Quarters ahead | 0 | 1 | 2 | 3 | 4 |
---|
GDP AR(1) | 1.550 | 2.013 | 2.276 | 2.473 | 2.461 |
FLUR-Okun | 1.183 | 1.800 | 2.073 | 2.216 | 2.203 |
(Diebold-Mariano z-test) | (− 1.178) | (− 1.742**) | (− 2.021***) | (− 1.553*) | (− 1.495*) |
SNB-BCI | 1.218 | 1.472 | 1.878 | 2.167 | 2.389 |
(Diebold-Mariano z-test) | (− 1.357*) | (− 2.106***) | (− 2.329***) | (− 1.807**) | (− 0.040) |
KOF Barometer | 1.567 | 1.774 | 1.961 | 2.129 | 2.216 |
(Diebold-Mariano z-test) | (0.084) | (− 1.107) | (− 1.572*) | (− 1.829**) | (− 1.900**) |
- Estimation period starts in 1991/Q1–1997/Q4 and increases iteratively up to 2021/Q4
- The Diebold and Mariano (1995) statistic is asymptotically normally distributed under H0
- GDP and FLUR-Okun, seasonally unadjusted; GDP, SNB-BCI and KOF Barometer, revised figures; RMSE, root-mean-squared error; H0, the difference resulting from subtracting the RMSE of an AR(1) model of GDP growth from that of the given leading indicator is zero, implying that the leading indicator provides no improvement over the AR(1) model; H1, that difference is negative, implying that the leading indicator is superior to an AR(1) of GDP growth
- *, **, *** denote rejection of H0 at the 10%, 5% and 1% level of statistical significance, respectively