HonestDiD when I am interested in the average over the two post-treatment periods

[Nicolas Campos]

Hi everyone,

I have the following problem when I try to run the honestdid package (I am using Stata 17). I am not using the default option of the first post-treatment period, but the average over the two post-treatment periods. The error that arises is:

4 child processes encountered errors. Throwing last error.
(note: error during parallel run; falling back on sequential execution)
*: 3200 conformability error
_honestRMConditionalCS(): - function returned error
HonestSensitivityHelper(): - function returned error
HonestSensitivityResults(): - function returned error
HonestDiD(): - function returned error
<istmt>: - function returned error

Here is an example with simulated data with the same structure than my own data (staggered treatment) illustrating my process: * Set the number of units and periods

 local num_units = 100 local num_periods = 10

* Create base data

 clear set obs `num_units' gen id = _n expand `num_periods' bysort id: gen time = _n

* Create the time variable as a date

 gen date = mdy(1, 1, 2000) + (time - 1) * 30 format date %td

* Randomly assign the treatment start time for each treated unit

 gen treatment_start = . replace treatment_start = 5 + int(runiform() * 5) if id <= `num_units'/2

* Create the post variable indicating if treatment is received

 gen post = 0 bysort id: replace post = 1 if time >= treatment_start & !missing(treatment_start)

* Create the empleo_formal (dummy) variable

 gen empleo_formal = 0

 replace empleo_formal = 1 if runiform() > 0.5

* Adjust empleo_formal based on treatment

 replace empleo_formal = 1 if post == 1 & runiform() > 0.3 replace empleo_formal = 0 if post == 1 & runiform() <= 0.3

* Create gvar (time of treatment)

 sort id time

 by id: gen first_change = post == 1 & post[_n-1] == 0 if _n == 1 | post[_n-1] == 0

 by id: egen period_start = min(cond(first_change == 1, time, .))

 drop first_change

 gen gvar = cond(period_start == ., 0, period_start)

 label variable gvar "Treatment activation period"

[CODE] tab gvar [/ CODE]

 drop period_start treatment_start

* List some observations to verify the data

 list id time date post empleo_formal gvar in 1/20

* Estimation with csdid2 and sensitivity with HonestDiD

 csdid2 empleo_formal, ivar(id) time(time) gvar(gvar) agg(event) long2

 estat event, estore(csdid)

 estimates restore csdid

 matrix define l_vec = 0.5 \ 0.5

 matlist l_vec

* But when I try this, the error above arises:

 honestdid, l_vec(l_vec) pre(3/8) post(9/14) mvec(0.2(0.1)1.5) coefplot xtitle(Mbar) ytitle(95% Robust CI)

Any help would be greatly appreciated!