Instrumental Variables

Question 1

when is an IV approach helpful?

Answer 1

1. instrument has predictive power for explanatory variable
   - relevance
   - first-stage regression
2. no feedback/endogeneity from outcome to the instrument and the instrument is uncorrelated with unobserved Xs
   - exogeneity/independence condition
3. instrument only affects the outcome through the endogenous variable
   - exclusion restriction

Question 2

identifying a causal effect

Answer 2

if the instrument fulfils the 3 conditions and also has a large impact on the outcome of interest, then it’s likely that there’s a causal effect