Continuing large-scale global trade and illegal trade of highly hazardous chemicals

The paper addresses the global trade of highly hazardous chemicals and the effectiveness of the Rotterdam Convention’s Prior Informed Consent (PIC) procedure in regulating transboundary movements. Chemical pollution poses severe risks to human health and biodiversity, constituting a planetary crisis alongside climate change and biodiversity loss. While many countries have domestic chemical management frameworks, their effectiveness is limited by international trade and environmental transport. The Rotterdam Convention (in force since 2004; EU and 164 countries as of Feb 2023) aims to facilitate informed import decisions and shared responsibilities but lacks a robust effectiveness evaluation mechanism and only recently established a compliance mechanism (2020). Prior studies using UN Comtrade-derived datasets had limited coverage and did not adequately address erroneous records. This study aims to provide an up-to-date, accurate, and comprehensive assessment of global trade and illegal trade of chemicals listed under the Convention from 2004 to 2019, considering evolving listings and ratifications, and to evaluate the Convention’s impacts and areas for improvement.

The Convention currently relies on parties’ submissions to assess impacts, which are limited. Núñez-Rocha and Martínez-Zarzoso (2019) evaluated the Rotterdam and Stockholm Conventions using the BACI database (1995–2012) but covered only 88 countries and did not remove erroneous/implausible records; mirror reconciliation in BACI does not correct over/underinvoicing errors. Broader context literature notes lack of compliance mechanisms (before 2020) and challenges in global chemical governance, including private governance interactions and the need for comprehensive strategies. UNEP and other reports highlight illegal chemical trade and data quality issues in UN Comtrade (asymmetries, outliers). Comparisons with Basel and Stockholm Conventions indicate stronger obligations and compliance tools in related regimes, and persistent illegal trade (e.g., mercury) implies under-enforcement across treaties.

Scope: Included 46 of 54 chemicals/groups listed under the Rotterdam Convention (as of Feb 2023) that have dedicated Harmonized System (HS) codes; excluded eight without specific HS codes or where HS codes cover broad chemical groups (e.g., 2909.30, 2903.99) that prevent precise attribution. Asbestos: treated crocidolite (own HS code) separately from other varieties under HS 2524.90 (which includes chrysotile not listed during study period). Data sources and period: Retrieved global trade records (imports/exports) from UN Comtrade for 2004–2019, integrating all reporting countries/regions and HS versions (2002, 2007, 2012, 2017, 2019). Excluded re-imports/re-exports (0.058% and 0.064% of global imports/exports). Countries/regions were grouped into 15 analytical regions (e.g., Western/Central/Southern Europe; China, USA, Russia, Canada as separate units) to examine interregional trade. Data treatment workflow: Starting from 79,194 records, conducted a two-step process to address data quality and asymmetries.

1. Error analysis (outlier detection on unit prices): Calculated unit prices (value/quantity) and applied (a) quartile-based statistical outlier detection to identify lower and higher outliers (including reciprocals for high-price outliers), and (b) k-means clustering on unit prices. Compared quartile outliers with cluster boundaries; conservatively flagged records as probably erroneous when outliers fell before/after specified cluster halves (detailed rules in Extended Data Fig. 9). Validated by known implausible entries (e.g., ammonium PFOS export from Kuwait in 2018 ~10.5 kt). Identified 13,038 probably erroneous records (1.2% of original trade volume) concentrated in specific chemicals (alachlor, ethylene dichloride, ethylene oxide, pentachlorophenol and derivatives) and regions (South/Southeast Asia, Central/Western Europe, USA, Middle East). PFOS and its salts, despite cleaning, still showed implausible large intra-Europe flows post-2017; thus excluded from final calculation for that substance.
2. Mirror analysis (reconciliation of bilateral flows): After removing outliers, reconciled import/export pairs: if only one side reported, mirrored to fill the gap; if both reported and unequal, retained the higher value; if equal, retained either. This produced a unified dataset with matched imports/exports (Extended Data Fig. 10). Nearly 10,000 missing records (esp. exports) were filled via mirroring. Final dataset contained 66,156 records. Illegal trade (defaulting) identification: Considered only trades between parties to the Rotterdam Convention and only for listed chemicals with dedicated HS codes. Using the official database of import responses (Article 10), flagged a transaction as defaulting when an export from a party to another party occurred after the importer’s ‘no consent’ decision year. Counted defaulting events per exporter (and counts of being defaulted per importer) as 1 per chemical per occurrence. The USA (a non-party) was excluded from illegal-trade counts; separate descriptive analysis quantified USA exports to parties with ‘no consent’. Compared EU trades reported under the EU PIC Regulation (in force since 2014) against UN Comtrade-based records to identify additional defaulting within the EU context (minimum estimate due to aggregation of origins/destinations in PIC reporting). Analyses: Computed cumulative and annual trade volumes by chemical class (pesticides, industrial chemicals, multi-use), chemical, and region; mapped interregional flows (2015–2019); examined time trends; and summarized defaulting counts and volumes over time and by region.

• Scale of trade: Cumulative trade of 46 listed chemicals reached 64.5 megatonnes (Mt) in 2004–2019. Major contributors: ethylene dichloride (55.3 Mt) and ethylene oxide (6.3 Mt). Pesticides/formulations: 2.4 Mt, dominated by pentachlorophenol (PCP) and its salts/esters. Annual totals rose over time largely due to ethylene dichloride.
• Legacy hazardous chemicals continue in trade: • Tetraethyl and tetramethyl lead: Despite global phase-out of leaded gasoline (finalized 2021), several thousand tonnes were still traded in 2019; 2019 volume ~10× lower than in 2012, with continued uses likely in aviation gasoline, farm equipment, and racing fuels. • Tributyltin compounds (TBTs): Constantly >4.5 kilotonnes (kt)/year traded since 2012 despite bans on anti-fouling paints; likely due to illegal anti-fouling uses and other uses (wood/material preservatives, slimicide). • Tris(2,3-dibromopropyl) phosphate: Averaged ~14 kt/year traded (2007–2019) despite phase-outs from skin-contact textiles, with ongoing uses in various polymers and interior components.
• Asbestos: Crocidolite (listed) declined markedly to ~0.3 kt in 2019 (2% of 2010 level), suggesting listing impact. Other asbestos varieties under HS 2524.90 (including chrysotile, 90–95% of global production) showed ~constant ~1 Mt/year post-2013, underscoring need to list chrysotile.
• Geographical patterns (2015–2019): Most of Asia (Central/Northeast and South/Southeast) were major importers of pesticides and multi-use chemicals; the USA and Middle East were major exporters of these classes. Central/Western Europe were both major importers and exporters of industrial and multi-use chemicals. The USA exported ~158 kt of listed pesticides and ~7 Mt of listed multi-use chemicals (mainly ethylene dichloride) to Asian countries, while importing ~71 kt of pesticides from Central Europe and Latin America and several kt of multi-use chemicals. Germany was a key exporter of TBTs and ethylene oxide.
• Illegal trade (defaulting): Among 48,342 HS-specific party-to-party records, 25,328 were defaulting (illegal) after import ‘no consent’ decisions, accounting for ~40% of total volume (25.7 of 64.5 Mt). Defaulting was prevalent globally, especially in Central/Western/Southern Europe, South/Southeast Asia, and the Middle East. Time series show decreasing defaulted volumes after ~2014 for pesticides and industrial chemicals, but persistently high (>1.5 Mt/year) for multi-use chemicals. Separate analysis (not counted as illegal due to USA non-party status) found ~122 kt of listed pesticides and ~3,989 kt of listed multi-use chemicals were exported from the USA to parties with ‘no consent’.
• Treaty interplay and compliance issues: Despite Stockholm Convention bans, small trades of aldrin, chlordane, heptachlor, and dieldrin were observed post-2004. Endosulfan imports into certain EU member states since 2017 were registered despite no Stockholm exemptions. PCP exception (Stockholm, 2015–2021) showed discrepancies: exports from Israel, Jordan, and USA to Asian countries exceeded Mexico’s registered exception exports to USA by >2×.
• Data treatment outcomes: Identified 13,038 probably erroneous records (1.2% of original volume); filled ~10,000 missing mirrored records. Over 70% of listed chemicals show decreasing trade trends, indicating some positive Convention impact; however, ethylene dichloride and TBTs increased.

The study provides a conservative, minimum estimate of global and illegal trade in highly hazardous chemicals under the Rotterdam Convention by rigorously treating UN Comtrade data and constraining illegal-trade identification to party-to-party trades after explicit ‘no consent’ import responses. Findings indicate that while the Convention correlates with decreasing trends for over 70% of listed chemicals (e.g., crocidolite), significant global trade persists for several hazardous substances, and illegal trade is widespread, representing approximately 40% of recorded trade volume for HS-specific party-to-party transactions. The persistence of trade in tetraethyl/tetramethyl lead, TBTs, and tris(2,3-dibromopropyl) phosphate suggests incomplete phase-out, possible ongoing legal uses, and suspected illegal applications. Defaulting is geographically widespread, with hotspots across Europe, Asia, and the Middle East, reflecting enforcement gaps, data/reporting limitations, and possibly increased reporting over time and new HS code assignments that improve detectability. EU PIC Regulation data show even higher defaulting counts than UN Comtrade-based estimates, indicating stricter reporting in the EU but continued illegal trade. Discrepancies with Stockholm Convention obligations (e.g., trades in banned POPs; PCP exception misalignments) further emphasize enforcement and monitoring deficiencies. Addressing data quality (outliers and asymmetries) improves confidence in trend and compliance assessments and highlights the need for improved global data availability, quality, and accessibility to monitor trade flows, detect smuggling/mislabeling, and evaluate treaty effectiveness. Overall, the Convention’s positive influence is evident but insufficient to prevent substantial ongoing and illegal trade without enhanced compliance mechanisms, swifter listings (e.g., chrysotile, paraquat), and coordinated international, regional, and national actions.

This work delivers a comprehensive, data-treated assessment of global trade and illegal trade in 46 highly hazardous chemicals under the Rotterdam Convention (2004–2019). The Convention has contributed to decreasing trade for the majority of listed substances, yet substantial trade continues in several legacy hazardous chemicals, and illegal trade is pervasive, constituting around 40% of HS-specific party-to-party trade volume identified. The findings underscore the need to: expand Convention membership; strengthen compliance and enforcement (e.g., verifying exports against import responses, penalties for illegal trade); expedite the listing of chemicals recommended but not yet listed (notably chrysotile asbestos and paraquat); and improve trade data quality, reporting, and accessibility for better monitoring and effectiveness evaluations. Future research should quantify illegal trade between parties and non-parties through domestic legislation reviews, investigate smuggling/mislabeling and black markets, analyze sectoral drivers in key exporting/importing regions, and evaluate policy instruments that reduce exports of highly hazardous chemicals and accelerate transitions to safer alternatives.

• Conservative illegal-trade scope: Only party-to-party trades after ‘no consent’ import responses were counted; trades between parties and non-parties (which should also cease upon ‘no consent’) were not included due to lack of globally compiled domestic legislation and reporting.
• Exclusion of clandestine flows: Smuggling, mislabeling, and black-market activities were not captured, likely underestimating illegal trade.
• HS code constraints: Chemicals lacking dedicated HS codes or with HS codes covering broad groups were excluded, impeding full coverage (e.g., some brominated substances; multiple asbestos varieties under HS 2524.90). Illegal-trade analysis excluded chemicals without unique HS codes.
• Data quality and reporting biases: Despite outlier removal and mirror reconciliation, residual errors may remain (e.g., PFOS and salts post-2017). Reporting asymmetries and changes in HS coding over time affect detectability, especially in early years (2004–2006). The mirroring rule (retaining higher of two reports) may overstate some flows but was applied consistently after removing likely erroneous data.
• USA non-party status: Trades from the USA to parties with ‘no consent’ were not counted as illegal under Rotterdam in this analysis, though they may be illegal domestically in importing countries.