2SLS with we use polynomial terms of birth cohort.

[Ayl Gnpr]

Hello,

I am trying to estimate a fuzzy RD. Fuzzy RD is equivalent to 2SLS (Hahn et al. 2001). So I am trying to run 2SLS. (I am not allowed to use rd or rdrobust commands.)
Compulsory schooling in Turkey increased from 5 to 8 years in 1997. Birth cohorts that are born after 1986 (not included) are affected by the reform. I want to control for birth-cohort polynomials up to fourth-order. I also include age dummies. However, when I run the command below, birth_year2 birth_year3 birth_year4 are omitted because of multicollinearity. Could you help me with my code?

xi: ivregress 2sls lwage (yrs_school = reform) birth_year birth_year2 birth_year3 birth_year4 i.age , first

Thank you very much

[Dung Le]

Hi Ayl,

I think you should read FAQ and provide your data example using -dataex- command and Stata outputs so that the forum members are able to help you.

By the way, it seems your fuzzy RD is not a standard setup. You may want to read this paper: https://www.mdrc.org/sites/default/f...nuity_full.pdf

After reading that reference, if you still need help, then post back with your data example, Stata code and outputs.

Hope this helps.