Very high F-statistic in my Huber M-estimator regression

[ASHWINI SANKAR]

I have a panel dataset with 117 units (districts) from 1998 to 2007. I have district level fixed effects (FE), year FEs, and district-specific linear trends as controls in the model. I have a main independent variable, and also 3 other control variables in this model. When I ran the model with all 3 controls + main independent variable + all FEs, I get N=144, F( 75, 68) = 5226.51, and main coefficient = 0.02 (p=0.01). However, when I run this model with any 1 less control variable from the previous specification, I get a much smaller and more reasonable result. For example, for one such specification, I got N=158, F(81,76)=135, and the main independent variable coefficient was -0.013 (p=0.749). The first model mentioned above seems improbable to me, but I don't know how to formally ascertain that. Can you please help? Thank you very much for your help! Regards, Ashwini