Forum: help

RE: SFA [ Reply ] By: Jennifer Beckensteiner on 2020-06-15 16:26	[forum:47848]
Dear Arne, I am using a Cobb-Douglas Production frontier with Z variables. I would like to include an input in my SFA that is a ratio. Does this need to be logged as other input variables? Thanks, Jennifer

RE: SFA [ Reply ]
By: Arne Henningsen on 2019-11-03 14:03

[forum:47117]

Regarding your first question: the SFA models with and without Z values have different specification of the inefficiency terms and, thus, give different efficiency estimates. The two types of models give the same results only if the Z variables are orthogonal to the X variables and the (estimated) coefficients of the Z variables are zero.

Regarding your second question: There should be efficiency estimates for all observations that were used in the estimation. If you post a minimal reproducible example that illustrates the problem, I can try to find out the problem (and fix the"frontier" package if this is caused by a bug in the package).

SFA [ Reply ]
By: Jennifer Beckensteiner on 2019-10-23 18:49

[forum:47060]

Dear Arne,

I am starting to use SFA BC95 in R with your package, I have two questions related to a production frontier, with and without Z variables included.

- Why are efficiency scores lower when including Z variables compared to the specification without Z variable (but same input set Xs/ output Y) ?

- For a specification of the model without Z variables, how is possible that the SFA function is not calculating any efficiencies for some DMUs while I do not have any NA's neither in my Y, neither in my Xs?

Thanks for any help on these two subjects,
Jennifer