Re: st: Cumulative Return Calculation for Recurring Periods

[Richard Herron <richard.c.herron@gmail.com>]

This generates rolling three-month returns.

***
clear
set obs 30
generate i = int((_n - 1) / 10) + 1
generate t = mod((_n - 1), 10) + 1
generate r = 0.0 + (0.00 + 0.10) * runiform()
xtset i t

generate sumlogr = sum(ln(1 + r))
generate r3 = exp(s3.sumlogr) - 1
***

If you use -xtset- then you can use -s.-, which is seasonal
difference, here three months. This approach "throws away" the first
month's return, but codes/computes quickly.

On Wed, Jun 19, 2013 at 2:27 PM, Edward Crawley <edward.crwly@gmail.com> wrote:
> Dear listers,
>
> I am trying to calculate cumulative returns for recurring periods in
> my panel dataset;
>
> date company return
> 2001m1 1  .
> 2001m2 1 .14425
> 2001m3 1 -.215709
> 2001m4 1 -.160343
> 2001m5 1 .390866
> 2001m6 1 -.014815
> 2001m1 2  .
> 2001m2 2 .315081
> 2001m3 2 -.392042
> 2001m4 2 -.127833
> 2001m5 2 .646627
> 2001m6 2 .133531
> 2001m1 3  .
> 2001m2 3 .184734
> 2001m3 3 -.036814
> 2001m4 3 -.29605
> 2001m5 3 .457318
> 2001m6 3 -.060292
> .
> .
> .
> .
>
> For instance, for each firm, I need to calculate cumulative returns
> for recurring 3 periods ( e.g. from 2001m1 to 2001m4 and from 2001m2
> to 2001m5 and ... etc.).
>
> I used the following code, but it didn't give the desired result;
>
> bys company (date): g c=sum(ln(return+1))
> replace c=exp(c)-1
> rename c cumreturn
>
> I appreciate any help,
>
> Best regards,
>
> Edward
> *
> *   For searches and help try:
> *   http://www.stata.com/help.cgi?search
> *   http://www.stata.com/support/faqs/resources/statalist-faq/
> *   http://www.ats.ucla.edu/stat/stata/
*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/faqs/resources/statalist-faq/
*   http://www.ats.ucla.edu/stat/stata/