While up and down movements of real economic output in a time period are defined as “economic fluctuations”, the indicators that give information about current and future course of fluctuations of real output are called “economic indicators”. Short term forecasting of fluctuations is regarded important in terms of predicting resessions nearby and taking precautions against to them. In this study, by using 29 monthly or quarterly indicators between 2002Q1-2016Q3 period, it was tried to establish a forecasting model of quarterly GDP. For this purpose, with two variables, one of which was GDP as dependent variable and the other was an economic indicator as independent variable, VAR models were forecasted. Then, forecast combination techniques were applied by taking arithmetic means and medians of forecasts derived from individual VAR models. Finally, the performances of forecasting models were measured by comparing RMSE’s of forecasts with each other and a benchmark AR model. In conclusion, it was found that the indicators such as BIST100 index, real exchange rate and consumer price index had valuable information for short term forecasting of GDP. Besides it was seen that forecast combination methods were outperformed many individual forecasts.
Keywords: Business cycle, leading indicators, forecast, VAR
|