State space model - Three factor model
Posted: Tue Nov 03, 2015 10:22 am
I am trying to estimate a state space model (three factor model) in Eviews 7 but I am having a hard time defining the state equations.
The signal equation is: i(t) = a + B*x(t) + n(t)
Where "i(t)" is a Qx1 vector of observed data; "a" is a Qx1 coefficient vector; "B" is Qx3 coefficient matrix; "x(t)" is a 3x1 state variable vector; "n(t)" is a 3x1 gaussian error vector.
The state equation is: x(t) = exp(K)*x(t-1) + z(t)
Where, "K" is a 3x3 lower triangular coefficient Matrix and "z(t)" is 3x1 error vector with variance equal to ∫Exp(Ks)*H*H'*exp(K')ds.
Where Exp(k) = ∑[(1/N!)*(K^N)]; and "H" is a diagonal matrix.
How can I define this transition equation in th Eviews?
**I have attached the equations from the original working paper.
The signal equation is: i(t) = a + B*x(t) + n(t)
Where "i(t)" is a Qx1 vector of observed data; "a" is a Qx1 coefficient vector; "B" is Qx3 coefficient matrix; "x(t)" is a 3x1 state variable vector; "n(t)" is a 3x1 gaussian error vector.
The state equation is: x(t) = exp(K)*x(t-1) + z(t)
Where, "K" is a 3x3 lower triangular coefficient Matrix and "z(t)" is 3x1 error vector with variance equal to ∫Exp(Ks)*H*H'*exp(K')ds.
Where Exp(k) = ∑[(1/N!)*(K^N)]; and "H" is a diagonal matrix.
How can I define this transition equation in th Eviews?
**I have attached the equations from the original working paper.