How to test for heteroskedasticity in a panel data-set

[Yasmin Mostafa]

Hello,

I have a panel data set with 180 countries observed during the period 2008-2017, I am trying to see the effect of interacting democracy measure (polity2) with different measures of government ideology on the environmental performance of the country. My supervisor advised me to use three different models: Pooled OLS with time fixed effects, FE and RE to see how the regression looks like in each of them.
The question is, I would like to test for heteroskedasticity under the three models. I have searched before asking and found that I should use xttest2 for FE. For the RE, I saw some suggestions for xttest0, but I am not sure if this is the right context. Also, I couldn't find what should I use to test in the pooled OLS.

Also, to control for heteroskedacity, I've understood that I should cluster standard errors at the country level using: vce, cluster(country) with xtreg as it is the same as vce, robust. In the pooled OLS model, I fail to understand if I should use the cluster or the robust option with regress.

I'd be very grateful if you could help.

Thanks and best regards.

[Mario Ferri]

There are several ways to test for heteroscedasticity, including
Breusch–Pagan's one, Brown–Forsythe's, Cook–Weisberg's, Harrison–McCabe's, Glejser's, Goldfeld–Quandt's, Levene's , Park's, and last but not least (perhaps, the most commonly used), the White's test.
But each one has its proper conditions of use. Breusch-Pagan and Whites Test is mostly used and popular for heteroscedasticity
See the help for each one

[Yasmin Mostafa]

Thank you Mario for your helpful response. I probably should have been more precise, I was asking how to test for heteroskedasticity using the Breusch-Pagan test, specifically for the Random Effects and Pooled OLS models, since as far as I know, xttest2 does the trick for Fixed Effects.

I would appreciate if you have some idea on that.

Thanks