Re: st: Re: Finding initial feasible values for xtfrontier, ti cost version, please help!

[Scott Merryman <scott.merryman@gmail.com>]

On Fri, Mar 26, 2010 at 1:32 PM, Chong, Qi Lin Andrew
<qchong@middlebury.edu> wrote:
> Hi all, many thanks to Rogelio and Scott for the tips about Winbugs and xtfrontier in STATA.
>
> To Scott, for the xtfrontier, would you know if i am using a translog function model of the form:
>
> xtfrontier logexpense logsum logpdep logpop logsum2 logdeppop logsumdep logsumpop, ti cost difficult from(b0, copy)
>
> logexpense = cost of firm
> logsum = total output of firm
> where logpdep = price of deposits (an input)
> logpop = price of operational input
> logsum2 = logsum*logsum
> logdepop = price of dep*price of operational input
> logsumdep = logsum*logdep
> logsumpop = logsum*logpop
>
> where I basically list all outputs, all input prices, all outputs interacted all outputs interacted with each other (in pairs), listing all input prices interacted with each other (in pairs), and then all possible pairings between outputs and input prices.
>
> 1) Should I put all the variables after logsum2 into the original regression as well?

If you want to use a translog function then you need .5 * the square
terms plus all the interaction terms.

> 2) How did you get the value of 1 of the -mu term?

I just a chose a value - though 0 might be a better starting point.

> 3) I've run the specifications with the addition of 1 and with all the variables above put into the original regression eqn to find initial values and into the xtfrontier command. I have been receiving equal log likelihoods of the same value (and "backed up) for many iterations. Should I just let it stay running? Is there anything I can do if I fail to get a solution that converges?

You could try different algorithms by using the -technique()- option,
such as -tech(nr 10  dfp 10 bfgs 10)-

Scott

*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/