Palm Oil
View or edit on GitHub
This page is synchronized from trase/data/indonesia/palm_oil/README.md. Last modified on 2025-12-13 00:30 CET by Trase Admin.
Please view or edit the original file there; changes should be reflected here after a midnight build (CET time),
or manually triggering it with a GitHub action (link).
Indonesia Palm Oil
Pre-processing steps and decision tree implementation for Indonesian palm oil.
1. Trade data
1.1 Export
We currently have customs data for 2013 - 2018, covering all CPO and RPO
under the HS code header of 1511 - Palm oil and its fractions; whether
or not refined, but not chemically
modified.
When compared to the official monthly statistics from BPS, the results look like so:
There are some notable discrepanices, particularly in 2013, however the annual volumes in all other years are within 0.1% and 3.6% of the official statistics in all years, with a mean discrepancy of 1.4%.
1.2 Domestic
In addition to the customs data, we also have per-shipment records for domestic trades of CPO and RPO in 2015. This follows a similar structure to the export records, including a date stamp, exporter name, importer name, volume, port of loading and port of destination.
The data contains information about 1990 shipments in 2015, accounting for 5.4 million tons of palm oil (21% of exports).
2. Production data
Our palm oil production data are derived from province level CPO statistics provided by the Directorate General of Estate Crops via BPS. These data are cut to the kabupaten level using Auriga’s 2016 palm cover map.
Much like our crop maps in Brazil, the palm cover map does not span the entire country. For Riau Island we have used the kabupaten level proportions provided by BPS for 2015.
3. Logistics data
3.1 Ports
We have identified and located all 149 ports in the trade data (export and domestic). They have each been provided with a unique ID and a set of coordinates.
3.2 Refineries & bulking facilities
Our refinery and bulking facility dataset combines inputs from Aid Environment, Maps & Globe Specialist and desk based research carried out by the Auriga team.
The dataset does not yet include comprehensive data on capacities, but the ownership and location information is key for utilising the information in the traceability reports (section 3.4).
3.3 Mills
The dataset that we’re using is an enriched version of the Universal Mill List (UML), a collaborative effort between the World Resource Institute, Rainforest Alliance, Daemeter and Proforest.
Thanks to the hard work of UCSB and Auriga, our list includes 1083 mills with verified locations, ownership data and capacity information.
The capacity values was obtained through reference to Disbun repots, the initial SIPERIBUN results and RSPO/ISPO filings. There were 118 mills for which we could not find a value. For these we used a spatial interpolation at the Kabupaten or Province level (depending on the availability of other mills). This was found to be more accurate than a kriging interpolation when tested on a sample of known values.
3.4 Plantations
To be completed.
3.5 Traceability reports
Between 2014 and 2018, 37 of the 123 refineries and bulking facilities in Indonesia published traceability reports, linking their operations to 980 mills. These connections provide important constraints for port - mill allocation.
3.6 Transport cost matrix
To be completed.
4. Companies
One of the central challenges in preparing the data for Indonesian palm oil has been aligning company information from multiple sources. We have compiled a dictionary of 2777 companies from an original list of 4775, allowing us to match multiple permutations of a given name.
In addition we have also drawn information from Indonesian corporate research (CDMI) and company filings (AHU) to ascertain the relationships between these various companies. Doing so has alloed us to gather exporters, processors and producers under the banner of a group, as can be seen below. In total our data includes 188 groups.
5. Indicators
To be completed - for now see the google sheet here.
6. Boundaries
We are using the 2016 Kabupaten boundaries defined by BIG, the Indonesian geospatial agency. These are paired with geocodes provided by BPS to provide a standard identifier for each Kabupaten.
7. SEI-PCS
The SEI-PCS model lives at https://github.com/sei-international/TRASE/tree/master/trase/models/Indonesia/Palm%20Oil.