Calculating GINI coefficients

[Stan Breetman]

Hi everbody,

I am new here. If you need something more like data or something else, tell me please. But for answering this question, I think you wont need the Data.

I have the wages of basketball players for each team in the NBA in the period of 2001 till 2014. My question is, how I can calculate GINI coefficient in Stata for every team in year x. I am writing because calculating it in Excel takes to much time, especially when i want to modify the wages later.

So for example, I need all the Gini coefficients for Team Cleveland in the year 2001, 2002, 2003, ..., 2014. And so for the next team.

I also installed the DASP. If anybody has suggestions how to calculate the GINI coefficients with this extra program for STATA or any suggestions how to calculate the GINI in Excel efficiently, or suggestions for another way, I would be very happy.

Thank you very much for a fast answer.

With kind regards,

Stan

[Friedrich Huebler]

findit gini

finds several options for calculating the Gini coefficient in Stata.

[Stan Breetman]

Thank you. But since I am a beginner in Stata, how can I use a command and tell Stata to calculate just for the same team name and the same year?

And when I type findit gini, what are these, some programms to download?

Maybe anybody knows the command for my problem?

[Friedrich Huebler]

For a beginner a good way to start would be the "Getting started with Stata" guide that is part of the PDF documentation that comes with Stata

help gs

.

[Stan Breetman]

Sorry Mr. Huebler, but thats no concrete help for my question.

[Carlo Lazzaro]

Stan:
you may want to consider the user-written programmes -descogini- or -ineqdeco-, that you can install by typing -search descogini- or -search ineqdeco- from within Stata.
A toy-example possibly related to your query:

. sysuse auto.dta
(1978 Automobile Data)

. descogini price foreign if rep78==3 & gear_ratio<=4

Gini Decomposition by Income Source

Total Income Variable: price
----------------------------------------------------------------------
Source           Sk           Gk          Rk        Share     % Change
----------------------------------------------------------------------
foreign        0.0000       0.9000     -0.2840     -0.0000    -0.0000
Total income                0.2625
----------------------------------------------------------------------

Please do not underestimate Friedrich's wise advice about getting yourself familiar with Stata bread and butter.
As per FAQ, posters are expected to search a solution for their own problem on Statalist, web or other sources.
Had you googled with the string -calculating gini index with Stata- you would have found out some promising entries yourself.
Besides (as per FAQ again), the best way to increase your chances of getting helpful replies is to post what you typed and what Stata gave you back and/or posting an example/excerpt of your dataset via -dataex- (type -search dataex- from within Stata to install it).
Put differently, as the whole forum works on goodwill, please help interested listers help yourself out by fulfilling the basic requirements of posting and Stata knowledge. Thanks.
We all passed through this seemingly unpleasant stage of devoting time to learn Stata language backbone (and I'm still learning it with interest after arbout 10 years), but, unfortunately, there's no faster approach.
I do hope that you enjoy staying with us despite some initial drawbacks.

[Stan Breetman]

Hi Carlo,

thank you very much. I appreciate your and Mr. Hueblers help really. I will try to do so as you have said.
Your code is interesting and I have already read in google and forums about these comands.
The thing is, I couldnt find any solution for calculating Gini Coefficient with the goal to have finaly the following output for example:
(I am using Stata 13) Something like: Stata should use the wages of the same team in the same year and so on..

Year Team GiniCoefficient (I am using numbers for teams, instead of team names )
---------------------------------------------------
2001 Cleveland 0.45
2002 Cleveland 0.35
...
2014 Cleveland 0.51
----------------------------------------------------
2001 San Antonio Spyrs 0.38
2002 San Antonio Spyrs 0.48
...
2013 San Antonio Spyrs 0.47
----------------------------------------------------
... and so on for every team


With the goal finally, not to giving around 420 comands (for every Team in the specific year to calculate), but rather to calculate the GINIs for all Teams in every year from 2001 to 2014 with one comand.

So my question is, if there is such a comand which one it is. If it would be the code suggested by Carlo, then my question would be, how can i define the code, so excel works with the mentioned criteria.

I hope somebody can help me soon. I will try to find out more .

Kind regards
Ivan

[William Lisowski]

In the made-up example below inspired by Carlo's post I use the user-written ineqdeco command to calculate "gini coefficients" for price in the auto dataset, separate for each combination of foreign/domestic and reputation(1 to 5). This is similar to calculating the gini coefficient for wage separately for each combination of team and year.

I think I understand that you want the results in Excel. It's my assessment that the easiest way of accomplishing that would be to copy and paste the table from the Stata results window into Excel.

. sysuse auto, clear
(1978 Automobile Data)

. egen bygroups = group(foreign rep78), label
(5 missing values generated)

. ineqdeco price, by(bygroups)
 
[[[ intervening output deleted ]]]

Subgroup indices: GE_k(a) and Gini_k

-----------------------------------------------------------------------
group(fore |
ign rep78) |     GE(-1)       GE(0)       GE(1)       GE(2)        Gini
-----------+-----------------------------------------------------------
Domestic 1 |    0.00330     0.00329     0.00328     0.00328     0.04048
Domestic 2 |    0.09696     0.10614     0.12448     0.15739     0.24586
Domestic 3 |    0.11674     0.11882     0.12844     0.14785     0.26851
Domestic 4 |    0.03164     0.03124     0.03155     0.03256     0.14063
Domestic 5 |    0.00138     0.00138     0.00138     0.00138     0.02622
 Foreign 3 |    0.02179     0.02219     0.02280     0.02363     0.11045
 Foreign 4 |    0.04081     0.03982     0.03980     0.04076     0.15974
 Foreign 5 |    0.07218     0.07305     0.07738     0.08585     0.21065
         . |                                                          
-----------------------------------------------------------------------

[[[ remaining output deleted ]]]

I will also say that with sufficient Stata expertise it's possible to do this more elegantly, looping over the data to compute each coefficient one-at-a-time and posting the results to an Excel spreadsheet. For a new user, the copy-and-paste route will get your results into Excel fastest.

[Friedrich Huebler]

I have already read in google and forums about these comands.

I would recommend that you start with Stata's own documentation, including the help files and PDF documents. Most user-written packages for Stata, such as those found with findit gini, also come with a help file that explains how they should be used. For the packages mentioned by Carlo and William, you can read the help files with the following commands:

help descogini
help ineqdeco

[Stan Breetman]

Thank you very much everybody.

I think I will try the way Mr. Lisowski and Carlos told me with ineqdeco.

Mr. Lisowski did you use the 'GINIDESC' module to install? Because there are several modules. I think thats the good one, but I would be happy if i get a confirmation.

[Stan Breetman]

Hi all,

I have to say it worked well with the suggestion of Mr. Lisowski. Thank you very much for everybody. I checked the Gini Coeffiecients and they seem to be correct what makes me happy .

Following steps to do for my mentioned problem:
1. Download 'GINIDESC' module
2. Follow the comands by William Lisowski
3. Copying the Gini Coefficients into Excel is very easy and you can sort them later as you want.

Kind regards,

Stan

[William Lisowski]

Actually, what I downloaded was the ineqdeco package returned as the fourth item from search ineqdeco.

ineqdeco from http://fmwww.bc.edu/RePEc/bocode/i
    'INEQDECO': module to calculate inequality indices with decomposition by
    subgroup / ineqdeco estimates a range of inequality and related / indices
    commonly used by economists, plus optional / decompositions of a subset of
    these indices by population / subgroup.  Indices estimated are:  members

But the good news is that the ginidesc package uses, and downloads, the ineqdeco command, so you got what you needed.

[Qurat ul Ain]

In the made-up example below inspired by Carlo's post I use the user-written ineqdeco command to calculate "gini coefficients" for price in the auto dataset, separate for each combination of foreign/domestic and reputation(1 to 5). This is similar to calculating the gini coefficient for wage separately for each combination of team and year.

I think I understand that you want the results in Excel. It's my assessment that the easiest way of accomplishing that would be to copy and paste the table from the Stata results window into Excel.

. sysuse auto, clear
(1978 Automobile Data)

. egen bygroups = group(foreign rep78), label
(5 missing values generated)

. ineqdeco price, by(bygroups)

[[[ intervening output deleted ]]]

Subgroup indices: GE_k(a) and Gini_k

-----------------------------------------------------------------------
group(fore |
ign rep78) | GE(-1) GE(0) GE(1) GE(2) Gini
-----------+-----------------------------------------------------------
Domestic 1 | 0.00330 0.00329 0.00328 0.00328 0.04048
Domestic 2 | 0.09696 0.10614 0.12448 0.15739 0.24586
Domestic 3 | 0.11674 0.11882 0.12844 0.14785 0.26851
Domestic 4 | 0.03164 0.03124 0.03155 0.03256 0.14063
Domestic 5 | 0.00138 0.00138 0.00138 0.00138 0.02622
Foreign 3 | 0.02179 0.02219 0.02280 0.02363 0.11045
Foreign 4 | 0.04081 0.03982 0.03980 0.04076 0.15974
Foreign 5 | 0.07218 0.07305 0.07738 0.08585 0.21065
. |
-----------------------------------------------------------------------

[[[ remaining output deleted ]]]

I will also say that with sufficient Stata expertise it's possible to do this more elegantly, looping over the data to compute each coefficient one-at-a-time and posting the results to an Excel spreadsheet. For a new user, the copy-and-paste route will get your results into Excel fastest.

Respected Sir
I was using these commands and i studied help comands as well,actually i have four provinces and 20 years data for each province and i wan to calculate yearly data for each province.but when i enter comands i either be sucessful in getting yearly data or just for provinces as whole.my variable is GDPPC and i want to calculate weighted gini GDPPC for each province and for every year.i want to get the results so that i copy it in excel.

i do this my id and year i got zero results
----------------------------------------------------------------------
group(id |
year) | GE(-1) GE(0) GE(1) GE(2) Gini
----------+-----------------------------------------------------------
1 1990 | 0.00000 0.00000 0.00000 0.00000 0.00000
1 1991 | 0.00000 0.00000 0.00000 0.00000 0.00000
1 1992 | 0.00000 0.00000 0.00000 0.00000 0.00000
1 1993 | 0.00000 0.00000 0.00000 0.00000 0.00000
1 1994 | 0.00000 0.00000 0.00000 0.00000 0.00000
1 1995 | 0.00000 0.00000 0.00000 0.00000 0.00000
1 1996 | 0.00000 0.00000 0.00000 0.00000 0.00000
1 1997 | 0.00000 0.00000 0.00000 0.00000 0.00000
1 1998 | 0.00000 0.00000 0.00000 0.00000 0.00000
1 1999 | 0.00000 0.00000 0.00000 0.00000 0.00000
if i do this by just year i got

Subgroup indices: GE_k(a) and Gini_k

----------------------------------------------------------------------
group(yea |
r) | GE(-1) GE(0) GE(1) GE(2) Gini
----------+-----------------------------------------------------------
1990 | 0.31453 0.22186 0.18524 0.17825 0.32842
1991 | 0.31402 0.22168 0.18519 0.17827 0.32841
1992 | 0.31351 0.22150 0.18515 0.17830 0.32840
1993 | 0.31300 0.22132 0.18510 0.17832 0.32839
1994 | 0.31249 0.22114 0.18505 0.17834 0.32838
1995 | 0.31199 0.22096 0.18501 0.17837 0.32837
1996 | 0.31148 0.22078 0.18496 0.17839 0.32837
1997 | 0.31098 0.22060 0.18492 0.17842 0.32836
1998 | 0.31786 0.22578 0.19016 0.18482 0.33316
but i want gini for each province and for each year

i would be thanful if you help me.

Regards
qurat ul ain

[Stephen Jenkins]

[no access to Stata, so guessing syntax]

egen year_pr = groups(year province), labels
statsby gini = r(gini), by(year_pr) saving(output.dta, replace): ineqdeco percapita_income
*now you have a Stata dataset
use output.dta, clear
export excel <...>

[Qurat ul Ain]

[no access to Stata, so guessing syntax]

egen year_pr = groups(year province), labels
statsby gini = r(gini), by(year_pr) saving(output.dta, replace): ineqdeco percapita_income
*now you have a Stata dataset
use output.dta, clear
export excel <...>

Respected sir

I have data in stata and i make subgroups by egen still the results are zero for year and provincebut when i calculate it only for year it gives me the results.

Subgroup indices: GE_k(a) and Gini_k

------------------------------------------------------------------------
group(provi |
nces year) | GE(-1) GE(0) GE(1) GE(2) Gini
------------+-----------------------------------------------------------
Punjab 1990 | 0.00000 0.00000 0.00000 0.00000 0.00000
Punjab 1991 | 0.00000 0.00000 0.00000 0.00000 0.00000
Punjab 1992 | 0.00000 0.00000 0.00000 0.00000 0.00000
Punjab 1993 | 0.00000 0.00000 0.00000 0.00000 0.00000
Punjab 1994 | 0.00000 0.00000 0.00000 0.00000 0.00000
Punjab 1995 | 0.00000 0.00000 0.00000 0.00000 0.00000
Punjab 1996 | 0.00000 0.00000 0.00000 0.00000 0.00000