Grouping SIC codes

[Irfan Ali]

Hello all,

I want to generate new variable for Industry dummies (say SIC). I want to further condense the industries based on their two digit SIC codes. For example, codes 23-27, 30-37, 40-49, 51-59, 60-69, 70-80, rest in category for OHTERS, making total of 7 industries for my analysis.

I would appreciate help in this regard.

Thanks in advance.

       SIC |      Freq.     Percent        Cum.
------------+-----------------------------------
         15 |         10        1.46        1.46
         23 |         10        1.46        2.93
         27 |         10        1.46        4.39
         28 |        110       16.11       20.50
         29 |         10        1.46       21.96
         30 |         20        2.93       24.89
         32 |         19        2.78       27.67
         33 |         38        5.56       33.24
         35 |         48        7.03       40.26
         36 |         16        2.34       42.61
         37 |         85       12.45       55.05
         38 |         10        1.46       56.52
         42 |         10        1.46       57.98
         45 |         20        2.93       60.91
         48 |         30        4.39       65.30
         49 |         18        2.64       67.94
         51 |         13        1.90       69.84
         53 |         10        1.46       71.30
         59 |         11        1.61       72.91
         60 |         10        1.46       74.38
         61 |         10        1.46       75.84
         62 |         10        1.46       77.31
         63 |         20        2.93       80.23
         65 |         56        8.20       88.43
         67 |          7        1.02       89.46
         73 |         23        3.37       92.83
         79 |         10        1.46       94.29
         80 |         20        2.93       97.22
         99 |         19        2.78      100.00
------------+-----------------------------------
      Total |        683      100.00

[Carole J. Wilson]

You can use -recode-:

recode SIC (23/27=1) (30/37=2) ... , gen(SIC_group)

[David Benson]

Irfan, Carole's code elegantly solves your problem.

I am posting here just because I find myself frequently trying to create 2- and 3-digit SIC codes from 4-digit SIC codes and thought I would post two different ways to accomplish that in Stata (for those coming along later).

* Creating 2- and 3-digit SIC codes (from 4-digit code)
* Also, could put in Excel and just use LEFT() function (and then merge in)

gen target_SIC_3digit = substr(string(target_SIC),1,3)  /* converts target_SIC to a string, then takes left 3 numbers */
gen target_SIC_2digit = substr(string(target_SIC),1,2)

destring target_SIC_3digit, replace
destring target_SIC_2digit, replace
order target_SIC_3digit target_SIC_2digit, after(target_SIC)  // just places them after target_SIC so all SIC code variables are together


* Creating 2- and 3-digit SIC codes (without converting to string)
gen target_SIC_3digit = int(target_SIC/10)  // int() or floor() creates integer by rounding down. So int(367 / 10) = 36
gen target_SIC_2digit = int(target_SIC/100)
order target_SIC_3digit target_SIC_2digit, after(target_SIC)