Dear Mr. Weiss,
thanks for your quick reply. I looked for a solution with an indiator
variable to signal that the target_cusips occur more than once within
the last five years. As STATA cannot compare the values of target_cusip
within a column, but only within rows, I found out that the "duplicate
tag target_cusip, generate(dupliucate_target)" command might be usefull.
If more than one observation with the same target_cusip occurs, it is
assigned a number how often it occurs (1,2,3,...) in the generated
inficator variable "duplicate_target". This is a convenient solution. My
problem is now that I want to look for duplicates only within the last 5
years, so I programmed the following loop:
sort target_cusip year_of_announcement
local j=1992
forvalues j=1992/2006 {
duplicates tag target_cusip if `j-5'<=year_of_announcement |
`j-4'<=year_of_announcement | `j-3'<=year_of_announcement |
`j-2'<=year_of_announcement | `j-1'<=year_of_announcement,
generate(duplicated_target)
}
Duplicates in terms of target_cusip
generate() must specify new variable
r(110);
It works fine except that also duplicated observations are assigned an
indicator if the consecutivly ordered year_of_announcement have a gap of
more than 5 years as well, which I do not want. Also sometimes no
indicator (0=no duplication, 1,2,3 for multiple duplications) is
assigned with "." as I get the error message "generate() must specify
new variable r(110);" even though the indicator variable did not exist
before in the dataset.
The next problem is to figure out how to search for duplicated cusips
between two different colums, the acquiror_cusip and the target_cusip.
Is it possible to put the cusips in one column by duplicating the
observations to have two otherwise identical observations that differ
only in the cusip's stored in the same column, the pervious
acquiror_cusip and target_cusip? This way I could search for the cusips
in just one column with the "duplicates tag cusip" command.
Thanks for your help and kind regards
Pascal
> -----Ursprüngliche Nachricht-----
> Von: "Martin Weiss" <[email protected]>
> Gesendet: Do. 16.10.08 (13:21)
> An: [email protected]
> Betreff: st: RE: Programming a loop to exclude M&As in the 5 years
prior to an M&A
>
> Not fully understanding your prob, let me comment on two aspects:
> Firstly,
> it is not clear to me why you include the first line and set -local
> j- equal
> to 1987. The next line tells Stata exactly which values -j- should
> acquire
> in the loop as in
>
> **********
> loc i=1
> forv i=3/5{
> di `i'
> }
> *********
>
> The first line is harmless, but also redundant.
>
> Secondly, it is not a good idea to -drop- observations left and
> right.
> Instead, create an indicator variable for your -if- qualifiers.
> Simply do
> something along these lines in your loop:
>
> ************
> g indicator=0
> replace indicator=1 if acquiror_cusip==acquiror_cusip &
> year_of_announcement==`j'-1
> ************
>
> Later you can condition your analyses on this indicator, but can go
> back to
> the original data if needed.
>
>
> BTW, the FAQ on such problems are quite comprehensive and helpful
> (http://www.stata.com/support/faqs/). Also check the contents of the
> Stata
> Journal archive, part of which is free:
> http://www.stata-journal.com/archives.html
>
>
> HTH
> Martin
>
> -----Original Message-----
> From: [email protected]
> [mailto:[email protected]] On Behalf Of
> [email protected]
> Sent: Thursday, October 16, 2008 12:03 PM
> To: [email protected]
> Subject: st: Programming a loop to exclude M&As in the 5 years prior
> to an
> M&A
>
> Hello Statausers,
>
> I am new to STATA and try to program a screening loop to exclude from
> my SDC M&A sample those M&As in which the acquiror or target of an
> M&A
> in a particular year_of_announcement==`j' with j=1987-2006 has been
> the target or acquiror in an M&A in one of the preceeding 5 years.
> This screening is needed to avoid the problem of overlapping events
> in
> the analysis of the stock returns after the M&A.
>
> My problem is how to program that STATA is supposed to look in a
> particular year for all acquirors and targets that are acquirors or
> targets in one of the past 5 years. The target_cusip and
> acquiror_cusip as identifiers are organized in seperate colums. So
> STATA has to go through these two colums in each year and look for
> duplicated cusips that occur already in the year_of_announcement
> looked at.
>
> So far I programmed the following loop, but do not know how to
> indicate that the year_of_announcement==`j' is the base year from
> which Stata has to go back year by year to look for the cusips
> occuring in year j that are duplicated in the previous years :
>
> local j=1987
> forvalues j=1987/2006 {
>
> drop if acquiror_cusip==acquiror_cusip & year_of_announcement==`j'-1
> drop if acquiror_cusip==target_cusip & year_of_announcement==`j'-1
> drop if target_cusip==target_cusip & year_of_announcement==`j'-1
> drop if target_cusip==acquiror_cusip & year_of_announcement==`j'-1
>
> drop if acquiror_cusip==acquiror_cusip & year_of_announcement==`j'-2
> drop if acquiror_cusip==target_cusip & year_of_announcement==`j'-2
> drop if target_cusip==target_cusip & year_of_announcement==`j'-2
> drop if target_cusip==acquiror_cusip & year_of_announcement==`j'-2
>
> drop if acquiror_cusip==acquiror_cusip & year_of_announcement==`j'-3
> drop if acquiror_cusip==target_cusip & year_of_announcement==`j'-3
> drop if target_cusip==target_cusip & year_of_announcement==`j'-3
> drop if target_cusip==acquiror_cusip & year_of_announcement==`j'-3
>
> drop if acquiror_cusip==acquiror_cusip & year_of_announcement==`j'-4
> drop if acquiror_cusip==target_cusip & year_of_announcement==`j'-4
> drop if target_cusip==target_cusip & year_of_announcement==`j'-4
> drop if target_cusip==acquiror_cusip & year_of_announcement==`j'-4
>
> drop if acquiror_cusip==acquiror_cusip & year_of_announcement==`j'-5
> drop if acquiror_cusip==target_cusip & year_of_announcement==`j'-5
> drop if target_cusip==target_cusip & year_of_announcement==`j'-5
> drop if target_cusip==acquiror_cusip & year_of_announcement==`j'-5
> }
>
> I gues a command with a varlist like forvalues j=1987/2006 of varlist
> year_of_announcement {.....}
>
> Kind regards
>
> Pascal Stock
>
> *
> * 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/
>
> *
> * 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/
>
>
>
> -----Ursprüngliche Nachricht Ende-----
#adBox3 {display:none;}
*
* 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/
