Comparing Two Sources of Retail Meat Price Data
by William Hahn
, Janet Perry, and Leland Southard
Economic Research Report No. (ERR-88) 42 pp, November 2009
USDA's Economic Research Service has a long history of
calculating meat prices at different stages in the marketing
system. ERS uses data collected by the U.S. Department of Labor's
Bureau of Labor Statistics (BLS) as the basis of its measure of
retail meat prices. The livestock industry uses this information to
make production decisions.
What Is the Issue?
Recent legislation required USDA to investigate the use of an
alternative source of data on retail meat prices. The purpose of
the legislation was to address livestock industry concerns
regarding the quality of retail meat price data used by ERS. This
report compares average retail prices calculated using data from
BLS with data from grocery stores using point-of-purchase scanners
to record prices. It analyzes the value of both data sets in
forecasting near-term market conditions.
What Did the Study Find?
Both BLS data and scanner data have relative strengths. BLS data
have several advantages over scanner data:
- BLS uses statistical sampling to select retail outlets to
survey. By contrast, scanner data are volunteered by stores and may
exclude certain retailers that BLS makes efforts to include.
- Because BLS uses sampling to select outlets, statistical theory
implies that price averages derived from BLS data ought to be
unbiased. To the extent that retailers whose price history is not
captured in the scanner data set have different price structures
than retailers who volunteer their data, scanner averages may or
may not be biased.
- BLS data are generally available 12-20 days after the end of
the month they are gathered. Because of processing issues, 7-8
weeks are required before scanner data become available.
Scanner data, in turn, have several advantages over BLS
- Scanner data include more meat cuts. The latest iteration of
scanner data showed 188 cuts, including domestic and imported lamb.
By contrast, the BLS database lists 32 meat and poultry cuts, some
of which have been discontinued.
- Scanner data provide some quantity measures. BLS data provide
none. Scanner data provide an index that compares a month's sales
of a particular meat cut with average monthly sales of the same cut
for a base year. Scanner data also show the share of meat cuts sold
at a discounted price.
- Scanner data provide a wider range of price-related statistics
than the BLS data. For example, scanner data are weighted by actual
sales and can be used to calculate standard deviations, not just
average prices. Standard deviations are a measure of the variation
In their current form, scanner data provide a different, but not
necessarily better, view of retail meat markets than BLS-based
data. Given the present production lag, scanner data for meat do
not appear to contribute value to the analysis.
How Was the Study Conducted?
ERS routinely uses BLS data to calculate retail composites for
Choice beef, pork, broilers, whole chickens, and whole frozen
turkeys. This study calculated the same composites using scanner
data and compared the two data sets. This study also used an
econometric analysis comparing scanner and BLS prices to determine
why scanner prices are more volatile than BLS prices. Both scanner
data and BLS data were then examined for their value in analyzing
current market conditions.