Table 10.
Recursive (expanding) estimation window.
| Forecasting models | QLIKE |
MSE |
||
|---|---|---|---|---|
| Range | SeimQ | Range | SeimQ | |
| Panel A: h = 1 | ||||
| MIDAS-RV | 0.1800 | 0.1729 | 0.2313 | 0.3566 |
| MIDAS-RV-OVX | 0.4488 | 0.4488 | 0.3444 | 0.3566 |
| MS-MIDAS-RV | 0.1800 | 0.1158 | 0.1977 | 0.3566 |
| MS-MIDAS-RV-OVX | 1.0000 | 1.0000 | 1.0000 | 1.0000 |
| Panel B: h = 5 | ||||
| MIDAS-RV | 0.0480 | 0.0769 | 0.1916 | 0.2481 |
| MIDAS-RV-OVX | 0.1348 | 0.2909 | 0.1916 | 0.2481 |
| MS-MIDAS-RV | 0.4190 | 0.4190 | 0.3071 | 0.3071 |
| MS-MIDAS-RV-OVX | 1.0000 | 1.0000 | 1.0000 | 1.0000 |
| Panel C: h = 20 | ||||
| MIDAS-RV | 0.0352 | 0.0380 | 0.1591 | 0.1163 |
| MIDAS-RV-OVX | 0.0202 | 0.0209 | 0.3483 | 0.4244 |
| MS-MIDAS-RV | 0.1343 | 0.1343 | 0.3483 | 0.4244 |
| MS-MIDAS-RV-OVX | 1.0000 | 1.0000 | 1.0000 | 1.0000 |
Notes: In this part, we apply recursive (expanding) estimation window to do the evaluation. The forecasts for different horizons h = 1, 5, 20 are shown in panel A, B and C, and the predictive results are shown in different columns. In this table, the MCS test p-value of 1indicates a better prediction.