Evaluating the forecast quality of GDP components

We assess and compare the quality of forecasts issued for Portugal, at several time spans. Our analysis, covering the 2002-2010 period, focuses on real GDP growth and the corresponding expenditure components. We use a scaled statistic to compare the forecast accuracy of GDP components with different volatility levels, and explore the contributions of expenditure components to the GDP forecast error. Moreover, we propose two new statistics – termed Mean of Total Weighted Absolute Error and Mean of Total Weighted Squared Error – to evaluate the overall accuracy of components' predictions. The results suggest that GDP forecasts are generally optimistic at longer horizons (1-year ahead predictions), mainly due to overly optimistic forecasts in investment and exports. At shorter horizons (same-year predictions), GDP forecasts are more accurate, but this is achieved with relatively large errors in components' predictions, whose effects tend to cancel out.