Reshape

[Edward Wu]

Hey!

I wonder how could I use reshape, to keep only one observation for every year, without PF and Treatment, and 4 POSTRET and 4 C5 per observation?

* Example generated by -dataex-. To install: ssc install dataex
clear
input float(year Treatment PF POSTRET C5)
1995 0  1   -.06258959   -.05443672
1995 1  1   -.04878007   -.01230235
1995 0 10     .0243682 -.0081219645
1995 1 10    .29053253   -.03238327
1996 0  1   -.02864548   .015554644
1996 1  1  -.008444818   .017122192
1996 0 10   -.00464885   .018017864
1996 1 10     -.174923    .01251025
1997 0  1  -.004441842  -.010664557
1997 1  1     .0945931    .05618503
1997 0 10    .16133273    .05300198
1997 1 10     .2606152   -.02023007
1998 0  1   -.10821836 -.0020010928
1998 1  1  -.027247274   -.01966048
1998 0 10    -.1923901  -.008912033
1998 1 10   -.02608099   -.03519965
1999 0  1    .12610069  -.064325556
1999 1  1    .04858679    .04201986
1999 0 10   -.05919228  -.018305905
1999 1 10     .4874617    .10914348
2000 0  1    .05259665   .009659555
2000 1  1   -.12960732   -.06425608
2000 0 10   .021079615    .04570518
2000 1 10  -.023501126   .025826465
2001 0  1   .062784225  .0022577462
2001 1  1   -.10950977   -.04774466
2001 0 10     .1361238    .01361336
2001 1 10     .3271786    .08065517
2002 0  1  -.035597265   -.02549423
2002 1  1   -.10617667   -.03976572
2002 0 10     .2094567    .05388877
2002 1 10    .07946272    .06824227
2003 0  1    .07292191    -.0452106
2003 1  1    .06298627  -.003759033
2003 0 10    .17196187    .04035956
2003 1 10    .18238416   .027393343
2004 0  1  -.004235355  -.021720765
2004 1  1   .017285408  -.006449955
2004 0 10  .0015900885 -.0090287635
2004 1 10    .03963603  -.007147728
2005 0  1    -.0215021  -.025990536
2005 1  1  -.005774237  -.013356016
2005 0 10    .08841917   .010603352
2005 1 10   .036350943   .021443876
2006 0  1   .032947898  -.026843885
2006 1  1    .08120047  -.015152384
2006 0 10  -.022264436  -.014675376
2006 1 10   .024896633    .02478188
2007 0  1   -.08709063  -.014743632
2007 1  1   -.02828323   -.04145284
2007 0 10    .03734744 -.0012899046
2007 1 10    .05865048  -.008247663
2008 0  1  -.035577077   -.05458454
2008 1  1   .014303045   -.06655561
2008 0 10  .0014968688    .00749701
2008 1 10  -.036107365   -.02814429
2009 0  1  -.008806608   -.01453006
2009 1  1    .07063906   .025290946
2009 0 10     .0871131   .031613152
2009 1 10    .04295475   .017743805
2010 0  1  .0007517777  -.014459757
2010 1  1   -.04635913   -.03373236
2010 0 10   .002602269    .01409031
2010 1 10   .007720996   .006256082
2011 0  1   -.09272193   -.02076576
2011 1  1   -.10027213  -.016807783
2011 0 10  -.007950167    .01116798
2011 1 10  -.010576516    .01448022
2012 0  1   -.06731945   -.02497123
2012 1  1    .04615171   -.02767306
2012 0 10 -.0082327025    .01082672
2012 1 10  -.063822806  -.006309513
2013 0  1    -.0479576  -.019090874
2013 1  1   -.02562972   .036441606
2013 0 10   .024701813   .028115595
2013 1 10   .018392062    .01514944
2014 0  1   -.05042549  -.019402636
2014 1  1   -.04476171   -.02302974
2014 0 10  -.023999644   .016111447
2014 1 10   -.05622108  -.008709985
2015 0  1   -.11762933  -.008463339
2015 1  1   -.10403471  -.014664032
2015 0 10   -.07629266    .01947039
2015 1 10   .037056226   .011590023
end

Thanks for your time!!

[Clyde Schechter]

I don't understand what you want. Any application of -reshape wide- to this data will necessarily give you four variables for each existing variable so that you retain the information in the four observations currently associated with each year. Perhaps you don't really want to -reshape- the data, but somehow combine the values in the four observations in some way? I think it would be best if you hand-worked what the results you want would look like for, say, two of the years you show in your example, and show that along with an explanation of how you arrived at it.

[Edward Wu]

Here are the variables I want:

year POSTRET11(Treatment=1 PF=1) POSTRET110(Treatment=1 PF=10) POSTRET01(Treatment=0 PF=1) POSTRET010(Treatment=0 PF=10) C11(Treatment=1 PF=1) C110(Treatment=1 PF=10) C01(Treatment=0 PF=1) C010(Treatment=0 PF=10)

Thanks a lot!

[Clyde Schechter]

This should do it:

gen suffix = string(Treatment)+string(PF)
drop Treatment PF
rename C5 C
reshape wide POSTRET C, i(year) j(suffix) string

[Edward Wu]

Excellent!!! Have a good one~~~