Do robust predictors improve the accuracy of inflation forecasts in moments of structural break?

The accurate prediction of inflation rates holds critical significance for both policymakers andeconomic agents. It is imperative to comprehend the limitations and strengths inherent in differentmodels and information sets used to forecast inflation across varying time horizons. This study seeksto enhance the existing literature on Brazilian inflation forecasting by assessing the predictive efficacyof predictors robust to structural breaks, with a particular emphasis on the methodology introducedby Martinez et al. (2022). The findings of this study indicate that robust predictors exhibit notablysuperior performance during periods of instability and structural change. In the Brazilian context,these predictors outperform expert forecasts specifically during the COVID-19 pandemic period, asindicated by the Focus Survey. However, it is noteworthy that in the immediately preceding period,these models do not outperform the aforementioned survey.