Portfolio variance calculations using mata

[Navruz Khotamov]

Dear Users,

I am using STATA 17.0. I am trying to calculate firm portfolio variance based on the demand growth (return) in the set of countries where these firms have affiliates weighted by sales share (weights). I have data for share of sales for each destination as weight of that destination (w_DEST), and growth rate demand of that destination (g_DEST). My data looks as follow:

* Example generated by -dataex-. For more info, type help dataex
clear
input long tk_code float(year w_CHN w_IND w_USA g_CHN g_IND g_USA)
10000016 2017          . . .18852627         . . .027771465
10000016 2018          . .  .2161741         . . .030423343
10000065 2010 .010309427 .         . .12010965 .          .
10000065 2011  .01036901 .         . .07751633 .          .
10000065 2012  .01096339 .         . .07946409 .          .
end

where tk_code is my firm_id.
I found a mata code written by Abraham Wolde-Tsadick as an answer for a similar questions. Link to that thread (https://www.statalist.org/forums/for...folio-variance).
The code I implement is as follow:

 

. egen fy = group(tk_code year)

. 
. gen VAR = .
. 
. mata
------------------------------------------------- mata (type end to exit) -----------------
: 
: mata clear

: 
: st_view(wt=.,.,tokens("w_ARE - w_ZWE"))

: 
: st_view(ex=.,.,tokens("g_ARE - g_ZWE"))

: 
: st_view(year=.,.,"year")

: 
: st_view(VAR=.,.,"VAR")

: 
: st_view(fy=.,.,"fy")

: 
: p = panelsetup(fy, 1)

: 
:
: 
: for (i=1; i<=rows(p); i++) {
> 
>     r1 = p[i,1]
> 
>     r2 = p[i,2]
> 
>     w = select(wt[r1..r2,.],  colsum(wt[r1..r2,.]:!=.) :== rows(wt[r1..r2,.])   )
> 
>     w = mean(w)
> 
>     s = select(ex[r1..r2,.],  colsum(ex[r1..r2,.]:!=.) :== rows(ex[r1..r2,.])   )
> 
>     s = (s :- mean(s))'(s :- mean(s)) :/ ( rows(s) - 1)
> 
>     VAR[r1..r2,1] = (w*s*w') * J(r2-r1+1,1,1)
> 
> }

: 
: end

The code executes without any errors, but the VAR variable does not change. It still retains all the missing values.
I have two questions:
1) What is wrong with my mata code?
2) In addition to calculation variance, I would like to calculate covariance matrix for a form portfolio depending on the set of countries for each year. How would that be possible?

Thank you very much for your support and guidance in advance.
I am relatively new to Statalist so apologies in advance, and your support is very much appreciated.