Real GDP and its structure are available within 70 days after the end of the reference quarter. By using leading indicators of higher frequency, it is possible to nowcast GDP in real-time.
With an assumption of unobserved factor driving the business cycle we estimate dynamic factor models for real GDP, its demand components, inflation, wages and employment using statistically significant domestic and foreign indicators. To ensure the consistency of out-of-sample forecasts for GDP and its components, past forecast deviations and correlation coefficients are used to adjust the forecast, which helps to reduce the bias of individual models. Forecasts using real-time database are carried out since the 1st January of 2017 using daily data vintages. Real-time forecasts display a reduction of forecasting error with the arrival of new data in the last month of the quarter until the official publication. The main role of nowcasting in CBR is to track the actual positive and negative macroeconomic risks of the Slovak economy in relation to the latest official national macroeconomic forecast by the Macroeconomic Forecasting Committee. Additionally, the nowcast models help to improve precision of estimates of initial conditions of the economy by bridging the short-term forecast and mid-term forecast.
Keywords: Nowcasting, dynamic factor model, business cycle, Kalman filter, Slovak economy, demand components, short-term forecasting
JEL classification: C22, C32, C53, E32, E37