The RATS Software Forum

hello,
I am interested in estimating a state space model of the form:

measurement equation: y'(t) = y(t) + u(t)
state equation: ||y(t),a(t),b(t)|| = ||m1,m2,m3||' + ||c1,c2,c3|0,c4,c5|0,0,c6||*||y(t-1),a(t-1),b(t-1)||'+||e1,e2,e3||'
ie. in my state equation the variables a(t) and b(t) are observed while the latent is the y(t). How can I set that the state variables are observed?
(that model is for example used in (Andersen et al 2005: A framework for exploring the macroeconomic determinants of systematic risk" American Economic Review).
thank you

manos_nko wrote:hello,
I am interested in estimating a state space model of the form:

measurement equation: y'(t) = y(t) + u(t)
state equation: ||y(t),a(t),b(t)|| = ||m1,m2,m3||' + ||c1,c2,c3|0,c4,c5|0,0,c6||*||y(t-1),a(t-1),b(t-1)||'+||e1,e2,e3||'
ie. in my state equation the variables a(t) and b(t) are observed while the latent is the y(t). How can I set that the state variables are observed?
(that model is for example used in (Andersen et al 2005: A framework for exploring the macroeconomic determinants of systematic risk" American Economic Review).
thank you

You have A=||c1,c2,c3|0,c4,c5|0,0,c6||,Z=||m1,m2,m3||,SW=covariance matrix of ||e1,e2,e3|| which describes your states. Your data are y=||yprime,a,b|| with C=%identity(3). You show that a and b are observed by making the measurement covariance matrix sv=||sigmausq|0.0,0.0|0.0,0.0,0.0||.