Random Effects Model +Autocorrelation
Posted: Sat Nov 21, 2009 12:15 pm
Hey guys,
I have a question for you all. I did a hausman test on my model and the p-value was 0.08. This is a very low p-value, so i did not know whether i should use a fixed or random effects model. Given that i have 30 cross sections and only 6 time periods I chose random. I also chose random because it was only one variable that drove the difference in the point estimates. This variable is in the model not even stat. significant.
Anyways, I tested for autocorrolation and my model suffers from it. I guess the idiosyncratic errorterm is corrolated. How do i correct for this in the random effects model? I thought about adding a variable called Uhat(-1) with Uhat=Resid.
It was very significant when i put it in. However, i do not know if i should since the random effects model is not supposed to have autocorrolation i guess.
How would you guys deal with this?
I have a question for you all. I did a hausman test on my model and the p-value was 0.08. This is a very low p-value, so i did not know whether i should use a fixed or random effects model. Given that i have 30 cross sections and only 6 time periods I chose random. I also chose random because it was only one variable that drove the difference in the point estimates. This variable is in the model not even stat. significant.
Anyways, I tested for autocorrolation and my model suffers from it. I guess the idiosyncratic errorterm is corrolated. How do i correct for this in the random effects model? I thought about adding a variable called Uhat(-1) with Uhat=Resid.
It was very significant when i put it in. However, i do not know if i should since the random effects model is not supposed to have autocorrolation i guess.
How would you guys deal with this?