gini-brazil

Land tenure data description

• We have built a land tenure map with public available georeference layers of private properties and rural settlements. From private properties we have included CAR, SIGEF and titled lands from Terra Legal Program.
• The layers were processed with the algorithm described in Pinto et al (under review)
• Rural properties with less than 1 hectar (335,484 in total) were considered as errors coming from layers compilation steps and were excluded;
• Rural settlements (7,333 in total), present in the land tenure map only by the external rural settlement boundary polygon, were processed to generate rural settlement properties based on the number of properties informed by INCRA (935,797 in total). This process was not done spatially explicit and rural properties were only generated in the database to calculate gini index. The head of the rural settlement database and the code used for this processing step are detailed below:

## # A tibble: 7,333 x 6
##    cd_sipra   cd_mun num_lotes area_lote cum_lotes   seq
##    <chr>       <dbl>     <dbl>     <dbl>     <dbl> <dbl>
##  1 GO0094000 5207253      5.00      39.6      5.00  1.00
##  2 SP0115000 3503307     14.0       14.0     19.0   2.00
##  3 SE0003000 2805406    211         29.7    230     3.00
##  4 AM0021000 1300706   1060        105     1290     4.00
##  5 RR0059000 1400209    158         67.7   1448     5.00
##  6 CE0188000 2303931     10.0       49.0   1458     6.00
##  7 RS0038000 4314605     35.0       28.0   1493     7.00
##  8 AL0206000 2707107     20.0       19.4   1513     8.00
##  9 MF0252000 2613909     46.0       48.5   1559     9.00
## 10 PI0438000 2211704     69.0       28.0   1628    10.0 
## # ... with 7,323 more rows

aru_inline <- data.frame(cd_mun=NA, cd_sipra=NA,cd_lote=NA,area_lote=NA)

for (i in unique(aru$cd_sipra)){
  tempDB = subset(aru, aru$cd_sipra == i)
  add_lote = tempDB$num_lotes
  l= (tempDB$cum_lotes - tempDB$num_lotes) + 1

  while(add_lote > 0){
    aru_inline[l,'cd_mun']    <- tempDB$cd_mun
    aru_inline[l,'cd_sipra']  <- tempDB$cd_sipra
    aru_inline[l,'cd_lote']   <- paste(tempDB$cd_sipra, as.character(l))
    aru_inline[l,'area_lote'] <- tempDB$area_lote

    l = l + 1
    add_lote = add_lote - 1
  }
  print(tempDB$cd_sipra)
}

Results

The presented results will consider the both situations, i.e., with and without considering rural settlements.

Lorenz curve

Gini Index Brazil

Considering rural settlements gini index for Brazil was 0.8.
Without considering rural settlements gini index for Brazil was 0.84.

Gini Index per State

Gini Index for Para State

For Para, the CAR database is available with the CPF/CNPJ of each rural property. Were identified 176,051 registries, being 154,740 related to CPF and 21,648 to CNPJ. CAR database was processed to aggregate rural properties owned by the same CPF or by the same company. The following steps describe the treatment applied to the data:

1. We first filtered the rural properties owned by only one owner (170,176), being it a CPF or a CNPJ. This filtered dataset was processed as described below:
   1. Company aggregation was based on the first 8 digits of the CNPJ number. For example, if two properties are owned by the CNPJ number 99.999.999/0001-99 and 99.999.999/0002-99 we considered that both properties are owned by the same CNPJ (number 99.999.999), and rural properties areas were aggregated. This step resulted in 1,111 registries, meaning that 14,325 CARs are owned by companies with more than one CAR;
   2. CPF agregation was based on the entire CPF number that is composed by 11 digits (e.g., 999.999.999-15). This step resulted in 133,092 registries, meaning that 21,648 CARs are owned by people with more than one CAR;
2. Rural properties owned by many people or companies (5,875) were considered to be unique, and no aggregation was performed.

We compared Gini Index for Para with three different steps:

• Without considering rural settlements or CPF/CNPJs treatments, Gini Index for Para was 0.81.
• Considering rural settlements in Para (261,595), Gini Index for Para was 0.71.
  
• Considering rural settlements in Para (261,595) and CPF/CNPJs aggregation, Gini Index for Para was 0.85.

The following graphs present the same results for Gini Index and also the Lorenz-Curve for each situation.