| Abstract |
: |
Gross Domestic Product (GDP) growth is one of the most important economic indicators since GDP figures provide comprehensive information regarding the economic activity. GDP data are published with considerable delay, so early estimates of GDP growth may be valuable. For this aim, we use an extended version of the Stock and Watson coincident indicator model that can deal with mixed frequency (such as quarterly and monthly variables), ragged ends (some indicators are published before others), and missing data (data may not be available at the beginning of the sample for some variables). As soft data we use PMI, and as hard data we use industrial production, import and export quantity indices. We perform simulated out of sample forecasting exercise by taking the flow of data releases for 2008Q1-2012Q2 into account. Results show that nowcasts obtained with a model including a soft indicator tracks the GDP growth relatively successfully. Also, the model outperforms benchmark AR model. |
