# Instruments

#### Anbrhave

##### New Member
I have a question concerning endogenous variables. I want to do the following OLS regression:
Grade Point Average=constant + φ1 work hours + φ2 study hours + φ3 Xi + ε
Where Xi included individual characteristics. The coefficient of interest is φ1.
However work hours will be endogenous as motivation/ability is included in ε. This will also be the case for study hours. I'm now interested if I need to do 2SLS for both work hours AND study hours as only work hours is of interest? Can't I just find an instrument for work hours? Or Should I do the following?:
Work hours i = α0+ Zi•α1 + α2•Xi + εi
Study hours i = φ0+ Zi• φ 1 + φ 2•Xi + ψi
GPA i = β0 + β1•Work hours i + β2•Study hours i + β3•Xi + μi
Where Zi is an instrument / are instruments.
I already thought of instruments as unearned income or/and disability for work hours as they both have (theoretical) a (negative) correlation/relationship with work hours but not directly with GPA. However, these do probably not work for study hours (?)