Hi all,
I am following the method of Borusyak, (2021) by running the
Code:
did_imputation
package. I want to examine whether staggered implementation laws affect a dependent variable y.
I got the result as below but I am wondering how to know whether the parallel test is satisfied ?

My code is:

Code:
did_imputation y i t  ei ,autosample fe( i t) controls($control_list) maxit(30000) tol(0.001)
And the result is as below. Its seems that the laws have impact on dependent variable y.
Code:
 
Number of obs = 100,244
------------------------------------------------------------------------------
y | Coefficient Std. err. z P>|z| [95% conf. interval]
-------------+----------------------------------------------------------------
tau | .0422128 .0170445 2.48 0.013 .0088061 .0756194
------------------------------------------------------------------------------
However, I do not know whether the parallel trend is satisfied, I run event_plot and joint null test as below, I am wondering whether they are properly used here? (The joint null test is the one I am creative, not from the guideline of Borusyak)
The code

Code:
did_imputation y i t ei ,autosample fe( i t) controls($control_list) allhorizon pretrend(3) maxit(30000) tol(0.001)

event_plot, default_look graph_opt(xtitle("Periods since the event") ytitle("Average causal effect") ///
 title(" (\Borusyak et al. (2021))\") xlabel(-3(1)4))

test (pre1 = 0) (pre2 = 0) (pre3 = 0)
The result is

Code:
 
Number of obs = 100,244
------------------------------------------------------------------------------
y | Coefficient Std. err. z P>|z| [95% conf. interval]
-------------+----------------------------------------------------------------
tau0 | -.0062363 .0073831 -0.84 0.398 -.020707 .0082344
tau1 | -.007813 .0112215 -0.70 0.486 -.0298067 .0141807
tau2 | .038708 .0130032 2.98 0.003 .0132222 .0641939
tau3 | .07724 .0156284 4.94 0.000 .0466089 .1078711
tau4 | .1038992 .0164318 6.32 0.000 .0716934 .1361049
tau5 | .0624891 .0220759 2.83 0.005 .0192212 .1057571
tau6 | .0821273 .0253541 3.24 0.001 .0324341 .1318205
tau7 | .0851133 .0283558 3.00 0.003 .029537 .1406896
tau8 | .0647298 .0303133 2.14 0.033 .005317 .1241427
tau9 | -.0118512 .0216589 -0.55 0.584 -.0543018 .0305995
tau10 | -.0693634 .0224063 -3.10 0.002 -.113279 -.0254479
tau11 | -.0252598 .0238936 -1.06 0.290 -.0720904 .0215708
tau12 | -.0373753 .0288603 -1.30 0.195 -.0939404 .0191897
tau13 | -.0405529 .0321011 -1.26 0.206 -.10347 .0223641
tau14 | -.0091321 .0334497 -0.27 0.785 -.0746922 .056428
tau15 | .0134297 .0365213 0.37 0.713 -.0581506 .08501
tau16 | .000612 .0341164 0.02 0.986 -.0662549 .0674789
tau17 | .0297979 .0406446 0.73 0.463 -.0498641 .1094599
tau18 | .0605124 .0454848 1.33 0.183 -.0286361 .1496609
tau19 | .0674091 .0475789 1.42 0.157 -.0258439 .160662
tau20 | .045582 .0509481 0.89 0.371 -.0542744 .1454384
pre1 | .0242929 .0118413 2.05 0.040 .0010845 .0475014
pre2 | .0177519 .0090836 1.95 0.051 -.0000516 .0355553
pre3 | .0091315 .0066814 1.37 0.172 -.0039639 .0222268
------------------------------------------------------------------------------
Array

Code:
 
. test (pre1 = 0) (pre2 = 0) (pre3 = 0)
( 1) pre1 = 0
( 2) pre2 = 0
( 3) pre3 = 0
chi2( 3) = 4.47
Prob > chi2 = 0.2147
Can I conclude that the regression above satisfies parallel trend test because p-value of joint null test (Prob>chi2) is higher than 0.1 ?