How to generate cumulative sum of changes of a variable in a group across time

[Dorothea Ramizo]

Dear All,

I'm new on Statalist and I'm still figuring out how to use dataex so apologies for my dataset example below. reporter_iso and partner_iso are string variables. Meanwhile, id is generated by egen=group(reporter_iso partner_iso). I need to generate the variable "sum" which cumulatively adds up the changes in "TA_envi_tot" across reporter-partner pairs and years.

I tried these codes but I don't get to generate the values in the "sum" column below:

bysort id (year): gen sum=TA_envi_tot[_n] + TA_envi_tot[_n+1] if TA_envi_sum[_n]!=TA_envi_tot[_n-1]
bysort id (year): replace sum = sum[_n-1] if missing(sum)

id	reporter_iso	partner_iso	year	TA_envi_tot	sum
3271	ATG	DEU	1981	0	0
3271	ATG	DEU	1982	0	0
3271	ATG	DEU	1983	0	0
3271	ATG	DEU	1984	36	36
3271	ATG	DEU	1985	36	36
3271	ATG	DEU	1986	36	36
3271	ATG	DEU	1987	67	103
3271	ATG	DEU	1988	67	103
3271	ATG	DEU	1989	67	103
4217	BDI	BEL	1981	3	3
4217	BDI	BEL	1982	3	3
4217	BDI	BEL	1983	3	3
4217	BDI	BEL	1984	35	38
4217	BDI	BEL	1985	35	38
4217	BDI	BEL	1986	35	38
4217	BDI	BEL	1987	35	38
4217	BDI	BEL	1988	36	74
4217	BDI	BEL	1989	36	74
4217	BDI	BEL	1990	36	74

Thanks a lot for your help!

[Nick Cox]

Cross-posted and answered at https://stackoverflow.com/questions/...-across-time-u

Please note our policy on cross-posting, which is that you are asked to tell us about it. https://www.statalist.org/forums/help#crossposting

[Dorothea Ramizo]

Hi Nick, Thanks a lot for the guidance. The policy on cross-posting is noted.