Summing a variable over a set number of observations

[Anthony Voyage]

Hi there,
I am certain this is a stupid question (but I can't figure it out!). I have 2 variables in my dataset: date, recorded monthly, and value.I want to make a new variable (call it var) that =1 if the sum of the values over the past 36 months are positive (and zero otherwise).

I am struggling with the 36 months part. I cannot figure out how to sum up the values without having something like:

gen yy = year(date)
gen mm = month(date)
sort yy mm

gen var = value[_n-1] +value[_n-2] + ... + value[_n-36]
replace var = 1 if var>0
replace var =0 if var!=1

Is there a simpler way to do this?

Thanks for any help.

[Richard Williams]

Do you want the first 35 months to be missing? Then month 36 would be based on months 1-36, month 37 would be based on months 2-37, 38 would be based on 3 through 38, etc? How many months does the data go?

[Anthony Voyage]

Correct. I am going to merge this data later with another file. The first 36 months of the data is not actually important (so missing values are fine). The 37th month will be based on months 1-36. That is, I would like to sum value from [_n-36] to [_n-1] inclusive for every value[_n].

The data goes from January 1985 to December 2013.

[Nick Cox]

The advice given you in answer to a previous question still applies. The sum you want is a difference between two cumulative sums. See post #9 in

http://www.statalist.org/forums/foru...hin-a-variable

[Paul T Seed]

Assuming -date- is a standard Stata date, it is also possible to save a couple of lines by using

sort date

instead of

gen yy = year(date)
gen mm = month(date)
sort yy mm

For the main problem, the following code will work (based on Nick Cox's previous post).

gen sum = sum(x)
gen sum_36 = (x-x[_n-36])
replace sum_36 = x if _n == 36
gen sum_36_pos = ( sum_36 >0 ) if  sum_36<.

[Nick Cox]

Paul: Typos there. You need to use sum once you have calculated it. So references to x after line 1 should all be to sum.

[Anthony Voyage]

Sorry Nick, I forgot about the previous question.
Thank you all for the help.