建议请用 dataex (先 ssc install dataex 并见说明) 将原始 Stata 资料中具有”代表性”的一部分资料列出，以供有意回答者实验之用，并能提供具体操作指令。并请参考 http://www.jianshu.com/p/9870080fe769,  https://bbs.pinggu.org/thread-5048204-1-1.html, 与 https://bbs.pinggu.org/thread-5917273-1-1.html。

* Example generated by -dataex-. To install: ssc install dataex
clear
input str13(id t) str32 ROA
"000001" "1998-06-30" ".010836"
"000001" "1998-12-31" ".019397"
"000001" "1999-06-30" ".006223"
"000001" "1999-12-31" ".012104"
"000001" "2000-06-30" ".003578"
"000001" "2000-12-31" ".007535"
"000001" "2001-06-30" ".00262"
"000001" "2001-12-31" ".003349"
"000001" "2002-03-31" ".001662"
"000001" "2002-06-30" ".002332"
"000001" "2002-09-30" ".002622"
"000001" "2002-12-31" ".002601"
"000001" "2003-03-31" ".000944"
"000001" "2003-06-30" ".001541"
"000001" "2003-09-30" ".001629"
"000001" "2003-12-31" ".001641"
"000001" "2004-03-31" ".001009"
"000001" "2004-06-30" ".001406"
"000001" "2004-09-30" ".002047"
"000001" "2004-12-31" ".001418"
end

by t,sort,keep if t strmatch(t,"12-31")

其实，我觉得也可以提取字符串 月份12，然后保留月份是12也行吧

試試