Hi everyone,
I am trying to generate a rolling sum over the last 12 observations in an unbalanced panel. I tried movingsumm and egen commands but did not succeed, thus I thought about creating a brief loop which also worked elsewhere in a similar intention. However, I am unable to retrieve the desired moving average. Note: I - tsset - my variables already prior to that calculus which are named pvar and tvar.
Data looks like this:
The code I tried is down below. It should calculate the sum over the last 12 months, which is 0.41 for sum_alpha in 2006m3 and 0.47 in 2006m4. I also tried replacing the missing values by 0 first, but this neither helped.
Is there something I am missing?
Alternatively, I tried this which neither worked (again I - tsset - the panel correctly).
It seems such a simple request that I must be missing something in my command lines. Any help is highly appreciated!! Many thanks in advance!
I am trying to generate a rolling sum over the last 12 observations in an unbalanced panel. I tried movingsumm and egen commands but did not succeed, thus I thought about creating a brief loop which also worked elsewhere in a similar intention. However, I am unable to retrieve the desired moving average. Note: I - tsset - my variables already prior to that calculus which are named pvar and tvar.
Data looks like this:
:
clear input byte pvar str10 tvar double alpha 1 "2005m1" . . 1 "2005m2" . . 1 "2005m3" . . 1 "2005m4" 0.03 1 "2005m5" 0.02 1 "2005m6" 0.01 1 "2005m7" 0.04 1 "2005m8" 0.08 1 "2005m9" 0.04 1 "2005m10" 0.08 1 "2005m11" 0.03 1 "2005m12" 0.04 1 "2006m1" 0.01 1 "2006m2" 0.02 1 "2006m3" 0.01 1 "2006m4" 0.09 2 "1996m3" . 2 "1996m4" . end
:
sort pvar tvar
gen sum_alpha = alpha
forvalues i = 1/11 {
by pvar: replace sum_alpha = sum_alpha+sum_alpha[_n-`i']
}
Alternatively, I tried this which neither worked (again I - tsset - the panel correctly).
:
sort pvar tvar by pvar: egen sum_alpha = sum(L1.alpha-L12.alpha)
