how to add values for 3 years for each observation

[shweta agarwal]

Hi my data set looks like this

gvkey	fyear	liva	n	liva_2	liva_3	liva_5
1004	1999	-0.81841	1
1004	2001	-0.12328	2	-0.94168
1004	2004	-0.29572	3	-0.41899	-1.2374
1004	2005	0.547	4
1004	2006	-0.0125	5			15
1004	2008	-0.26287	6
1004	2010	0.059042	7
1004	2011	-0.35232	8
1004	2013	0.271471	9
1004	2014	-0.06726	10
1004	2015	-0.06743	11
1004	2016	0.211439	12
1004	2017	-0.03207	13
1004	2018	0.107273	14
1013	1999	8.703646	1
1013	2001	-26.258	2
1013	2002	-5.24997	3
1013	2003	0.170867	4
1013	2004	-1.53502	5
1013	2005	0.201077	6
1013	2006	-2.42768	7
1013	2008	-0.94143	8
1013	2009	-0.24918	9
1034	1999	-0.64264	1
1034	2000	0.686527	2
1034	2001	-1.01673	3
1034	2002	-1.24141	4
1034	2003	0.404838	5
1034	2005	0.769777	6
1034	2006	-0.72259	7
1034	2007	-0.49277	8

I want to generate liva_2, liva_3 and liva_5 and as the the name goes, for liva_2 for each gvkey i want to add the current year and prior year liva.
for liva_3 I will need to add current year and last 2 years liva ( for each gvkey)
for liva_5 I will have to add current year nd last 4 years liva.

please help me with the code.

[Andrew Musau]

xtset gvkey year
*CURRENT YEAR + 1 YEAR LAG
gen wanted1= liva+ L.liva

*CURRENT YEAR + 1 YEAR LAG + 2 YEAR LAG + 3 YEAR LAG
gen wanted2= liva+ L.liva + L2.liva + L3.liva

For an extended lag period, see a trick in the linked thread by Nick Cox attributed to Michael Blasnik.
https://www.stata.com/statalist/arch.../msg00278.html

[Nick Cox]

The data example shows many gaps which are likely to confuse if not handled directly. I would tend to use rangestat from SSC in more or less this way:

* Example generated by -dataex-. For more info, type help dataex
clear
input int(gvkey fyear) float liva
1004 1999  -.81841
1004 2001  -.12328
1004 2004  -.29572
1004 2005     .547
1004 2006   -.0125
1004 2008  -.26287
1004 2010  .059042
1004 2011  -.35232
1004 2013  .271471
1004 2014  -.06726
1004 2015  -.06743
1004 2016  .211439
1004 2017  -.03207
1004 2018  .107273
1013 1999 8.703646
1013 2001  -26.258
1013 2002 -5.24997
1013 2003  .170867
1013 2004 -1.53502
1013 2005  .201077
1013 2006 -2.42768
1013 2008  -.94143
1013 2009  -.24918
1034 1999  -.64264
1034 2000  .686527
1034 2001 -1.01673
1034 2002 -1.24141
1034 2003  .404838
1034 2005  .769777
1034 2006  -.72259
1034 2007  -.49277
end

rangestat (count) count3=liva (sum) sum3=liva, int(fyear -2 0) by(gvkey)

list, sepby(gvkey)

    +------------------------------------------------+
     | gvkey   fyear       liva   count3         sum3 |
     |------------------------------------------------|
  1. |  1004    1999    -.81841        1   -.81840998 |
  2. |  1004    2001    -.12328        2   -.94168998 |
  3. |  1004    2004    -.29572        1   -.29572001 |
  4. |  1004    2005       .547        2    .25127998 |
  5. |  1004    2006     -.0125        3    .23877998 |
  6. |  1004    2008    -.26287        2   -.27537001 |
  7. |  1004    2010    .059042        2   -.20382801 |
  8. |  1004    2011    -.35232        2   -.29327799 |
  9. |  1004    2013    .271471        2   -.08084899 |
 10. |  1004    2014    -.06726        2      .204211 |
 11. |  1004    2015    -.06743        3      .136781 |
 12. |  1004    2016    .211439        3      .076749 |
 13. |  1004    2017    -.03207        3      .111939 |
 14. |  1004    2018    .107273        3      .286642 |
     |------------------------------------------------|
 15. |  1013    1999   8.703646        1    8.7036457 |
 16. |  1013    2001    -26.258        2   -17.554354 |
 17. |  1013    2002   -5.24997        2   -31.507969 |
 18. |  1013    2003    .170867        3   -31.337102 |
 19. |  1013    2004   -1.53502        3    -6.614123 |
 20. |  1013    2005    .201077        3    -1.163076 |
 21. |  1013    2006   -2.42768        3    -3.761623 |
 22. |  1013    2008    -.94143        2     -3.36911 |
 23. |  1013    2009    -.24918        2     -1.19061 |
     |------------------------------------------------|
 24. |  1034    1999    -.64264        1   -.64263999 |
 25. |  1034    2000    .686527        2    .04388702 |
 26. |  1034    2001   -1.01673        3   -.97284293 |
 27. |  1034    2002   -1.24141        3    -1.571613 |
 28. |  1034    2003    .404838        3    -1.853302 |
 29. |  1034    2005    .769777        2     1.174615 |
 30. |  1034    2006    -.72259        2    .04718697 |
 31. |  1034    2007    -.49277        3   -.44558302 |
     +------------------------------------------------+

Given so many gaps, it's a decision for the researcher whether means work better than sums.

Different lags require different runs.