Page 1 of 1

state space model

Posted: Wed Apr 06, 2011 3:12 am
by iloverats
:wink:
y_t=x1_t*b1_t+x2*b2+u_t,
b1_t=beta*b1_t-1+v_t,

how can i use dlm to estimate this state sapce model?

thank you

Re: state space model

Posted: Wed Apr 06, 2011 10:30 am
by TomDoan
iloverats wrote::wink:
y_t=x1_t*b1_t+x2*b2+u_t,
b1_t=beta*b1_t-1+v_t,

how can i use dlm to estimate this state sapce model?

thank you
That's a partially fixed, partially varying TVP model. You might want to check out:

http://www.estima.com/forum/viewtopic.php?f=5&t=631

http://www.estima.com/forum/viewtopic.php?f=5&t=932

The simplest way to handle it is to also treat B2 as a state, with a unit coefficient in the state matrix and a zero variance. So your A would be ||beta,0.0|1.0,0.0|| and SW would be ||var(v)|0.0,0.0||