Abstract
This paper investigates whether consumer confidence indices can improve the forecasts of aggregate consumption in Brazil, taking into account information on economic fundamentals contained in lagged financial indicators and growth rates of GDP and credit to households. In this context, we allow different lag structures for the potential predictors of consumption, which gives rise to a large space of potential models. Thus, we apply Bayesian model averaging techniques as an agnostic strategy to deal with the inherent uncertainty about the model. This approach allowed us to investigate which predictors can be considered robust. The in-sample results suggest that GDP, credit to households, return on stocks and consumer confidence indices have a robust predictive potential. Finally, the out-of-sample results suggest a not insignificant role for consumer confidence indices in forecasting the growth rate of the aggregate consumption in Brazil, especially for short-term forecasting horizons.
Keyword:
Aggregate consumption; Consumer confidence indices; Forecast; Bayesian model averaging
Thumbnail
