DOI: 10.32725/978-80-7394-976-1.31
In this paper, we examine the accuracy of the forecast models of the selected commercial banks in Slovakia. We focus on GDP and inflation forecasts in three different forecast horizons. The data is taken from an ongoing survey administered and regularly updated by the National Bank of Slovakia. The sample consists of nine of the largest commercial banks in Slovakia. Firstly, we see how well the banks compare to each other based on RMSE. Secondly, we estimate different baseline models, namely random walk and ARIMA models, for each horizon and compare each bank’s accuracy to the accuracy of these baseline models. Data is taken from the OECD revisions database, Eurostat and the database of the National Bank of Slovakia. Thirdly, we take different types of averages of the individual banks’ forecasts, as forecast averaging should increase accuracy based on contemporary forecasting scientific literature. We find that the models of these commercial banks are generally more accurate than our baseline models, except for one bank, which systematically provides less accurate forecasts in the survey. What is more, we find that individual models are capable of being more accurate than the averages for each and every horizon.
pages: 210-214