Applications of state space models in finance
Applications of state space models in finance
Applications of state space models in finance
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
6.2 Model<strong>in</strong>g conditional betas 97<br />
Table 6.6: Parameter estimates for Kalman filter <strong>models</strong>.<br />
sectors. All estimates for σ 2 , σ 2 η,<br />
This table reports the estimated parameters for the Kalman filter based <strong>models</strong> for the eighteen DJ Stoxx<br />
σ 2 ζ and ¯ β are significant at the 1% level. For φ and the test statistics that refer to the observation errors, *** means significance at the 1% level<br />
(**: 5%, *: 10%).<br />
Sector Model σ 2 × 10 4<br />
σ 2 η × 10 2<br />
σ 2 ζ × 10 2<br />
φ β¯ 2<br />
log L BIC R JB a<br />
Q(12) b<br />
LM(6) c<br />
Automobiles RW 3.20 0.80 2289.9 −5.09 0.72 39.65*** 13.63 23.15***<br />
MR 2.61 16.61 0.55*** 1.15 2304.5 −5.11 0.81 84.93*** 13.79* 13.02**<br />
MMR 2.50 19.96 0.22 0.01 2309.5 −5.12 0.82 115.11*** 14.26** 10.03<br />
GWR 0.02 0.25 4668.2 −10.39 0.69 9.60*** 14.01 4.30<br />
Banks RW 1.06 0.19 2792.9 −6.21 0.86 368.83*** 13.31 97.17***<br />
MR 0.75 9.38 0.41*** 1.03 2808.8 −6.23 0.92 1186.10*** 25.16*** 15.65**<br />
MMR 0.75 7.44 0.10 0.00 2825.1 −6.27 0.92 1052.70*** 21.14*** 13.50**<br />
GWR 0.02 0.07 4664.6 −10.39 0.85 186.35*** 17.28** 1.10<br />
Basics RW 3.22 0.47 2297.7 −5.11 0.62 417.28*** 25.31*** 116.97***<br />
MR 3.07 2.08 0.92*** 0.96 2303.6 −5.11 0.65 525.94*** 22.91*** 112.63***<br />
MMR 2.77 11.51 0.23 0.00 2308.5 −5.12 0.71 759.67*** 27.08*** 80.00***<br />
GWR 0.02 0.13 4665.4 −10.39 0.70 20.71*** 23.04*** 4.18<br />
Chemicals RW 1.85 0.23 2547.9 −5.67 0.73 287.48*** 16.34* 70.59***<br />
MR 1.78 0.89 0.94*** 0.91 2551.5 −5.66 0.75 360.87*** 15.81** 63.37***<br />
MMR 1.62 5.40 0.10 0.40* 2555.4 −5.67 0.79 578.40*** 20.31*** 52.85***<br />
GWR 0.02 0.05 4653.5 −10.36 0.73 28.35*** 19.46** 8.76<br />
Cont<strong>in</strong>ued on the next page. . .<br />
a JB is the Jarque-Bera statistic for test<strong>in</strong>g normality. The relevant critical values at the 95% (99%) level are 5.99 (9.21).<br />
b Q(l) is the test statistic <strong>of</strong> the Ljung-Box portmanteau test for the null hypothesis <strong>of</strong> no autocorrelation <strong>in</strong> the errors up to order l. In<br />
a structural model, the Q-statistic is asymptotically χ 2 distributed with l − w − 1 degrees <strong>of</strong> freedom, where w denotes the total number <strong>of</strong><br />
estimated parameters (Harvey 1989, p. 259).<br />
c LM(l) is the LM-statistic <strong>of</strong> Engle’s ARCH test for the null hypothesis <strong>of</strong> no ARCH effects up to order l. The test statistic is asymptotically<br />
χ 2 distributed with l degrees <strong>of</strong> freedom. The relevant critical values at the 95% (99%) level are 12.59 (16.81).