Or you could do it by logs if your variables takes on all positive values -
e.g.
k y
1. 1 5
2. 2 9
3. 2 3
4. 3 1
5. 3 5
6. 3 6
7. 4 9
8. 4 1
9. 4 10
10. 4 2
11. 5 1
12. 5 5
13. 5 10
14. 5 8
15. 5 10
. gen double z=log(y)
. sort k
. egen sz=sum(z),by(k)
. gen p=exp(sz)
Then p contains the products, by k (subject to rounding errors).
. list k y p
k y p
1. 1 5 5
2. 2 9 27
3. 2 3 27
4. 3 1 30
5. 3 5 30
6. 3 6 30
7. 4 9 180
8. 4 1 180
9. 4 10 180
10. 4 2 180
11. 5 1 3999.999
12. 5 5 3999.999
13. 5 10 3999.999
14. 5 8 3999.999
15. 5 10 3999.999
Al Feiveson
-----Original Message-----
From: Nick Cox [mailto:[email protected]]
Sent: Thursday, June 12, 2003 11:37 AM
To: [email protected]
Subject: st: RE: Product within a variable
Mario Gamboa-Cavazos
> I am struggling to compute a constant of products within a variable.
> What I am trying to do is something very similar to what the 'egen'
> command with the option 'sum' and 'by(.)' does, except that
> I would need
> something like a multiplication option.
>
> My dataset is a panel of monthly stock returns and I want
> to compute the
> follwing:
>
> - For each company and for each year: The product of
> (1+r_t) where r is
> the monthly stock return in month t. More specifically, my dataset
> looks like:
>
> Company Year Month Return What I need:
> Pi^12_1(1+r_t)
> 1 1990 1 .33
> constant
> for 1990 firm 1
> 1 1990 2 .44
> constant
> for 1990 firm 1
> . . . .
> constant
> for 1990 firm 1
> . . . .
> constant
> for 1990 firm 1
> . . . .
> constant
> for 1990 firm 1
> 1 1990 12 .55
> constant
> for 1990 firm 1
> 1 1991 1 .23
> constant
> for 1991 firm 1
> 1 1991 2 .43
> constant
> for 1991 firm 1
> . . . .
> constant
> for 1991 firm 1
> . . . .
> constant
> for 1991 firm 1
> . . . .
> constant
> for 1991 firm 1
> 1 1991 12 .43
> constant
> for 1991 firm 1
> 2 1990 1 .87
> constant
> for 1990 firm 2
> 2 1990 2 .98
> constant
> for 1990 firm 2
> . . . .
> constant
> for 1990 firm 2
> . . . .
> constant
> for 1990 firm 2
> . . . .
> constant
> for 1990 firm 2
> 2 1990 12 .65
> constant
> for 1990 firm 2
> . . . . .
> . . . . .
> . . . . .
>
> and so on for nearly 2,000 companies and ten years.
>
> If any of you has any good suggestions I would greatly
> appreciate it,
> especially if it can deal with missing return data.
I don't know what white magic you want for missing
values. I think you will need to spell that out.
I would try something like this:
bysort firm year (month) : gen product = return[1]
by firm year : replace product = return * product[_n-1]
But for a fuller implementation with more bells and
whistles, Phil Ryan has -egen- functions which should help.
Type
. findit egen product
Nick
[email protected]
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
