Data and code from: False reporting undermines the integrity of supply chain sustainability initiatives
Data files
Aug 04, 2026 version files 209.68 KB
-
Audit_survey_v7.1_anonymised.xlsx
25.57 KB
-
cheating_wide_anonymised_v2.csv
46.73 KB
-
cheating_wide_anonymised_v2.dta
55.20 KB
-
codebook.xlsx
9.80 KB
-
collusion_analysis_v3.do
71.08 KB
-
README.md
1.30 KB
Abstract
The supply chains of agricultural commodities are major drivers of tropical deforestation and forest degradation. Sustainability standards are fundamental to improve sustainability in these supply chains, but they rely on field data collected by actors who may benefit from favorable outcomes—creating potential for collusion. This study is the first to quantify such collusion and test a design to reduce it. We study a tree planting program for cocoa farmers in Côte d’Ivoire and find that collusion between farmers and data collectors occurs in 25% of field verification reports. A randomized controlled trial shows that withholding information about the compliance threshold reduces collusion to 11%. These findings highlight that monitoring and traceability alone are insufficient to detect compliance unless reporting accuracy is also ensured. Improving reporting accuracy, while supporting farmers, is key to ensuring both the effectiveness and equity of sustainability standards, including the EU Deforestation Regulation.
Dataset DOI: 10.5061/dryad.v9s4mw77q
Description of the data and file structure
Files and variables
File: cheating_wide_anonymised_v2.csv
Description: CSV anonymized dataset to replicate all analysis presented in the paper. Empty cells identifies missing data
File: codebook.xlsx
Description: contains all metadata required to understand data in cheating_wide_anonymised_v2.csv
File: cheating_wide_anonymised_v2.dta
Description: Stata formatted dataset already including labelling and formatting
File: collusion_analysis_v3.do
Description: contains all script required from the analysis. The first rows contain the code to label and format cheating_wide_anonymised_v2.csv
File: Audit_survey_v7.1_anonymised.xlsx
Description: survey used to administer the experiment
All variables are described in the codebook file
Code/software
All data is in the .do file
Human subjects data
Data referring to human participants are anonymized, informed consent was obtained to publish anonymized data. Anonymization was obtained by removing the observation identifier and shuffling rows in the dataset
