Finding optimal measures of core inflation in the Kyrgyz Republic

The paper examines alternative measures of inflation caused by monetary factors in the Kyrgyz Republic and determines the best method of measuring the core (or underlying) inflation. Use of the most appropriate method may improve the efficiency of monetary policy and budget forecasting.

The main reasons for inflation in recent years in the Kyrgyz Republic have been budget deficit, administrative changes of prices and tariffs, an increase in imported energy prices due to changes of external markets, and price increases on imported consumer goods. These non-monetary factors together with various supply shocks cause the high volatility of a consumer price index (CPI), a widely used measure of inflation.

The author argues that a core inflation index, which takes into account long-term and steady inflation while ignoring short-term price movements, is better suited as an inflation measure for monetary policy purposes than CPI. The author asserts that the core inflation measure allows for a more effective decision-making, as short-term shifts in prices do not require reaction from policymakers, nor are they within the policymakers’ area of control.

In order to find the best method for measuring core inflation in the Kyrgyz Republic the author analyses annualised monthly data for CPI for the period of July 1995 to December 2000, and finds that:

• exclusion method:significantly improves the inflation measure but excludes observations of the most volatile components in the consumer basket, such as agricultural produce, imported consumer goods, energy and rental fees, which constitute a large share of families’ budgets and thus, may omit some important information
• trimmed means method:allows one to lower the undesirable properties of a sample mean, such as maximal dispersion, bias and abnormal distribution, however, due to non-normal, not known and varying over time population distributions of CPI, finding a good estimator is problematic
• standard deviation trimmed means method:also improves the behaviour of the inflation measure, however, it excludes prices regardless of the sources of noises, and thus may lose important information on price expectations in the future
• percentile method:takes into account all the available observations and therefore may be a preferred method.

The paper does not recommend any one method as the best suited for the Kyrgyz Republic. However, of the four methods above the exclusion method and the standard deviation trimmed means method are found to yield better results than the other two. However, they also have the disadvantage of a high probability of losing important information.