[Nick Cox]

Please bear in mind that as someone outside your field I don't have any precise knowledge of what portfolio, event, etc. mean at all.

But levelsof may help.

But why do you need the company identifiers any way? They're the companies in one of your bins: isn't the bin identifier enough?

[Huw Simpson]

My apologies, this is an event study analysis where i look at what happens to stock returns (the abnormal returns) after a shock to the market (marked by an event day).
The company's (10,14,...122) together form the portfolio/group of firms which i want to total abnormal returns for, on each of the days 1->20 after the event date.

Ideally I'd want to use something like the following, but the syntax is invalid.

keep company_id if percentiles==10

And then run the code

forval d = 0/20 {      
capture {          
total abnormal_return if days==`d'          
local D : di %02.0f `d'          
estimates store CAR`D'1W

[Nick Cox]

Sorry, but I don't see what that syntax would even do if it were valid. keep with variables keeps only those variables. You don't want to throw out all variables other than company identifier so far as I can see.

Conversely, if you keep just one decile bin using an if condition, then the companies left are those you want any way, are they not? Equivalently your last block of code doesn't depend on, or even use, the information on which companies are included.

[Huw Simpson]

The deciles (top and bottom) are made on day 0, and then the same companies which make up those deciles are needed on each of the following days 1 to 20. If i keep just one decile bin, that keeps only the observations for the company identifiers on day 0. While i want only those companies for all the days.

Since i want to run this over 207 events, I guessed that the best way to code the problem would be to have stata recognise which companies belong to the top/bottom decile bins, then cut down the dataset to only those selected firms, before totaling the abnormal returns. But i'd like it to be able to do this automatically without me inputting the company identifiers for the top and bottom deciles each time.

[Nick Cox]

OK; that helps. I've already suggested levelsof in #16.

[Huw Simpson]

Thank you for all your help so far Nick,
I've spent the last day playing around with 'levelsof' and become rather stuck again.

Following Clyde's reccomendations on "https://www.statalist.org/forums/forum/general-stata-discussion/general/103576-use-elements-of-a-local-as-argument-for-inlist-function"

I tried the egen function anymatch(), but i have the problem of the invalid numlist (since the company_ids aren't separated by commas).

egen var3 = anymatch(company_id), values(r(winner))
*values() invalid -- invalid numlist   r(121)*

I tried to use the local macro 'winner', and then transform it so that i would be able to
use the 'inlist' which needs a list of arguments seperated by commas. But i get the following error:

levelsof company_id if percentiles==10, local(winner)
replace r(winner) = subinstr(r(winner), " ", ",",.)
*r101 factor-variable and time-series operators not allowed*

I'm at a loss as to how i can solve the latest error!

[Nick Cox]

In general the only way to change r-class results is to overwrite them with an r-class command. replace is for variables in your data (only).

But a local macro isn't an r-class result any way.

I suggest

levelsof company_id if percentiles==10, local(winner)  
local winner : subinstr local winner " " ",", all

This should also work for modest numbers of values:

local winner = subinstr(`"`winner'"', " ", ",", .)

[Huw Simpson]

Thank you again for all your help. That coding works perfectly.

[Katrine Marie Jensen]

Hi all,

I'm working with panel dataset containing 195 countries (see example below).

However, I only need a subset of countries, which is why i tried dropping some of the countries using

drop if inlist(country,"Albania","Algeria","Andorra","Angu illa","Antiqua_and_Barbuda","Australia","Austria", "Azerbaijan")

I do not get an error code, but 0 observations are deleted:

. drop if inlist(country,"Albania","Algeria","Andorra","Angu illa","Antiqua_and_Barbuda","Australia","Austria", "Azerbaijan")
(0 observations deleted)

Has anyone experienced a similar problem and know the solution?

Thank you in advance!

* Example generated by -dataex-. To install: ssc install dataex
clear
input int year str25 country float rainfallmm
1988 " Afghanistan"     27.25257
1989 " Afghanistan"     26.55535
1990 " Afghanistan"    30.222525
1991 " Afghanistan"     36.28749
1992 " Afghanistan"     34.01302
1993 " Afghanistan"    26.423775
1994 " Afghanistan"     28.51853
1995 " Afghanistan"    25.074846
1996 " Afghanistan"    26.750185
1997 " Afghanistan"    27.604815
1998 " Afghanistan"     29.87177
1999 " Afghanistan"     24.60539
2000 " Afghanistan"    16.817682
2001 " Afghanistan"    15.403734
2002 " Afghanistan"     24.32237
2003 " Afghanistan"     27.39291
2004 " Afghanistan"     24.06963
2005 " Afghanistan"     29.60997
2006 " Afghanistan"     27.12046
2007 " Afghanistan"     25.32968
2008 " Afghanistan"    21.505365
2009 " Afghanistan"    31.106476
2010 " Afghanistan"    22.452065
2011 " Afghanistan"     25.53845
2012 " Afghanistan"     31.52742
2013 " Afghanistan"     28.91377
2014 " Afghanistan"    25.967264
2015 " Afghanistan"    33.449856
2016 " Afghanistan"    28.248095
1988 " Albania"        66.524025
1989 " Albania"         63.91075
1990 " Albania"         67.68621
1991 " Albania"         76.48907
1992 " Albania"         66.47468
1993 " Albania"        71.772896
1994 " Albania"         83.85102
1995 " Albania"         89.73432
1996 " Albania"         95.78157
1997 " Albania"         70.55569
1998 " Albania"         80.19603
1999 " Albania"         88.70529
2000 " Albania"         62.89793
2001 " Albania"         64.84759
2002 " Albania"         87.40131
2003 " Albania"         79.38502
2004 " Albania"         86.85091
2005 " Albania"         84.01238
2006 " Albania"         81.72401
2007 " Albania"         82.57442
2008 " Albania"         81.57327
2009 " Albania"        105.86436
2010 " Albania"        107.84811
2011 " Albania"         59.83802
2012 " Albania"         81.53146
2013 " Albania"         83.96725
2014 " Albania"         106.4032
2015 " Albania"         78.13724
2016 " Albania"         91.31202
1988 " Algeria"         6.719759
1989 " Algeria"         5.770886
1990 " Algeria"         7.478644
1991 " Algeria"         7.344397
1992 " Algeria"          7.33942
1993 " Algeria"         5.201967
1994 " Algeria"         7.242153
1995 " Algeria"         6.266937
1996 " Algeria"         8.494614
1997 " Algeria"          7.31943
1998 " Algeria"         5.362187
1999 " Algeria"         7.439788
2000 " Algeria"          4.36128
2001 " Algeria"          5.03195
2002 " Algeria"         5.633926
2003 " Algeria"         8.935742
2004 " Algeria"          8.43283
2005 " Algeria"         5.777586
2006 " Algeria"         8.015256
2007 " Algeria"         7.325529
2008 " Algeria"         6.621873
2009 " Algeria"          7.65311
2010 " Algeria"         7.745368
2011 " Algeria"         7.957712
2012 " Algeria"         6.898958
2013 " Algeria"         7.169009
2014 " Algeria"         6.193111
2015 " Algeria"         6.511808
2016 " Algeria"         5.449697
1988 " American Samoa" 274.34576
1989 " American Samoa"  240.0255
1990 " American Samoa" 198.81413
1991 " American Samoa" 264.01498
1992 " American Samoa"   268.401
1993 " American Samoa"  252.2328
1994 " American Samoa" 300.87372
1995 " American Samoa"  248.7933
1996 " American Samoa"  271.2207
1997 " American Samoa" 209.18135
1998 " American Samoa" 113.25059
1999 " American Samoa"  237.8423
2000 " American Samoa" 246.33046
end

[Rich Goldstein]

your data has spaces in front of the name but your code does not; see

help trim

to get rid of the spaces

[Katrine Marie Jensen]

Thank you so much Rich - sometimes you don't see the most obvious errors yourself .

[Nick Cox]

There is a leading space in your data, but not in your search strings.

[Elena Califano]

Hello everyone,

I have a time series with X years. I want to drop only the observations for variable year in year < 2013 & year > 2019.

I tried: drop if year<2013 & year>2019 but I got r(109) - type mismatch

Any suggestion?

[Rich Goldstein]

it appears that your "year" variable is a string (which means you. need quotes around the values) but, in addition, year cannot be both <2013 and>2019 at the same time; you can still do as is:

keep in in range(year, "2014","2018")

or you could -destring- the variable and use any of a number of ways to do it

the above makes assumptions about your data, assumptions that could not be tested because you did not supply a data sample; see

h dataex

and also please read the FAQ, esp. about CODE blocks and -dataex-