Predictability of competing measures of core inflation : an application for Peru

A central element of an inflation targeting approach to monetary policy is a proper measure of inflation. The international evidence suggests the use of core inflation measures. In this paper we claim that core inflation should be measured as the underlying trend of inflation that comes from nominal shocks that have no real effect in the long term. However, most of the time core inflation is computed zero weighting observations at the tail of the inflation distribution. Quah and Vahey (1996) proposed a method of computing core inflation imposing theory restrictions to a SVAR specification. In this paper we present estimation for Peruvian data and compare the predictability properties of competing measures of inflation following an idea of Diebold and Killian (1997).