How to fix error r(1400) when running XTPMG

[Ngoc Anh Nguyen]

Dear Statalists,

I currently have a panel data including N = 9, T = 21.

Dependent variable is Log(FDI), 4 control variables also transformed in logs and 1 key explanatory variable.

So my sample is actually quite small. However, when I run the command xtpmg I encounter the error r(1400), the message is ''initial values not feasible''

I have tried to reduce the numbers of explanatory variables to give more degree of freedom but it does not help. Don't quite know how to fix this problem now.

So please give me some advices if you know how to fix it.

Thanks a lot and wish you all a great day.

Anh Nguyen.

[Phil Bromiley]

You didn't get a quick answer. You'll increase your chances of a helpful answer by following the FAQ on asking questions - provide Stata code in code delimiters, readable Stata output and sample data using dataex. We don't even know exactly what you ran.

Do any of your variables have zero as a legitimate observation in the original data set? If so, logging things artificially creates missing values. Some suggest taking the square root or cube root.

The documentation for xtpgm explicitly says it is for data with large N and large T. You have small N. Whether this creates the problem, I don't know. It does raise questions why you'd use this technique.

Diagnosing "initial values not feasible" is almost impossible without being able to replicate the problem. What happens if you use xtreg on these data? Sometimes this will reveal a data problem.