Causal Analysis Flashcards
causal relationship only if _ assumption holds
exogeneity
unobserved factors (ui) unrelated to regressor X
exogenous regressor
Xi and ui correlated
endogenous regressor
techniques to overcome endogeneity
randomisation and IVs
limitations of randomisation in OLS
unethical, estimate only those who chose to take part, limited duration
assumptions of an IV
relevance and exogeneity
relevance assumption
Z must be correlated with X
exogeneity assumption (exclusion restriction)
Z must not be correlated with error term u
instrument is used to
observe exogenous variation in Xi through movements in Zi to identify causal effect of Xi on Yi
IV estimation can be done by
2SLS
2SLS stages
- regress X on Z to get predicted Xs
2. regress Y on predicted Xs
weak instruments problem
2SLS estimator will be biased towards OLS estimator if correlation is close to zero
doesn’t explain sufficient variation in endogenous variable X
testing strength of instrument
F-stat
F-stat (actual value, not p-value)
<10 = weak >10 = strong
testing validity of instrument
can only use IV if same number of instruments and endogenous regressors