Translate page with Google

Journalist Resource Publication logo October 4, 2023

How We Revealed That China’s Giant State-Owned Food Importer Is Fueling Deforestation

a farm in Brazil

Sasha Chavkin will examine China’s exposure to commodities linked to deforestation, including...


Most reporting on the global supply chains driving rainforest destruction focuses on multinational corporations based in the U.S. and Europe. These companies play an important role, but the country that imports the most products that contribute to deforestation is China. China’s imports accounted for a bigger share of tropical deforestation caused by international trade than the U.S. and Europe combined, a 2021 report by the WWF found.

This investigation explored how such massive amounts of forest-risk commodities were entering China, and what the Chinese government was doing to address this problem. To examine these questions, we focused on the China Oil and Foodstuffs Corporation, which goes by the name Cofco. Cofco describes itself as “the leader of the Chinese agricultural industry” with annual revenues of more than $110 billion. Its commodity trading wing, Cofco International, imports huge volumes of soybeans, palm oil, and other products from Brazil, Indonesia, and other rainforest nations. As a state-run company, Cofco’s actions directly reflect the policies of the Chinese government.

Cofco warehouses in Mato Grosso, Brazil, received soybeans planted in deforested areas in the state through contracts with intermediaries. Image by Sasha Chavkin/Repórter Brasil.

Cofco also describes itself as a leader in sustainability and has received billions of dollars in reduced-interest loans tied to promises to clean up its supply chain. We wanted to investigate whether Cofco—and by extension the Chinese government—was living up to these promises.

As a nonprofit journalism organization, we depend on your support to fund journalism covering underreported issues around the world. Donate any amount today to become a Pulitzer Center Champion and receive exclusive benefits!

Our hypothesis was that despite its public promises, Cofco’s supply chains for soybeans and palm oil were still contaminated with deforestation. We suspected that Cofco was not doing enough due diligence on its suppliers to screen out farms that had cut down rainforests and that it was not adequately enforcing the strict supply chain policies it had committed to on paper. We also questioned whether the traceability pledges that had allowed Cofco to obtain billions of dollars in reduced-interest loans were genuinely changing its buying practices on the ground. 

Data Sources

Ibama: embargoes and fines for environmental offenses by the Brazilian Federal Environmental Agency

GeoPortal Sema: data and metadata portal of Mato Grosso’s environmental agency providing georeferenced information

Sema/MT: documents with terms of embargo and permits for legal deforestation issued by Mato Grosso’s environmental agency

SIGEF, CAR: property records in Brazil from its Land Management System and Rural Land Registry

Prodes: deforestation data based on satellite monitoring by Brazil’s National Institute for Space Research (INPE)

MapBiomas: recent deforestation, fire alerts and analysis of land cover, based on satellite monitoring by a network of NGOs and universities

Global Forest Watch: satellite-based analysis of land cover by the World Resources Institute

Planet: satellite imagery from Planet Labs (special access through the NICFI program)

CruzaGrafos (limited access): system created by Abraji (Brazilian Association of Investigative Journalism) to search for a person's name and see companies 

JusBrasil: Brazilian court cases against companies and individuals

Cofco: palm oil supply mill lists published by Cofco International on its website

Unilever: Suspended or banned palm mill lists published by Unilever on its website

Connecting the Company to Deforestation

1. Researching Cofco’s supply chains

We found the first signs that Cofco’s supply chains were contaminated by deforestation from publicly available documents and reports by environmental groups.

A number of farms that supplied soybeans to Cofco in Brazil were linked to recent cases of deforestation in reports by AidEnvironment, a supply chain monitoring group. We also compared Cofco’s publicly available list of palm oil supply mills with a list of mills that had been banned or suspended by Unilever, the giant British consumer goods company, for sourcing palm oil fruits from recently deforested land. We found that some of the mills blocked by Unilever remained on Cofco’s supplier lists, including one mill linked to reports of deforestation in a protected national park in Aceh, Indonesia that has been described as the “orangutan capital of the world.”

But Cofco—and China as a whole—imported much higher volumes of soybeans from Brazil than any other forest-risk product. We wanted to take a deeper look at Cofco’s exposure to deforestation in these critical soy supply chains.

To do this, we scraped publicly available documents that showed us a sample of the sales by soy producers to Nutrade Comercial Exportadora, an intermediary company that delivered grains to Cofco International and other traders in Brazil. These documents made it possible to identify the producer and/or the farm of origin of the soybeans, even though Cofco does not disclose its soy suppliers. In most of these cases, Cofco International was buying from what specialists call indirect suppliers. The farmers did not have a direct relationship with Cofco; instead, they sold the product to local traders who in turn sold them to Cofco.

2. Identifying case studies

Once we had the list of Cofco’s indirect suppliers from scraping these sales documents, we started investigating the environmental crimes related to these producers. We did this by looking up the suppliers in public databases such as the list of embargoes imposed by the federal environmental agency IBAMA and embargoes imposed at a state level by the State Environmental Agency of Mato Grosso (SEMA). We also researched the names of these indirect suppliers on land information databases, such as Sigef, the land register data system of Incra, Brazil's land agency and CAR, the Rural Environmental Register of properties in Brazil. These datasets provided map files for the properties we were investigating. Since some suppliers owned more than one property, this allowed us to see whether the deforestation had occurred on the same properties that sold grains to Cofco. 

Our analysis of Cofco and 86 of its soy suppliers in Mato Grosso found that 15 suppliers had been fined or embargoed by SEMA or IBAMA for recent violations of state or federal rules. At least five and as many as 11 of those 15 suppliers were under an active embargo during the time Cofco did business with them.

Another important tool we used was Global Forest Watch’s “soy-planted area” feature. This allowed us to see whether soy was being cultivated within areas that had been deforested. In one of our key cases involving the landowner Joao Luiz Lazarotto, this feature provided us with strong indications that he had not only cleared forests (red polygon) but also that he had converted that land into soy plantations (olive green in right image).

João Luiz Lazarotto’s farm Fazenda União II in Mato Grosso was deforested 650 hectares (1,600 acres) in 2017 (red polygon, above), as seen on the satellite imagery platform Planet. The deforested area was cultivated with soybean (highlighted in green, right) in the following year, according to the Global Forest Watch database. Image by Planet Labs & Global Forest Watch.

It was also important to check whether the embargo was active at the time the soybeans were purchased—otherwise, the case would not represent a failure of Cofco's due diligence. In another case study involving the farm Uggeri Agropecuária, Mato Grosso’s state environmental agency imposed an embargo in 2015 that prohibited the farm from cultivating in the areas it had illegally cleared. Satellite images showed that Uggeri Agropecuária continued to grow soybeans within the embargoed area, and it indirectly sold them to Cofco in February and March 2021, while the embargo was still in effect. However, this embargo was later lifted during the course of our investigation. It was therefore important to keep track of the list of farms embargoed by the environmental agency of Mato Grosso in order to correctly report the status of the area.

In response to the investigation, Cofco stated that "we are working to increase traceability of indirect purchases, which will lead us to strengthen our controls and risk monitoring for this part of the supply chain."

3. Checking Cofco’s promises

Cofco describes itself as a global leader in sustainability and has received billions of dollars in sustainability-linked loans, which allow the company to pay lower interest rates if it meets pre-agreed environmental targets. The exact benchmarks for these loans are not public, but Cofco has said they are connected with tracing its soy and palm oil supply chains to avoid buying from farms that cut down rainforests. It has also announced a series of supply chain policies, including specific rules for soy and palm oil suppliers.

We reviewed Cofco’s supply chain policies, annual sustainability reports, and public statements and then compared them with the cases we had identified. We found the cases appeared to violate various commitments by the company, including its Code of Conduct, Supplier Code of Conduct, Sustainable Soy Sourcing Policy, and Sustainable Palm Oil Sourcing Policy.

4. Reporting in the field

To confirm our analysis and report on conditions on the ground, we visited three farms in Mato Grosso that we had identified as Cofco suppliers involved in deforestation. We observed and photographed the properties, spoke with local agricultural groups, and requested interviews with the landowners allegedly responsible for deforestation.

One of our key interviews was with Lazarotto, a prosperous landowner who, satellite images showed, had cleared 650 hectares of forests on his property in 2017. Lazarotto had legal permission to clear this land, but his actions violated the conditions of the Soy Moratorium, an agreement among leading soy traders not to buy soybeans grown on recently deforested land in the Amazon. Cofco is a signatory to the Soy Moratorium, which meant it had committed not to buy soybeans from farms that engaged in this conduct.

When we spoke to Lazarotto, he openly admitted to clearing the land in question. He said the Soy Moratorium was unjust because it went further than Brazilian law and made unreasonable demands on farmers. He complained that his biggest buyer, Grupo Amaggi, refused to buy the soybeans grown on the recently deforested tract of his property because of the moratorium. He contrasted Amaggi with his second largest buyer, Cofco—which he said he was still doing business as usual and had not asked him to remove the soybeans grown on deforested land.

Sasha Chavkin interviewing Joao Luiz Lazarotto at the offices of Fazenda Lazarotto, his farm in Mato Grosso, Brazil. Image by Giovanny Vera.

Chavkin's visit to the property of Uggeri Agropecuaria, one of the farms investigated for the story. Image by Giovanny Vera. Brazil.

Responding to Company Denials

One of our key challenges emerged when we reached out to Cofco to get its response after we had completed our supply chain research, case studies, and field reporting.

To our surprise, Cofco responded to our initial inquiry by claiming that it had never bought soybeans from any of the three farms in question. When we sent receipts showing its purchases, Cofco shifted its position. It said it had bought from those landowners, but that its purchases came from sections of their property that were registered separately in Brazil’s Rural Land Registry (CAR) from the deforested zones. It said these differently registered sections of property were in fact different farms, and that therefore it had not purchased soybeans grown on deforested land or violated its supply chain commitments.

When we asked experts about Cofco’s response, they told us that its claims didn’t make sense if the farms in question were still a single operation. Dividing the land into different registrations, they said, was a strategy for gaming the system. It allowed landowners who cut down forests to sell products to big commodity traders that had promised to avoid deforestation simply by marking down their sales as coming from a “clean” section of the farm. Since landowners themselves filled out the sales receipts, this tactic was easy to use unless the traders conducted due diligence to confirm the origin of products (as Amaggi Group did with Lazarotto).

But we still needed to show evidence that these farms were a single operation. To do this, we relied on satellite images to visually demonstrate that the properties were really one farm. One important piece of confirmation was the lack of borders between the differently registered sections. We found that there were no fences or other boundaries separating the “clean” and deforested areas, and that sometimes the same fields straddled both sides of the border claimed by the landowner. 

Another key piece of evidence was grain silos. For both Lazarotto and Uggeri Agropecuaria, we found the only grain silos on the property were in the “clean” section of the farm—meaning that the grains were likely mixed together and all marked as originating from the compliant zone. The images of Lazarotto’s property below show that its only grain silos are in the upper left corner of the farm (see a closeup of silos in the image at right).

Images courtesy of Planet Labs.

Supply Chain Tracking in Your Reporting

The soy-planted area feature of Global Forest Watch was critical for identifying areas covered by government embargoes or the Soy Moratorium where soy was still being planted. However, field reporting was still necessary to confirm the findings of this analysis. 

We also found that satellite images were a key tool for refuting dubious claims by trading companies. In particular, the images allowed us to disprove the argument that buying from different CARs within a single operation meant the buyer was not exposed to deforestation. 

To use this method effectively, journalists must have a clear standard for supply chain contamination. In our view, when a company buys from a farm with an active violation of local laws or company policies against the destruction of forests, and which is growing crops on the deforested land, the buyer is exposed to deforestation. A lower standard, such as buying from a farm with any history of deforestation, would not reflect whether a farm had returned to compliance or whether its deforested zone was connected to the supply chain. On the other hand, only considering a supply chain to be contaminated if an individual shipment can be traced from a deforested zone to a buyer would allow producers and traders to use tactics such as separate CAR registrations to avoid accountability for deforestation.


logo for the Rainforest Investigations Network


Rainforest Investigations Network

Rainforest Investigations Network


a yellow halftone illustration of a truck holding logs



navy halftone illustration of a halved avocado


Food Security

Food Security