looping over observations to identify positions

[Annalisa Russino]

looping over observations to identify positions

Today, 00:49
Dear All,

I am using stata14. I need to identify observation position. In particular I need to loop over the observations and to retrieve the positions corresponding to the case where variable n3 is not missing. Once I get that position I need to change the value of one variable if a set of condition are satisfied. Here there is a sample of my database, in the database group indicates firms and SH_D_cf4 the stakes of the shareholders by year,

* Example generated by -dataex-. To install: ssc install dataex
clear
input float(group Year SH_D_cf4 VarSH MeanSH n3)
1 2014 80 . 80 1
2 2010 36.1 12.428596 20 .
2 2010 11.5 12.428596 20 .
2 2010 11.3 12.428596 20 .
2 2010 10.3 12.428596 20 .
2 2010 30.8 12.428596 20 5
2 2011 36.1 10.73623 25 .
2 2011 26.9 10.73623 25 .
2 2011 26.7 10.73623 25 .
2 2011 10.3 10.73623 25 .
2 2011 . 10.73623 25 5
2 2012 36.1 10.73623 25 .
2 2012 26.9 10.73623 25 .
2 2012 26.7 10.73623 25 .
2 2012 10.3 10.73623 25 .
2 2012 . 10.73623 25 5
2 2013 36.1 10.73623 25 .
2 2013 26.9 10.73623 25 .
2 2013 26.7 10.73623 25 .
2 2013 10.3 10.73623 25 .
2 2013 . 10.73623 25 5
2 2014 36.1 10.73623 25 .
2 2014 26.9 10.73623 25 .
2 2014 26.7 10.73623 25 .
2 2014 10.3 10.73623 25 .
2 2014 . 10.73623 25 5
3 2009 12 2.73252 14.666667 .
3 2009 12 2.73252 14.666667 .
3 2009 18 2.73252 14.666667 .
3 2009 . 2.73252 14.666667 .
3 2009 14 2.73252 14.666667 .
3 2009 14 2.73252 14.666667 .
3 2009 18 2.73252 14.666667 7
3 2010 12 2.73252 14.666667 .
3 2010 12 2.73252 14.666667 .
3 2010 18 2.73252 14.666667 .
3 2010 . 2.73252 14.666667 .
3 2010 14 2.73252 14.666667 .
3 2010 14 2.73252 14.666667 .
3 2010 18 2.73252 14.666667 7

With the following code I get the positions and VarSH is correctly taken at the selected positions

gen Flag_Year_stable=0

su group, meanonly
forvalues i=1/`r(N)'{
if n3[`i']!=.{
local pos=`i'
display `pos'
display `pos'+1
display VarSH[`pos'], VarSH[`pos'+1]
}
}

I need to identify years with a stable ownership structure, So I want to change the value of Flag_Year_stable to one if a set of conditions are satisfied. In particular, for each identified position, I need to

replace Flag_Year_stable=1 if VarSH[`pos'+1]==VarSH[`pos'] & MeanSH[`pos'+1]==MeanSH[`pos']

I tried several posiibilities, but I did not get the result. For instance I tried



by group: gen n2=_n
by group: egen max_n2=max(n2)

su group, meanonly
forvalues i=1/`r(N)'{
if n3[`i']!=.{
local pos=`i'
replace Flag_Year_stable=1 if n2[`pos' ]<max_n2 & VarSH[`pos'+1]==VarSH[`pos'] & MeanSH[`pos'+1]==MeanSH[`pos']
replace Flag_Year_stable=1 if n2[`pos' ]==max_n2 & VarSH[`pos'-1]==VarSH[`pos'] & MeanSH[`pos'-1]==MeanSH[`pos']
}
}

Any suggestion?

Thank you very much

[Carole J. Wilson]

I'm not certain, but perhaps this gives you what you want, or at least starts you in the right direction:

bysort group: gen next_varsh=VarSH[_n+1] if !mi(n3)
bysort group: gen next_meansh=MeanSH[_n+1] if !mi(n3)
bysort group Year: gen flag=cond( VarSH== next_varsh & MeanSH==next_meansh, 1, 0)
list, sepby(group)

[Annalisa Russino]

Thank you, your suggestion works with an adjustment for the last observation of each group that must be compared with the previous block.