supervisor coordinates a set of field representatives (which NASS calls interviewers).
Those field representatives can be assigned to conduct routine or special surveys in the state.
The NASS alternative would entail having NASS field representatives collect prices from
the rural PSU's on a regular basis (for example, bimonthly). According to NASS, it would
plan to assign one representative to a PSU to collect price data. BLS or some other orga-nization
would then provide it with a listing of the outlets from which prices would be
collected for each of the PSU's.
To ensure that the rural food price index resulting from the prices collected by NASS
representatives will be comparable to the CPI food-at-home index, BLS methods and proce-dures
would have to be followed. Hence, representatives assigned by NASS would be
provided training by BLS comparable to mat provided by BLS to its own field representatives.
(To be determined is whether the NASS supervisors need to go for the full period of training.)
BLS would also supply NASS with the specification sheets or checklists containing a descrip-tion
of the food hems and with other collection materials as needed. It would also provide
NASS with all updates to reference materials, changes in specifications and procedures and
overall pricing schedules. That will keep NASS' price collection procedures up-to-date and
synchronized with the price collection BLS is doing for the CPI food-at-home index.
Under NASS' standard procedures, die field office supervisor initially accompanies the field
representative as the field work gets under way. Subsequently, the supervisor would do mat
V-l
■I
3?
at occasional intervals. The supervisor would coordinate the field representative's assignments.
The supervisor would also act as a backup for the field representative.
The initiation of outlets and the routine collection of prices would be built into the regular
survey assignments of field staff. Because of the location of field offices in each state, the col-lection
for the PSU's would be carried out through the respective field offices. The field office
would assign the field staff who would cover the PSU. It would men collect the completed
checklists and send them to a central location (one field office) for data entry. The electronic
files of the price data would then be sent to BLS for verification, checking, and analysis. If
BLS wanted to build in some initial data edit into mis process, NASS could accommodate that
in the data entry. Alternatively, NASS could forward the original paper checklists directly to
BLS.
Most of the process outlined is likely to be implemented and coordinated through BLS and
NASS. The major exception is the POPS to identify the outlets. That would be carried out
by Census or a private organization.
Once the POPS was completed, NASS field representatives would initiate the prices and
items at the outlets. It would then collect the prices at regular intervals using BLS procedures.
By using the same checklists (and/or computer programs) that BLS uses, NASS field
representatives will collect prices the same as do BLS field representatives. Thus, when NASS
sends its price data to BLS' Washington office, the data will be in the same form in which all
CPI data arrive. Consequently, the data will be in a format that can be easily merged into
BLS* CPI system. Furthermore, they will undergo the same set of verification procedures as
other data used in the CPI. BLS (and/or NASS) would then have to verify the data and to
construct the food price index for rural areas. After BLS has verified the data, they men will
be put through the same calculation procedures as data used for the food-at-home CPI, and will
then compute the food price index for rural households.
V-2
</0
The Economic Research Service (ERS) was queried about whether it currently does any
construction of price indexes from raw price data. The response was that it does not Instead,
it uses the BLS constructed indexes. The procedure is that BLS provides the detailed food
item price indexes to ERS, and ERS uses that information both in the construction of reports
it issues and in the projections of future food price changes.I3
V.B. Organizational and Operational Factors
The basic information on NASS and NASS' operations was provided by Robin Roark. He
works in the Commodity Surveys Section and with the national survey administration unit His
office is responsible for coordinating the logistics of NASS surveys and for coordinating all
outside requests.
According to information provided by NASS, it maintains field staffs, organized in state
field offices to conduct surveys in agricultural and rural areas. Currently its operations cover
all of the counties of the 28 non-metropolitan C and D PSU's. Its field staffs are relatively
stable and maintain relatively constant workloads. The current staffs are likely to have the
available time to add survey work of the type proposed.
For special or non-recurring surveys, NASS may add temporary staff either in the state
office or in the field. The ongoing nature ofthe rural price collection would, however, be built
into its regular staff operations and would not be viewed as "temporary." NASS anticipates
work on the order of 2-4 days per month, once the price collection work gets underway.
One potential concern was the effect on NASS if, as is anticipated, NASS assumes
responsibility for the Census of Agriculture. The Census of Agriculture, which is run once
every five years, is next scheduled for Spring 1998. The NASS staff person who was
"Annette Clausen is the individual in the Food Markets Branch of the Food and Consumer
Economics Division who specifically uses the price index information from BLS. She
assembles the food price numbers used within ERS and uses the BLS data for analysis.
V-3
Hi
contacted (Robin Roark) thought that the Census of Agriculture work would not present a
problem for the rural food price collection work. That is because the Agricultural Census is
primarily a mail survey and would be handled by adding temporary staff in the state offices.
Another issue was NASS' experience with computer assisted personal interview devices.
The NASS field enumerators have had some experience with such devices. If BLS changes
to computer assisted data collection procedures, as it is expected to do over the next several
years, NASS staff have the experience to accommodate that change.
The need for NASS to maintain comparability of data collection procedures would require
relatively close coordination between it and BLS. For example, it is conceivable that there
would be some overlap between the stores for the urban and rural indexes if urban and rural
households both shopped at the same places. To reduce the burden on those stores, that sort
of overlap would need to be addressed by coordinating BLS and private contractor price
collection.
V.C. Budgetary Factors
NASS worked with the information that was provided to it about the number of outlets and
PSU's and quotes. The budgetary factors it provided can be found in Appendix 6. The cost
estimates NASS provided are for price collection in 28 PSU's on a bimonthly basis.
Additional cost estimates were then made from those estimates for other components to make
cost estimates for configurations resembling the ones presented for the BLS-only alternative.
How that was done also can be found in Appendix 6. The cost estimates are intended to be
comparable to those of die BLS alternative.
The POPS costs for NASS are the same as for BLS insofar as they do not depend on
NASS; and, as with the BLS-only system, the POPS costs will be ongoing. As for the training
and initiation costs, those have been presented jointly because NASS did not supply separate
figures for mem. Also, the training costs do not include BLS charges for training. Insufficient
information was available to derive BLS charges for training. The element labelled "USDA
V-4
%~
travel costs" covers charges to USDA for travel and per diem. It is based on the assumption
that all trainees come to Washington for training. It should also be noted that NASS only
assumed eighty hours of training and that may be too low. (See especially Appendix 4.)
TABLE V-l
COST ESTIMATES FOR BLS/NASS PRICE COLLECTION SYSTEM
Census
Private Contractor
Private Contractor
Number of quotes (est.)
First Year (est)
Additional Year
USDA travel costs for training
First Year
Additional Year
Monthly
Bimonthly
Quarterly
BLS(est)
Verification
Computation
Noto: NASS'Mti WK of
"USDA
did DM
com for
NUMBER OF PSU'J
21 14
(thousand of dollars)
POPS COST
10
706-470 353-235 252-168
314 157 112
439 220 157
5900 2900 5900 2100
A B
TRAINING A INITIATION
32 16 27 11
It 9 14 6
112 56 56 40
22 11 11 8
ANNUAL COST OF
ROUTINE PRICE COLLECTION
150 75 88 54
75 38 44 27
50 25 35 18
VERIFICATION AND COMPUTATION
109
6
109
6
109
6
109
6
10 Waal***. IDote cotfi «e entered in toe Ik*
Abo toe ttaainf com do not include my BLS chmjes for
V-5
«
VL THE COMBINATION BLS/PRIVATE CONTRACTOR ALTERNATIVE
VI.A. Summary Description of the BLS/Prrvate Contractor Alternative
The hypothetical data collection system referred to as the combination BLS/private
contractor alternative is based on the concept that the tasks necessary for calculating a rural
food price index would be divided between BLS and a private organization. The personnel
training, price verification and index calculation tasks of the rural food price index calculation
would be contracted to BLS. The recruiting of field representatives and the price collection
tasks (including store initiation) would be contracted to a private organization.
The way this alternative is envisioned, USDA would contract with a private organization
to recruit field representatives who would do the price collection for the rural index. The
private organization alternative could follow one oftwo main models: 1) a large geographically
dispersed staff model; or 2) a small dedicated staff model.
Under the large-staff model, price collection could take place nearly simultaneously in all
of the areas where prices are being collected for the rural food price index or it could be
staggered. Under the small-staff model, price collection for the different areas would have to
be staggered to allow a relatively small group of field representatives to travel from area to
area to collect all the prices needed
To ensure that the rural food price index resulting from either of those two price collection
processes is comparable to the CPI food-at-home index, BLS methods and procedures would
have to be followed. BLS would provide training on food price collection to the private
organization's field representatives mat is comparable to mat provided to its own field
representatives. BLS would also provide the private organization with all updates to reference
materials, changes in food item specifications and procedures, and overall pricing schedules.
That would keep the private organization's price collection procedures up-to-date and synchro-nized
with the price collection BLS is doing for the CPI food-at-home index.
VI-1
r<
The private organization would also need to create a system to convert the price information
collected by their field representatives to a form that could be entered easily into BLS' data
verification system for the CPI. Once BLS had the data, it would use the same set of
verification procedures on the prices as it uses for the CPI food-at-home data and men calculate
the rural food price index.
VLB. Organizational aad Operational Factors
VTIfl frlwinliny
Under mis alternative, BLS would have the responsibility for the overall pricing schedule
in mat it would determine when price collection needed to take place in order to meet USDA's
timeframe for index publication. The private contractor would have the responsibility for
scheduling all travel, etc. to see mat the price data can be forwarded to BLS on schedule.
VLB.2. Recnntma
The private organization would have all the responsibility for recruiting an appropriate
number of field representatives to conduct the price collection on the schedule detennined by
BLS. The number of field representatives needed would depend on the number of areas in
which price collection would take place, the frequency of the price collection and the structure
of the organization itself.
The large-staff model would be good for an organization whose employees are
geographically dispersed and are working on several different types of tasks for the private
organization. Employees that are in geographical areas near where the rural price collection
will occur would be trained to follow BLS methods and procedures. Then, when price
collection is required, each would be sent to one of the areas to collect prices. Under that
scenario, price collection in all of the rural areas could be undertaken virtually simultaneously.
The amiU-ataff option would be more appropriate for a private organization whose
employees are very localized and who do not reside near the rural price collection areas. The
private organization would assign a small group of employees to travel from one price
VI-2
!&
collection area to another to collect die prices needed for die rural food price index. Unlike
the large-staff alternative, which would make price collection a periodic task in a broader range
of tasks mat a large group of employees does, the small-staff would make price collection a
more nearly full-time dedicated task for a small group of employees.
YTB1 Training
Under either die small or large-staff model, the private organization would have the
responsibility of ensuring that its field representatives were fully versed in BLS methods and
procedures and had completed all of the required BLS training. Hence, die training of the
private organization's field representatives would have to be done by BLS and would have to
be the same training that BLS provides its field representatives. That will ensure comparability
of price collection in rural areas with collection for food-at-home prices used in the CPI. (As
with die NASS alternatives, insufficient information was available to derive BLS charges for
training.)
Logistically there may be one problem with die large-staff model and BLS training. Over
die next few years, BLS will be retraining all of its staff to handle changes in procedures that
will result from die revision to die CPI and a change to computer-assisted data collection. That
increases die burden on BLS' training staff significantly. Consequently, they may not have die
resources to train a large number of private organization field representatives in addition to
retraining their own field representatives. The extent to which that affects rural price collection
will depend on when USDA would like rural price collection to begin.
Store initiation, price collection and conversion ofprice information to a format compatible
with BLS' data verification systems would be die responsibility of die private organization.
Data verification and index calculation would be the responsibility of BLS.
VI-3
rt
MAI I— Concerning Operational B—1mm
The most complicated operational procedures would be those associated with price
collection. However, depending on whether the contractor uses a small-staff model or a large-staff
model, those procedures could have an effect on BLS procedures as well.
The large-staff model is the one that would be most like BLS* own operating methods for
the CPI. Because mere are many geographically dispersed field representatives under this
model, price collection would take place in all the rural pricing areas nearly simultaneously.
That would allow all of the data for the index to be forwarded to BLS at one time and the data
verification and index calculation could be done sequentially.
The small-staff model would require slightly different operational procedures. Under mat
option, the price data could not be collected simultaneously from all of the rural pricing areas.
The frequency of the index calculation and the number of areas in which prices for the rural
index will be collected will determine the number of field representatives needed and how
many areas each will be visiting. However, it is possible that several batches of data would
be sent to BLS for verification (a task mat needs to be done as close to price collection as
possible) before the index calculation is done. BLS will be able to handle data sent in batches
without a problem if the frequency is monthly or bimonthly. There is some uncertainty about
mat if the frequency is quarterly.
Regardless of which model is followed, the need to maintain the comparability of data
collection procedures would require relatively close coordination between BLS and the private
organization. For example, it is conceivable that there would be some overlap between the
stores for the urban and rural indexes if urban and rural households both shopped at the same
places. To reduce the burden on those stores, mat sort of overlap would need to be addressed
by coordinating BLS and private contractor price collection.
VM
VLC. Budgetary Factors
Table VI-1 presents die budgetary factors for the private contractor alternative for both the
large-staff and small-staff variants of that alternative. Some of the costs in the table were not
supplied by the private organization. They were based on estimates developed for this study.
See Appendixes 4 and 7 for how various cost elements were derived. The training costs in
Table VI-1 do not include BLS charges for training. The costs in the table are intended to be
comparable to the costs of the BLS alternative.
In comparing the two models, what becomes apparent is the expense of the small-staff
model. That arises from the expense of air travel. Offsetting that expense is the extra
experience that will be accumulated by the small number of individuals who will be doing the
price collection in comparison with the staff of the large-staff model where each field
representative will be collecting prices only a few days a month at best The continuing
experience of the field representatives of the small-staff model as well as the constant use of
the same individuals to collect the prices, ensures mat the same procedures will be followed
consistently during each price collection. While those characteristics do not guarantee that
results from the small-staff model will precisely match those of the BLS, they should,
nevertheless, be expected to produce results that are more nearly comparable.
VI-5
<ff~~
TABLE VM
COST ESTIMATES FOR BLS/PUVATE CONTRACTOR
PRICE COLLECTION SYSTEM
NUMBER OF PSU's
14 10
(thousand of dollars)
POPS COST
Zaaus 706-470 333-238 252-168
•rrvate Contractor 314 179 112
•rivate Contractor 439 220 1S7
tumber of Quotes 5900 2950 5900 2100
Larne-ataff Model
TRAINING
first Year 164 82 39
Additional Year 32 16 12
■J Coat for Training
Fast Year -
Additional Yew -
INITIATION
tint Year 56 28 47 20
Additional Year 30 17 23 11
ftMlhaff Mftdri
TRAINING A INITIATION
tont Year 144 101 170 90
Additional Year 70 37 89 49
ANNUAL COST OP
ROUTINE PRICE COLLECTION
MgajUx
192 168 120
368 288
III
Modal 168 84 98 60
184 213 144
112 36 63 40
240 132 143 107
VERIFICATION AND COMPUTATION
109 55 109 39
",h^T<*T,IJ-r'Tiw<
VI-6
/o
vn. THE SCANNER ALTERNATIVE
VILA. Summary Description of the Scanner Alternative
The hypothetical data collection system referred to as the scanner alternative is based on
the concept mat all tasks except index calculation would be the responsibility of a company
(hereinafter referred to as the scanner company) collecting retail price information from scanner
terminals in retail outlets. The personnel training and price verification would be carried out
by the scanner company. The recruiting of field representatives and the price collection tasks
(including store initiation) would be the responsibility of the scanner company. Additionally,
the scanner company would have to run its own POPS from its household data base. The
POPS would move the scanner company towards using its scanning data base in a manner
consistent with scientific sampling. Finally, an organization other than BLS would do the price
index computation.
The way this alternative is envisioned, USDA would contract with a scanner company that
collects sales and price information using scanners. Because such companies do extensive
market research, they generally have two data bases. One contains information about
household buying patterns. The other has information from retail outlets, including price data,
that comes from the scanning systems. From the household data base, the scanner company
would conduct a POPS. The POPS would lead to the selection of outlets at which prices will
be collected. Some of the outlets selected will be in the scanner company's data base or in
samples in place, while other outlets will not be. For outlets in its data base, price data will
come electronically right off the data tapes supplied by the outlets to the scanner company.
For the outlets not in the data base, but which have scanning equipment and which give the
scanner company permission to read their data tapes, the data could also be transferred
electronically. For other outlets, the scanner company will have to use field representatives to
collect prices. It could do mat by using its existing staff or by expanding its staff of field
representatives. Some cost savings on price collection for outlets not in the data base could
be effected by having the scanner company substitute comparable outlets in its data base for
outlets not in the data base. Comparable outlets would be those of the same size in terms of
vn-i
s*
dollar volume and selling to die same individuals having comparable characteristics. That may
prove to be difficult insofar as only 10 percent of supermarkets are in the scanner company's
database.
Two factors stand out about the scanner alternative as compared with other alternatives.
The first is mat the scanner company's price collection methods will be very different from
BLS' collection methods. The second is mat the scanner company's household data base is
configured very differently from Census' data (which is by PSU). One consequence is that
there is no assurance mat a rural food price index based on prices collected using scanner
techniques can be compared to the CPI for food at home. Thus, to be able to answer the
question posed in Section II, about how a rural food price index moves with regard to an urban
price index for food, an urban food price index would also have to be computed from prices
collected using scanner techniques.
VTLB. Organizational and Operational Factors
VHB.1. Scheduling Rimming. Training Store Initiation Price Collection. Data Verification
«iri CMlrnlmtinn
Under this alternative, the scanner company would have the responsibility for deterrnining
and scheduling price collection. It would determine when price collection needs to take place
in order to meet USDA's timeframe for index publication. It would also have the
responsibility for scheduling all travel and data collection. Finally, it would have all the
responsibility for recruiting an appropriate number of field representatives.
For outlets without scanning equipment, there would need to be an initiation comparable to
the kind used when BLS-type procedures are followed. Similarly, for those outlets, the price
collection and data verification procedures used by the scanner company would probably have
to follow BLS procedures (although the scanner company could choose to follow its own
unique procedures in those cases). For outlets in the scanner company's data base, or outlets
not in the data base but with sramring equipment and which agree to share electronic data with
vn-2
L. S?
the scanner company, initiation and price collection could be carried out electronically.
Although these two approaches are somewhat different, they emulate the BLS spirit of using
the best data available when initiating outlets.
Not all hems are scanned in a supermarket In particular, items whose weight and size are
not uniform, such as fresh meats and produce, are not in the scanner company's data base. For
those hems, the scanner company sends field representatives into the outlet to collect prices.
Nonetheless, because initiation costs are significant, it may be advisable to restrict the prices
included in the food price index to food prices mat can be collected by «*.«miing systems.
Those currently comprise about 60 percent of the CPI food price index.
Index calculation would have to be carried out by an organization other than BLS. That is
because the price collection procedures of the scanner company do not easily fit into BLS'
current procedures. Also, BLS cannot be expected to produce two urban price indexes for
food. What is important about the index calculation is mat the organization that does the
calculation follow exactly the same procedures for the urban index as for the rural index. The
organization that computed the indexes could be the scanner company, or a subcontractor to
the scanner company, or h could be a separate contractor with USDA.
VILC. Budgetary Factors
For a number of reasons a budget was not estimated for the scanner alternative. One was
mat USDA/FCS would have to get involved in estimating an urban food price index as well
as a rural food price index An urban food price index is already produced by BLS, though,
and having two competing government indexes does not seem feasible. There is also the
expense of setting up with USDA a whole infrastructure to monitor the index production and
mat also does not seem feasible.
VH-3
VOL PROS AND CONS OF THE ALTERNATIVES
VELA. Intradnctkm
This report has jented four alternatives for estimating a CPI-type price index for food at
home purchased by rural households. One of the alternatives has two variants. To summarize
them, they are as follows:
1. BLS-only system:
USDA contracts with BLS to administer every aspect of the price index system; Census
or private organization conducts POPS in rural areas selected for sample.
2. BLS/NASS:
NASS administers the price collection with a large staff while BLS trains field representa-tives,
provides hem specifications, verifies the data and computes the index; Census or
private company does POPS in rural areas selected for sample.
3.a. BLS/PC: «m«ll.«fflr grij^rUaTPA contracts with a private company to administer the
price collection with a small staff while BLS trains the field representatives, provides hem
specifications, verifies the data and computes the index; Census or private company does
POPS in rural areas selected for sample.
3.b. BLS/PC: large-staff variant-USDA contracts with a private company to administer the
price collection with a large staff while BLS trains the field representatives, provides hem
specifications, verifies the data and computes die index; Census or private company does
POPS in rural areas selected for sample.
Scanner system:
USDA contracts *vith a private company which will use its own household data base to
conduct a POPS, and will then adminiUrr the price collection system in outlets with
scanner equipment where possible, or conduct on she price collection where not; the
scanner company will also verify the data; h will compute the price index, or h or USDA
will contract with another company for the index coiiyia'ation; the index would have to
be compiled for urban and rural
The intent of each of the systems is to answer the main question posed for mis feasibility
retort, namely:
vm-i
JO
How does the movement in prices offood items typically bought by rural households
compare with the movement in prices of food items typically bought by urban households?
In answering that question, the intent is to minimize the effect of operational differences in
price collection and index computation between urban and rural food price indexes. The
purpose of minimizing the effect of those differences is to be able to determine whether in fact
rural prices behave differently from urban prices without the finding being compromised by
differences in collection and computational procedures across the organizations involved.
Vm.B. The BLS-type Alternatives
mm 0—I Advantages and Disadvantages of AH BLfcfia Alternatives
The goal for the BLS-type systems is to produce an index for food at home in rural areas
that is constructed in the same way as is the CPI for food at home. All the BLS-type
alternatives should yield an index comparable to the CPI. The advantages of using a BLS-only
system is that it wili yield an estimate for a rural food price index that methodologically is
most comparable to the CPI for food at home. Other BLS-type systems will emulate the BLS
system for the CPI. Consequently, they will differ in some measure from the BLS system and
will not receive the imprimatur of BLS—as signified by BLS publication of the index—even if
BLS does the training and final index computation. Nonetheless, the other systems need to be
considered to illuminate the trade-offs between comparability and cost
For USDA, all BLS alternatives have the advantage of minimizing the management burden
in USDA except for general oversight, even in alternatives involving price collection by an
organization other than BLS. For the BLS alternative, the need for only minimal USDA
management should be apparent For the other alternatives, mat conclusion stems from
discussions with each of the organizations mat would potentially collect prices. They indicated
that they would conform to BLS schedules and standards, and would provide the data to BLS
in a way mat BLS could most easily use the data to compute the index. BLS, for its own
scheduling purposes, would work with the other organization to determine the various
vra-2
at
parameters concerning price collection and publication. There is one caveat winch must be
kept in mind when using private contractors. It is that BLS' willingness to compute a price
index using outside data will depend on agreement on language concerning the labelling of the
index.
By having BLS decide scheduling and other issues, USDA is relieved of most of the
managerial burden for producing the index. Some managerial decisions still remain for USDA.
Those include such things as the number of PSU's in the sample, the frequency of collection
and the number of quotes per outlet Once those decisions have been made, though, there are
really no other managerial decisions outstanding.
For all the BLS-type alternatives described here for monitoring prices in rural areas, two
issues have emerged that need to be considered. The first stems from operational methods for
the price collection. All the variants outlined here, including the BLS alternative itself, are
based on collecting prices during only one week of each month. The reason for that is to keep
down the cost for a rural index. By collecting prices within a one-week timeframe, field
representatives will have to visit the PSU only once. That coutiaals with BLS' current
procedure of spreading price collection within a PSU (or PSU half-sample) over a three-week
timeframe. It does mat to allow the CPI to pick up anomalous mtra-monthry price changes
Because of its one-week timeframe, the rural food price index will be unable to do that Any
differences, though, in the two indexes because of anomalous intra-monthly price changes
should wash out and the CPI and rural price index should then be comparable; of course,
monthly comparisons of the indexes will probably be affected by intra-monthly price changes
But mat points up the recommendation made elsewhere in this report namely that the
comparisons between the CPI for food at home and the rural price index should be baaed on
index changes mat are over a longer time period, such as a year, rather than on changes that
are for one month at a tune.
The second issue that has emerged for all BLS-type alternatives has to do with the length
of time under which comparison will have to be made between the CPI and rural food price
vra-3
index. Probably at least four years will have to elapse before strict comparisons can be made
between urban and rural price movements. One reason is that BLS plans to resample product
categories on a four-year rotation cycle. That procedure will probably apply to food. A four-year
cycle implies mat on average new items and some new outlets will be fully reflected in
the CPI only after a four-year delay. In comparison, the rural food price index will be more
up-to-date at its inception and for several years thereafter.
Another important reason for basing analysis of movement between the two indexes on
several years of data is that any conclusions to be drawn about differences in food price
movements will reflect differences in items purchased, differences in expenditures on those
items and differences in the movement in prices of the items selected for pricing. To test
whether those differences are significant will require an adequate timespan. Of course, if
differences do emerge, further analysis will be needed to determine their sources.
VIII.B.l.a. The BLS Alternative
The primary advantage of the BLS alternative has already been described, namely mat if
BLS constructs the rural price index, mat index will be methodologically comparable to the
CPI. Part of the advantage stems from the use of BLS personnel who (obviously) have been
trained in BLS procedures and who are well-versed in price collection. Another part of the
advantage stems from having personnel who are already in the field either in or near most of
the PSU's from which prices will be collected. That serves to reduce the travel cost associated
with this alternative. Finally, as Table VIII-1 indicates, the BLS alternative is competitive with
the costs of the other alternatives.
VIII.B.i.b The NASS Alternative
The advantages of the NASS alternative are based on NASS' having personnel in the field
who are dispersed throughout the U.S. and who have experience in conducting surveys.
Because those personnel have experience in surveys, they may not need extensive training, and
because they are dispersed, travel costs will not be a large element in NASS' costs. Indeed,
NASS is quite competitive with BLS in terms of costs. (See Table VIII-1.)
Vffl-4
&>
There are several disadvantages ofthe NASS alternative. One :s the expense of training the
personnel in BLS practices. (Even those who have experience in surveying will need training
in BLS practices.) Another disadvantage of the NASS variant is that its personnel will be
engaged only infrequently in collecting prices (at best for only a few days at a time, with the
specific frequency depending on the frequency of price collection). Hence, the staff will not
be as adept at price collection as typical BLS personnel who collect prices on an ongoing basis.
Another disadvantage of the NASS alternative is systemic to any large-staff model. It is a
potential problem scheduling training by BLS. The problem stems from two factors. One is
that the upcoming revision of the CPI may not allow BLS to allocate resources to training
outside personnel. As indicated in Section IV, though, that may not prove to be a significant
problem based on the timeframe outlined there. The second problem is BLS' intention to
switch from using paper forms to hand-held computers for data entry at the outlets. That will
apparently require BLS to do a fair amount of staff-retraining. Again, the consequence is that
BLS may be unable to provide training until it has retrained its own staff. Further discussion
with BLS may clarify the issue.
VIII.B.1.C The PC/Large-Staff Alternative
The private contractor alternative utilizing a large staff has some of the advantages of die
NASS alternative. It is that it will have personnel in the field dispersed throughout the U.S.
Because they are dispersed, travel costs will not be a large element in the variant of mis
alternative.
On the other hand, mere is some uncertainty about the personnel who would be used for
collecting prices—whether they would resemble NASS personnel who have experience in
conducting surveys or whether they would be individuals hired for the purpose of collecting
prices. If they do not have experience, then they will need more extensive training than NASS
personnel.
vra-5
-T7
For this variant of the private contractor alternative, the disadvantage of the NASS
alternative applies as well. Those include the low frequency with which field personnel will
be engaged in collecting prices, and the potential problem about training by BLS.
A final disadvantage is one of cost There seems to be no particular cost advantage
associated with the large-staff variant of the private contractor model. (See Table VIII-1
below.)
VIII.B.l.d The PC/Small-Staff Alternative
The private contractor alternative utilizing the small-staff variant is based on having a small
staff whose members travel on a regular basis from PSU to PSU to collect prices. The primary
advantage of this alternative is the frequency with which the staff will be collecting prices.
That will allow the staff to acquire a fair amount of expertise in collecting prices. Another
advantage is that training costs will be very modest if only because, according to information
garnered for mis report, the staff will be recruited from the Washington, D.C. area and because
the size of the staff will be quite modest The disadvantage of the private contractor alterna-tive,
small-staff variant is one of cost The cost of travelling from PSU to PSU is apparently
high enough to make this alternative the most expensive of all the BLS alternatives examined.
(See Table VIII-1 below.)
Vm.B.2. Advantages and Disadvantages of Various POPS Alternatives
The cost of the POPS has turned out to be a significant factor in any of the BLS-type
alternatives examined for mis feasibility study. That will be true whether the POPS is
conducted by Census or by a private organization, but will be all the more so if conducted by
Census. There are a number of ways of minimizing the cost of the POPS. The purpose of the
discussion here is to outline some of those ways, as well as the advantages and disadvantages
of using Census versus a private organization.
One of the ways to minimize the POPS cost is to conduct the survey less frequently than
is standard practice for the CPI. Currently, BLS plans to conduct a POPS for the CPI four
vm-6
times a year. Based on consultation with BLS, it appears that a twice a year POPS will be
sufficient for the rural price index and will not risk comparability with the CPI for food at
home. Another way of reducing the POPS cost is to reduce the number of PSU's in which
prices will be collected. Again, based on consultation with BLS, it appears necessary to collect
prices in 14 PSU's rather than in 28 PSU's. To offset the effect on the overall number of
quotes collected because of the reduction in PSU's, BLS suggested doubling the number of
quotes to be collected from the outlets. That yields the same number of quotes for 14 PSU's
as would be the case for 28 PSU's but with a significant cost savings.
Another way of reducing the cost of the POPS is to have a private organization conduct it
instead of Census. Two organizations were contacted and asked how much they would charge
for a POPS. Their cost estimates were lower than Census'. Of course, it may be that Census'
costs will be reduced when it considers a formally presented, detailed proposal, but at the
moment that is speculative. (Other considerations such as ability to differentiate rural from
urban households do not come into play here because Census appears to have the ability to do
that as do the private organizations.)
Assuming, men, that Census' costs remain unchanged, cost considerations alone would
dictate using a private organization to conduct the POPS. Offsetting that consideration are two
factors. One is mat by using an organization other man Census, comparisons between the rural
food price index and the CPI for food at home could become less precise than they otherwise
would be. Whether that will be the case would depend on how well the private organizations
can duplicate Census' efforts. Any degradation of or, for that matter even, improvement on
Census' efforts might introduce an effect mat reduces the comparability of the two indexes.
The second factor that needs to be considered is the position that BLS will take if Census
does not do the POPS. While BLS at mis time is unsure of its position on the matter, it
outlined two possibilities. One would be not to publish the rural food price index in any of
its standard publications. (USDA would of course be free to publish the index, but language
on the index published would have to be worked out with BLS.) The second position is mat
Vffl-7
BLS would publish the index, but would footnote it to indicate that the index is not based on
a POPS conducted by Census.
Obviously, cost considerations are important USDA, though, has to weigh the tradeoffs
between having Census do the POPS as opposed to having a private organization do it
Vm.B.3 Comparative Cost of the Various Alternatives
For the several BLS-type alternatives examined, costs for the initiation period and for
subsequent years were computed. They appear in Table VIII-1. The basis for the cost
comparison is the 14-PSU configuration BLS thought would be adequate for a rural food price
index, with twice the normal quotes collected per outlet and with the POPS to be conducted
twice a year instead of the normal four times a year the POPS is conducted for the CPI.
Finally, the estimate is for bimonthly collection rather man monthly collection. The period for
the first initiation and POPS will probably have to be two years. The verification costs should
be assigned only to the second of those two years.
In comparing the costs, it appears that both BLS and NASS are the low end of the cost
rutimates, while the private contractors are the high end. The primary advantage of the BLS
alternative derives from the absence of training costs.
vra-8
6*6
TABLE Vffl-1
SUMMARY OF COST ESTIMATES OF BLS-TYFE ALTERNATIVES
14-PSU CoaflfaratioB, 5900 Qootes aid Biaoathty Collection
BLS
Initinfon Pmrt
FC-NASS
Large
FC-Saall
FC-BLS
NASS Saall
FC-Large
»OPS (Census) 294 294 294 294 294 294 294 294
naming - 56 82 — - 11 16 —
framing ft Initiation — - - 170 - - — 89
oitiatton 42 27 47 - 22 14 25 —
Routine Price Collection - - - — 70 44 98 215
Verification ft
109 109 109 109 115 115 115 115
IOTAL 445 486 532 573 501 478 548 713
VHLC. A Scanner-type System
BLS-type alternatives have been conr.Hered because they pennit food prices to be collected
and compiled in a way that will make it possible to produce a rural food price index
constructed the same as the CPI for food at home. The use of scanner methodology requires
a different approach which makes it necessary to collect prices in bom urban and rural areas
so as to produce urban and rural food price indexes that will be comparable. The reason why
the CPI cannot be used as a comparator is mat collection methods based on scanner techniques
can differ greatly from those based on BLS techniques. If the behavior of the rural index
should be found to differ from that of the CPI, it will not be known whether it was because
of the different collection methods or because of real differences in the behavior of urban and
rural food prices.
The need to make two indexes if a scanner company is used appears to be one disadvantage
to using the scanner alternative. A second disadvantage is the managerial burden on FCS (or
USDA). In contrast, for the other alternatives, the managerial burden is slight That is because
BLS and the organization collecting prices would coordinate and schedule the production of
the index. For the scanner alternative, though, FCS (or USDA) would have to produce the
vra-9
<r1
schedule and probably oversee coordination between the scanner company and the company
producing the index.
The advantage of scanner techniques really lies in the future. As scanners become more
universal and as the number of scannable items increases, it will be possible to do more data
collection electronically. That should reduce the need for a staff of field representatives and
their attendant costs, and it may reduce the need for data verification. The net effect could be
a considerably reduced cost for data collection and verification at some point in the future.
vra-io
^
0) ro
APPENDIX 1 - MATRIX
P) <o « «
D. (boa MM.)
Irving ■ Waft* MRU ■MM^Hl
bOMM Mod
iWriM aM7
Do (too* tt.) I
(A DoM-Do.,
a.
far feed
iaaWCn
how foot
m nnl amaT
of CM for food
•crvW ircar? (CowM aao W
M faoa <■p tnnianlli)
Do too*
MM
• faacfcoa afarka <■
k, —(—-o-M I (MM.) MM <• 0A
I (MM <nM M faoa
(••*•> Total US . ofMmi c ml TaM US. o. -W. aai ml Total US. or arka aai rani US TaMi US. of 0*01 a»J i To*J US. a ate MI ml
Typoaf
Pmof ay<*»a«--)
wo* DM afCFl fooa Mai
Koo. M m aaaa arw a*
■TM waaaalara-awl
■a" CP1
of MM ■ UA
FT*. M. »? «M. MM.) of MM .boa
Fooa BM amove* FoaJi
l(aT) CPT)
■toad anal (arijaj aai ml)
Pofcy bfaoa h-cn *. CaiCn< Dafaat
GMMk
•«.-M-M-I
M" MM III II Wy Hill lliMlMrfTTf
lafCTl
»a«.
Iwrui Jod Popkin and Comrjajry
<7 Al-1
APPENDIX 2
THE IOWA DATA COLLECTION FIELD STUDY
One of the unknowns that existed when this project began was how long it would take to
collect prices in rural areas. A major concern was the amount of travel time between outlets
as well as the amount of time that might have to be spent in the outlets. Hence, part of mis
feasibility study was to conduct a trip to several rural PSU's to see how much time was
involved in travel and price collection. Three PSU's in Iowa near Waterloo were designated
as the PSU's where the field study would take place.
•
Prior to the field study, information was gathered on where grocery stores and convenience
stores were located in each of the three PSU's. The location information is from the Iowa
Business Directory, published by American Business Directories (Omaha). The stores visited
were only the rural ones (although clearly urban outlets would be included in any rural price
index if households shopped mere). Before they were visited, their permission was sought to
visit them. About IS percent of the outlets contacted would not provide that permission.
Nine to eleven stores were visited in each PSU. The representative who canvassed the
PSU's went to the stores with a pre-selected list of 66 food hems. On visiting the store, he
recorded whether the item was available. He also mentally noted the price of the item and took
enough time with each hem to simulate writing down its price. (One of the criteria for getting
permission to visit the stores was that prices would not be recorded.) The representative took
a day to canvass each of the PSU's.
Tables A2-1 to A2-3 contain the information on the schedule used to visit the outlets in each
PSU and the time spent at the outlets. The tables also contain mileage and travel times
between outlets. Table A2-4 contains summary statistics for three PSU's and is based on the
data found in Tables A2-1 to A2-3. Table A2-4 contains the figures used in various
simulations of costs for the BLS alternative. In particular, note the travel time in minutes per
outlet (22.9 minutes) and the travel miles between outlets (19.3).
A2-1
4/
Another important datum for this report is the amount of time it takes to record the price
of an hem. To know that, it is necessary to know how many hems mere are per outlet for
which prices were "recorded" and the amount of time spent in an outlet recording mem. The
way to determine mat datum is by combining data from Table A2-4 with data from Table A2-
5. Table A2-5 provides information for determining the number of hems per outlet That is
determined by summing the percentages in the first column of the table. The sum (divided by
100) is 44.8 which is the average number of hems per outlet for which prices were "recorded."
The mean visitation length in minutes for an outlet can be found in Table A2-4. It is 26.1
,niini*rf The amount of time spent per hem (in minutes) is determined by dividing 26.1 by
44.8. *That comes to 0.6 minutes per hem. The figure of 0.6 was not used in the simulation,
though, because of substitution needed for disappearing hems. The actual figure used was 0.8
imnules per item.
In addition to determining those critical data, the person canvassing the outlets was asked
to determine whether the store had a scanner, the size of the store in square feet, the number
of registers, etc. A by-product of the visit was determining the distribution of non-packaged
goods in typical rural outlets. That information can be found in Tables A2-6 to A2-8.
A2-2
TABLE A2-1
USDA Iowa Trip TRAVEL STATISTICS
BAY I
Minates Minutes
DottaabM Town Departart Time Miles Reads Traveled Road Quality Arrival Tha* Traveling ia Outlet
Grocery 1 Clarksvilte 08:35 am 30.7 2ii, 3, in good 09:10 am 35 21
Grocery 2 Clarksvilte 09:40 am 0.5 IU good 09:42 am 2 31
Grocery 3 Parkersburg 10:15 am 22.4 IM, 3, 14, 20 good 10:38 am 23 34
Convenience 1 Parkersburg 11:15 am 0.5 20 good II :20 am 5 14
Grocery 4 Applinfton 11:35 am u 20 good 11:47 am 12 34
Grocery 5 Welltburg 12:26 am 14.1 20,119 good 12.51 pm 25 35
Convenience 2 Welbborg 01:21 pm 1 214 good 01:33 pm 5 14
Convenience 3 Grundy Center 01:50 pm 13.1 214, 175 good 02:05 pm IS 16
Grocery 6 Grandy Center 03:03 pm 0.5 175 good 03:05 pm 2 37
Grocery 7 Dike 03:44 pm 16.5 14,20 good 04:09 pm 25 29
Convenience 4 Dike 04:40 pm 0.5 20 good 04:43 pm 3 12
Return to: Waterloo 04:59 pm 21.2 20 good 05:25 pm 26
Summary Total Miles 127.9 Total minutes
Minutes per
171 284
Miles per outlet 11.63 outlet 16.18 25.12
standard deviation 10.06 standard
deviation 11.09 9.76
pet owlet" i. pottMMoftaMc the -MMMtts Travetliaf" canaaa ii the travel tmc pa owlet, ml mta the 'MiiaUi ia Owlet" M ■• the •venae lie: ia the oatkt
A2-3
S£
TABLE A2-2
DAY 2
Mlaales Mlaates
DONHNH Town Departare Time MHet Reads Traveled RoadQasUty Arrival Time TraveMaaj iaOatkt
Grocery! Dysart MM am 30 21.1 good 08:35 am 35 3$
Convenience S Dysart 09:20 am 0.5 Mam Street (OOd 0921 am 1 16
Grocery 9 Oladbioolr 09:41 am 24.4 8,63.96 good 104)6 am 25 33
Convenience 6 Toledo 10:41 am 202 96.6.1 good 114)3 am 22 13
Convenience 7 Toledo II :20 am \2 63 good 11:22 am 2 14
Grocery 10 Clntier 11:40 am IS 63, e43 good-many bends 11:51 am II 26
Grocery 11 Btaintowa 12:30 pm 29.6 36. 14, 30.12 good-82 bumpy 01:03 pm 33 29
Grocery 12 NCWfMll 01:43 pm 12.2 82. 30. 217 good 01:51 pm IS 36
Convenience 1 P4CWIMII 02:38 pm 0.5 217 good 0239 pm 1 17
Convenience 9 Atkins 02:58 pm 10 •44 good 03:08 pm 10 15
Grocery 13 SheHsourg 03:24 pm t e24 good 03:15 pm II 36
Rotate: Waterloo 04:15 pm 58.3 w26. 380, 20 good 054)9 pm 54
Summary Total MHes 209.9 Total minutes 227 270
Minutes per
Mies per outlet 19.08 outlet 20.6 24.6
ttMMHnl dCVMDOn 15.93 standard
feviation 15.33 9.65
NMK -MBtfa>p«r«nM-iaai^rrMrfiM •*.*. Mi-«» Travcttaf- c* xnactaat iadMaantt
c? A2-4
TABLE A2-3
PAY 3
Mantes
n'" "" Turn Departarc Ttroe Mats Roads Traveled ReedQaaary Arrival Time Travearog MOattet
Grocery 14 Oxford Junction 09 00 am 411 30. k64 |0Od 09 51am Si 29
0—1— io Oxford J.mcticm 10 21 an 1 ■.<■■■■>■ ■ ■! food 10.30 am 2 12
0—1— u Crocaae 10 44 «n 266 136 pood 11:13 M 29 17
Grocery IS Crocade IIJI am 2 136 good llJSm 4 31
Groovy 16 St. Porot 12:15 pro 53.1 136. 20. 31. 3 tood 01:16 pro 61 37
0—1— 12 St Point 01 54 pm \2 136 pood 01:57 pro 3 16
Grocery 17 Winrorop 02:14 pro 2SJ 3,117, 939 pood 02:44 pro 30 35
Grocery It tan* 03:21 pro MJ 939 pood 03:47 pro 26 39
Grocery 19 Fattens: 04:29 pro 111 v62 pood 04:41 pro 19 33
Hem to: Cedar Rapids OS:22 pm m v62.20.3M pood 06:35 pro 73
SumrMfY Total Mites 261.4 Total naaaaa
Mmutes per
305.0 256.0
Miles per outiet 29.04 outlet 3319 2144
standard deviation 23.70 standard
deviation 24.33 1051
TOTAL TRAVEL STATISTICS
Total Miles 599.2
Mies per outlet I9J3
I deviation 11.15
tponioaofMMc nifecnvdi
Total minutes 710.0 •10.0
Travel Mmutes
per outlet 22.90 26.13
MroTHMTU
deviation 11.49 9.76
■ Otter mliiMi. Hiittivi•ajtHMtaaaiMM.
A2-5
roT
TABLE A2-4
SUMMARY STATISTICS OF IOWA FIELD STUDY
Put A: Travel Statistics
Total miles traveled 599.2
Miles per outlet 19.3
Standard deviation 18.2
Total travel minutes 710.0
Minutes per outlet 22.9
Standard deviation 18.5
Part B: Store Statistics
All Stores Convenience Grocery
Number of stores
visited 31 12 19
Mean visitation
length in minutes 26.1 14.7 33.4
A2-6
H
TABLE A»
DISTRIBUTION OF FOOD ITEMS
(AD stores and by type of store)
ALL STORES CONVENIENCE GROCERY
ITEM (percent of stars (percent of stores (percent of stores
carrying each Hem) carrying each item) carrying each
Hem)
n-3I n-12 n-19
FROZEN FOODS
From broccoli (spun) 58.1 8J 89.8
Frozen orange juice 87.1 66.7 100.0
Frozen pies 54.8 0.0 89.5
ice cream (1 gal) 77.4 50.0 94.7
Frozen Turkey 32.3 0.0 52.6
Frozen Pizza 96.8 91.7 100.0
frozen nsri 64.5 16.7 94.7
MEATS
uncooked ground beef 58.1 0.0 94.7
uncooked beef roast 452 0.0 73.7
urtfooked beefsteak 58.1 0.0 94.7
uncocked veal 0.0 0.0 0.0
baco* 83.9 58.3 100.0
nam 71.0 25.0 100.0
pork chops 48.4 0.0 78.9
pork roast 22.6 0.0 36.8
frankfurters (pack) 80.6 50.0 100.0
bologna (pack) 80.6 58.3 94.7
lamb coops 0.0 0.0 0.0
chicken (whole fryers) 32J 0.0 52.6
fresh catfish 0.0 0.0 0.0
DAIRY PRODUCTS
margarine (stick) 80.6 50.0 100.0
buoer(sock) 77.4 41.7 100.0
CCC5 (niCulUDl 1 OO-ZCfl) 87.1 75.0 94.7
oik (1 gallon) 96.8 91.7 100.0
cream cheese 71.0 33.3 94.7
cottage cheese (!6oz) 64.5 25.0 89.5
yogurt 80.6 50.0 100.0
FRESH FRUITS * VEGETABLES
apples 64.5 13 100.0
58.1 U 89.5
oranges SS.1 oo 94.7
grapes (while seedless) 51.6 oo 842
potatoes 64.5 u 100.0
lettuce 61.3 13 94.7
-—rr*~- 58.1 OO 94.7
encasaban 41.9 0.0 68.4
CANNED GOODS
peas(15oz) 80.6 50.0 100.0
tammaenp ^ 93J 91.7 94.7
onves (black, medium) 54.8 u 842
tuna (oil 6oz) 96.8 91.7 100.0
A2-7
?*
(«*.)
TABLE A2-5
DISTRIBUTION OF FOOD ITEMS
(AM jjjgw aad by lype of »IDW)
ITEM
DRINKS
apple juice
cola (2 liter)
snappte-type drink
coffee (caffeinated, 8oz)
. (16 Iwcs)
SWEETS
non-dairy creamer (pint)
pound cake
oatmeal cookies (pack)
doughnuts (loose or packed)
chewing gum (pack)
licorice (black or red)
chocolate bar
BREADS
white (1 lb)
Kaiser rolls
OTHER ITEMS
Italian dressing (8 I oz)
vegetable oil (4Soz)
flour (white all purpose, 51>a)
com flakes
rice (white, 2fcs)
salt (iodized)
BBQ sauce (16 oz)
rcJun
baby (bod (purged carrots)
chocolate syrup (24oz)
0»)
San of (dWdad by If) - Al
ALL STORES CONVENIENCE GROCERY
(percent of stores (percent of stoics (percent of stores
carrying each hem) ctfiying each item) carrying each
hem)
96.1 100.0 94.7
1000 100.0 100.0
80.6 100.0 6*4
80.6 50.0 100.0
T7.4 41.7 100.0
77.4 50.0 94.7
3-2 0.0 5J
87.1 83 3 89 5
90J 100.0 842
903 83.3 94.7
100.0 100.0 100.0
100.0 100.0 100.0
100.0 100.0 100.0
T7.4 41.7 100.0
22.6 0.0 36.8
90J 91.7 89.5
71.0 25.0 100.0
71.0 25.0 100.0
67.7 16.7 100.0
71.0 25.0 100.0
67.7 16.7 100.0
71.0 25.0 100.0
77.4 41.7 100.0
•0.6 50.0 100.0
US 25.0 89.5
742 33.3 100.0
100.0 100.0 100.0
0.0 0.0 0.0
•41 Mi HJ
A2-8
7/
rxscrnr
SUMMARY STATISTICS OF IOWA FIELD STUDY
All Store Convenience Grocery
Percent of Stores with:
Fresh fruits and vegetables 6