Table 1 Pseudo-real-time evaluation. Root-mean-squared errors (RMSE) for forecasts on days with a new quarterly GDP release. A lower RMSE implies higher predictive accuracy. h=0 (h=1) denotes the forecast for the current (next) quarter. We use three benchmarks. First, we use the first quarterly release of the corresponding quarter (panel a). Second, we use an AR(1) model (panel b). Third, we use the KOF Economic Barometer (panel c). The Diebold-Mariano-West (DMW) test provides a p value for the null hypothesis of equal predictive accuracy against the alternative written in the column header (Diebold and Mariano 2002; West 1996). We assume a quadratic loss function
(a) Real GDP growth: First release vs. f-curve | ||||
|---|---|---|---|---|
RMSE | RMSE | Relative RMSE | DMW test (p value) | |
First release | f-curve | First release/f-curve | First release <f-curve | |
h=0 | 0.45 | 0.57 | 0.79 | 0.177 |
h=1 | 0.45 | 0.7 | 0.64 | 0.042 |
(b) Real GDP growth: f-curve vs. AR(1) | ||||
RMSE | RMSE | Relative RMSE | DMW test (p value) | |
f-curve | AR(1) | f-curve/AR(1) | f-curve < AR(1) | |
h=0 | 0.57 | 0.7 | 0.82 | 0.039 |
h=1 | 0.7 | 0.7 | 0.99 | 0.458 |
(c) Real GDP growth: f-curve vs. KOF Economic Barometer | ||||
RMSE | RMSE | Relative RMSE | DMW test (p value) | |
f-curve | Barometer | f-curve/Barometer | f-curve < Barometer | |
h=0 | 0.57 | 0.6 | 0.95 | 0.162 |
h=1 | 0.7 | 0.65 | 1.07 | 0.832 |