Timing when specifying state-space model
Posted: Fri May 01, 2026 4:38 pm
Hi
Say I specify a state-space model in the following way:
@signal y = x + [ename=e1, var = (sigma(1)+dummy*sigma(2))^2]
@state x = x(-1) + [ename=e2, var = (sigma(1)+dummy*sigma(2))^2]
where dummy is 1 in period t1. Am I right in thinking that when "x" is written in a signal equation, it means "x(t)", but when "x" is written in a state equation, it means "x(t+1)"? But "dummy" always means "dummy(t)", whether it's in a state or signal eq?
So as written, the signal-equation dummy will affect y in period t1, but the state-equation dummy will affect x and y in period t1+1, even though it's the same variable?
Thanks!
Say I specify a state-space model in the following way:
@signal y = x + [ename=e1, var = (sigma(1)+dummy*sigma(2))^2]
@state x = x(-1) + [ename=e2, var = (sigma(1)+dummy*sigma(2))^2]
where dummy is 1 in period t1. Am I right in thinking that when "x" is written in a signal equation, it means "x(t)", but when "x" is written in a state equation, it means "x(t+1)"? But "dummy" always means "dummy(t)", whether it's in a state or signal eq?
So as written, the signal-equation dummy will affect y in period t1, but the state-equation dummy will affect x and y in period t1+1, even though it's the same variable?
Thanks!