Hello, I am having a hard time dealing with endogeneity problems in my data, and recently I found a control function approach.
I am in the middle of the learning process, and I have many questions that I'm trying to find answers to..
One of the questions that I couldn't find the answer to is about the instrument variable.

I understand that the CF approach requires an instrument variable which needs to be used in the reduced form model, but not in the original, structural model.
But the thing is I am having a very hard time finding good instrument variables for my endogenous variable y, where not many previous studies exist.
So I am considering using a lagged variable (l.y) as an instrument in the reduced form model. But I am not very sure if this is okay because I cannot find much literature regarding this matter.
will this be a problem? Is there any article that I can refer to?

If I can get any advice, it'll be a great help for me.
Thank you!