topic Re: Standard deviation of ROA with a rolling window of the past five years in SAS Programming

Standard deviation of ROA with a rolling window of the past five years

joon1 — Thu, 21 Mar 2019 13:50:44 GMT

Dear Madam/Sir,

I would like to construct standard deviation of ROA with a rolling window of the past five years.

After referring several comments in this forum, I made the following command and had the error messages (attached).

proc sql;

create table m24 as select *, (std(a.roa) from have where a.fyear between

a.fyear-6 and a.fyear-1 and a.gvkey) from m23 as a;

quit;

Obs	gvkey	FYEAR	roa
1	1004	1999	0.04745
2	1004	2000	0.02640
3	1004	2001	-0.08299
4	1004	2002	-0.01807
5	1004	2003	0.00494
6	1004	2004	0.02110
7	1004	2005	0.03592
8	1004	2006	0.05494
9	1004	2007	0.05517
10	1004	2008	0.05710
11	1004	2009	0.02973
12	1004	2010	0.04098
13	1004	2011	0.03084
14	1004	2012	0.02574
15	1004	2013	0.03314
16	1004	2014	0.00673
17	1004	2015	0.03308
18	1004	2016	0.03756
19	1013	2000	0.21864
20	1013	2001	-0.51514
21	1013	2002	-1.00070
22	1013	2003	-0.05914
23	1013	2004	0.01148
24	1013	2005	0.07212
25	1013	2006	0.04077
26	1013	2007	0.06023
27	1013	2008	-0.02181
28	1013	2009	-0.35301
29	1013	2010	0.04205

The error logs are attached.

Any of your help will be highly appreciated.

Sincerely,

Joon

Re: Standard deviation of ROA with a rolling window of the past five years

novinosrin — Thu, 21 Mar 2019 14:26:20 GMT

Hi @joon1 Can you please post a sample of the expected Output to your sample. This may help somebody test their code against your code and make it reusable for the wider community.

Re: Standard deviation of ROA with a rolling window of the past five years

joon1 — Thu, 21 Mar 2019 16:04:02 GMT

Thanks for your response, novinosrin.

The expected output is attached.

Your help will be highly appreciated.

Joon1

Re: Standard deviation of ROA with a rolling window of the past five years

novinosrin — Thu, 21 Mar 2019 16:47:33 GMT

Hi @joon1 See if this works

data have;
input Obs gvkey FYEAR roa;
drop obs;
cards;
1	1004	1999	0.04745
2	1004	2000	0.0264
3	1004	2001	-0.08299
4	1004	2002	-0.01807
5	1004	2003	0.00494
6	1004	2004	0.0211
7	1004	2005	0.03592
8	1004	2006	0.05494
9	1004	2007	0.05517
10	1004	2008	0.0571
11	1004	2009	0.02973
12	1004	2010	0.04098
13	1004	2011	0.03084
14	1004	2012	0.02574
15	1004	2013	0.03314
16	1004	2014	0.00673
17	1004	2015	0.03308
18	1004	2016	0.03756
19	1013	2000	0.21864
20	1013	2001	-0.51514
21	1013	2002	-1.0007
22	1013	2003	-0.05914
23	1013	2004	0.01148
24	1013	2005	0.07212
25	1013	2006	0.04077
26	1013	2007	0.06023
27	1013	2008	-0.02181
28	1013	2009	-0.35301
29	1013	2010	0.04205
;
proc sql;
create table want as
select gvkey,Fyear,roa,ifn(n=6,std(t),.) as standard_dev 
from
(select a.*,ifn(max(b.fyear)=b.fyear,.,b.roa) as t,n(a.fyear) as n  
from have a inner join have b
on a.gvkey=b.gvkey and   a.fyear-5<= b.FYEAR<= a.fyear
group by a.gvkey,a.fyear,a.roa)
group by gvkey,fyear,roa
order by gvkey,fyear;
quit;

Re: Standard deviation of ROA with a rolling window of the past five years

joon1 — Thu, 21 Mar 2019 23:44:18 GMT

Thank you so much, novinosrin. I have the following error log.

115 proc sql;

116 create table want as

117 select gvkey,Fyear,roa,ifn(n=6,std(t),.) as roastd

118 from

119 (select a.*,ifn(max(b.fyear)=b.fyear,.,b.roa) as t,n(a.fyear) as n /*std(calculated t) as

119! roastd*/

120 from have a inner join have b

121 on a.gvkey=b.gvkey and a.fyear-5<= b.FYEAR<= a.fyear

122 group by a.gvkey,a.fyear,a.roa)

123 group by gvkey,fyear,roa

124 order by gvkey,fyear;

ERROR: File WORK.HAVE.DATA does not exist.

It will be highly appreciative if you can guide me to fix this error.

Thanks

Joon

Re: Standard deviation of ROA with a rolling window of the past five years

novinosrin — Fri, 22 Mar 2019 00:20:08 GMT

HAVE is the sample I created using your sample. Please review my post carefully

Re: Standard deviation of ROA with a rolling window of the past five years

novinosrin — Fri, 22 Mar 2019 02:57:46 GMT

@joon1 The following approach is simpler and faster

data have;
input Obs gvkey FYEAR roa;
drop obs;
cards;
1	1004	1999	0.04745
2	1004	2000	0.0264
3	1004	2001	-0.08299
4	1004	2002	-0.01807
5	1004	2003	0.00494
6	1004	2004	0.0211
7	1004	2005	0.03592
8	1004	2006	0.05494
9	1004	2007	0.05517
10	1004	2008	0.0571
11	1004	2009	0.02973
12	1004	2010	0.04098
13	1004	2011	0.03084
14	1004	2012	0.02574
15	1004	2013	0.03314
16	1004	2014	0.00673
17	1004	2015	0.03308
18	1004	2016	0.03756
19	1013	2000	0.21864
20	1013	2001	-0.51514
21	1013	2002	-1.0007
22	1013	2003	-0.05914
23	1013	2004	0.01148
24	1013	2005	0.07212
25	1013	2006	0.04077
26	1013	2007	0.06023
27	1013	2008	-0.02181
28	1013	2009	-0.35301
29	1013	2010	0.04205
;

data want;
do n=1 by 1 until(last.gvkey);
set have;
by gvkey;
array t(9999);
if n>=6 then std_dev=std(of t(*));
t(n)=roa;
if n(of t(*))>5 then do;k=n-5; call missing(t(k));end;
output;
end;
drop k n t:;
run;

Re: Standard deviation of ROA with a rolling window of the past five years

Ksharp — Fri, 22 Mar 2019 12:25:02 GMT

data have;
infile cards expandtabs truncover;
input Obs gvkey FYEAR roa;
drop obs;
cards;
1	1004	1999	0.04745
2	1004	2000	0.0264
3	1004	2001	-0.08299
4	1004	2002	-0.01807
5	1004	2003	0.00494
6	1004	2004	0.0211
7	1004	2005	0.03592
8	1004	2006	0.05494
9	1004	2007	0.05517
10	1004	2008	0.0571
11	1004	2009	0.02973
12	1004	2010	0.04098
13	1004	2011	0.03084
14	1004	2012	0.02574
15	1004	2013	0.03314
16	1004	2014	0.00673
17	1004	2015	0.03308
18	1004	2016	0.03756
19	1013	2000	0.21864
20	1013	2001	-0.51514
21	1013	2002	-1.0007
22	1013	2003	-0.05914
23	1013	2004	0.01148
24	1013	2005	0.07212
25	1013	2006	0.04077
26	1013	2007	0.06023
27	1013	2008	-0.02181
28	1013	2009	-0.35301
29	1013	2010	0.04205
;
proc sql;
create table m24 as 
select *, (select std(roa) from have where fyear between
a.fyear-6 and a.fyear-1 and gvkey=a.gvkey) 
from have as a;
quit;

Re: Standard deviation of ROA with a rolling window of the past five years

joon1 — Fri, 22 Mar 2019 15:52:39 GMT

Thank you so much, novinosrin. I greatly appreciate your help.

Have a great day

Joon1