Introduction

This is an R Markdown document investigating the need for additional soybean in Argentina to solve the SEI-PCS model 1.1 (which could be seen as stock + imports), where this additional soybean might be needed and what its magnitude might be between 2015 and 2019.

This investigation follows initial results of Argentina soy SEI-PCS (v.1.1) where production could not meet demand in Rosario. As a fix, we have implemented an “infinite bucket” solution to allow the allocation step to solve We can then compare the magnitude of that depleted bucket with our predicted storage + import information.

In a first step, we look at the soybean balance as Production + Imports - Domestic Demand - Exports and then compare to the amount of soybean needed for the allocation step. Then we look specifically at some ports to see if we can apply the infinite bucket method in those areas: Rosario, San Lorenzo and Quequen ports in particular.

Argentina’s soybean balance

We calculate the balance as:

\(Soybean ~balance = Production + Imports - Demand - Exports\)

Where all variables are in soybean equivalents (e.g. demand and exports include respectively soy cake and soy cake + oil converted into soybean equivalents through the equivalence factors). Production data is from GLAD, Imports from the FAO, Demand is estimated and Exports are from the Customs Declarations.

##   YEAR PROD_TONS SOY_IMPORTS SOY_DEMAND_TONS SOY_EXPORTS   BALANCE
## 1 2015  62306159         566         6302466    69565973 -13561715
## 2 2016  56123337      853849         6225864    67548476 -16797154
## 3 2017  56405478     1897831         5958253    63566244 -11221189
## 4 2018  36105845     6443248         6477214    45342446  -9270567
## 5 2019  53294539     4547682         6349398    71199859 -19707036

FAOSTAT data provides opening stocks of soybean in Argentina at (as unofficial figures) and stock variations:

2015: 21.7 Mtons (stock) and 5.4 Mtons (stock variation)
2016: 27.1 Mtons and 87,000 tons
2017: 27.2 Mtons and -160,000 tons
2018: 27.0 Mtons and -3.3 Mtons

These are not idle stocks, but likely some “backed up” soybean that hasn’t been crushed yet.

Soybean crushing capacity

We estimate the crushing capacity using the values obtained by CIARA (and published in JJ Hinrichsen, but available on the CIARA website http://ciaracec.com.ar/ciara). This data is also estimated by the FAO.

CIARA data provides a theoretical maximum daily crushing capacity. If we use this value as the theoretical total, we obtain:

2015: 86.3 Mtons compared to 63 Mtons of capacity and 44 Mtons of actual crushing (CIARA) - this value can be compared to processed beans by the FAO: 40.7 Mtons
2016: 86.5 Mtons compared to 63 Mtons of capacity and 48 Mtons of actual crushing (CIARA) - compared to 45.9 Mtons (FAO)
2017: 85.4 Mtons compared to 63 Mtons of capacity and 45 Mtons of actual crushing (CIARA) - compared to 43.8 Mtons (FAO)
2018: 85.4 Mtons compared to 63 Mtons of capacity and 40 Mtons of actual crushing (CIARA) - 39.4 Mtons (FAO)
2019: 88.7 Mtons compared to 63 Mtons of capacity and 45 Mtons of actual crushing (CIARA) - no estimate for FAO at this time.

The CIARA values are the total crushing capacity considering all crops (soy, sunflower, peanut, linen, cotton, safflower and rapeseed).

According to CIARA, the crush utilization rates are 71% (2014/15), 76% (2015/16), 71% (2016/17), 64% (2017/18), and 72% in (2018/19). We can apply some corrections between the theoretical crush of CIARA and the reported crush from CIARA:

2015: 0.51, with which we estimate 43.7 Mtons of crushing capacity
2016: 0.55, with which we estimate 47.4 Mtons
2017: 0.53, with which we estimate 45.2 Mtons
2018: 0.47, with which we estimate 40.3 Mtons
2019: 0.51, with which we estimate 45.1 Mtons
2020: to do

This means that the difference between soybean production + imports and estimating crushing in any given year is:

2015: 62.3 Mtons or a surplus of 18.4 Mtons
2016: 56.1 Mtons or a surplus of 9.4 Mtons
2017: 56.4 Mtons or a surplus of 13.1 Mtons
2018: 36.1 Mtons or a surplus of 2.3 Mtons
2019: 53.3 Mtons or a surplus of 12.9 Mtons
2020: to do

The surplus represents the additional soy bean available in the country which can be compared to soybean exports in a given year to then estimate the actual stock of beans in the country:

2015: 27 Mtons of bean exported leaves -8.7 Mtons of potential stock
2016: 21 Mtons of bean exported leaves -11.6 Mtons of potential stock
2017: 22.8 Mtons of bean exported leaves -9.7 Mtons of potential stock
2018: 11.3 Mtons of bean exported leaves -9 Mtons of potential stock
2019: 28.9 Mtons of bean exported leaves -16 Mtons of potential stock

We then compare these numbers to the soybean balance predicted by the SEI-PCS allocation step using the “infinite bucket” method

SEI-PCS additional soy bean requirements

All products

The allocation step in SEI-PCS v.1.1. uses an “infinite bucket” which can be used to ensure that export demands are met. In this first test, we check the size of this bucket (or amount of extra soybean) used by the allocation step and compare to official numbers.

We first compare to the soy balance estimated as Production + Imports - Domestic Demand - Exports

##   YEAR   BALANCE SEI_PCS_SOY_NEEDS     DIFF
## 1 2015 -13561715           9808023 23369738
## 2 2016 -16797154          14026814 30823968
## 3 2017 -11221189           8927301 20148490
## 4 2018  -9270567          11608333 20878900
## 5 2019 -19707036          18821422 38528458

The above is the expected stock from bean and oil + cake as needed for the allocation step (SEI-PCS_SOY_NEEDS) and the predictions from Production + Imports - Domestic Demand - Exports in soybean equivalents.

If we look at differences between bean and oil + cake we see the following differences, starting with bean.

BEAN comparison

We can compare these values to what we predicted would be available nationally:

##   YEAR STOCK_TYPE PROD_TONS SOY_IMPORTS CRUSHING_CAP BEAN_EXPORTS
## 1 2015       BEAN  62306159         566     52661178     27028102
## 2 2016       BEAN  56123337      853849     57086073     20999183
## 3 2017       BEAN  56405478     1897831     54632032     22837652
## 4 2018       BEAN  36105845     6443248     48656653     11308183
## 5 2019       BEAN  53294539     4547682     54086138     28922830
##   BEAN_SEIPCS_TONS_STOCK BEAN_TONS_BALANCE
## 1                1806340         -17382555
## 2                 313399         -21108071
## 3                 500686         -19166375
## 4                 499520         -17415743
## 5                1056314         -25166748

I am still not 100% sure what to make of these numbers:

BEAN_SEI-PCS_TONS_STOCK is the amount of bean taken from the “infinite bucket” to satisfy the allocation step (beans only)
BEAN_TONS_SOY_NEEDS is the amount of bean stock estimated by Production + Imports - Crushing - Exports

I find it reassuring that the need for beans in the allocation step does not overly exceed the expected stock, but the variations across years is a little worrying. The crushing capacity (or processed bean) reported by the FAO is slightly lower than what is reported above (estimated using CIARA) with about 1-2 Mtons lower values reports and the amount of soybean needed to meet export demand does not exceed 2 Mtonnes, given the possible error in the estimates (production, crushing, etc.) this is acceptable.

Important factors to consider in these calculations:

Trade data: there were some discrepancies between the pre-processed trade data and the official numbers. Differences in the total soy exports are about 18% (2015, 8 Mtons differences with COMTRADE), 17% (2016, 8 Mtons difference), 6% (2018, 2.7 Mtons different), 9% (2018, 2.9 Mtons difference), 21% (2019, 9.5 Mtons difference)
Production: there are some discrepancies between official numbers and remote sensing information, generally 1-3 Mton difference

OIL/CAKE needs

We then look at the size of the “SEI-PCS bucket” to meet the soy demand needs for both oil and cake:

##   YEAR STOCK_TYPE PROD_TONS SOY_IMPORTS CRUSHING_CAP OIL_CAKE_EXPORTS
## 1 2015   OIL_CAKE  62306159         566     52661178         42537871
## 2 2016   OIL_CAKE  56123337      853849     57086073         46549292
## 3 2017   OIL_CAKE  56405478     1897831     54632032         40728593
## 4 2018   OIL_CAKE  36105845     6443248     48656653         34034263
## 5 2019   OIL_CAKE  53294539     4547682     54086138         42277028
##   SOY_DEMAND_TONS OIL_CAKE_SEIPCS_TONS_STOCK OIL_CAKE_TONS_BALANCE
## 1         6302466                    8001683              13466387
## 2         6225864                   13713415               4202029
## 3         5958253                    8426615              11616463
## 4         6477214                   11108813               2037616
## 5         6349398                   17765108               9215794

Location where additional soybean is required

Additional soybean is required in the provinces of Buenos Aires and Santa Fe mainly (with the exception of Pampa in 2019, but closely related to Buenos Aires through Siogranos zone 2).

First we show the sources for BEAN (focusing on the Zone of origin as identified through the SEI-PCS method):

Then the sources for OIL AND CAKE:

Siogranos analysis

We look at the Siogranos grain movement from zones to ports to better understand the issues identified in the allocation step at:

Port of Rosario in the ROSARIO S Siogranos zone
Port of San Lorenzo in the ROSARIO N Siogranos zone
Port of Necochea in the QUEQUEN Siogranos zone

These are common port zones where soybean demand exceed production (as seen above)

Rosarios North and South

First, let’s compare the crushing capacity and the Siogranos grain movement to the ports of interest. We add about 10% to the volume of Siogranos (arbitrarily) to make up any missing volume that couldn’t be attributed during pre-processing of the Siogranos data.

## # A tibble: 18 x 6
## # Groups:   YEAR [6]
##     YEAR ZONE_DESTINATION SIOGRANOS_TONS CRUSH_CAPACITY_TONS SIOGRANOS_TONS_CORR
##    <dbl> <chr>                     <dbl>               <dbl>               <dbl>
##  1  2015 ROSARIO S              5365328.            5677575             5901861.
##  2  2016 ROSARIO S              3947178.            6142950             4341896.
##  3  2017 ROSARIO S              4299987.            5956800             4729986.
##  4  2018 ROSARIO S              2364356.            5305275             2600792.
##  5  2019 ROSARIO S              4028229.            5677575             4431052.
##  6  2020 ROSARIO S              2779130.                 NA             3057043.
##  7  2015 ROSARIO N             16768165.           35145302.           18444982.
##  8  2016 ROSARIO N             13944341.           38074245            15338775.
##  9  2017 ROSARIO N             11670185.           36873760            12837204.
## 10  2018 ROSARIO N              9617831.           32840692.           10579614.
## 11  2019 ROSARIO N             17761394.           37371802.           19537533.
## 12  2020 ROSARIO N              8575769.                 NA             9433345.
## 13  2015 QUEQUEN                2009501.             445300             2210452.
## 14  2016 QUEQUEN                1660862.             481800             1826948.
## 15  2017 QUEQUEN                1181846.             467200             1300031.
## 16  2018 QUEQUEN                 673899.             416100              741289.
## 17  2019 QUEQUEN                1299873.             445300             1429860.
## 18  2020 QUEQUEN                 529818.                 NA              582800.
## # ... with 1 more variable: DIFF <dbl>

In the above table DIFF refers to the differences between the zone’s crushing capacity (CRUSH_CAPACITY_TONS) and the volume of soybean arriving in the zone (SIOGRANOS_TONS_CORR).

So the estimated crushing capacity at Rosario generally exceeds the amount of soybean received at the ports (from our data). There is more discrepancy between soybean grain arriving in Quequen and the crushing capacity because grain is generally exported at Necochea (not oil or cake).

We then compare Siogranos grain movement to the region to total exports from both zones with customs office Rosario (Rosario S), San Lorenzo (Rosario N) and Villa Constitución (Rosario S).

##    YEAR ZONE_DESTINATION TONS_EXPORT CRUSH_CAPACITY_TONS SIOGRANOS_TONS_CORR
## 1  2015        ROSARIO S  4995635.62             5677575           5901860.6
## 2  2016        ROSARIO S  6196458.62             6142950           4341896.1
## 3  2017        ROSARIO S  5664491.02             5956800           4729985.7
## 4  2018        ROSARIO S  3641113.27             5305275           2600791.7
## 5  2019        ROSARIO S  5049691.73             5677575           4431051.7
## 6  2015        ROSARIO N 35081994.54            35145303          18444982.0
## 7  2016        ROSARIO N 36694040.66            38074245          15338774.6
## 8  2017        ROSARIO N 29552814.80            36873760          12837203.7
## 9  2018        ROSARIO N 24684363.64            32840692          10579613.8
## 10 2019        ROSARIO N 34478364.53            37371803          19537533.4
## 11 2015          QUEQUEN   224363.82              445300           2210451.5
## 12 2016          QUEQUEN   271808.30              481800           1826948.0
## 13 2017          QUEQUEN   192550.30              467200           1300030.6
## 14 2018          QUEQUEN   104175.00              416100            741289.2
## 15 2019          QUEQUEN    54091.13              445300           1429860.3
##        DIFF CAPACITY_DIFF
## 1   -224286       -681939
## 2   1801054         53509
## 3   1226814       -292309
## 4   2704483      -1664162
## 5   1246523       -627883
## 6  16700320        -63308
## 7  22735470      -1380204
## 8  24036556      -7320945
## 9  22261079      -8156329
## 10 17834269      -2893438
## 11 -1765152       -220936
## 12 -1345148       -209992
## 13  -832831       -274650
## 14  -325189       -311925
## 15  -984560       -391209

In the above table, CAPACITY_DIFF refers to the difference between the volume of soy oil or cake exported (TONS_EXPORT) and the crushing capacity in the zone (CRUSH_CAPACITY_TONS). Some observations:

In ROSARIO S and QUEQUEN, crushing capacity exceeds the exports of soy oil and cake by up to 2 Mtons
In ROSARIO N, there was way more exported oil and cake than crushing capacity in the zone (with the exception of 2018 where there was a drop in production).

There might be some movement of oil across the Rosario zones (which are close together). So we can combine them to see the joined capacity

##   YEAR TONS_EXPORT CRUSH_CAPACITY_TONS SIOGRANOS_TONS_CORR     DIFF
## 1 2015    44078185            82536355            73040528  9495827
## 2 2016    45823582            89397990            59042012 30355978
## 3 2017    37277381            86595520            52701568 33893952
## 4 2018    29903232            77124135            39541216 37582919
## 5 2019    44720659            86989355            71905755 15083600
##   CAPACITY_DIFF
## 1     -38458170
## 2     -43574408
## 3     -49318139
## 4     -47220903
## 5     -42268696

So now there is enough crushing capacity compared to the combined exports of ROSARIO N and S zones.

Argentina Stock Balance

Michael Lathuilliere

24/03/2022