Temporal decomposition - Function using data from 2 different rows of a table

[Rutger Hoekstra]

I want to do a temporal decomposition using STATA. Assume the following data below where x= y*z

year	x	y	z
2015	8	4	2
2016	12	3	4
2017	16	4	4

It is not necessary to show the entire decomposition but a crucial step is calculating the changes in the variables:

year	x	y	z	delta x	delta y	delta
2015	8	4	2
2016	12	3	4	4	-1	2
2017	16	4	4	4	1	0

But this involves subtracting one row from the previous row . However, nearly all the STATA functions refer to expression within a row. How can I create a loop which calculates and generates the variables delta x,y,z?

Thanks!

Rutger

[FernandoRios]

Hi Rutget
Im going to assume that you ONLY have time series data. If that is the case, the following should do:

tsset year
gen dx=d.x
gen dy=d.y
gen dz=d.z

You can also see more option for time series operations at "https://www.stata.com/manuals/u11.pdf#u11.4.4"
HTH
Fernando

[Rutger Hoekstra]

Thanks Fernando!

It is in fact a decomposition for 140 countries in the world. i.e. the data has the structure below.

countrycode	year	x	y	z
FRA	2015	8	4	2
FRA	2016	12	3	4
FRA	2017	16	4	4
USA	2015	100	10	10
USA	2016	121	11	11
USA	2017	150	15	10

etc, etc

What would be the best way to loop over the countries and then do the years?

Kind regards,

Rutger

[FernandoRios]

Well, that is a slightly different problem but with the same solution.
You need to set up the data as panel. So instead of using tsset, you need xtset countrycode year.
This assumes that countrycode is numeric.
HTH

[Rutger Hoekstra]

Thanks, worked great!