Maximum likelihood invariably works! (if you believe it)
I seem to recall that with 2 or sevreal Markow matrices it ends up with
something that is asymptotically like the Wishart & Bartlett tests for
comparing covariance matrices. It should be fairly straight forward.
However, I would not believe it too much.
When comparing two or several social mobility matrices, which some people
viewed as Markovian, it was often useful to compare corresponding cells and
look at the pattern of signs - this often suggested useful patterns for
structuring the differences. THis might work in your case. (THe SRUCTURE of
differences is usually more informative and useful than the simpler question
of whether differences exist)
JOHN BIBBY
> -----Original Message-----
> From: A UK-based worldwide e-mail broadcast system mailing list
> [mailto:[log in to unmask]]On Behalf Of Mark Coleman
> Sent: 12 August 2005 14:57
> To: [log in to unmask]
> Subject: Comparing two Markov matrices
>
>
> Greetings,
>
> I'm working on a problem in quantitative finance and I was hoping I
> might get some ideas about how to conduct a certain statistical test
> comparing two Markov matrices. In particular, can anyone point me to
> a statistical test of equality for two n x n Markov matrices?
>
> Thanks,
>
> -Mark
>
> --
> No virus found in this incoming message.
> Checked by AVG Anti-Virus.
> Version: 7.0.338 / Virus Database: 267.10.8/71 - Release Date: 12/08/2005
>
--
No virus found in this outgoing message.
Checked by AVG Anti-Virus.
Version: 7.0.338 / Virus Database: 267.10.8/71 - Release Date: 12/08/2005
|