Inflation Forecast: Just use the Disaggregate or Combine it with the Aggregate

Abstract

Using data from India, the paper provides three stylize facts about the inflation forecasting: (a) using disaggregate data helps to achieve gains in forecast accuracy relative to forecasting the aggregate inflation directly; (b) using weights derived from spillover index for component forecasting compared to the official weights or the criterion suggested by Bates and Granger further improves efficiency; (c) combining disaggregates along with aggregate data is beneficial for forecasting inflation. Results also highlights the fact that inclusion of too many disaggregates might result in efficiency loss in short-term forecasting but definitely results in gain for the medium-term.