LATE Difference-in-Differences

[Nicolas Orgeira]

Hi all,

I'm trying to estimate a generalized DID (two periods three groups)'s local average treatment effect. The DID code used was:

svy: regress OUTCOMEVAR i.time##i.intervention

Could you please confirm whether the following code is correct:

svy: ivregress 2sls OUTCOMEVAR i.time (i.intervention i.time#i.intervention = i.instrument i.time#i.instrument)

where time is a dummy variable for time, intervention=0/1/2 indicating the allocated treatment and instrument=0/1/2 indicating the actual treatment received.

Also, I'd like to look at heterogenous effects on compliers (i.e. female=0/1). Would the following code make sense?

svy: ivregress 2sls OUTCOMEVAR i.time i.female i.time#i.female (i.intervention i.intervention#i.female i.time#i.intervention i.time#i.intervention#i.female = i.instrument i.instrument#i.female i.time#i.instrument i.time#i.instrument#i.female)

Thank you

[Nicolas Orgeira]

Hi all,

I'm now experiencing a new issue using the basic IV-DID regression:

 
 svy: ivregress 2sls OUTCOMEVAR i.time (i.intervention i.time#i.intervention = i.instrument i.time#i.instrument)

The regression worked fine when using the original instrument distribution - however, I identified a small error and had to change the instrument for 10 observations. The regression no longer works (message "equation not identified; must have at least as many instruments not in) and I struggle to understand why.

Attached is the logfile. I would be very grateful if someone could help me understand.

Thank you