Forecasting Oil Price by Hierarchical Shrinkage in Dynamic Parameter Models
Table 6
Measures of forecast performance for log return of Brent with the full models.
Constant variance
Stochastic variance
MAFE
MSFE
MLPL
MAFE
MSFE
MLPL
Model (h=1)
LASSO on constant and TVPs
0.072
0.009
2.886
0.088
0.013
2.126
LASSO only on constant coeff.
0.061
0.007
3.377
0.075
0.010
2.896
LASSO only on TVPs
0.112
0.021
1.970
0.107
0.018
1.696
TVP regression model
0.100
0.016
2.222
0.114
0.019
1.589
Constant coeff. model
0.100
0.016
2.189
0.106
0.017
1.658
Model (h=12)
LASSO on constant and TVPs
0.373
0.223
0.545
0.407
0.265
0.414
LASSO only on constant coeff.
0.389
0.224
0.468
0.463
0.340
0.325
LASSO only on TVPs
0.678
0.809
0.308
0.710
0.842
0.232
TVP regression model
0.648
0.707
0.295
0.709
0.833
0.226
Constant coeff. model
0.658
0.717
0.289
0.689
0.809
0.227
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.
