Forum: help

RE: Stochastic Trandlog Ray frontier [ Reply ] By: Yvette Zen on 2017-05-16 07:38	[forum:45133]
sorry, i meant of course exogenous. thank you for your answer!! best, yvette

RE: Stochastic Trandlog Ray frontier [ Reply ]
By: Arne Henningsen on 2017-05-15 17:22

[forum:45130]

The 'Z variables' in the 'efficiency effects frontier' specification need to be *exogenous* (not endogenous).

The 'production frontier' should usually be defined by all inputs and all outputs, i.e. including both desirable 'good' outputs and undesirable 'bad' outputs. Hence, you need to use the undesirable 'bad' outputs in the production frontier specification and *not* as Z variables in the (in)efficiency model.

RE: Stochastic Trandlog Ray frontier [ Reply ] By: Yvette Zen on 2017-05-15 11:11	[forum:45129]
and i speak about usual SFA with Z variables of inefficiency

RE: Stochastic Trandlog Ray frontier [ Reply ] By: Yvette Zen on 2017-05-15 11:09	[forum:45128]
dear Mr. Henningsen, one small question, if you would be so nice. you mentioned that one of the option to take into account the "bad" output to include it in input. is it possible to put "bad" output from production in inefficiency variables Z? i know that these variables should be endogenous and not be a function of output, but may be it is still possible? best regards yvette

RE: Stochastic Trandlog Ray frontier [ Reply ] By: Yvette Zen on 2017-04-06 12:50	[forum:45082]
thank you, dear Mr. Hennnigsen, for your answer! ehhh - is about only positive outputs, but i supposed it. i have also read some works where undesirable output is made as input. thank you for your advice! best regards, yvette

RE: Stochastic Trandlog Ray frontier [ Reply ]
By: Arne Henningsen on 2017-04-06 08:41

[forum:45081]

Dear Yvette

When using the Translog Ray Production (Frontier) Function, all "outputs" need to be "positive" (desirable). However, you can model "undesirable" outputs as inputs. We model an undesirable output as an input in paper [1] (although in a DEA framework) and give arguments for this approach in section 4. Regarding some issues when estimating the Translog Ray Production (Frontier) Function, I highly recommend that you take a look at paper [2].

[1] http://dx.doi.org/10.5278/ijsepm.2015.7.7

[2] http://econpapers.repec.org/RePEc:foi:wpaper:2017_01

Best,
Arne

Stochastic Trandlog Ray frontier [ Reply ]
By: Yvette Zen on 2017-04-04 11:14

[forum:45075]

Dear Mr. Henningsen,

first of all, thank you for your work, especially package in R. it makes life of people that are not that smart, but try to be - much more easier!

May be you will be so polite to answer shortly, if in your package "Frontier" the function "translog ray froniter" suggests that it is possible to make as outputs negative and positive? (in your example with german farms you have two positive outputs; crop and animal). and then if it is possible, does the placement of negative or positive output play role?

thanks a lot!

best regards, yvette