Short Term Forecasting System of Private Demand Components in Armenia
WP/14/3/08
Central Bank of Armenia
Working Paper 14/3/08
Short Term Forecasting System of Private Demand Components in Armenia
Narek Ghazaryan
Monetary Policy Department
April 2014
Abstract
In this paper a system for the short term forecasting of private consumption and private investments is described used in the CBA. Large amount of time series are used in the system to produce conditional forecasts, giving analysts the opportunity to use all the available information in real time for the assessment of the private demand dynamics before the official estimates are published by the statistical office. The main forecasting methods used are BVAR and FAVAR. In sample quasi-real time recursive forecast evaluation shows that pooled forecasts outperform individual model forecasts and conditioning improves forecast quality.
The analysis of the forecast errors confirms that BVAR and FAVAR models produce reliable forecast for 2-3 quarters and hence are good tools for nowcasting and near term analysis of private demand components in CBA.