Endogenous variables

Summary

Typo at 0.44 in lecture: the final ββ should be β4β4

Example 1

  • True model:

E(xi,2,xi,3,xi,4)=β1+β2xi,2+β3xi,3+β4xi,4,i=1,,nE(xi,2,xi,3,xi,4)=β1+β2xi,2+β3xi,3+β4xi,4,i=1,,n

  • We assume that

E(xi,2,xi,3,xi,4)=β1+β2xi,2+β3xi,3,i=1,,nE(xi,2,xi,3,xi,4)=β1+β2xi,2+β3xi,3,i=1,,n

  • and specify

yi=β1+β2xi,2+β3xi,3+εi,i=1,,nyi=β1+β2xi,2+β3xi,3+εi,i=1,,n

  • If an included variable ( x2x2 or x3)x3) is correlated with the missing variable ( x4x4 ), then it is endogenous
  • If an included variable is independent of the missing variable, then it is exogenous

Example 2

  • True model: yi,1,yi,2yi,1,yi,2 are determined jointly from:

yi,1=β1+β2xi,2+β3yi,2+εi,i=1,,nyi,1=β1+β2xi,2+β3yi,2+εi,i=1,,n

yi,2=γ1+γ2xi,2+γ3yi,1+νi,i=1,,nyi,2=γ1+γ2xi,2+γ3yi,1+νi,i=1,,n

  • Then

E(xi,2,yi,2)=β1+β2xi,2+β3yi,2,i=1,,nE(xi,2,yi,2)=β1+β2xi,2+β3yi,2,i=1,,n

  • cannot hold in general.
  • If we consider the first equation on its own, then yi,2yi,2 is endogenous.
  • xi,2xi,2 will be exogenous if E(εi|xi)=E(νi|xi)=0E(εi|xi)=E(νi|xi)=0

Example 3

  • True model:

E(zi,xi)=β1+β2zi,2+β3zi,3E(zi,xi)=β1+β2zi,2+β3zi,3

  • where xi,2xi,2 is a measurement of true zi,2zi,2 and xi,3xi,3 is a measurement of true zi,3zi,3 .
  • We run yy on x2x2 and x3x3 .
  • If a variable has measurement errors, it will be endogenous.
  • If a variable does not have measurement errors, it will be exogenous.