Theory question - dummy variables, chi-squared, robust standard errors

[Gabriele More]

Hello, I am running a simple ols regression with industry and year dummies. Each observation is defined uniquely for one year and one industry. When i use the command robust se, the Chi-square disappear completely from the model statistics. If I remove one year dummy and one industry, the chi-square reappears even with robust se. My question: is this the dummy variable trap? How can I choose the year and industry to drop? Thank you in advance.

[George Ford]

between the industry and year dummies, haven't you saturated the model?

Also, robust does not work properly when you have dummy variables = 1 for one observation (the t-stat can be heavily inflated).

[Gabriele More]

Dear George, thank you for answering. I think I was not clear: there are 215 firms (each with a GICS sector dummy) scattered between 2019-2022. I don't think the model is saturated?

[George Ford]

help j_robustsingular