Economic interests cloud hazard reductions in the European regulation of substances of very high concern

The study examines whether REACH’s procedures for identifying and regulating substances of very high concern (SVHCs) effectively prioritize hazard reduction over economic interests. Within REACH, the Candidate List is the first step toward the Authorization List, which can impose binding limits or bans on SVHCs. Despite predictions that about 1,500 SVHCs would be addressed, by February 2020 only 303 substances were on the Candidate List and 86 on the Authorization List. The research investigates how toxicological properties (CMR and environmental hazards), economic factors (production/import volumes and breadth across EEA countries), and the availability of scientific knowledge influence inclusion on the Candidate List, thereby assessing whether political economy dynamics bias regulatory outcomes away from the most hazardous substances produced or imported in the EEA.

Prior work indicates that industry and business interests are better represented and more successful in EU environmental policymaking than diffuse-interest groups such as NGOs and local authorities. Lobbying success tends to increase with the number of aligned interest groups, suggesting substances produced in high volumes or across many countries may face stronger opposition to regulation. Historical analyses show policy action often lags behind scientific early warnings and that precaution is inconsistently applied, with consultation processes enabling stakeholders to highlight scientific uncertainty to delay regulation. These insights motivate examining whether economic interests and the extent of scientific evidence shape SVHC listing under REACH.

The authors compared chemicals across multiple lists: all substances registered under REACH (22,425); the Candidate List (CL; 303); the Authorization List (AL; 86); ChemSec’s SIN list (999); and the Swedish Chemicals Agency’s PRIO ‘phase-out’ list (1,938). Data were collected in February 2020. Variables were constructed as follows: (1) Toxicological properties from ECHA’s GHS-based classification and labelling inventory (147,633 registered compounds; 146,597 unique EC numbers). A CMR Score summed the percentages of firms assigning C, M, and R hazard codes, weighting ‘danger’ twice relative to ‘warning’, and normalized to 0–1. An Environmental Score summed acute and chronic aquatic hazard notifications (same signalling word) and was normalized to 0–1; the GHS environmental classes indirectly reflect persistence and bioaccumulation thresholds. (2) Economic variables from REACH registration dossiers: tonnage band (log10 transformed using the upper bound of the reported band; maximum across multiple entries; zero for no production or only intermediate use; missing for confidential tonnages) and number of EEA countries with active or inactive registrants (square-root transformed). Production versus imports could not be disentangled; inactive registrants indicate ceased activity. (3) Knowledge variable: publication rank (0–4) based on counts of occurrences of compound names in toxicology/ecotoxicology journals indexed in PubMed (36 selected international journals, 2000–2019). Ten most common names per compound were taken from PubChem; names were searched in titles, abstracts, keywords, and chemical lists across 194,018 papers, yielding 111,284 papers with 248,907 occurrences for 4,088 unique compounds. Publication rank: 0 (no papers), 1 (1–10), 2 (11–100), 3 (101–1,000), 4 (>1,000). Logistic regressions modeled inclusion on CL (and separately AL) as a function of toxicity scores, economic variables, and knowledge, with all predictors normalized to 0–1. Models were fitted for CL versus REACH, CL versus SIN, and CL versus PRIO, and analogously for AL. Highly correlated variables were removed prior to modeling. Model fit used adjusted McFadden pseudo R². Variable significance was tested against null models. Analyses were conducted in R 3.5.1 with ggplot2, tidyverse, and stringr. Additional regressions replaced composite toxicity scores with individual indicators for carcinogenicity, mutagenicity, reproductive toxicity (danger/warning), and acute/chronic aquatic toxicity to disentangle specific hazard contributions.

• Hazard profiles: Candidate List substances have significantly higher mean CMR and Environmental Scores than the average REACH-registered substances. Compared to CL, SIN substances have higher CMR scores; PRIO substances have higher Environmental Scores. No significant differences in CMR or Environmental Scores between CL and AL. - Economic and knowledge characteristics: Relative to SIN, CL substances are produced in lower volumes and by fewer EEA countries; relative to PRIO, CL substances have higher volumes and more countries. Publication rank is higher for CL substances than for SIN and PRIO, indicating better-studied chemicals are more likely to be listed. No significant differences between CL and AL on economic or knowledge parameters. - Logistic regressions (CL vs REACH): Major positive driver: higher CMR Score. Major negative driver: number of EEA countries producing/importing—a large breadth of production/import strongly reduces odds of CL inclusion. Knowledge availability (publication rank) increases inclusion odds. Quantitatively, at the extremes: CMR Score at maximum vs minimum increases odds by ~184-fold; if all EEA countries produce/import, odds decrease by ~520-fold vs none; lack of publications reduces odds by ~46-fold vs very well-studied substances. Tonnage band had smaller or non-significant effects in this comparison. - Logistic regressions (CL vs SIN): Tonnage band and number of producing/importing countries are key discriminators, with more widely produced/imported and higher-volume SIN substances less likely to appear on CL. CMR properties have a relatively larger positive effect in CL vs PRIO comparisons than vs SIN. - Specific hazard components: Reproductive toxicity is the strongest toxicity predictor of CL inclusion (vs REACH), followed by carcinogenicity (danger) and chronic aquatic toxicity. Compared to SIN and PRIO, carcinogenicity danger appears under-represented in CL, while reproductive toxicity is prioritized. - Authorization List: Patterns mirror CL but are more pronounced. Lack of EEA production/import is the most important determinant of inclusion on AL, with effects roughly 50 times larger than in CL. Of 86 AL substances, 36 had zero active or inactive registrants as of February 2020 despite not yet reaching their sunset dates. Knowledge availability consistently increases odds of inclusion on AL. - Scope observation: Nearly half of CL substances (144/303) were not produced or imported in the EEA, suggesting listing of ‘low-hanging fruit’ with minimal immediate economic impact. - Overlaps: 142 substances are common to REACH, CL, SIN, and PRIO; over 2,000 SIN/PRIO substances are not on CL; 47 CL substances are not on SIN or PRIO; all AL substances are on CL.

The findings reveal a tension between hazard reduction goals and economic interests in REACH’s SVHC listing. While higher CMR and environmental hazards significantly increase the likelihood of inclusion on the Candidate List and reproductive toxicity is emphasized, the strongest determinant of listing is the absence of current or past production/import within the EEA. The breadth of EEA production/import substantially reduces the odds of listing, consistent with political economy expectations that broader, more concentrated economic interests impede regulation. The positive effect of scientific knowledge underscores the enabling role of evidence in advancing regulation, yet many chemicals remain understudied. For the Authorization List, the economic-interest pattern is even stronger, reflecting higher stakes of binding restrictions. Together, the results suggest that listing processes have focused on substances with lower immediate economic repercussions (including many not produced/imported in the EEA), potentially limiting realized hazard reductions and delaying action on widely used hazardous substances.

This study quantifies the relative roles of toxicological hazard, economic footprint, and scientific knowledge in shaping inclusion of chemicals on REACH’s Candidate and Authorization Lists. Although hazard indicators, particularly reproductive toxicity, are important, the most influential factor is the lack of EEA production/import, indicating that political-economic considerations strongly shape listing outcomes. The work highlights the need to bolster the scientific evidence base for understudied chemicals and suggests that reforms may be needed to ensure that hazardousness—rather than limited economic impact—drives SVHC identification and progression to authorization. Future research should examine the interaction between REACH’s Authorization and Restriction programs, causal mechanisms within the political process of listing, and the differential impacts of production versus import on regulatory resistance.

• Causality not established: The logistic models identify associations, not causal effects. - Unobserved political process: Detailed dynamics of stakeholder interactions and negotiations during listing are not measured and may influence outcomes. - Production vs import: The economic data cannot distinguish domestic production from imports; opposition dynamics may differ between them. - Data coverage: Publication-based knowledge proxy is limited to selected toxicology/ecotoxicology journals (2000–2019) and may miss relevant evidence elsewhere. - Confidentiality/missing data: Some tonnage data are confidential or missing; inactive registrants indicate past activity but not timing relative to listing. - Focus on Authorization pathway: The study does not analyze the Restriction program, which also regulates hazardous substances; interactions between programs remain unexamined.