Skip to content

Down to the Core


Michael T. Owyang , Kristie M. Engemann
Friday, July 1, 2005

Turn on the news and you might hear the term "core inflation," as it is considered to be a better predictor of future price movements than all-inclusive inflation. Core inflation removes items that have frequent price fluctuations, such as food and energy.

In this article, we discuss two measures of core inflation using the Consumer Price Index: inflation excluding food and energy, and weighted median inflation.

The Consumer Price Index measures what consumers pay for a given basket of goods relative to a base year. For example, the same basket of market goods costing $100 in 2000 would have cost almost $110 during 2004. The inflation rate is the percentage change in the index from the same period of the previous year. Thus, an inflation rate of 3.0 percent for May 2004 implies that the index was 3.0 percent higher for May 2004 than for May 2003.

Inflation excluding food and energy is the measure of core inflation most often cited in the media. This measure is calculated similarly to overall inflation; however, food and energy are eliminated from the basket of goods.

Calculating weighted median inflation involves three steps. First, one calculates the percentage change in price from the previous month for each component–not the entire basket. Next, the components are ordered by their inflation rate for that month and weighted by the percentage of the total amount spent on that component. Finally, one determines the inflation rate where the median exists–half of the components have price changes higher (or equal) and half are lower (or equal).

The following example shows a market basket ordered by monthly inflation rates. Component A has an inflation rate of 0.3 percent and a weight of 10 percent. Component B has an inflation rate of 0.6 percent and a weight of 18 percent. When you add the weights of components A and B, the cumulative weight is 28 percent. Thus, 28 percent of the goods in this basket have an inflation rate of 0.6 percent or lower. Here, a cumulative weight of 50 percent is associated with Component D, and the weighted median inflation rate is 1.2 percent.

Component Infl. Rate (%) Weight Cumulative
A 0.3 0.10 0.10
B 0.6 0.18 0.28
C 0.7 0.17 0.45
D 1.2 0.11 0.56
E 1.5 0.25 0.81
F 1.9 0.19 1.00

As our first example illustrates, the weighted median does a good job of measuring core inflation because it eliminates components that have large and small price changes, which are not likely to reflect a trend.

Consider a second example where one component has an abnormally large spike in price. In this second basket, Component Z is computer supplies, and two of the other components are food and energy.

Component Infl. Rate (%) Weight Cumulative
V 0.2 0.15 015
W 0.4 0.17 0.32
X 0.5 0.19 0.51
Y 0.8 0.33 0.84
Z 3.0 0.16 1.00

Suppose the factory that makes the majority of computer chips has a fire that temporarily disrupts production, causing a price spike. We include Component Z when calculating inflation excluding food and energy, but eliminate it when calculating weighted median inflation rate. Thus, the weighted median inflation rate seems a truer reflection of price trends.

Inflation excluding food and energy eliminates only food and energy every month, regardless of the price changes for each good. In contrast, the weighted median inflation rate may exclude different components from one month to the next based on price fluctuations. Our second example shows that inflation excluding food and energy may not reflect price trends, as it may include components that have short-lived, extreme price changes. Weighted median inflation ignores those components, making it a better measure of core inflation.