Discussion Paper
No. 2019-22 |
March 08, 2019
VC - A method for estimating time-varying coefficients in linear models
Abstract
This paper describes a moments estimator for a standard state-space model with coefficients generated by a random walk. This estimator does not require that disturbances are normally distributed, but if they are, the proposed estimator is asymptotically equivalent to the maximum likelihood estimator.
Comments and Questions
See attached file
Thank you for your kind comments, for using VC and for sharing your experience!
In the meanwhile I became aware of some errata. I attach the file with the errata marked and hyperlinked with apologies especially to the referees.
I found another mistake: I gave again wrong dimensions for the vectors v and v_i. I attach a corrected version with my apologies.
If a reader or a referee finds some problems like that, you may ask in an anonymous comment and I will quickly respond (and correct, if necessary).