Statalisters,
After ample attempts at constructing a nlsur quaids function evaluator program to estimate a QUAIDS demand model with demographic scaling and censoring and even more browsing through the statalist, I have hit a brick wall, particularly with the demographic scaling. I would like to apply the demographic scaling approach applied by Poi (in Stata Journal (2012) 12, No. 3) but am not able to estimate the parameter rho (ρ) needed to calculate the mbar function which is necessary to scale total expenditure.
The main question is how can one arrive at rho within the framework of nlsur? Having studied the ado files for Stata's -quaids command I can see it produces a 1xn matrix (where n = number of demographic variables included) for rho, yet my mata skills are not strong enough to figure out how this matrix is populated. It should be noted that this is different from the n x g matrix for the eta's (where g is the number of goods). Unfortunately all of the examples included in this string of posts estimate the eta's yet ignore rho and none have actually taken the steps necessary to estimate the cofp and mbar functions needed for the scaling.
I would appreciate any help or suggestions you may provide.
Many thanks.
After ample attempts at constructing a nlsur quaids function evaluator program to estimate a QUAIDS demand model with demographic scaling and censoring and even more browsing through the statalist, I have hit a brick wall, particularly with the demographic scaling. I would like to apply the demographic scaling approach applied by Poi (in Stata Journal (2012) 12, No. 3) but am not able to estimate the parameter rho (ρ) needed to calculate the mbar function which is necessary to scale total expenditure.
The main question is how can one arrive at rho within the framework of nlsur? Having studied the ado files for Stata's -quaids command I can see it produces a 1xn matrix (where n = number of demographic variables included) for rho, yet my mata skills are not strong enough to figure out how this matrix is populated. It should be noted that this is different from the n x g matrix for the eta's (where g is the number of goods). Unfortunately all of the examples included in this string of posts estimate the eta's yet ignore rho and none have actually taken the steps necessary to estimate the cofp and mbar functions needed for the scaling.
I would appreciate any help or suggestions you may provide.
Many thanks.
Comment