Hi,
I am trying to find an estimator and a corresponding stata command to help me estimate a count model (with lots of zeroes and really large counts) with multiple equations, and also two sets of fixed effects. Is there anything available in Stata to do so?
Alternately, I am trying to transform the DV into a normally distributed variable by using first differences , which also gets rid of one of the fixed effects. Is there a way to check the skewness/kurtosis of the resulting variable, with a "within" measure, using the other fixed effect? The summ <var>,d option does not take into account the fixed effects, and the xtsum command does not take the d option, so I cannot see what the within-group skewness/kurtosis is for the first-differenced variable.
Any help most appreciated, and this is my first post on this forum-my sincere apologies for any errors.
Best,
Mukund Chari.
Update: It looks like I need the Quasi Poisson Generalized Least Squares estimator, or the Semiparametric generalized least squares estimator, described in Cameron and Trivedi-Regression Analysis of Count Data , 2013, pp 325-327. Any Stata provided/user written programs that implement SUR using the above estimators?
I am trying to find an estimator and a corresponding stata command to help me estimate a count model (with lots of zeroes and really large counts) with multiple equations, and also two sets of fixed effects. Is there anything available in Stata to do so?
Alternately, I am trying to transform the DV into a normally distributed variable by using first differences , which also gets rid of one of the fixed effects. Is there a way to check the skewness/kurtosis of the resulting variable, with a "within" measure, using the other fixed effect? The summ <var>,d option does not take into account the fixed effects, and the xtsum command does not take the d option, so I cannot see what the within-group skewness/kurtosis is for the first-differenced variable.
Any help most appreciated, and this is my first post on this forum-my sincere apologies for any errors.
Best,
Mukund Chari.
Update: It looks like I need the Quasi Poisson Generalized Least Squares estimator, or the Semiparametric generalized least squares estimator, described in Cameron and Trivedi-Regression Analysis of Count Data , 2013, pp 325-327. Any Stata provided/user written programs that implement SUR using the above estimators?
