Summary of UNECE Presentation. Kiev, April 2009
The UNECE provided a brief overview of the CPI Manual and other international recommendations and materials, and then focused on calculation methods for elementary price indices and the treatment of seasonal products in the CPI; some possible future challenges in the compilation of the Ukraine CPI were also mentioned.
Concerning the calculation of elementary indices it was mentioned that the Carli and Jevons indices are independent of the price level while the Dutot index depends on the initial price level. Price changes in the Dutot index are implicitly weighted by the (relative) price level of the individual observations in the reference period so that more weight is ascribed to price changes for relatively expensive items. One way to reduce this effect is to group only homogenous items in the same elementary aggregate so that price dispersions are reduced. However, this may not always be possible in practice. It was mentioned that Jevons in general is recommended as the preferred index in the CPI, PPI and XMPI Manuals because of better statistical properties.
Seasonal products raise a number of methodological and practical problems in the CPI calculation in all countries. A fixed weights approach, as in the current Ukraine CPI was recommended, where prices in the out-of-season periods are imputed. It was underlined that the method in which seasonal products are treated should be “self-correcting”. That is, if a product after the out-of-season period reappears at the same price as it was introduced in the previous in-season, the method should ensure that this would result in an index of 100. If the price reappears at a 10% higher price, the method should give an index of 110 etc.
The following possible future challenges in the compilation of the Ukraine CPI were mentioned: Review of index calculation methods; review and optimization of the sample; the development of a Ukraine HICP; implementation of methods to deal with seasonal products and quality changes; and dissemination of data and metadata.