Hi. I'm running opreg with Stata11 on an unbalanced panel with 1155996 firm-level observations. I have: Value added as dependent var, Labour and Materials as free vars, Investment as proxy and Capital and Age as state vars. I built an 'exit' variable on the basis of the non-missing cases for Investment and ran the regression. Stata dropped 478957 obs due to missing values for the proxy but estimated the model.
As i'm not sure about my results, I tried inverting Capital and Investment. That is, Capital is the proxy and Investment one of the state vars. I changed the 'exit' var, connecting it to Capital. Now Stata tells me "an error occurred when bootstrap executed opreg" and reports r(2000), which means no observations. In this case 525229 obs with missing values were dropped.
What went wrong? I don't seem able to find anything similar reported anywhere; it looks as if nobody ever had the same problem. Am i making some very silly mistake?
Thanks for any suggestion