Abstract:

 in this paper we show the implementation of an automatic economic nowcasting and short-term forecasting procedure using midas approach in gretl. We also demonstrate possibilities of gretl's interface for direct downloading macroeconomic time-series from db.nomics database. In empirical illustration we show how to predict quarterly gross domestic product using a mixture of quarterly and monthly explanatory variables. In all cases, the midas generates more accurate forecasts comparing to benchmark model. Results of our study confirm that midas approach may have better forecasting performance than other adl specifications.