Re: st: Time Series Operators and monthly returns after collapse

[Scott Merryman <scott.merryman@gmail.com>]

Are you asking if the -collapse- f6.lret6- is the 6 month return or 1
month return?  Isn't it just the average of the last 6 months?  In the
following simple example, is the mean of lnret what you expect?

clear
set obs 12
gen t = _n
tsset t
set seed 12304
gen x = ceil(runiform()*10)
gen  lnret = ln(x/ l6.x)
preserve
collapse (mean) f6.lnret
l
restore
sum lnre
sum f6.lnre

Scott


On Wed, Aug 22, 2012 at 11:33 AM, Daniel Brodback <schmani@gmx.de> wrote:
> Dear all,
>
> just a quick question because I got quite a bit confused in my analsis with time series operators and monthly returns. I am ranking stocks based on their prior 6 month return and then invest for an additional 6 months.
>
> I generate the log-return of the previous 6 month with the following command: generate lret6 = ln(price / l6.price)
>
> Later, I collapse in a foreach-loop:
> collapse (mean) f6.lret6, by(date `var')
>
> in order to obtain the mean return of my panel.
>
> Now my question: Do I now have the 6month-return or rather the 1 month return? So far I assumed this would be the 6 month return and hence divided by 6 in order to obtain a monthly return. But after I rethought my code I am not sure whether this was the right thing to do.
>
> Any input to get me back on the right track is highly appreciated.
>
> Thanks,
> Daniel

*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/