I have panel data with a time-invariant variable across firms. However, when I tested the xtsum after setting the panel data, there was very little (8.66e-17) within variance. I checked over and over in the data, but it was invariant within every firm over the time period of 2005-2014.

How do I deal with this? Is there a way to see which observation is causing this variance, if I had missed any? If not, how can I explain this variance in the test results?

  • 1
    $\begingroup$ That will be zero to rounding/truncation error. It's not a matter of any observation "causing" it. $\endgroup$ – Glen_b Dec 23 '16 at 4:55

Your concern is that the variance is not precisely zero? That variance is effectively zero, especially if you're using floats rather than doubles to store the variable (I assume from your reference to "xtsum" that you're using Stata).

If the panel data was assembled, for example, from a series of cross-sectional data, and the data weren't read into Stata in exactly the same way for each period, I can imagine that you'd get some nonzero but negligible variation in the values of that variable within each panel. But it's awfully hard to say without having the data at hand.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.