Tobit is often used because efficiencies are on a defined range 0 to 100% so least squares regression as a second stage does not make sense as the noise term can not result in values above 100 or less than 0.

However, doing the analysis in multiple stages is itself problematic.

A simple google scholar search on the terms

Tobit Regression in Second stage DEA Banks

Results in both example papers using this method on Banks and the papers explaining the potential problems.

https://scholar.google.co.uk/scholar?hl=en&as_sdt=0%2C5&as_vis=1&q=Tobit+Regression+in+Second+stage+DEA+banks&btnG=

Brilliant.... thanks for your valuable contribution on new research.

Dr. Paul explained well, I agree with him.