Sample size for state space modeling
Posted: Tue Feb 07, 2017 7:28 am
Hello everyone,
I was wondering if some of you were aware of a "rule of thumb" concerning state space model's sample size?
I work on a monthly dataset, with 9 observations (inputs in my signal equations), and I have 37 parameters to estimate, I would like to employ a "rolling" SSPACE, but do not know if 3 years of data would be enough (9*12*3=324 obs). What do you reckon? Any materials treating that question would be much appreciated.
Thanks a lot in advance
I was wondering if some of you were aware of a "rule of thumb" concerning state space model's sample size?
I work on a monthly dataset, with 9 observations (inputs in my signal equations), and I have 37 parameters to estimate, I would like to employ a "rolling" SSPACE, but do not know if 3 years of data would be enough (9*12*3=324 obs). What do you reckon? Any materials treating that question would be much appreciated.
Thanks a lot in advance