Tests for regression analysis (panel data)

[Pia Donner]

Hello,

I'm analysing the effect of R&D expenditure on firm's ESG score and the moderating effect of institutional quality.
I have one IV, and one DV (both firm-level), one moderator (country-level) and 6 control variables (5 firm-level, 1 country-level).
I'm using panel data from 2014 until 2019 for around 320 companies.

If I'm correct, I first need to perform several test: normality (Shapiro-Wilk test), heteroskedasticity (Breusch-Pagan test), multicollinearity (VIF) and autocorrelation (Durbin Watson test).
How do I perform these for panel data? I found -swilk-, -estat hettest-, -estat vif-, -dwstat-
However, I do not know how these commands differ when using panel data.

I would highly appreciate it if someone could help me with some clarification here.

[Carlo Lazzaro]

Pia:
welcome to this forum:
1) normality is a weak requirements for the u and e components of the error term. With a 320 companies the CLT rules and you do not have to worry about thsi issue;
2) you do not say if you're running -xtreg,fe- or -xtreg,re- (see community-contributed modules for testing heteroskedastciity and autocorrelation od the epsilon error);
3) as per https://www.hup.harvard.edu/catalog....=9780674175440, Chapter 23, do not care about multicollinearity; test your regression for endogeneity and misspecification of the functional form of the regressand, instead.