Core inflation

There are several measures of inflation. The official measure - that is, the rate of change in the Consumer Price Index (CPI) in most countries - is also referred to as "headline inflation" owing to its ability to make news headlines. Headline inflation, however, is often subject to large and temporary fluctuations arising from supply shocks, for example production declines due to unfavourable weather conditions or external factors affecting the prices of one or more consumer goods imported into a country. Another measure of inflation aims at removing these volatile components from headline inflation. The concept of core inflation is based on the idea of identifying the underlying persistent trend of inflation.

There are multiple approaches to derive core inflation from headline inflation. Among them, the most widely used approach is excluding selected groups of items from the basket used to compute headline inflation and recalculating the weighted change of prices of the remaining items in the basket. Food and energy items and interest charges are the most popular exclusions. The exclusion method is used by central banks more frequently than other methods, as that method is computationally simple, easy to understand and derivable without any time lag.

In the United States, the Bureau of Labor Statistics has published the movements of an index excluding food and energy items from the CPI since 1977. This measure was first systematically analysed by Gordon (1975). Since the movements in food and energy prices in the United States during the 1970s often reflected the developments outside of the country's domestic demand and supply factors, this measure of core inflation was useful to explain the inflation generated by domestic aggregate demand. Over the subsequent decade, economists tended to believe that food and energy price move- ments, being relatively volatile in the short to medium run, would make only transitory impacts on headline inflation. It was observed that large rises in these prices were often followed by large decreases in them, and vice versa. Volatilities in food and energy price movements, frequently caused by unusual weather conditions, were generally found to be self-correcting. In the countries located far from the equator, in particular, inclement weather conditions often lead to temporary food shortages and increases in demand for household fuels. In that framework, the core inflation measure compiled by excluding food and energy items served its intended purpose for several countries for a long period. The rationale for excluding food and energy items is the volatility in their prices.

However, analyses of the movements of prices by several researchers have revealed that only some (seasonal) components of food are more volatile (see Cutler, 2001). Many countries have excluded a part of the food basket, instead of the entire basket, together with energy items, for that very reason. Some of these countries exclude fresh food, unprocessed food or agricultural food without considering the statistical behaviour of price movements, while some other countries have picked the food items to be excluded based on the volatility measured using statistical properties of price movements. First- round effects of indirect taxes are another popular exclusion in several countries. If not excluded, the one-off impact of the tax change on inflation rates could obscure the long- run trend of the inflation time-series.

The most popular alternative approach to exclusion-based methodologies is the "trimmed mean" measure proposed by Bryan and Cecchetti (1994). The trimmed mean removes the items with extreme price changes; that is to say, the two ends of a histo- gram of price changes. The selection of upper and lower points, beyond which data are truncated, is a matter of judgment. As for its economic rationale, the trimmed mean has the potential to eliminate all relative price changes and thereby isolate the com- ponent of aggregate price change expected to persist (Clark, 2001). Among the other statistical measures used less often are the weighted median approach also proposed by Bryan and Cecchetti (1994), the volatility-weighted measure, and the exponentially smoothed measure (see Colgey, 2002). Despite the qualitative superiority of the statistical approaches over the exclusion method, central banks use them less commonly than the exclusion method, primarily because the statistical approaches are not easy to explain to the public and are difficult to replicate.

See also:

Consumer price indices; Inflation; Inflation measurement; Inflation targeting.