stochastic volatility using kalman filter

[saizal87]

Hi, i m interested in estimating stochastic volatility of stock return (y) using kalman filter. But i m not so sure with the process. Currently im using this coding from the internet:
@signal y = -1.27 + s + [var=4.9348]
@state s = c(1) + c(2)*s(-1)+[var=(3)*2]
param c(1) -10.8720 c(2) 0.2736 c(3) 4.9532

where;
-10.8720 = parameter for constant through ols
0.2736 = y(-1) parameter through ols
4.9532 = s.e of ols egression.

Can anyone explain either -1.27 and 4.9348 are standard for kalman. And what does it represent? When i estimate, i have insignificant (c1, c2 n c3)! estimation, does it suggests that the model does not hav stochastic volatility? I am new with eviews and i really hope that someone could help me. Thanks in advance. Saizal.

[trubador]

You should be asking these questions to the author of the code.

If you had done a search in the forum, you would have found the following: http://forums.eviews.com/viewtopic.php? ... 744#p24051

[terrya]

Hi, i m interested in estimating stochastic volatility of stock return (y) using kalman filter. But i m not so sure with the process. Currently im using this coding from the internet:
@signal y = -1.27 + s + [var=4.9348]
@state s = c(1) + c(2)*s(-1)+[var=(3)*2]
param c(1) -10.8720 c(2) 0.2736 c(3) 4.9532

where;
-10.8720 = parameter for constant through ols
0.2736 = y(-1) parameter through ols
4.9532 = s.e of ols egression.

Can anyone explain either -1.27 and 4.9348 are standard for kalman. And what does it represent? When i estimate, i have insignificant (c1, c2 n c3)! estimation, does it suggests that the model does not hav stochastic volatility? I am new with eviews and i really hope that someone could help me. Thanks in advance. Saizal.

Hi

You need to look at the paper by Harvey, Ruiz and Shephard, Multivariate stochastic variance models, Review of Economic Studies, 1994, Vol 61, 247-264, to find the reasoning for the formulation.

[terrya]

Hi, i m interested in estimating stochastic volatility of stock return (y) using kalman filter. But i m not so sure with the process. Currently im using this coding from the internet:
@signal y = -1.27 + s + [var=4.9348]
@state s = c(1) + c(2)*s(-1)+[var=(3)*2]
param c(1) -10.8720 c(2) 0.2736 c(3) 4.9532

where;
-10.8720 = parameter for constant through ols
0.2736 = y(-1) parameter through ols
4.9532 = s.e of ols egression.

Can anyone explain either -1.27 and 4.9348 are standard for kalman. And what does it represent? When i estimate, i have insignificant (c1, c2 n c3)! estimation, does it suggests that the model does not hav stochastic volatility? I am new with eviews and i really hope that someone could help me. Thanks in advance. Saizal.

Trubador's approach works. I've used a variant of his in the past. However, I estimated the starting values of the parameters by using a Box-Jenkins AR1 model.

[saizal87]

Thanks