| author | Koren, Miklós (https://koren.dev) |
|---|---|
| date | 2026-10-13 |
| version | 0.10.1 |
| title | XT2TREATMENTS - event study with two treatments |
| description | Computes the average treatment effect on the treated (ATT), where the control is another treatment happening at the same time. |
| url | https://github.com/codedthinking/xt2treatments |
| requires | Stata version 18 |
xt2treatmentsvarname(numeric) [if] [aw=exp], treatment(varname numeric) control(varname numeric), [pre(#) post(#) baseline(string) weighting(string) cluster(varname) graph]
xt2treatments estimates average treatment effects on the treated (ATT) when there are two treatments. The first treatment is the treatment of interest, and the second treatment is the control.
The package can be installed with
net install xt2treatments, from(https://raw.githubusercontent.com/codedthinking/xt2treatments/main/) replace
| Option | Description |
|---|---|
| treatment | Dummy variable indicating the treatment of interest. |
| control | Dummy variable indicating the control treatment. |
| aw | Nonnegative analytical weights. Missing and zero weights are excluded. Analytical weights affect baseline means, cohort-time ATT estimates and their standard errors, and aggregation across cohorts and times. |
| pre | Number of periods before treatment to include in the estimation (default 1) |
| post | Number of periods after treatment to include in the estimation (default 3) |
| baseline | Either a negative number between -pre and -1 or average, or atet. If -k, the baseline is the kth period before the treatment. If average, the baseline is the average of the pre-treatment periods. If atet, the regression table reports the average of the post-treatment periods minus the average of the pre-treatment periods. Default is -1. |
| weighting | Method to weight different cohorts in the estimation. |
| cluster | Variable to cluster standard errors. If not specified, the panel identifier is used. |
| graph (optional) | Plot the event study graph with the default settings of hetdid_coefplot. |
Let m0 and m1 denote the control and treated mass in a cohort-time cell. Without analytical weights, these are observation counts. With analytical weights, they are sums of the analytical weights.
| Method | Description |
|---|---|
| equal (default) | Each estimable cohort-time cell is weighted equally. |
| proportional | Cohort-time cells are weighted by m0 + m1. |
| optimal | Cohort-time cells are weighted by (m0 * m1) / (m0 + m1). |
use "xt2treatments_testdata.dta", clear
xtset i t
xt2treatments y, treatment(treatmentB) control(treatmentA) pre(1) post(3) weighting(equal)
Analytical weights use native Stata weight syntax:
generate double aw = cond(i <= 50, 2, 1)
xt2treatments y [aw=aw], treatment(treatmentB) control(treatmentA) pre(1) post(3) weighting(optimal)Panel variable: i (strongly balanced)
Time variable: t, 1 to 10
Delta: 1 unit
Event study relative to -1 Number of obs = 1,000
------------------------------------------------------------------------------
y | ATET Std. err. z P>|z| [95% conf. interval]
-------------+----------------------------------------------------------------
-1 | 0 (omitted)
0 | .464904 .0179099 25.96 0.000 .4298014 .5000067
1 | .4581741 .0177579 25.80 0.000 .4233694 .4929789
2 | .4108288 .0173002 23.75 0.000 .3769211 .4447366
3 | .3221394 .0199 16.19 0.000 .2831362 .3611426
------------------------------------------------------------------------------
xt2treatments y, treatment(treatmentB) control(treatmentA) pre(3) post(3) weighting(optimal) graph
Panel variable: i (strongly balanced)
Time variable: t, 1 to 10
Delta: 1 unit
Event study relative to -1 Number of obs = 1,000
------------------------------------------------------------------------------
y | ATET Std. err. z P>|z| [95% conf. interval]
-------------+----------------------------------------------------------------
-3 | .0188954 .0377553 0.50 0.617 -.0551037 .0928944
-2 | -.01291 .0288885 -0.45 0.655 -.0695304 .0437105
-1 | 0 (omitted)
0 | .2940147 .0263712 11.15 0.000 .2423281 .3457014
1 | .2639324 .0261562 10.09 0.000 .2126672 .3151977
2 | .270967 .0253986 10.67 0.000 .2211866 .3207474
3 | .283611 .029183 9.72 0.000 .2264135 .3408086
------------------------------------------------------------------------------
xthdidregress estimates ATT against various control groups. However, it does not allow for two treatments.
When the control group is another treatment happening at the same time, the ATT is the difference between the treatment and the control.
The command returns, as part of e(), the coefficients and standard errors. See ereturn list after running the command. Typical post-estimation commands can be used, such as outreg2, estout, or coefplot.
- Miklós Koren (Central European University, https://koren.mk), maintainer
You are free to use this package under the terms of its license. If you use it, please cite the software package in your work:
- Koren, Miklós. (2024). XT2TREATMENTS - event study with two treatments [Computer software]. Avilable at https://github.com/codedthinking/xt2treatments