Question about PLR.summary

[vekeywang]

Dear DoubleML team,
I am sorry if my question is a bit dumb.I'm wondering if the standard errors in the summary of the PLR model are robust standard errors that take heteroscedasticity into account. If not, I'm curious about how I could obtain the standard errors that consider heteroscedasticity.I would truly appreciate it if you could spare a moment to offer some guidance on this matter. Even a brief pointer or a reference to relevant documentation would be of great help to me.

[SvenKlaassen]

Hello @vekeywang,

this is a good question. In short, yes.

You can find the complete conditions for the PLR model original Double Machine Learning Paper. The PLR model is discussed in Section 4, where Assumption 4.1 states the regularity conditions.
The assumption 4.1 (d) $‖ E[U^2|X]‖_{P,\infty} \le C$ still allows for heteroscedasticity.

[vekeywang]

Thank you for your prompt reply, really appreciate it!