View Full Version : new distribution
03-09-2011, 09:51 AM
Hi, I am an Italian researcher in Economics. I need to know if it is possible introduce new and different distribution for the error component in frontier model with panel data. I think that it is necessary write some instruction. How can I do this?
I hope someone can help me.
03-09-2011, 10:34 AM
what you mean by "new and different" distribution? One can always construct a model with any distribution.
03-09-2011, 10:51 AM
Dear Vinux, thank you for your prompt reply.
The issue is that xtfrontier estimates cost/production function forcing the error components to be normally distributed. My interest is to estimate a cost function and modelling the errors with different distributions by using the copula approach. At this stage of my knowledge of stata, I realize that the only useful command is xftrontier with the above limitation to run regressions assuming normal distribution.
I hope to be clearer than before, in the sense that new distribution states for distribution other than normal.
Thank you in advance for your cooperation in this matter.
Powered by vBulletin™ Version 4.1.3 Copyright © 2013 vBulletin Solutions, Inc. All rights reserved.