[R] Tobit Regression with unbalanced Panel Data

Vanessa Romero vanrome54 at gmail.com
Wed Jan 4 10:28:08 CET 2017


Hello,

I am doing Tobit Regression in R, because my dependent variable is censored
at 0. I have unbalanced panel data, for 6 years, 107 companies. I use
package CensReg.

I have imported my database(T1).

I use pdata.frame to specify the structure of my panel data. Like:


*mydata<- pdata.frame (T1, index = c("firm", "year")) *
Afterwards:

*Tob <- censReg(formula=Imp ~ Bath + CEOTurnover + ChangeOCF + E + Sales +
ROE + GTA + Size , data = mydata, method="BHHH") *
(as explained here:
https://cran.r-project.org/web/packages/censReg/vignettes/censReg.pdf)

I got here error message:


*Warnmeldung: In log(rEff$ercomp$sigma$id) : NaNs wurden erzeugt*

Another error message when *summary(Tob)*





*Call: censReg(formula = Imp ~ Bath + CEOTurnover + ChangeOCF + E + Sales +
ROE + GTA + Size, data = mydata, method = "BHHH") Observations: Total
Left-censored Uncensored Right-censored 606 469 137 0 Coefficients: Fehler
in printCoefmat(coef(x, logSigma = logSigma), digits = digits) : 'x' must
be coefficient matrix/data frame*

I am new to statistics and to R, what could be the problem or would you
suggest using other package.

Thank you,
Vanessa

	[[alternative HTML version deleted]]



More information about the R-help mailing list