st: Re: Portfolio returns per month, drop return duplicates

["M@rk " <krap_m@hotmail.com>]

Thanks for the replies! @ Nick: I now have the data that I need to proceed. 
Its probably not the way to delete duplicates, but it worked for me. 
@Sergiy: Collapse works, but I have multiple portfolios. Not only PRB1, but 
als PRB2, PRB3 etcetera (120 portfolios).  It uses mean values, which is no 
problem as the return figures in a month are the same for each observation. 
Can I combine multiple portfolios (/variables) when using this command? Or 
is it only applicable for 1 variable?

Kind Regards,

Mark Krap

-----Oorspronkelijk bericht----- 
From: Sergiy Radyakin
Sent: Monday, November 25, 2013 5:52 PM
To: statalist@hsphsun2.harvard.edu
Subject: Re: st: Portfolio returns per month, drop return duplicates

On Mon, Nov 25, 2013 at 7:27 AM, M@rk <krap_m@hotmail.com> wrote:
> Dear Users,
>
> For my research I need to have portfolio returns for each month from 1926
> till 2012. I already calculated the weighted returns for each portfolio 
> and
> summed them up (by date) by creating a  new variable (PRB1). The only
> problem is that this sum is not given once for each month, but is given 
> for
> each stock observation in a month. So if the total of the weighted returns
> is, let's say, 0.026 in March 2006 its mentioned for each stock 
> observation
> in March 2006. I tried to downsize this by using the duplicates drop [in]
> command, but I get the error message 'invalid obs no'. Does anyone know 
> how
> I can get the portfolio returns once per month, as for my research I only

Besides what Nick has already suggested:

#1
sort year month
by year month: keep if _n==1

#2
sort year month
collapse PRB1, by(year month)

note that you need year unless your month is cmc.


In the command that you quoted "duplicates drop PRB1" you are relying
on a random fact that your portfolio returns would vary uniquely by
time, which is commonly true, but not necessary (e.g. in the period of
stability you could have returns for e.g. 3 months being same).You
should seek duplicates in terms of time, not values!

Best, Sergiy

> need portfolio returns per month from 1926 till 2012 in order to do the
> regressions. I am using Stata 12 SE (Windows 7 64-bit).
>
> I look forward to the reactions.
>
>
> Kind Regards,
>
> Mark Krap
>
> *
> *   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/ 

*
*   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/