|
Changes in the Consumer Price Index (CPI) are widely used to measure inflation (changes in the general price level) in the U.S. economy. The CPI for food at home is the principal indicator of changes in retail food prices and is closely followed by industry analysts, food market participants, and policymakers, both public and private.
Analyses by the Bureau of Labor Statistics (BLS) and ERS have suggested that the CPI for food at home overstates inflation of food prices. Their evidence shows that the gap was around 0.5 percent per year between 1978 and 1995. More recent work using data for the years 1998-2001 found about a 10 percent upward bias in the CPI for food at home due to the increasing share of food purchases made at alternative (non-traditional) retail stores such as supercenters, mass merchandisers, and club stores (see CPI Bias from Supercenters: Does the BLS Know that Wal-Mart Exists? a paper presented at the Conference on Research in Income and Wealth (CRIW) Conference on Price Index Concepts and Measurement).
What are the sources of this gap? Broadly, part of the gap in the
food-at-home CPI arises from the challenges of tracking changes
in food consumption patterns, and a larger part arises from problems
in the construction of the index. The researchers cite four problems
associated with CPI construction problems.
- Consumers' broad spending patterns change gradually over time.
For example, consumers have more recently devoted smaller shares
of household budgets to meats and alcoholic beverages than they
did during the currently used 1993-95 base period and larger shares
to cereal and bakery products. Indexes must be adjusted periodically
to reflect changing consumption patterns, and adjustments require
investment in large-scale surveys. People generally spend a smaller
share of their income on food today than they did three or four
decades ago, and a smaller share of food spending goes for food
at home.
- Prices for some food items, particularly fresh fruits and vegetables
but also fresh meats, occasionally fluctuate widely. If not accounted
for, these fluctuations would create an inflationary drift by
implicitly attaching greater weight to price increases than to
decreases—for example, a $0.50 increase on a dollar price
is a 50-percent increase, but a return to the dollar price is
a 33.33-percent decrease.
- Price indexes must keep up with the introduction of new food items and new stores because the new items and stores may have different price levels and changes than those already in the index. In introducing new items and stores to the index, index designers also face the challenge of capturing a representative sample of new items and attaching appropriate weights to them. Each year, approximately 10,000 new items are introduced in supermarkets. To keep up with changing retail purchasing patterns, BLS has relied on a sophisticated system of household and store sampling.
- While the fixed-weight version of price indexes is easier to
understand and simpler to calculate, a calculation that assumes
fixed weights does not allow for the possibility that consumers
may change the quantity or level of purchases when faced with
changing prices. This potential bias, substitution bias, results
from the fixed-weight price index assigning too much weight to
goods and services with rising prices and not enough weight to
goods and services with declining prices. Although a real potential
bias, the general consensus seems to be that the bias is small,
especially compared with the other biases.
Bureau of Labor Statistics' Recent and Future Efforts to
Address These Problems
BLS identified problems with introducing new stores into their price
surveys, and in January 1995, it changed its procedures for estimating
the weights to be assigned to specific price observations at particular
stores. In January 1999, BLS introduced new methods of index construction
to limit problems associated with sharply fluctuating prices. The
agency estimates that, had the new policies been in effect between
1978 and 1995, the CPI for food at home would have risen 0.5 percent
per year less, or 8.8 percent less for the entire 17 years. Effects
would be substantially larger for fresh fruits and vegetables and
somewhat larger for fresh meats. Looking forward, users should expect
smaller CPI increases than in the past because of the change in
index construction. (For more detail, see James M. MacDonald, "Consumer
Price Index Overstates Food-Price Inflation," FoodReview,
September-December 1995.)
In the past, expenditure weights were updated every 10 years or
so. Starting in 2002, however, these weightsare updated every
2 years. For example, in 2002, the CPI for all Urban Consumers (CPI-U)
and the CPI for Urban Wage Earners and Clerical Workers (CPI-W)
use expenditures from the 1999-2000 period, and replace
the current 1993-95 weights used in the index. Effective with the
release of CPI data for January 2004, CPI expenditure weights are updated to the 2001-02 period.
By using more current expenditure weights, BLS is able to make
the CPI more nearly reflect current inflation. The more current
weights should ensure that the relative importance of CPI item categories,
such as food away from home, college tuition, or medical care services,
more accurately reflects how consumers currently allocate their
spending.
Since January 1999, the BLS has used a geometric mean formula to calculate most basic indexes within the CPI, which means the prices within most item categories are averaged using that formula. However, because the geometric mean formula is used only to average prices within item categories; it does not account for consumers substituting between item categories. In August of 2002, BLS began to publish a new chained index (C-CPI-U) to more closely approximate a cost-of-living index by reflecting substitution among item categories. More information on this new chained index can be found on the BLS website. Current index numbers are also available.
|
