How to write shorter

[MIchael Jefferson]

Hi reader,

I would like to know how I could make these formulas a bit shorter, hope you can help.

Formula 1:

retX and retcX are in columns, firms in rows

gen ar2=ret2-retc2
gen ar3=ret3-retc3
gen ar4=ret4-retc4
gen ar5=ret5-retc5
gen ar6=ret6-retc6
gen ar7=ret7-retc7
gen ar8=ret8-retc8
gen ar9=ret9-retc9
gen ar10=ret10-retc10
gen ar11=ret11-retc11
gen ar12=ret12-retc12
gen ar13=ret13-retc13
gen ar14=ret14-retc14
gen ar15=ret15-retc15
gen ar16=ret16-retc16
gen ar17=ret17-retc17
gen ar18=ret18-retc18
gen ar19=ret19-retc19
gen ar20=ret20-retc20
gen ar21=ret21-retc21
gen ar22=ret22-retc22
gen ar23=ret23-retc23
gen ar24=ret24-retc24
gen ar25=ret25-retc25
gen ar26=ret26-retc26
gen ar27=ret27-retc27
gen ar28=ret28-retc28
gen ar29=ret29-retc29
gen ar30=ret30-retc30
gen ar31=ret31-retc31
gen ar32=ret32-retc32
gen ar33=ret33-retc33
gen ar34=ret34-retc34
gen ar35=ret35-retc35
gen ar36=ret36-retc36
gen ar37=ret37-retc37
gen ar38=ret38-retc38

Formula 2:

gen ret3year= ((1+ret2)*(1+ret3)*(1+ret4)*(1+ret5)*(1+ret6)*(1+r et7)*(1+ret8)*(1+ret9)*(1+ret10)*(1+ret11)*(1+ret1 2)*(1+ret13)*(1+ret14)*(1+ret15)*(1+ret16)*(1+ret1 7)*(1+ret18)*(1+ret19)*(1+ret20)*(1+ret21)*(1+ret2 2)*(1+ret23)*(1+ret24)*(1+ret25)*(1+ret26)*(1+ret2 7)*(1+ret28)*(1+ret29)*(1+ret30)*(1+ret31)*(1+ret3 2)*(1+ret33)*(1+ret34)*(1+ret35)*(1+ret36)*(1+ret3 7))-1

gen retc3year= ((1+retc2)*(1+retc3)*(1+retc4)*(1+retc5)*(1+retc6) *(1+retc7)*(1+retc8)*(1+retc9)*(1+retc10)*(1+retc1 1)*(1+retc12)*(1+retc13)*(1+retc14)*(1+retc15)*(1+ retc16)*(1+retc17)*(1+retc18)*(1+retc19)*(1+retc20 )*(1+retc21)*(1+retc22)*(1+retc23)*(1+retc24)*(1+r etc25)*(1+retc26)*(1+retc27)*(1+retc28)*(1+retc29) *(1+retc30)*(1+retc31)*(1+retc32)*(1+retc33)*(1+re tc34)*(1+retc35)*(1+retc36)*(1+retc37))-1

Formula 3:

I need to count days per firm, since there are over 1000 different firms, what would be a smart command to generate the counted days?

Appreciate the help

[Nick Cox]

Not the answer you expected, but my guess is that the underlying problem here is your data set-up. My guess is that you have a wide layout (sometimes called format or structure) in which data for different times are in different variables.

reshape wide

and almost all your syntax can and will be much shorter.

[Nick Cox]

Indeed there is, but sorry, I won't offer code in support of bad ideas.

[Jorrit Gosens]

As Nick explains, your life will be easier if you have your data long.

*some data
clear
set obs 100
gen companyID=_n
forvalues i = 1(1)38{
gen ret`i'=runiform()
gen retc`i'=runiform()
}

*your formulas for a long data structure, roughly:
reshape long ret retc, i(companyID) j(time)
gen ar=ret-retc
gen retplus1 = ret+1
gen retcplus1 = retc+1
bys companyID (time): gen runningret = retplus1*retplus1[_n-1]
bys companyID (time): replace runningret = retplus1 if _n==1
bys companyID (time): gen runningretc = retcplus1*retcplus1[_n-1]
bys companyID (time): replace runningretc = retcplus1 if _n==1
replace runningret = runningret-1
replace runningretc = runningretc-1

If there is a need to have the data back into wide again after, because it is sometimes easier to read, you could do:

reshape wide ret retc ar retplus1 retcplus1 runningret runningretc, i(companyID) j(time)

However, for most reporting purposes, e.g., graphs, regression analyses, you'll want to keep your data long.
I also doubt this is really easier to read with this many reporting periods and variables

[Attaullah Shah]

I would agree with Nick the majority of the Stata's commands work great with the long data format. The formulas you have posted above seem to calculate cumulative returns over a period of time. I have a blog post on this, which might be of interest to you, see https://fintechprofessor.com/2017/10...ct-of-returns/
I this post, I use asrol (can be downloaded from SSC).