RE: st: k-step Markov

["SJ Friederich, Economics" <S.Friederich@bristol.ac.uk>]

Dear all,

Re. my request below, and in case someone else on the List is interested, I 
realise from the code in xttrans.ado that a one-step Markov analysis 
involves little more in principle than something like:

. tab var F.var, row nof [nok]

Extending this to k-steps is therefore straightforward.

In practice, -tab- doesn't like TS operators, so one needs to create a 
separate variable. One also needs to take the panel dimension of the data 
into account (i.e. not use the last obs of a unit and the first one of the 
next unit, etc.) In principle, there is also the "missing time periods" 
issue discussed on p. 230 of [xt].

Sylvain


--On 05 March 2004 10:04 -0500 <mganz@hsph.harvard.edu> wrote:


> And as a follow-up, is there any Stata code to estimate worklife
> expectancy  using employment/mortality probability transistion data?
>
> Thanks,
> Michael
>
> > ===== Original Message From <S.Friederich@bristol.ac.uk> =====
> > Dear all,
> >
> > -markov- and -xttrans- produce one-step ahead transition probabilities.
> > Would anyone have worked on a generalisation to produce k-step
> > probabilities?
> >
> > Many thanks for advice or pointers.
> >
> > Sylvain
----------------------
Sylvain Friederich
Department of Economics, University of Bristol
8 Woodland Road
Bristol BS8 1TN
Tel +44(0)117 928-8425
Fax +44(0)117 928-8577


*
*   For searches and help try:
*   http://www.stata.com/support/faqs/res/findit.html
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/