NaN results for covariance

[AlexandreSollaci]

Hello,

I am finding a NaN result for the covariance between firm and person effect (likely due to the plug-in variance being = 0; see below). I'm not really sure what might be happening, as the program does run as expected with the test data, while the actual data has shown no problems in other programs (e.g., lfe in R, or reghdfe in Stata). Any guesses on what might be going wrong?

Plug-in Variance Components:
Plug-in Variance of firm Effects: 0.0
Plug-in Variance of person Effects: 0.0027085154074460043
Plug-in Covariance of person-firm Effects: 0.0
Correlation of person-firm Effects: NaN
Fraction of Variance explained by person-firm Effects: 0.0

Bias-Corrected Variance Components:
Bias-Corrected Variance of firm Effects: 0.0
Bias-Corrected Variance of person Effects: 0.001744772197125726
Bias-Corrected Covariance of person-firm Effects: 0.0
Bias-Corrected Correlation of person-firm Effects: NaN
Bias-Corrected Fraction of Variance explained by person-firm Effects: 0.12032312957523913