Summary

This document investigates the intra-trade information contained within the siogranos database for soy beans, soy oil and corn between 2015 and 2020. The analysis should provide a better understanding of the links between departments of production and port zones where products are being shipped.

The final output of this analysis is a set of rules (the “siogranos rules”) that are to be used in Argentina soy and maize SEI-PCS as a linear programming constraint.

The siogranos data is pre-processed and then a matching tool (openrefine) was used to match the names of the source localities with the departments of Argentina. Siogranos information only shows locality/province connections to siogranos zones of destination. In other words, source departments are missing and need to be found to be able to show the connections between source regions (and siogranos zones of origin) with the destinations in siogranos.

More details behind the cleaning of the siogranos files see the R script on Github

Pre-processing and names matching (to find source departments) steps results in >87% of commodity volumes mapped in each of the years of interest, more specifically:

This notebook carries out the following steps:

Step 1: check the content of the siogranos data

The data was downloaded according to the date of operation (Fecha de operación) or concertation (Fecha de concertación), and not the date of delivery (Fecha de entrega). You will notice that each operation is linked to a specific harvest (Cosecha) together with a harvest year.

Some contracts are for future deliveries, meaning that the operation is recorded in one year (e.g. 2015) for a future delivery (e.g. Cosecha 2017). We therefore need to understand how many of the annual operations are linked to the year of harvest, and which are linked to past or future harvests.

The first version of SEI-PCS used an incomplete version of the siogranos data for soy only with the following filters:

PRODUCTO == "SOJA" & OPERACION %in% c("CONTRATO", "ANULACION", "AMPLIACION", "ANULACION FIJACION") & DESTINATION_TYPE == "EN DESTINO" & DEPARTMENT %in% dep_prod & (as a department of production) ESFINAL == "SI"

We can use the same filters for this analysis, removing the department of production filter in this particular analysis.

We look at the details of the operations in any given year to check the overlap between the year of operation and harvest year. All values below represent a percentage of total volume matched (meaning a fraction of the >87% matched - see details above):

SOY BEAN

SOY OIL

MAIZE

Observation: for any given year, there are contracts for delivery of harvest prior and following the operation year. To reach 90% of the trade of soy we should consider both years. This likely has to do with actual contracts of stored product. Filtering to the proper “Cosecha” also aligns the dates for the contracts with the dates of delivery (some deliveries are made in January in the following year, and are still included).

Decision: keep all intra-trade flows that represent > 1% of total product volume, and more than 10 trips per year to be used in the siogranos rules. This seems to represent best the flows linked to export.

Step 2: map the siogranos zone connections

Soy bean

We filter the siogranos data considering harvests that include the year of operation of the contract (e.g. 2015 contract year will include the 2014/15 harvest and the 2015 harvest (which are separate entries in Siogranos)). This filtering option also aligns the delivery dates with the export years of interest for SEI-PCS.

The first observation is that the amount of soybean traded in Argentina are within the total annual production in the country (following the selections made with the data, and keeping in mind that a lot of the soybean may be stored from previous years and traded):

  • in 2015, siogranos reports 43 Mtons for a production of 61 Mtons (Ministry of Agroindustry)
  • in 2016, siogranos reports 35 Mtons for a production of 59 Mtons (Ministry of Agroindustry)
  • in 2017, siogranos reports 30 Mtons for a production of 55 Mtons (Ministry of Agroindustry)
  • in 2018, siogranos reports 22 Mtons for a production of 38 Mtons (Ministry of Agroindustry)
  • in 2019, siogranos reports 39 Mtons for a production of 55 Mtons (Ministry of Agroindustry)
  • in 2020, siogranos reports 22 Mtons for a production of XX Mtons (FAOSTAT) - data not available at the time of writing

A first check on Siogranos trade flows was carried out to try and devise some rules to select the links between zones keeping in mind the input from the consultant in 2019. It seems that a selection of flows using annual volume > 1% and > 10 trips per year seems to be the right selection to try and isolate the connections that would be more linked to exports. It would be worth doing a sensitivity analysis on this choice and how it may impact the SEI-PCS results in all years.

We could also think about some discounting for the rest of the rules based on the remaining percentages.

Maps below refers to Siogranos 2020 data for soy bean and highlights the intra-trade volume sourced from zones that are connected to the destination zone (in blue).

Soy oil

We filter the siogranos data considering harvests that include the year of operation of the contract as well as the year prior to the contract (e.g. 2015 contract year will include the 2014/15 harvest and the 2015 harvest (which are separate entries in siogranos)).It seems that about 20% (on average) of soybean in one year gets crushed up to two years later.

The first observation is that the amount of soybean traded within Argentina is much smaller than the production (following the selections made with the data). This is in line with the in-country knowledge that soy bean is generally crushed at the export location:

  • in 2015, siogranos reports 0.2 Mtons for a production of 8 Mtons (FAOSTAT)
  • in 2016, siogranos reports 0.2 Mtons for a production of 9 Mtons (FAOSTAT)
  • in 2017, siogranos reports 0.2 Mtons for a production of 8 Mtons (FAOSTAT)
  • in 2018, siogranos reports 0.3 Mtons for a production of 7 Mtons (FAOSTAT)
  • in 2019, siogranos reports 0.2 Mtons for a production of X Mtons (FAOSTAT) - data not available at the time of writing
  • in 2020, siogranos reports 0.1 Mtons for a production of X Mtons (FAOSTAT) - data not available at the time of writing

It is difficult to make a call on whether all of this trade is specific to international trade. Additional research is needed to see where the locations of the crushing facilities are (e.g. see CIARA).

Maps below refers to siogranos 2020 data for soy oil and highlights the intra-trade volume sourced from zones that are connected to the destination zone (in blue).

Maize

We filter the siogranos data considering harvests that include the year of operation of the contract as well as the year prior to the contract (e.g. 2015 contract year will include the 2014/15 harvest and the 2015 harvest (which are separate entries in siogranos)).

The first observation is that the amount of maize traded in Argentina are within the total annual production in the country (following the selections made with the data, and keeping in mind that a lot of the maize may be stored from previous years and traded):

  • in 2015, siogranos reports 18 Mtons for a production of 34 Mtons (FAOSTAT)
  • in 2016, siogranos reports 17 Mtons for a production of 40 Mtons (FAOSTAT)
  • in 2017, siogranos reports 21 Mtons for a production of 50 Mtons (FAOSTAT)
  • in 2018, siogranos reports 18 Mtons for a production of 43 Mtons (FAOSTAT)
  • in 2019, siogranos reports 33 Mtons for a production of 57 Mtons (FAOSTAT) - data not available at the time of writing
  • in 2020, siogranos reports 20 Mtons for a production of X Mtons (FAOSTAT) - data not available at the time of writing

Step 3: graph time series of siogranos data

There has been a decline in soybean and soy oil trade to the main ports in Argentina between 2015 and 2020. The trade of maize has experienced ups and downs. We need to keep in mind that this is the “filtered” siogranos information and that this trade does includes some product from other years (perhaps stored on site).

What is striking though is the sharp decline in deliveries of soy bean and oil to Rosario N which might be linked to production, but also the fact that an increasing amount of soy bean is crushed directly in Rosario (this is where the crushing capacity information needs to come in).

Rosario also remains an important hub for exports of maize which may also be linked to crushing capacity at this location.

The following graphs show the influx of products into Argentina’s main locations and ports of export:

Step 4: create the siogranos rules

The siogranos rules need to reflect the connections between zones of origin and zones of destination to be used in soy and corn SEI-PCS.

The proposed filter to use for the siogranos data is:

PRODUCTO == "SOJA" & OPERACION %in% c("CONTRATO", "ANULACION", "AMPLIACION", "ANULACION FIJACION") & DESTINATION_TYPE == "EN DESTINO" & ESFINAL == "SI"

Using the department of production in deforestation free together with the following selections:

  1. harvest year as defined above (according to product)
  2. representing > 1% of total trade in the given year (together with 3.)
  3. representing > 10 trips between zones in the given year (together with 2.)

The siogranos rules could potentially use a percentage distribution based on the intra-trade volume percentage identified in the dataset.

Special note: The siogranos rules are mean to reproduce the general tendencies of soy bean, oil and corn movement in Argentina.

Following initial runs of SEI-PCS v.1.1, additional connections were removed when creating the Siogranos rules:

## [1] "saved at C:/Users/MICHAE~1.LAT/AppData/Local/Temp/RtmpEBSYbS/file1a587c94c5d/SIOGRANOS_RULES_SOY.csv"
## [1] "saved at C:/Users/MICHAE~1.LAT/AppData/Local/Temp/RtmpEBSYbS/file1a58167bc6d/SIOGRANOS_RULES_OIL.csv"
## [1] "saved at C:/Users/MICHAE~1.LAT/AppData/Local/Temp/RtmpEBSYbS/file1a582af23a53/SIOGRANOS_RULES_MAIZE.csv"

Important considerations

When working on SEI-PCS some of the following should be considered: