: This paper discusses in some detail various existing approaches of measuring core inflation, evaluating their potential advantages and disadvantages. Then a variety of measures of core inflation for India based on three methods are constructed. Among these measures, three are based on conventional ex-food and energy principle and one measure excludes 15 of the most volatile components. While constructing exclusion-based indices of core inflation, measures are constructed in such a way that only a small weight gets excluded from the index of the core inflation. The other two core measures that are variations of ‘Neo-Edgeworthian index’ are constructed by reweighting 69 disaggregated components series of WPI. Then another class of core measures is computed based on weighted exponential smoothing which was primarily developed by Cogley (2002). Estimates of core inflation based on their indexes are then calculated for the period from 1995 to 2007 (on monthly basis). Subsequently, an empirical evaluation of these estimated core measures is performed. While choosing criteria for empirical evaluation, the following were considered: first, forward-looking monetary policy framework, viz., tracking trend inflation, and second, an attractor property of core inflation and their relative predictive power for future transient movement in inflation. The empirical findings show that the weighted exponential smoothing-based measures of core inflation perform far better compared to all other measures of core inflation in terms of similarity in means, lower volatility, tracking trend inflation and an ability to predict future transient movement in headline inflation at both 12-month and 24-month horizons.
