Short-term forecasting Romanian GDP growth using a limited selection of monthly indicators

We apply a bridge equation model to forecast short-term GDP growth for Romania using a small number of commonly employed indicators with a monthly frequency. The monthly indicators are forecast to the time horizon of interest through an autoregressive process (AR). The data is aggregated to quarterly frequency and each independent variable is paired with the dependent variable (GDP growth). For each pair a distributed lag model is applied, and the forecast is obtained as the average of the forecasts produced by all pairwise models. The idea of using indicators with a higher frequency to forecast quarterly GDP data has been applied to the Euro Area and countries from Western Europe. Despite this, its application to Eastern Europe remains limited. We test our simple model on current quarter (nowcast) and quarterahead forecasts under two scenarios. In the first scenario only car-registration data are available for the first month of the current quarter, whereas in the second all data are available for the current quarter. We find that our model produces more accurate forecasts compared to a firstorder AR model using only GDP data. As expected, the accuracy of the forecast improves under the second scenario.