cumulative returns

[Felipe Damasceno]

I'm using stata 13 with windows 10 OS.

I want to calculate cumulative returns by firm in a certain period of time. I thought about calculate the sum of ln(1+ret), but as I have negatives, I don't think this is the way to go.
Does anyone have any idea about how to do this?
My date variable represents the date for the last earnings announcement. I want to calculate the cumulative returns for this date.
Following is a example of my data

* Example generated by -dataex-. To install: ssc install dataex
clear
input long permno float(abn_ret date_ea)
10001 .0009003075 20528
10001  -.01267928 20528
10001  .006637205 20528
10001 -.007484646 20528
10001  -.02332831 20528
10001  .011132848 20528
10001   .02164412 20528
10001  -.01230608 20528
10001   .01596631 20401
10001 -.027130686 20401
10001  .004668585 20401
end
format %td date_ea

[Jesse Wursten]

What kind of returns are they? Percentages? Log returns? How come you have multiple returns for the same day?

[Felipe Damasceno]

They are percentages.

date_ea refers to the date of the next earnings announcement. I want to calculate the cumulative returns until this date, therefore I created this variable that gives the same date for all the returns I want to put together.

[Jesse Wursten]

gen cumReturn_a = 1
gen cumReturn_b = 1

bysort permno (date_ea): replace cumReturn_a = cumReturn_a[_n-1] * (1+abn_ret) if _n > 1
bysort permno date_ea  : replace cumReturn_b = cumReturn_b[_n-1] * (1+abn_ret) if _n > 1

bysort permno (date_ea): gen totalReturn_a = cumReturn_a[_N] - 1
bysort permno date_ea  : gen totalReturn_b = cumReturn_b[_N] - 1

I'm not entirely sure what you want, so I coded two options.

_a simply calculates the total return for this permno
_b calculates the total return per permno per date_ea

cumReturn_* can best be seen as a stock variable, normalized to one. I.e. it shows your stock now relative to what you started with.
totalReturn_* is the actual percentage change between start and end. I think this might actually be an incorrect way to do it and it should in fact be a division of some sorts, but you probably know that better than I do (I'm not a finance guy)?

Does that help?

[Felipe Damasceno]

it does, but can you explain what the if option does?

[Jesse Wursten]

it does, but can you explain what the if option does?

The if qualifier prevents all the observations from being set to missing.

When within a <by: statement>, _n provides the number of the observation within the group. Hence, if _n>1 stops the command from doing anything on the very first observation within a group. This is necessary because *Return_*[_n-1] doesn't exist if _n == 1.

Does that make sense? You can try it without the _n > 1 and you'll see the difference, specifying _n == 1 will show you what's going wrong.

EDIT: Changed option to qualifier as the third word as per #7

[Nick Cox]

Note if is not an option; here it's a qualifier.

[Felipe Damasceno]

Jesse,

Here is the thing. With this code I´m getting the cumulative result at that point. I´d like to get the cumulative result for all observation of the same firm with the same date_ea.

[Jesse Wursten]

Jesse,

Here is the thing. With this code I´m getting the cumulative result at that point. I´d like to get the cumulative result for all observation of the same firm with the same date_ea.

Isn't that what you get in totalReturn_b? If not I don't really understand what you're asking for... Can you perhaps include a simple example?

[Felipe Damasceno]

For what I understood, the totalReturn is the percentual difference between two observations. But, I don't got price to calculate returns. I have returns for each day and want to calculate the cumulative return for the period. What I did is to create a new variable called date_ea that gives the same date to every return I want to accumulate together.
Here is a better example of my data. I'd like to calculate all returns together that have the same data_ea and. if it's possible, get the value in the EPS date.

* Example generated by -dataex-. To install: ssc install dataex
clear
input long permno float(abn_ret anndats date_ea EPS)
11691    .012172948 17465 17465 1.14
11691  -.0089746155 17468 17567    .
11691   -.013763692 17469 17567    .
11691     .02741313 17470 17567    .
11691    .011868282 17471 17567    .
11691    .006654933 17472 17567    .
11691    .003023942 17475 17567    .
11691   .0031912194 17476 17567    .
11691     .00923176 17477 17567    .
11691   .0034949144 17478 17567    .
11691   -.014358364 17479 17567    .
11691   -.006314079 17482 17567    .
11691   -.008196113 17483 17567    .
11691  -.0008487114 17484 17567    .
11691  -.0008843668 17485 17567    .
11691   -.009579462 17486 17567    .
11691  -.0009021742 17489 17567    .
11691   -.013464108 17490 17567    .
11691    .007748195 17491 17567    .
11691   -.008166206 17493 17567    .
11691    .009179472 17496 17567    .
11691   -.000940704 17497 17567    .
11691    .019943254 17498 17567    .
11691  -.0020121557 17499 17567    .
11691    .024047587 17500 17567    .
11691   -.004227172 17503 17567    .
11691    .007689217 17504 17567    .
11691    .000648663 17505 17567    .
11691 -.00044146396 17506 17567    .
11691      .0148687 17507 17567    .
11691     .00501804 17510 17567    .
11691    .004268157 17511 17567    .
11691   .0021716324 17512 17567    .
11691    .005686784 17513 17567    .
11691   -.004705402 17514 17567    .
11691   -.005461689 17517 17567    .
11691   .0008598414 17518 17567    .
11691    .006586414 17519 17567    .
11691   .0022872891 17520 17567    .
11691   -.013708348 17521 17567    .
11691   .0013388155 17524 17567    .
11691   -.016903091 17526 17567    .
11691    .016134791 17527 17567    .
11691     .01082256 17528 17567    .
11691   -.022595154 17531 17567    .
11691   -.014054383 17533 17567    .
11691    -.01058653 17534 17567    .
11691    -.01157745 17535 17567    .
11691      -.014192 17538 17567    .
11691   -.008841429 17539 17567    .
11691    -.02037946 17540 17567    .
11691   -.013391983 17541 17567    .
11691   -.014078217 17542 17567    .
11691    .024091717 17545 17567    .
11691   -.016190942 17546 17567    .
11691   -.015647493 17547 17567    .
11691    .004940304 17548 17567    .
11691    .012980293 17549 17567    .
11691   -.018685527 17553 17567    .
11691    .004865461 17554 17567    .
11691    .008396044 17555 17567    .
11691    .011103337 17556 17567    .
11691   -.015268028 17559 17567    .
11691     .01705285 17560 17567    .
11691    .011386285 17561 17567    .
11691   -.013333277 17562 17567    .
11691   -.013327525 17563 17567    .
11691   -.012675485 17566 17567    .
11691     .04067366 17567 17567  .85
end
format %td anndats
format %td date_ea

[Jesse Wursten]

I'm sorry but I still have no idea what it is you need. Can you give a simple example? E.g. two date_ea's with three returns each and then your manual calculation (with intermediary steps) on how to get the cumulative returns you want.