xtile and egen

[Mohammad Khodadadi]

I want to categorize a variable on the basis of any set of cutpoints, not just percentiles. At the same time, since I have a panel, I want to do it for each point of time separately. In the following code and data, I categorize my firms on the basis of percentiles but when I use cutpoints(cpt) instead of nq() , it shows this error to me: option cutpoints() not allowed.

How can I categorize my firms in terms of cpt at each point of time? (Preferably, without using loops, because my data set is large)

* sample generated data
clear all
input time firm value cpt
    
1        1        3        4    
1        2        4        8    
1        3        5        10    
1        4        6        .    
1        5        7        .
1        6        8        .
1        7        9        .
1        8        10        .
1        9        11        .
2        1        4        6    
2        2        6        10    
2        3        7        12    
2        4        9        .
2        5        12        .
2        6        13        .
2        7        15        .
end

* finding quartiles for each time
egen perc = xtile(value), by(time) nq(4)

[Clyde Schechter]

You are confusing the official Stata command -xtile- with the -egen- function -xtile()- (from the -egenmore- package). The former accepts a -cutpoints()- option, but the latter does not. Unfortunately, you also want a -by()- option, which the latter supports and the former does not! See -help xtile- and -help egenmore-.

So you need to do this in a loop over time:

foreach t of local times {
    gen current_cpt = (cond(time == `t'), cpt, .)
    xtile group_perc = value if time == `t', cutpoints(current_cpt)
    replace perc = group_perc if time == `t'
    drop group_perc current_cpt
}

[Mohammad Khodadadi]

Dear Clyde,

Thanks a lot for your help. :-)

Mohammad