Restricted forecasts with multiple time series models and their application to evaluate macroeconomic policy targets in Mexico

Abstract

This work applies the restricted forecasting methodology to monitor the attainment of targets announced by the Mexican Government for some relevant macroeconomic variables. This method yields scenarios that are in line with the expectations underlying the proposed targets. Some multiple time series models of the VAR, VEC and BVAR type are first built and their predictive abilities are analyzed. The method is then applied with the best forecasting model, which was found to be a BVAR model. With the scenarios so derived it is possible to decide whether the targets are feasible or not and the moment when statistically significant evidence has arisen to declare the targets unfeasible.