Forecasting Inflation from Disaggregated Data: The Colombian Case

• Activity Number: 310
• Type: Contributed
• Date/Time: Tuesday, August 6, 2013 : 8:30 AM to 10:20 AM
• Sponsor: Business and Economic Statistics Section
• Author(s): Wilmer Martinez*+ and Eliana Rocio Gonzalez
• Companies: Central Bank of Colombia and Central Bank of Colombia
• Keywords: Disaggregated information ; Forecast inflation ; Jonckheere Terpstra test ; Rank forecasting models

Abstract

Empirically I try to determine whether disaggregated information of the CPI is useful to forecast inflation and which level of disaggregation of the CPI basket is more useful to generate forecasts of total inflation and some aggregate CPI components. Additionally, I compare different forecasts obtained from both disaggregated information and the aggregate series directly and order them according to their forecasting performance. To achieve this we consider all levels of aggregation of the CPI basket and for each level an ARIMA model is estimated for each item, so that forecasts are generated and grouped using the CPI weights to obtain a forecast for the aggregate components: total inflation, food, tradable, nontradable and non food inflation. Then a scheme to evaluated forecasts based on the procedure proposed by Martínez and Hernandez (2012), which is emphasized in the Jonckheere Terpstra test, is implemented in order to rank the different forecasting models based on disaggregates and comparing then to more traditional models based on aggregate series and other models currently in used by the Colombian central Bank.