The Volatility Forecasting Power of Financial Network Analysis
Table 6
Forecasts of realized volatility of North and Latin American stock market indices using out-of-sample analysis with monthly data (π ≡ P/R = 0.4).
(1)
(2)
(3)
(4)
(5)
(6)
(7)
(8)
(9)
Benchmark model
S&P 500—USA
Nasdaq—USA
Toronto composite—Canada
BMV IPC—Mexico
IBOVESPA—Brazil
IPSA—Chile
MERVAL—Argentina
IGBVL—Peru
Panel A VPMFGL model
AR (3)
−0.401
−0.435
−0.147
−0.567
0.053
−0.250
0.132
−0.814
AR (6)
−0.425
−0.517
−0.201
−0.441
0.046
−0.580
−0.030
−0.816
AR (3) VVIX(1)
0.358
0.284
−0.152
0.766
3.278
−0.708
0.738
−0.215
AR(6) VVIX(1)
0.631
0.732
−0.020
1.577
2.922
−0.916
1.513
−0.118
Panel B VMSTL model
AR(3)
−0.330
−0.330
−0.036
−0.641
−0.081
−0.479
−0.002
−0.891
AR(6)
−0.294
−0.439
−0.102
−0.494
−0.020
−0.843
−0.154
−0.880
AR(3) VVIX(1)
−0.442
−0.327
−0.286
0.470
2.794
−0.894
0.763
−0.202
AR(6) VVIX(1)
−0.270
0.041
−0.207
1.265
2.695
−1.100
1.448
−0.205
10%, 5%, and 1% critical values are 0.685, 1.079, and 2.098, respectively, when there is only one excess parameter. P represents the number of one-step-ahead forecasts and R the sample size of the first estimation window. The AR (3)-VVIX (1) benchmark corresponds to model 1. ,, and . Source: authors’ elaboration.