cleaning the dataset for diff and diff model

[Chul-Kyoo Jung]

Hi, I have the following unbalanced data. I want to clean my data to apply simple diff-in-diff model so that data may look like:

I want to drop all the firms that do not have any year before 1997 and then let each firm id to have years only 1996 and 2005 so that I can compare those two years from difference in difference view.
'chaebol' is my treatment variable.

Can you help?
Thanks

* Example generated by -dataex-. To install: ssc install dataex
clear
input str6 kiscode int year byte chaebol double asset
"350257" 1980 0  18285348000
"500607" 1980 0 131784602000
"351253" 1980 0  62865288000
"320498" 1980 1  33838416000
"351156" 1980 0   6360425000
"360252" 1980 1 541330024000
"360317" 1980 0  22984591000
"360228" 1980 0 132764602000
"380911" 1980 1 213239161000
"381152" 1980 0   6541576000
"310140" 1980 0  49222411000
"310166" 1980 0  14139029000
"370274" 1980 0  21186479000
"310026" 1980 1  24339944000
"350648" 1980 0  28434461000
"310123" 1980 0  34758175000
"350117" 1980 0  17158988000
"380806" 1980 1  15005232000
"320129" 1980 1 217819358000
"340251" 1980 1  25009066000
"370118" 1980 0  58045333000
"370177" 1980 0  36533137000
"370150" 1980 0  41099893000
"360287" 1980 0  57162344000
"360090" 1980 0  83874586000
"380784" 1980 0  44872170000
"351350" 1980 0  45377311000
"370053" 1980 0  28931137000
"310190" 1980 0  30997921000
"381160" 1980 0   6330526000
"380938" 1980 0  50391070000
"380733" 1980 1  16471520000
"310590" 1980 0  44011174000
"351067" 1980 1 130526305000
"310476" 1980 1  47629107000
"350451" 1980 0  16930929000
"310174" 1980 1  76886460000
"340324" 1980 1  29793379000
"310433" 1980 1 110228999000
"350320" 1980 0   4901957000
"381250" 1980 0  13125490000
"310131" 1980 0  32708711000
"320099" 1980 0  46811999000
"340154" 1980 1  26297430000
"320218" 1980 1 159598422000
"350591" 1980 1  36722653000
"380989" 1980 0  14065113000
"380954" 1980 1 299788090000
"380474" 1980 0   4559409000
"360155" 1980 0  16989303000
"320293" 1980 0   5749583000
"350729" 1980 0  11320038000
"320536" 1980 0   7566338000
"350940" 1980 1 3.753688e+11
"320242" 1980 1  72924040000
"380725" 1980 1 255463903000
"320692" 1980 1  36229805000
"380377" 1980 1  84495668000
"370070" 1980 0 147094202000
"320730" 1980 0 134657126000
"350478" 1980 0  18782347000
"380407" 1980 1 253086691000
"370088" 1980 0  13385963000
"350745" 1980 0  16256468000
"350354" 1980 0  23786698000
"350893" 1980 1  47589241000
"370355" 1980 0   8763943000
"370266" 1980 0 187501484000
"350281" 1980 0   6170263000
"320234" 1980 1  93930236000
"350826" 1980 0  91398637000
"351245" 1980 1 791687216000
"330019" 1980 1  30833118000
"350664" 1980 0  25927689000
"100013" 1980 0  33738516000
"320102" 1980 0  38081496000
"350605" 1980 0  20336333000
"381519" 1980 0  74751893000
"320234" 1981 1 191479919000
"320536" 1981 0  10653474000
"381497" 1981 0   6876018000
"350427" 1981 0  13974417000
"310476" 1981 1  46282129000
"370169" 1981 0  17031189000
"350940" 1981 1 363957318000
"360090" 1981 0 108913679000
"380377" 1981 1  99681110000
"320307" 1981 0  12349372000
"351067" 1981 1 153999084000
"310239" 1981 0  63909181000
"380989" 1981 0  19277127000
"360341" 1981 1  81424045000
"350672" 1981 0  11976339000
"320692" 1981 1  40404825000
"381225" 1981 0  13215386000
"310026" 1981 1  25694216000
"500607" 1981 0 218146702000
"380784" 1981 0  50509180000
"370177" 1981 0  46809912000
"360279" 1981 0   8186674000
end

[Fei Wang]

Chul-Kyoo, I think the following codes are what you want.

keep if year == 1996 | year == 2005
bysort kiscode: keep if _N == 2

[Chul-Kyoo Jung]

Thank you so much!