ECON 151 Lecture 33: ECON 151 L33
Document Summary
Si = 0 + 1zi + i (7) > 0 is a test of the relevance assumption . Importantly, the independence/exogeneity assumption cannot be rigorously tested and should be assume or discussed. Typically, iv are rare and hard to identify. B1 = causal effect of schooling on wage with 1 = e (wagei |z ), the coefficient from the reduced from regression ; 1 = e (si |z ), the coefficient from the first stage; 1 is identified i. e. can be estimated without bias. Structural model estimating log wage on college attendance = return to education. Fs: college attendance = a0 + a1*distance + e column 1 + 2. Rf: wage = g0 + g1*distance + e column 3 + 4. Ivreg(outcome ~ endogenous | iv, data = ps5) Result should be the same as the wald ratio. Still can use it even if it"s bad. Comparing impact of treatment (s=1) vs no treatment (s=0), comparing the two means.