Open JohnCRuf opened 1 year ago
Is there any random feature of the setting that I can use to exploit this more than just with TWFE? Eg. Assume contagiousness of government approval, if so then wouldn't vote share for incumbent in nearby precincts outside of ward be a valid IV for incumbent sentiment in this ward?
Suppose that voters can only imperfectly attribute spending to an alderman, then using nearby spending as an IV for support would also be valid. However, this can be a substitute for spending in border precincts.
Remove old, drastically underpowered BLP models and replace them with modestly underpowered BLP models exploiting voting precinct data.
There are 46 runoffs in the dataset. 40 precincts per ward/year mean a BLP model with ~1800 observations rather than ~200. No longer drastically underpowered, now merely modestly underpowered.