Do Underlying Measures of Inflation Outperform Headline Rates? Evidence from Australian Data

Abstract

Many central banks often focus on underlying measures of inflation when assessing inflation trends. This paper compares the accuracy of underlying measures of inflation relative to the headline rates by using Australian data. It is found that the underlying measures did have smaller errors in predicting the long-term trend in inflation than the quarterly headline rate in the sample period 1977 to 2001 and the inflation targeting subperiod starting from 1993, due to the large volatility of the headline rate. As compared to the year-ended headline rate, the statistical test results , however, support only the measure of market prices excluding volatile items, not the others. There is some weak evidence of the weighted median measure outperforming the headline rate in the subperiod after 1993. With respect to directional accuracy, the test statistics cannot reject the null hypothesis of an equal probability correctly predicting the moving direction of the inflation trend, though the headline rates have a higher probability.