Cross-section vs time fixed effects

[Jandh]

Hey,

I'm investigating whether the capital structure of 29 chosen firms changed during the financial crisis with a dataset from 2006 until 2013.
I ran the Hausmann test, revealing that the fixed effects model was more efficient than the random with a p-value less than 1%.
However, I wonder how I can choose between cross-section and/or time fixed effects?
I've googled around and read in textbooks. My overall impression is two ways of "testing" this;
1) Through the Maximum Likelihood (..) test (above the Hausmann test alternative): With this test, I got that cross-section fixed was significant with a p-value of zero, period fixed was significant at the 5% level, and both cross-section&time were significant with a p-value of zero. Does this mean that I should use a model with both time and cross-section fixed features? Be aware that I did not create any dummies..
2) The other way I've tried to sort this out, is through dummy variables. I saw a clip on youtube, and the guy in the movie wrote the regression equation as, say, leverage=c(1)+c(2)*size+c(3)*profitability+c(4)*D1+.. (as would be some of the variables in my regression), but he did not choose fixed effects in the "panel option" tab. Is this correct? Further, he tested whether the dummies all were zero in a Wald test. With a p-value of zero the fixed effects model should again be chosen. But I still don't understand whether time and/or cross-section should be chosen! Can you please help?


In advance, thank you!