A Bayesian Structural Time Series Approach to Forecast Mexico’s Consumer Index Cover Image
  • Price 10.00 €

A Bayesian Structural Time Series Approach to Forecast Mexico’s Consumer Index
A Bayesian Structural Time Series Approach to Forecast Mexico’s Consumer Index

Author(s): Mario A. GARCÍA-MEZA, Francisco Venegas-Martínez
Subject(s): Economy, National Economy, Business Economy / Management
Published by: Reprograph
Keywords: Bayesian structural time series; state space models; consumption index; forecasting;

Summary/Abstract: A Bayesian structural time series model is used to forecast the current value of the consumption index in Mexico, where correlated searches in Google's search engine are used to determine the weekly value of the index. Mexico’s Institute of Statistics and Geography releases the consumer index, along with other important macroeconomic variables, in a three-month span from their recollection. This can be improved by using information that is readily available just a few days from the end of each month to create an estimate with Bayesian methods. For this we set the time series model in state space mode, which allows us to use a big set of regressors as predictors of the current value of Mexico’s consumer index. The main finding is that the use of the search queries improves significantly the accuracy of prediction.

  • Issue Year: XIII/2018
  • Issue No: 57
  • Page Range: 615-626
  • Page Count: 12
  • Language: English