Dummy Variables in Cross-Country Regressions

I post the following at the request of a current student at Marquette.  If you have any insights for him, please post a comment or email me at jason.czarnezki@marquette.edu.

"I am writing a paper on economic growth and want to look at the effects, if any, that the mixture of civil law and common law in Louisiana has had on that state's rate of growth.  Theory and past research using cross-country regressions show that nations with civil law origins have lower rates of economic growth, all else constant.  I want to use a cross-state comparison (LA, MS, AL, and AR) and include a variable that captures the affects of LA's mixed legal system of the state's rate of growth.  Most of the research I have looked at uses a simple binary choice dummy variable to distinguish the legal origin.  However, as you probably know, this dummy can pick up influence from any number of variables not included in the model and that are unrelated to legal origin.  Therefore, I have been trying to come up with some way to capture the legal origin distinction, but with a proxy variable that provides a greater quantitative distinction for legal origin than the dummy.  Some quantifiable measure that would theoretically move with the ebb and flow of civil law influence on the state's legal system, which has waxed and waned since LA became a state.  There may be no real good solution for this problem (which I imagine is why dummy variables exist in the first place), but I would appreciate greatly any helpful comments or ideas." 

Comments

[Alex]

How about percentage French speakers?

[Sara Benesh]

This may be silly, but would size of the state's code disentagle this? (Expecting that civil law codes will be longer...) I don't know much though, so feel free to laugh at me! :)