The evaluation of the forecasts of any economic variable is a key element, both for analysts who generate these predictions and for those who make decisions based on them. For those who make the predictions, the evaluation of the predictive performance is a fundamental aspect to determine the adjustment of the model or method used to represent a relevant economic process, and provides valuable information on the adequacy of the models to the data, as well as on problems that They have not been detected in the model specification phase.
Therefore, the evaluation of inflation predictions constitutes an input of the modeling process and is especially important for those who make use of the predictions. Among them, it is particularly relevant for those who design monetary policy and for users in contract price adjustments. As an example, currently the salary adjustments in Uruguay take into account the inflation projections of macroeconomic analysts that the Central Bank of Uruguay (BCU) relieves.
This paper proposes methodological guidelines for the evaluation of inflation forecasts released by the BCU from January 2004 to the present. The proposed scheme and of which an application is presented, is based on the calculation of a set of descriptive statistics on prediction errors, in particular the RMSE-h statistic proposed by Cecchetti et al (2000). This is especially relevant in the practice of predictive evaluation for policy makers as it allows for a more accurate evaluation of the performance of inflation forecasts, taking into account corrective measures that decision-making agents could take in the light of forecasts more credible about the evolution of the variable.
Evaluation of analysts’ forecasts: the BCU Outlook Survey
