Hausman Test important to choose fixed effect or random effect model. It basically null hypothesis (Ho) Random Effect Model is consistent. If p-value of the is greater than 0.05, we accept the null hypothesis. If p-value is less than 0.05, we reject the null hypothesis. In this case, Fixed Effect Model is more consistent.

The decision to use either FEM or REM is determined by the Hausman Test.

If the value of hausman is greater than 5% :

REM is more consistent than FEM

The Hausman specification test is usually used to identify the best model between RE and FE. Read the method of this paper if you need reference

https://link.springer.com/article/10.1007/s11846-019-00344-5

Actually, without Hausman Test, we can already say that your estimation will tend to suggest RE. This is because you choose 5 out of the whole population. However, I am curious about the perfect (100%) p-value, which is in my understanding because of no difference between coefs from RE and FE. If we say that there is no data issue or software problem, this means the errors within cross-sections are zero. I don't think this could happen econometrically. So, I suggest you estimate again using other software to confirm the Hausman test result.

Carlos Mangunsong Thanks for answer.

Sir, what do you mean the error within cross section are zero?

Regarding to your advice, I have tried Stata and the outcome is same as Eviews and also there is not difference between coefs, std error, t-stat and probability from RE and FE. I am just guessing is that the problem arise from the dataset? Because I used same figure for the macroeconomic variables to regress the 5 bank stock price

Agree with you. I think this is a data problem. Which is exactly to what I meant about zero error cross section.

Carlos Mangunsong Sir, I not really understand what does zero error cross section mean. Do you mind to explain to me or recommend me some journals to refer?

Dear Low,

Kindly check your data set by carring out preliminary tests first.

Then continue with the Hausman test to predict either fixed or random effect for continuity.

Regards

Kehinde Mary Bello Dear miss Bello, is it the preliminary test is refer to the diagnostic test such as heteroscedasticity, autocorrelation and multicollinearity test? Besides, is that I need to do the diagnostic tests with all the three model (panel OLS, REM and FEM) or simply run the diagnostic test with only OLS model.

Sorry for asking lot of questions since I really not familiar with econometrics

Dear Low.

Preliminary test are test carried out to avoid spurious regression such as unit root test, cointegration test and multicollinearity test. These can be used to test your variables for analysis to avoid spurious results.

Diagnostic test are carried out on a particular model.

So that's a problem with the model, your P-value should not be 1.

There I recommend @ Zunaira Zahoor

Regards