Standard deviation over 3 and 5 years

[Abdullah Baba]

Hi all,

I have been able to calculate t-1 and average of the previous 2 years (t-1 + t-2). However, I would need SD over 3 and 5 years of
Webit_asset
Wni_asset
i2a_bhy_l

Can anyone help me with that?

My code is the following:

tab year, gen (yd_)

destring sic, replace

gen sic2=int(sic/100)
label var sic2 "2-digit SIC code"

gen sic3=int(sic/10)
label var sic3 "3-digit SIC code"

tab sic2, gen(id_)

sort gvkey year

by gvkey: gen lag_at = at[_n-1]
label var lag_at "totals assets t-1"

by gvkey: gen lag_at_t2 = at[_n-2]
label var lag_at_t2 "totals assets t-2"

by gvkey: gen lag_at_avg = (lag_at + lag_at_t2)/2
label var lag_at_avg "totals assets average 2 years"

gen ln_at=log(lag_at)
label var ln_at "LN(Total assets) in t-1"

gen ln_at_avg=log(lag_at_avg)
label var ln_at_avg "LN(Total assets) average 2 years"


by gvkey: gen ebit_asset = ebit[_n-1]/at[_n-1]
by gvkey: gen ebit_asset_t2 = ebit[_n-2]/at[_n-2]
by gvkey: gen ebit_asset_avg = (ebit_asset + ebit_asset_t2)/2

by gvkey: gen ni_asset = ni[_n-1]/at[_n-1]
by gvkey: gen ni_asset_t2 = ni[_n-2]/at[_n-2]
by gvkey: gen ni_asset_avg = (ni_asset + ni_asset_t2)/2

winsor ebit_asset, gen(Webit_asset) p(0.01)
winsor ebit_asset_avg, gen(Webit_asset_avg) p(0.01)

winsor ni_asset, gen(Wni_asset) p(0.01)
winsor ni_asset_avg, gen(Wni_asset_avg) p(0.01)


by gvkey: gen ly_bhr = y_bhr[_n-1]
label var ly_bhr "BHR t-1"

by gvkey: gen ly_bhr_t2 = y_bhr[_n-2]
label var ly_bhr_t2 "BHR t-2"


* Weighted industry return
* 2-digit level

egen t_2_mv = sum(mkvalt), by(sic2 year)
gen w_2_sic = mkvalt/t_2_mv
gen w_2_bhr = w_2_sic*y_bhr
egen ind2_bhr = sum(w_2_bhr), by(sic2 year)

by gvkey: gen l_ind2_bhr = ind2_bhr[_n-1]

gen i2a_bhy_l = ly_bhr-l_ind2_bhr
label var i2a_bhy_l "Lagged BHR adjusted by MCAP weighted mean BHR at 2-digit"


by gvkey: gen l_ind2_bhr_t2 = ind2_bhr[_n-2]

gen i2a_bhy_l_t2 = ly_bhr_t2-l_ind2_bhr_t2
label var i2a_bhy_l_t2 "t-2 BHR adjusted by MCAP weighted mean BHR at 2-digit"


gen i2a_bhy_l_avg = (i2a_bhy_l + i2a_bhy_l_t2)/2
label var i2a_bhy_l_avg "Average 2 years BHR adjusted by MCAP weighted mean BHR at 2-digit"

[Abdullah Baba]

anyone here, who can help me

[Carlo Lazzaro]

Abdullah:
you posted a very log code supported by no data (please, see the FAQ on how to share excerpt/example of your dataset via -dataex-. Thanks).
As you can easily envisage, even the most interested lister couldn't reply to your query.

[Abdullah Baba]

Hi Carlo, thank you very much - you are rihgt. It is a quite long code. However, this would be easier now hopefully.

I did already the calculation of 2 years average.

I need to calculate the SD over 3 years and 5 years of the following:
ebit_asset

by gvkey: gen ebit_asset = ebit[_n-1]/at[_n-1]
by gvkey: gen ebit_asset_t2 = ebit[_n-2]/at[_n-2]
by gvkey: gen ebit_asset_avg = (ebit_asset + ebit_asset_t2)/2

Same for
ni_asset

by gvkey: gen ni_asset = ni[_n-1]/at[_n-1]
by gvkey: gen ni_asset_t2 = ni[_n-2]/at[_n-2]
by gvkey: gen ni_asset_avg = (ni_asset + ni_asset_t2)/2

[Carlo Lazzaro]

Abdullah:
thanks for providing clearer codes.
Perhaps you can get what you're after by tweaking the following toy-example (tailored on a two-year timespan):

use "http://www.stata-press.com/data/r15/nlswork.dta"
bysort idcode: g flag=_n
bysort idcode: egen var1=mean(year) if flag<=2
bysort idcode: egen var2=sd(year) if flag<=2
bysort idcode: sum var1 var2

[Abdullah Baba]

Thank you Carlo. I tried another way and hope it will work somehow

[Nick Cox]

See also rangestat (SSC): search for mentions here.

[Abdullah Baba]

Thanks a lot Nick!