Forecasting Oil Price by Hierarchical Shrinkage in Dynamic Parameter Models
Table 11
Measures of forecast performance for log return of Brent with WTI forecasting.
Constant variance
Stochastic variance
MAFE
MSFE
MLPL
MAFE
MSFE
MLPL
Model (h=1)
LASSO on constant and TVPs
0.075
0.009
2.629
0.093
0.014
1.883
LASSO only on constant coeff.
0.063
0.007
3.131
0.080
0.011
2.476
LASSO only on TVPs
0.137
0.029
1.626
0.129
0.029
1.390
TVP regression model
0.115
0.022
1.891
0.111
0.021
1.413
Constant coeff. model
0.114
0.022
1.912
0.112
0.021
1.432
Model (h=12)
LASSO on constant and TVPs
0.333
0.202
0.612
0.388
0.265
0.422
LASSO only on constant coeff.
0.340
0.192
0.529
0.482
0.427
0.419
LASSO only on TVPs
0.642
0.732
0.308
0.696
0.859
0.218
TVP regression model
0.630
0.766
0.291
0.651
0.812
0.208
Constant coeff. model
0.629
0.771
0.293
0.673
0.856
0.204
Note. The value noted in bold and underlined text indicates a model performing the best out of all models, while the bold and italic text represents a model performing the worst. MSFE, MAFE, and MLPL refer to the mean squared forecast error, mean absolute forecast error, and mean log predictive likelihood, respectively.
