Unbalanced panel and high p value when running Hausman test

[Emma Anderson]

I have a dataset of 22 variables, including my identifier variables for panel regression and roughly 10000 observations.

I set my dataset as panel.

xtset hid year
panel variable: hid (unbalanced)
time variable: year, 2012 to 2016, but with gaps
delta: 1 unit

I then ran regress for the fixed effects and random effects model and then proceeded to run the Hausman test.

I ran this using the following code:

hausman fe re

Below this code it is noted

Note: the rank of the differenced variance matrix (16) does not equal the number of coefficient being tested (17); be sure this is what you expect, or there may be problems computing the test. Examine the output of your estimators for anything unexpected and possibly consider scaling your variables so that the coefficients are on a similar scale.

I do not know how to interpret this note and I am unaware if this has an effect on the results and the comments following.

The results are returned in a table however below it is noted:

b = consistent under Ho and Ha; obtained from xtreg
B = inconsistent under Ha, efficient under Ho: obtained from xtreg

Test: Ho: difference in coefficients not systematic

chi2(16)=(b-B)'[(V_b-V_B)^(-1)](b-B)= 16.57
prob>chi2= 0.4140

From prob=0.4140 I would have to accept the null hypothesis since this value is greater than 0.05 and therefore choose the random model however I feel the fixed effects model is more appropriate. Does the fact that my panel is unbalanced and that the rank of the variance matrix vs the number of coefficients being tested is different have an effect on this result? Or is it safe to assume this is accurate and accept the null hypothesis of the random model?

[Carlo Lazzaro]

Emma:
-hausman- comppains quite often when applied to real data.
That said:
- -hausman- outcome points you toward -re- specification;
- feeling that a -fe- specification is better than another one (despite the -hausman- indication) is something that I find difficult to comment on;
- perhaps you can give it one more try with the community-contributed module -xtoverid- (type -search xtoverid- to spot and install it).

[Emma Anderson]

thank you for your help