Well I am trying to estimate a translog cost function for banking competition, however two of my explanatory variables are highly correlated (lets say greater than 90%). How to deal with it?
One approach suggested is to go for the GLS estimation, however that does not work for me.
Dear Abdur,
If the correlation is between linear and square terms then expressing variables in mean deviations could help mitigate collinearity.
In general, in a translog function I would not mind dropping one of the non linear terms if they are highly correlted with linear ones.
Best,
Piero
It would help to know the variables you were including. Moreover, it would help to know what you are trying to estimate beyond cost functions. Specifically, what is the connection between cost and competition that you mentioned in your question? I am not trying to be critical; instead, this will really help in offering suggestions.
If you get a chance, you might wish to take a look at our paper on "economies of scale and scope in Kansas banking" ... using translog functions. We think when economies of scale &/or economies of scope can be realized at relatively small scales of production, as we found, it opens the opportunity for more competition. However, I am not sure how you are attempting to include competition in your translog function. This is fascinating to me.
I hope this helps.
Rob
- Badges
- Science topic
- Similar topics
- Linguistics
- Composition Studies
- Publication