Abstract
This paper implements different econometric models (Bridge, MIDAS, factor-augmented versions, MF-BVAR models and their combination) to nowcast Costa Rican quarter-to-quarter GDP growth. I exploit a comprehensive set of macroeconomic indicators to conclude that models ARIMA, Factor-VAR, unrestricted MIDAS and Bridge are consistently more precise than other specifications. Furthermore, I find that production-related variables have higher predictive power (mainly the IMAE), controlling for seasonality adds biases to the model’s forecasts, and structural breaks in the series do not affect the nowcasts. I recommend to use these models and their combination in order to have up to date information for policy making decisions.
Type
