Working Paper: CEPR ID: DP6746
Authors: Elena Angelini; Gonzalo Cambamendez; Domenico Giannone; Lucrezia Reichlin; Gerhard Rünstler
Abstract: This paper evaluates models that exploit timely monthly releases to compute early estimates of current quarter GDP (now-casting) in the euro area. We compare traditional methods used at institutions with a new method proposed by Giannone, Reichlin and Small, 2005. The method consists in bridging quarterly GDP with monthly data via a regression on factors extracted from a large panel of monthly series with different publication lags. We show that bridging via factors produces more accurate estimates than traditional bridge equations. We also show that survey data and other `soft' information are valuable for now-casting.
Keywords: Factor model; Forecasting; Large datasets; Monetary policy; News; Real time data
JEL Codes: C33; C53; E52
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| monthly economic indicators (E66) | quarterly GDP growth (E20) |
| bridging approach with factors (C38) | accurate estimates of GDP growth (E20) |
| soft data early in the quarter (C88) | forecasting performance (C53) |
| hard data availability later in the quarter (C82) | accuracy of forecasts (C53) |