03-08-2017 11:11 AM - edited 03-08-2017 11:43 AM
I have a general question about running panel OLS model.
Here are the steps I took:
1. check if variables are stationary. (Is this necessary in panel data?)
2. check cointegration.
3. I have Y and X1 that are unit-root and no co-integration, so I first-differenced variables, and they are stationary. (Instead, can I run fixed or random effect?)
4. I ran ranone, and I have Pr>m of 0.56 for Hausman test; I ran ranone, and I have DF=0 / m value=0 / Pr>m= .
Do I run pooled ols, fixone, fixtwo, ranone, or rantwo?
Ranone suggest using random effect, but I am not sure how to interpret general picture of the analysis.