Independent monitoring of zero deforestation commitments through supply chain data

Bullet-point draft

Abstract

Max 150 words.

Introduction

ca. 800 words.

• More than 150 countries and multinational companies have made public zero deforestation commitments (ZDCs) – commitments to eliminate deforestation from the production of agricultural commodities {cite Forest 500}.
• These commitments include the NYDF, Amst decl… etc.
• If implemented, these commitments offer a powerful lever for reducing deforestation, biodiversity loss, and agricultural greenhouse gas emissions.
• In practice, however, the potential impact of ZDCs is undermined by weaknesses in their definition and implementation - not least a lack of transparency (Garret et al., Lambin et al., NYDF progress report)
• Most ZDCs do not include a clear blue-print for their implementation, monitoring, and the reporting of their impacts - i.e. how the risk of deforestation in supply chains is changing over time.
• Without transparency, it is impossible to differentiate whether they are actually leading to on-the-ground improvements in sourcing, or are paper commitments used to improve the image of companies. (The wording of this needs improvement - we could also discuss the risk that they are a distraction from other froms of conservation efforts).
• Transparency is also key to enable third parties and shareholders to hold committed companies to account, and to pressure them to implement these commitments.
• Commodity supply chains are, however, long and complex, and some of this lack of transparency stems from a paucity of data and mechanisms for reporting these deforestation risks.
• The reality of commodity trading is that many of the agricultural products associated with deforestation are often sourced in the spot market, and traded in bulk. At various stages in the supply chain - at storage facilities, and processing plants, maritime vessels - products from multiple sources may be mixed. The exact identity of the original source is therefore lost.
• In cases where companies do have information about the location of production, as for some certified supply chains, they are often reluctant to publish such high-resolution information.
• Here we therefore present a middle-ground approach to monitor the impact of ZDCs (Godar et al., 2016). By stitching together official per-shipment trade records with comprehensive asset ownership registries, sanitary inspection records, sub-national commodity production statistics and remote sensing data on agricultural activities and land use change, it is possible to quantify the commodity-associated deforestation occurring in the localities from which different actors source, and thereby observe the deforestation trends in the supply chains of companies and countries both with and without ZDCs.

Soy in Brazil

• We demonstrate this approach for ZDCs made by soy export traders operating in Brazil and by countries sourcing soy from Brazil. We focus on these actors and on the Brazilian soy sector for four reasons.
  • First, soybean production, along with beef, oil palm, and wood products, is one of the leading causes of commodity-driven deforestation in the tropics along with beef, oil palm, and wood products (Henders et al., 2015).
  • Second, Brazil is the world’s largest exporter of soy and has experienced a massive recent expansion in soy production. Since 2000, the area of soy has grown 2.5-fold, today covering an area of 34 Mha in Brazil (a land area larger than Finland) (IBGE, 2017). This expansion has been a major direct and indirect driver of the loss of forests and other natural vegetation – notably in Brazil’s Amazon and Cerrado biomes {reference – Gibbs}.
  • Third, soy trading is typical of agricultural commodities, in being characterised by long and complex supply chains. Soy is a traded in bulk. Traceability is made even more challenging because soy is traded both as a raw bean and as a processed product. Soy is milled and sold as soybean meal, used for animal feed, and as soy oil which is primarily used as a biofuel. Each of these products are consumed both within Brazil and abroad. X, Y, and Z % of Brazil’s raw beans, meal, and soy oil are exported each year, with major markets in Europe and Asia.
  • Finally, the Brazilian soy sector is also an exemplary case by being the subject of several ZDCs.
  • While some consumer-facing brands have made ZDCs, we focus on ZDCs by traders and international governments because of data constraints and because of their potential leverage to improve the sustainability of international supply chains.
  • Supply chains are often hourglass-shaped (Lyons-White and Knight, 2018), with a small number of traders acting as intermediaries between tens of thousands of producers and millions of consumers. If the ZDCs of traders are implemented and effective, this would likely have a multiplier effect on the deforestation risk of both upstream farmers and downstream consumers.

Summary paragraph

• By stitching together novel datasets, we find that companies and consumer markets that have made ZDCs have historically had high soy deforestation risk. Since making their commitments, there has been no noticeable reduction in their deforestation risk. We end by reflecting on the prospects of acheiving zero deforestation commodity production.

Methods

ca. 650 words + supplementary material.

• Explanation of SEIPCS - Ben and I have a schematic figure ro summarise SEIPCS that we would like to try; I think he’s alerady discussed this with Toby.
• Calculation of 5-yr soy deforestation risk, and our different metrics of soy deforestation risk. Justify our focus on the Amazon/Cerrado, citing Hansen group’s recent article.
• We make use of the Brazilian government’s official PRODES dataset; as such our estimates are aligned with the Brazilian government’s efforts to enforce the Forest Code and define legal vs illegal clearance.
• Proportional allocation of soy deforestation risk.
• Assessment of who does/doesn’t have a commitment. Assessed commitments according to the criteria for effectiveness from Garrett et al.

Results & Discussion

ca. 1400 words + supplementary material.

Opening paragraph summarising the commitments:

• We identified eight ZDCs which apply to the Brazilian soy sector, summarised in Table 1. (Note: Table 1 is shared in a separate Word document).
• Most of these are recent (post-2014).
• Brief description of commitments - noting cut-off dates, and that these companies are the five largest in terms of market share (Figure 6 - in the supplementary material).
• There is an apparaent trade-off between accountability and scope - the SoyM has a historic cut-off date and clear monitoring and reporting; the others are global but lack transparent mechanisms (Table S1 will describe ZDCs in more detail).
• In general, I want to keep text here to a minimum and direct the reader to the SOM.

Table 1 to be inserted here - it is saved in a separate Word document

ZDC Coverage has increased but is uneven.

• SoyM was signed in 2006 and covers ca. 90% of soy exported from the Amazon, but doesn’t apply to the Cerrado (Figure 1).
• As of 2017, only around 30% of soy exported from the Cerrado was traded by companies who have made commitments.
• This low coverage is problematic because the Cerrado is where most soy-associated deforestation is taking place. It also has lower protection than the Amazon under Brazilian law.

Regional variation in the soy land use dynamic

Before discussing the effect of ZDCs, I think we need some text explaining soy deforestation trends in Brazil - we don’t really want readers to leave with the misconception that soy deforestation is on a permanent downward trajectory. * As a whole, soy deforestation peaked in the late 2000s, and has been stable or even declined since then - though there are marked differences in land use dynamic between regions. Direct clearance of forests for soy has fallen in old soy frontiers in the Cerrado - notably Mato Grosso (Figure 2 or 8 in SOM) - and in the Amazon biome since the introduction of the Soy Moratorium in 2006. Soy, however, continues to be associated with deforestation in the Matopiba region, where over the last decade 0.5-0.8 Mha of soy each year has been planted on recently deforested land (Figure 2/8).

Figure 1: (A) ZDC coverage is higher in the Amazon biome, where the Soy Moratorium applies, than in the Cerrado, where most soy deforestation is taking place (B)

Note 1: I did look at regionalised ZDC coverage in the Cerrado, but it paints no clear picture: the proportion is ca. 0.45 in Matopiba vs 0.6 in MT vs 0.42 in the rest of the Cerrado - this is shown in the document of alternative plots. In the maintext, I favour inclduing a simple figure which doesnt’ differentiate ZDC coverage per region, since the message is much clearer. Let me know if you disagree.

Note 2: Here is an alternative to the above plot, which differentiates cerrado clearance per region. This plot is more complex visually, but I think this information is important. Alternatively, we could put the regionalised deforetation plot in the SOM (currently it’s repeated there as Fig 8).

Figure 2: (A) ZDC coverage is higher in the Amazon biome, where the Soy Moratorium applies, than in the Cerrado, where most soy deforestation is taking place (B)

No evidence of commitments having effect

• These commitments are on the most part recent, and it is too early to say that they will not have any effect, but our data show little evidence of impacts on the deforestation risk of committed companies.
• The companies adopting ZDCs have historically had high soy deforestation risk (Figure 2 & Figure 6 in SOM). Though the relative soy deforestation risk of Bunge, for example, has fallen over the past decade, Bunge has consistently had a soy deforestation risk above the market average (Figure 2b) and was associated with more than a third of all soy deforestation risk (XX ha) - which is disproportionate to its <20% market share (Figure 6). Since making their commitment in 2015, their soy deforestation risk remains close to the market average (Figure 2b).
• More than being shaped by commitments, companies’ soy deforestation risk is strongly determined by the geographies in which they have made infrastructure investments and procure soy. Bunge, Cargill, and ADM all source from within Matopiba, where most soy deforestation is taking place (Figure 7 - in SOM). Amaggi and Louis Dreyfus however have operations concentrated in Mato Grosso and the south-west Cerrado, respectively (Figure 8 - in SOM), areas that are older soy frontiers. While Amaggi had high soy deforestation risk in the mid-2000s, this has since fallen, reflecting the pattern of soy deforestation in the state, where soy has already occupied much of the landscape and expansion has slowed (Figure 7 - in SOM).

Figure 3: (A) The soy deforestation risk (ha/thousand tons soy) of major traders with and without ZDCs.(B) Soy deforestation risk, relative to the market average for each year (shown as a grey line). Points above (below) the line have above (below) average market risk.

• A similar picture emerges for commitments by national governments.
• The NYDF and Amsterdam declaration countries have had higher-than-average soy deforestation risk. If anything, this has worsened since their commitments were made (Figure 3) - which bodes poorly for their 2020 commitment of eliminating commodity-associated deforestation.

Figure 4: (A) The soy deforestation risk (ha/Mton) of different consumer markets. (B) The soy deforestation risk, relative to the market average in each year (the grey line).

Using supply chain data to monitor zero deforestation commitments.

• These data, of course, only paint a partial picture of the progress of these commitments. Soy is but one commodity, and Brazil one producer – though a major one. The NYDF applies to all sources of deforestation and the Amsterdam declarations apply to both soy and oil palm production, and a complete picture of their impacts must analyze deforestation trends across multiple commodities – ideally also accounting for indirect effects.
• We believe that our method for bringing together supply-chain specific data can be extended to other commodities and geographies, though the wording of some ZDCs can undermine big-data approaches to monitoring. The coverage of ZDCs is, for example, higher in the oil palm sector than in soy (estimates range from 60-96% of oil palm is traded under some form of ZDC (CLUA, 2014; Hurowtiz, 2014), and the vast majority of these prohibit the clearing of “High Conservation Value” forest {cite Garrett et al. forthcoming}. The designation of High Conservation Value areas, however, requires labor intensive on-the ground assessment; it cannot be measured using satellite imagery, which makes large-scale monitoring of these commitments challenging.

Prospects for deforestation-free soy

• Given what we know about the implementation of zero deforestation commitments, it is perhaps unsurprising that we see no noticeable change in the soy deforestation risk of commited actors.
  • Cargill is focussed on illegal deforestation rather than zero deforestation.
  • ADM the most transparent on implementation - but their initiatives are fluffy.
  • Bunge set out three-phase effort to increase traceability and claim that 90% of their direct sourcing in high risk regions (i.e. Matopiba) is traceable to the farm. They say nothing about third-party suppliers. They report that in 2017/18, 98 monitored farms were flagged for deforestation within their boundaries. Nine farms were suspended as of October 2018 based on their failure to engage further with Bunge. Despite this, their deforestation risk has remained about the market average.
• There is a similar picture for the multi-lateral commitments by governments, NYDF and Amsterdam declaration countries:
  • While 36 national governments signed the NYDF, the NGO and civil society effort to follow up on this commitment has tended to focus on the multinational companies who also signed.
  • The follow-up on the Amsterdam declarations appears to be at the engagement, rather than implementation stage. After signing, each signatory government set up a national roundtable to engage businesses on the zero deforestation agenda, though there remain few examples of concrete initiatives to change the sourcing patterns of companies supplying their markets. Here, efforts by Norway and France are perhaps the most advanced. {Provide details of each, and link to EU action plan}.
• Though the Brazilian context is in some way favourable to the implementation of ZDCs - there are, for example, high quality remote rensing products and a rural property cadaster available (Table S1) - our assessment suggests that these commitments as currently defined are unlikely to lead to significant reductions in deforestation.
• First, several do not set deadlines for eliminating deforestation (Table 1).
• Second, they do not have transparent mechanisms for enforcing their commitments (Table S1).
• Third, they are undermined by several structural factors {these could be framed as different kinds of leakage - though Patrick might disagree! - alternatively, we can simply call them challenges}:
  • Leakge between actors - while Amaggi & LD have made commitments, they have a joint-venture ALZ Graos, which is associated with high deforestation risk. Explain the history of ALZ Graos, and point out that if included in their figures it increases their soy deforestation rsik 5x and 10 x respectively (Figure 4).
  • Leakage because of low coverage. Committed traders make up ca. 50% of the soy export market - the proportion of soy deforestation associated with other companies and markets is however growing (Figure 5). Similarly, the importance of Amsterdam dec & NYDF signatories is decreasing as their market share declines. If only a segment of the market demands ZDC, it risks creating a differentiated market (where deforestation-free goes to EU market and other soy flows elsewhere), with no effect on overall deforestation. Here we should make reference to multi-stakeholder initiatives like the Cerrado Working Group, and the fact that China buys 60% of Brazilian soy. Chinese buyers currently more focussed on legality than zero deforestation, though COFCO’s president has called for Moratorium in the Cerrado and the meat industry has a ZDC.
  • Leakage between commodities - though the commitments by individual traders are in theory global, they do not address the soy-beef dynamic. A sustainable soy sector requires a holistic approach to governing agriculture. {Nb this may be too similar to Javier’s article, and is not key to the messaging, so we could cut this}.

Figure 5: The soy deforestation risk (ha) of Amaggi, Louis Dreyfus, and their joint-venture ALZ Grãos.

Figure 6: Traders and consumer markets not covered by ZDCs account for an increasing a proportion of soy deforestation risk

Conclusion

• Something along the lines of:
  • We are little over a year away from the NYDF’s 2020 deadline for zero deforestation commodities, at a time where global demand continues to increase and political winds are shifting in favor of agricultural expansion, rather than forest conservation {cite one of Soares Filho’s recent pieces about Bolsonaro}.
  • In the current political climate, corporate commitments may become increasingly important for efforts to deliver sustainable supply chains – but only if they are implemented and held accountable.
  • Here we present a pragmatic middle-ground approach which can help deliver accountability to actors along international supply chains and provide independent monitoring of their progress in delivering on promises for zero deforestation commodities.

Plots in the supplementary material

In the supplementary material, I think we should also include a plot of the deforestation risk of each of the top 30 traders. I did this for the last version, but haven’t copied the plots over here yet.

Figure 7: Comparison of the market share and share of soy deforestation risk of top-10 soy traders in Brazil (data from 2008-2017)

Figure 8: While soy deforestation has decreased in consolidated regions of the Cerrado, such as Mato Grosso, clearance in Matopiba continues.

Figure 9: Soy sourcing of companies with ZDCs. Note this needs to be improved a fair bit - e.g. playing with the scale