How to calculate spatial GINI coefficient

[DavidMeek]

Hello all,

I am interested in calculating an index of agricultural land inequality for Brazil. I'm thinking that a GINI coefficient would be a good approach given traditional uses of GINI, but I'm open to other approaches.

I have a data set for all municipalities in Brazil that consists of four columns for each municipality (rows): 1) number of family farms, 2) area occupied by small farm land use; 3) number of non-family farms; 4) area occupied by non-family farms.

What's I'm really hoping to attain is a value that can represent the relations between percent area occupied by non-family farms in comparison with family farms.

I can obtain the total area of the municipality from municipality shape file, but I don't think it makes sense to have a simple ratio of non-family farm area/municipality area as there will be various other forms of land use.

  Any suggestions on how to calculate a spatial GINI using this data set, or a different statistic that makes more sense would be greatly appreciated

[XanderBakker]

Hi David Meek , thanks for sharing the data and the additional explanation.

There is an important point that triggers me; you seek to generate a Gini coefficient value for each municipality. As far as I understand, calculating the Gini coefficient based on all municipalities in Brazil will generate a single value for Brazil. I believe it is a method for aggregation, but you could calculate the coefficient for all municipalities in each region and visualize the result per region.

Another possibility is to determine the relative position of each municipality (the area fraction) on the variation of the fraction in Brazil. This would indicate the "situation" of each single municipality, but this would no longer be a Gini coefficient.

[XanderBakker]

OK, I'm getting somewhere...

The data in you Excel has a mix of 4 levels of data (regions, to municipalities).I added some fields to distinguish them (find attached the adapted Excel). The graph is based on the highest level (level 4, municipalities I guess). This would yield a single Gini coefficient, which can be useful to compare the result to others countries if this coefficient would be available. However, if you want to visualize the inequality distribution you would have to use the fraction (familly farm area over total farm area).