{smcl}
{marker xt2treatments-estimates-event-studies-with-two-treatments}{...}
{title:{cmd:xt2treatments} estimates event studies with two treatments}
{marker syntax}{...}
{title:Syntax}
{text}{phang2}{cmd:xt2treatments} varname(numeric) [{it:if}], {bf:treatment}(varname numeric) {bf:control}(varname numeric), [{bf:pre}(#) {bf:post}(#) {bf:baseline}({it:string}) {bf:weighting}(string) {bf:graph}]{p_end}
{pstd}{cmd: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.{p_end}
{pstd}The package can be installed with{p_end}
{p 8 16 2}net install xt2treatments, from(https://raw.githubusercontent.com/codedthinking/xt2treatments/main/) replace
{marker options}{...}
{title:Options}
{marker options-1}{...}
{dlgtab:Options}
{synoptset tabbed}{...}
{synopthdr:Option}
{synoptline}
{synopt:{bf:treatment}}Dummy variable indicating the treatment of interest.{p_end}
{synopt:{bf:control}}Dummy variable indicating the control treatment.{p_end}
{synopt:{bf:pre}}Number of periods before treatment to include in the estimation (default 1){p_end}
{synopt:{bf:post}}Number of periods after treatment to include in the estimation (default 3){p_end}
{synopt:{bf:baseline}}Either a negative number between {cmd:-pre} and {cmd:-1} or {cmd:average}, or {cmd:atet}. If {cmd:-k}, the baseline is the kth period before the treatment. If {cmd:average}, the baseline is the average of the pre-treatment periods. If {cmd:atet}, the regression table reports the average of the post-treatment periods minus the average of the pre-treatment periods. Default is {cmd:-1}.{p_end}
{synopt:{bf:weighting}}Method to weight different cohorts in the estimation.{p_end}
{synopt:{bf:graph} (optional)}Plot the event study graph with the default settings of {cmd:hetdid_coefplot}.{p_end}
{synoptline}
{marker weighting-methods}{...}
{dlgtab:Weighting methods}
{synoptset tabbed}{...}
{synopthdr:Method}
{synoptline}
{synopt:{bf:equal} (default)}Each cohort is weighted equally.{p_end}
{synopt:{bf:proportional}}Cohorts are weighted linearly by the number of observations, (n0 + n1), where n0 is the number of controls, n1 is the number of treated units.{p_end}
{synopt:{bf:optimal}}Cohorts are weighted by the inverse of the standard error of the treatment effect estimate of the cohort, (n0 * n1) / (n0 + n1).{p_end}
{synoptline}
{marker examples}{...}
{title:Examples}
{p 8 16 2}use "xt2treatments_testdata.dta", clear
{p 8 16 2}xtset i t
{p 8 16 2}xt2treatments y, treatment(treatmentB) control(treatmentA) pre(1) post(3) weighting(equal)
{p 8 16 2}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
------------------------------------------------------------------------------
{p 8 16 2}xt2treatments y, treatment(treatmentB) control(treatmentA) pre(3) post(3) weighting(optimal) graph
{p 8 16 2}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
------------------------------------------------------------------------------
{marker background}{...}
{title:Background}
{pstd}{cmd:xthdidregress} estimates ATT against various control groups. However, it does not allow for two treatments.{p_end}
{pstd}When the control group is another treatment happening at the same time, the ATT is the difference between the treatment and the control.{p_end}
{marker remarks}{...}
{title:Remarks}
{pstd}The command returns, as part of {cmd:e()}, the coefficients and standard errors. See {cmd:ereturn list} after running the command. Typical post-estimation commands can be used, such as {cmd:outreg2}, {cmd:estout}, or {cmd:coefplot}.{p_end}
{marker authors}{...}
{title:Authors}
{text}{phang2}Miklós Koren (Central European University, {browse "https://koren.mk"}), {it:maintainer}{p_end}
{marker license-and-citation}{...}
{title:License and Citation}
{pstd}You are free to use this package under the terms of its {browse "https://github.com/codedthinking/xt2treatments/blob/main/LICENSE"}. If you use it, please cite the software package in your work:{p_end}
{text}{phang2}Koren, Miklós. (2024). XT2TREATMENTS - event study with two treatments [Computer software]. Avilable at {browse "https://github.com/codedthinking/xt2treatments"}{p_end}