Probit estimation with clustered standard errors

[chanikjo]

Hello everybody~

I am now making my master thesis.

My data is panel data and I use probit estimation.

In order for my data set to be robust to heteroscedasticity, I try to estimate probit with clustered standard errors(Rogers standard errors).

However, in estimating probit model, eviews only provide Huber/White standard errors or GLM.

By contrast, if I estimate using OLS model, there is White period(Panel option --> coefficient covariance method). I heard that 'White period' is same with clustered standard errors.

Is there any way to estimate probit with clustered standard errors(rogers standard errors)?

Is Huber/White standard errors same with clustered standard errors?

Please help me~ T.T

[EViews Glenn]

EViews built-in probit models do not allow for clustering in the standard errors. You could use the results from the probit equation to compute your own standard errors. It is not particularly difficult, but does require a bit of knowledge and the ability to perform matrix operations.