Assessing validity of multiple regression models with fixed effects

[maria lesablo]

Hello everyone! I am unsure about which methods to employ when testing for validity of the MLR models with fixed effects that I run for my dissertation.

I investigate the impact of technology on the size of informal economy for two groups of countries (low and high income), so I introduce an interaction between the technology variable and the income level dummy.

There are standard procedures reported (testing for multicollinearity, autocorrelation of residuals, linearity & homoscedasticity), however, from my research I understood that many believe that these procedures are outdated and not particularly useful.

Please let me know what you think I should include, and let me know if you need more information

[Carlo Lazzaro]

Maria:
welcome to this forum.
As per your statement, it would seem that in your research field you can simply go plain vanilla -xtreg,fe- and live with what the outcome table gives you back.
Set aside quasi-extreme multicollinearity (that can be an issue under some conditions only), the possible misspecification of the functional form of the regressand, endogeneity, autocorrelation and heteroskedasticity should be investigated.
Eventually, a word of advice: please read the FAQ on how to post more effectively. It's up to posters to provide all the necessary information, not to interested listers to ask, for the trivial evidence that interested listers do not know your data(set). Thanks.

[maria lesablo]

Thank you, Carlo! I will definitely follow your advice (and check the FAQ as well). Have a great day!