The Precautionary Principle in EU and US Chemicals Policy: A Comparison of Industrial Chemicals Legislation

Abstract

In this chapter, the precautionary principle will be considered as the ­starting point for decision-making on chemicals in cases of scientific uncertainty. The principle will serve as the reference point for an analysis and a comparison of chemicals­ policies and, in particular, of legislation for industrial chemicals in the European Union and the United States of America. In the second section, the precautionary principle will be described on a general level and operationalised with respect to chemicals management. The third section will focus on EU precautionary and chemicals policy and, in particular, on the recently adopted REACH regulation. A similar analysis will be made of US policies in the fourth section, with a focus on the Toxic Substances Control Act. In the fifth and concluding section, the results from the analyses will be compared and discussed with the aim to identify measures that could improve the management of chemicals under uncertainty.

1 Chemicals and Complex Risks

Society would not look the same without man-made chemicals, being of utmost importance in medicine, industry and agriculture and for the daily welfare of citizens (European Commission 2009). At the same time, the production and use of many chemicals are causing severe health and environmental problems, including allergies, cancer and decline of biodiversity (see e.g. EEA 2007; EEA 1998); in fact, 70% of new chemical substances assessed under EU law have at least one property that is dangerous to the environment (European Commission 2003). Adverse effects result from the continued use of well-known hazardous substances and from new chemicals that are introduced without much control, not least ­chemicals found in various articles (i.e. goods). Comprehensive risk assessments have only been carried out for a few of the several tens of thousands of substances in use. Data about intrinsic properties and exposure conditions are lacking for most chemicals on the market (Allanou et al. 1999; Roe et al. 1997); it is even unclear how many substances are in use.¹ Results from toxicological and eco-toxicological studies are difficult to interpret and extrapolate from, and combination effects are seldom studied, meaning that the impact of exposure to mixtures of chemicals is more or less completely unknown (Cairns and Smith 1996). Furthermore, risks that would normally be considered as unacceptable are very difficult to detect for ­statistical reasons, even in well-designed epidemiological studies.² In addition, knowledge and data within the field have more often than not been heavily disputed.

Against this background, it is clear that scientific uncertainty will characterise knowledge about chemicals for the foreseeable future. A central challenge for chemicals management – and the object of this article – is, therefore, to answer the question of how to deal with uncertainty, particularly when the issues are considered controversial by various stakeholders.³

2 Core Elements of the Precautionary Principle

In spite of much debate about the precautionary principle and its more precise ­meaning, it is clear that precautionary decisions have been taken in environmental policy for a long time and that the precautionary principle is widely adopted today in policy and legislation, not least when it comes to chemicals policy (see overviews in e.g. de Sadeleer 2007a; Ashford 2007; Karlsson 2005; Sandin 2004; Lökke 2004; Tickner 2003a; O’Riordan et al. 2001; Applegate 2000). The political and academic criticism of the principle has been shown to be quite weak (Ahteensuu 2007; Gardiner 2006; Sandin et al. 2002). The principle is often described with reference to the Rio Declaration, Principle 15, which states that ‘… lack of full scientific ­certainty shall not be used as a reason for postponing cost-effective measures to prevent environmental degradation’ in cases of threats of serious or irreversible environmental damage. This definition is quite weak, though, since it merely discredits lack of scientific proof as an argument for stalling preventive measures (UNCED 1993).

Much legislation and many binding international environmental agreements stipulate stricter and more action-oriented versions, which Sandin (1999) has summarised in the formula that ‘if there is a threat, which is uncertain, then some kind of action is mandatory’. On the basis of this interpretation, I have previously ­elaborated five core elements of the precautionary principle in relation to chemicals policy (Karlsson 2006).⁴ As a starting point for my operationalisation, I reasoned that ‘a threat’ needs to respond to some level of seriousness, including effects on health or the environment that are severe, spatially or temporally dispersed, irreversible or non-linear. In the same manner, I consider ‘uncertainty’ to require some level of qualification, such as science-based suspicion, well-grounded practical experience or lay knowledge of threats. Obviously, any interpretation must be contextual.

Once the precautionary principle is invoked, the question of what kind of action is called for follows: Which additional measures need to be taken, in addition to traditional risk management? In the following, I will briefly outline the five core elements that I believe answer that question.

Group Classification: Generating more data is of key importance. In a scientific endeavour, though, a strong correlation relating cause to effect is required for making a claim, and scientists would rather miss true links than consider false incidents to be true. In precautionary policies, the opposite is justified, and unknown substances in a chemical group should be classified as the most hazardous known substance in that group, or as the ‘worst-case’, which is reasonably imaginable.

Management Based on Intrinsic Properties: The complex life cycles of many substances prevent proper estimations of exposure. Therefore, precautionary policies primarily focus on intrinsic hazardous properties, i.e. assume that critical exposure is the case. Since hazardousness (e.g. chronic toxicity) is difficult to estimate, also substances classified as ‘persistent’ and ‘liable to bioaccumulate’ should be treated as hazardous and as action targets.

Preventive Measures: Unless voluntary measures are proven sufficient, the precautionary approach requires anticipatory regulation. If a substance can be substituted with a suitable, better known and less hazardous substance or non-chemical option, then that should be the first priority. Secondly, partial or full restrictions can be implemented on substances that are precautionarily classified as being persistent, bioaccumulative, toxic or otherwise hazardous.

Maximin Decision-Makin: In the analysis of whether or not to require a specific preventive measure, it is common (in chemicals policy) to aim for maximised utility based on costs and benefits. When uncertainty about risks caused by a substance prevents that, precautionary decisions – based on reasoning and avoidance of obviously absurd requirements – favor measures that cause the least possible loss if the ‘worst-case’ turns out to be true.⁵

Reversed Burden of Proof: In each of the four core elements operationalised above, the precautionary principle favours placing the so-called ‘burden of proof’ on the operator, which thereby bears the responsibility for convincingly showing that statutes are being followed and, for example, that a regulatory decision on ­classification or a preventive measure is factually illegitimate or obviously ­unreasonable from a likely cost–benefit point of view.

These core elements of precautionary chemicals management will next be used as an analytical framework in relation to chemicals policy in general and legislation on industrial chemicals in particular in the EU and the US.

3 The Precautionary Principle and EU Chemicals Policy

In the following, the history of the precautionary principle within the EU and its member states will be briefly described, followed by a concise overview of EU chemicals policy and a more detailed description and analysis of the REACH regulation and its relation to the precautionary principle.

3.1 The Precautionary Principle in the EU

The precautionary principle is far from being a new component of environmental and health policies in Europe. Some of the core elements of the principle, as elaborated above, can be traced back in Swedish legislation centuries ago (Karlsson 2006), and the idea of reversing the burden of proof in environmental law in the Nordic countries goes back more than four decades (Sand 2000).⁶ The strong standing of precautionary policies in, for instance, Germany and the Netherlands led to the inclusion of the principle in the EC Treaty in conjunction with the Maastricht Treaty (1992), followed by an interpretation by the European Commission (2000) in a so-called ‘Communication’, which was relatively well endorsed by EU heads of states and governments in 2000 (European Council 2000).

In the Communication, the Commission interprets the principle as being applicable­ when a hazard is identified but when a full scientific evaluation of the risk at hand cannot be completed. It requires that precautionary management is proportional, non-discriminatory, consistent with previous measures, based on an examination of potential benefits and costs and subject to review. However, these aspects are stated always to be relevant for risk management and do not help much in interpreting the specifics of the precautionary principle. On the contrary, some elements in the Communication can be seen to oppose common interpretations of the principle, for instances by relating the principle purely to risk management and not to risk assessment and by giving cost–benefit determinations a central role (Hansen et al. 2007; Karlsson 2005). However, the Communication acknowledges that the burden of proof may be reversed in some cases (see also the analysis by Rogers 2003a). Regardless of interpretation, though, it is clear that the precautionary principle nowadays is commonly applied in secondary EU law (see e.g. Krämer 2006) and in the practice of the EC courts (de Sadeleer 2007b), even though it has been claimed that the courts have recently increased demands on scientific justification for regulation (Stokes 2008).

3.2 The Development of Chemicals Policy in the EU

The chemicals policy of the European Union emerged in the 1960s, with a 1967 directive on classification, packaging and labelling of chemical substances, which had the purpose of harmonising market legislation (EEC 1967). Since then, EU chemicals legislation has been substantially developed and broadened, with several amendments to the early directives and new directives on restrictions (EEC 1976; EC 2003a), chemical preparations (EEC 1988), waste (EC 2003b) and specific types of chemicals, such as pharmaceuticals, cosmetics and pesticides.⁷

In 1979, the central classification directive introduced a differentiation between ‘existing’ – as of September 18, 1981 – and ‘new’ substances, the latter being thereafter only possible to list if certain basic data on the properties of the substance were reported. For the 100,106 substances categorised as ‘existing’, no new data were required, but following a 1993 regulation, 141 prioritised substances eventually became targets of risk assessments (EEC 1993). Due to a lack of data, high complexity and controversies, this process did not deliver what it promised for most chemicals (European Commission 1998).⁸ Consequently, in 1998 and 1999, the EU ministers of environment expressed concerns and called for a stricter chemicals policy based on, among other aspects, the precautionary principle (see e.g. Environment Council 1999). Thereafter, the European Commission published a White Paper on a Strategy for a Future Chemicals Policy (European Commission 2001), which also emphasised the precautionary principle (see e.g. Rogers 2003b). The White Paper became highly contested, and the subsequent lengthy legislative process was probably the most controversial in the history of the EU (Fisher 2008; Selin 2007). In 2006, however, the REACH Regulation was finally adopted (EC 2006a). The regulation explicitly refers to the precautionary principle in the Preamble (9, 60), and in particular in the aim, stating that ‘Its provisions are underpinned by the precautionary principle’ (Article 1).⁹

3.3 The REACH Regulation

Before REACH, EU legislation on industrial chemicals was disparate, and different provisions applied to similar types of chemicals. The system did not provide ­sufficient data or sufficient protection for human health and the environment and did not charge companies with the main responsibility for chemicals management. The White Paper formulated seven objectives for the chemicals legislation reform: protection of health and the environment, enhancement of the competitiveness of the EU chemicals industry, a functioning internal market, international integration and conformity, increased non-animal testing and increased transparency (European Commission 2001). Implementation of the precautionary principle was mentioned as one of the cornerstones of the new legislation. The resulting REACH regulation replaces some 40 pieces of legislation and gradually harmonises the regulation of existing and new chemical substances, even though several groups of chemicals, such as pesticides and cosmetics, are still regulated separately. REACH will, thus, end the previous system that gave preferential treatment to existing substances and thereby created disincentives for the development of new substances.

In the following, the four cornerstones of REACH – registration, evaluation, authorisation and restrictions of chemical substances – as well as a number of other important aspects of REACH will be described and analysed.

3.3.1 Registration (Title II)

REACH charges each manufacturer and importer with responsibility to submit a ­registration to the new European Chemicals Agency (ECHA) for each substance, either on its own or in preparations or articles, that is manufactured or imported in a volume of at least 1 t per year and company (Article 6). For articles, registration for new uses is required for quantities over 1 t per producer or importer per year, but only if the substance in question either is intended to be released or meets the criteria for being placed on the candidate list for authorisation (see below) and is present in more than 0.1% by weight (Article 7).¹⁰ With quite a number of statutory exemptions (including many polymers and intermediates, see Annexes IV and V), a substance not registered according to the provisions is not allowed to be manufactured or put on the market in the EU (Article 5). REACH thereby introduces the concept of ‘no data, no market’.

A substance that was previously notified as ‘new’ is automatically considered to be registered under REACH (Article 24). For the group previously called ‘existing substances’, now called ‘phase-in substances‘, data requirements and deadlines for registration primarily depend on the quantity of the substance in question, though to some extent also on intrinsic properties (Articles 10, 12–14, 23; Annexes VII–X).¹¹ The greater the quantity, the sooner the registration deadline and the higher the data requirements. The registration deadline for substances at or above 1,000 t, as well as for substances that, for example, are carcinogenic, mutagenic or toxic to ­reproduction (so-called CMRs) at or above 1 t, is 30 November, 2010, whereas it is 31 May 2013 for 100–1,000 t and 31 May 2018 for the span from 1 to 100 t. The latter deadline is set nearly 20 years after the ministerial call for a new chemicals­ policy in the EU. Above 10 t, a ‘Chemical Safety Report’ with data on a relatively large set of parameters, including data on intrinsic properties, exposure scenarios and risk management measures, is to be included in the registration (Article 14, Annex 1). For substances in quantities of 1–10 t, a ‘technical dossier’ with more basic data (Article 10) is stated to be sufficient. Concerning so-called ‘non-phase-in substances’, i.e. basically those not being produced or marketed before REACH, the registration provisions entered into force on 1 June, 2008 (Article 141).

Compared to pre-REACH legislation on ‘new chemicals’, the data requirements have been lowered for all quantities. However, with regard to existing substances, for which no general data requirements were stipulated in previous law, REACH provides completely new requirements over time.¹² Furthermore, data requirements have been completely abolished for quantities below 1 t, thereby leaving the clear majority of all industrial chemicals outside of REACH registration (the previous notification limit was 10 kg). Also in the span 1–10 t, data requirements are very low.

The registration phase as such does not necessarily lead to reduced risks, and the quality of the data generated is not guaranteed. The ECHA is merely required to make a completeness check within three weeks (Article 20), after which the chemical in question can be used.

3.3.2 Evaluation (Title VI)

It is first in the evaluation phase that dossiers and substances are qualitatively investigated. In the ‘dossier evaluation’ part (Chapter 1), the ECHA checks how a small portion of the dossiers and reports received complies with the registration requirements, for instance concerning data and proposed risk management measures (Article 41). The ECHA also examines testing proposals, not least in order to prevent unnecessary animal testing, and decides on possible further measures (Article 40). Evaluation is not a prerequisite for starting manufacturing of substances, not even in cases where there are obvious needs for further testing. The registration and evaluation in REACH thus operate along quite independent tracks (EDF 2007a), as was the case under previous EU chemicals legislation. Most substances will not be evaluated at all – the ECHA is not obliged to carry out compliance checks on more than 5% of the registration dossiers for each tonnage band (Article 41).

The second form of evaluation involves substances that for one reason or another are considered or assumed to be problematic (Chapter 2). Both the ECHA and the Member State’s Competent Authorities will continuously evaluate suspected ­substances, on the basis of a risk-oriented rolling action plan and criteria for ­prioritisation developed by the ECHA and member states (Article 44). If needed, further information can be requested from registrants (Article 46). If an evaluation indicates that further risk management measures are needed, it may result in the application of the authorisation or restriction procedures under REACH, as well as measures under other laws.

3.3.3 Authorisation (Title VII)

Authorisation is focused on so-called ‘substances of very high concern’ (SVHCs), which is a new element as compared to previous chemicals legislation. These include substances that meet the criteria for being classified¹³ as carcinogenic, mutagenic or toxic for reproduction (CMRs) in categories 1 and 2, substances that are persistent, bioaccumulative and toxic (PBT substances, according to Annex XIII) or very persistent and very bioaccumulative (vPvBs, as defined in Annex XIII) and other substances that cause equivalent concern, based on scientific evidence of probable serious adverse effects, for instance endocrine disrupters (Article 57). The Commission or member states may identify substances meeting these criteria, and the Agency is to – after a process that may involve a Member State Committee, interested parties and the Commission – establish a ‘candidate list’ for substances that may be included in Annex XIV (Article 59). After taking the opinion of the Member State Committee and the viewpoints of interested parties into account, the ECHA is to recommend, in line with its capacity, to the Commission a set of ‘priority substances’ to be included in Annex XIV, first in line being substances with PBT or vPvB properties, or with wide dispersive use or high volumes (Article 58).

The final decision rests with the Commission but is to be taken after Committee procedure (Article 133), commonly known as comitology, in this case with the ‘Regulatory procedure with scrutiny’ (EC 2006b), which basically means that the Commission has the unique power to draft proposals but that a member state committee, the Council or the European Parliament, given certain conditions, can block the proposals. In summary, these processes mean that there is no guarantee at all that a substance, which definitely meets the criteria in Article 57, will be listed.¹⁴ In practice, the listing may depend on resources and viewpoints among member states. At present, the candidate list contains 15 substances, of which 7 have been placed on ECHA’s first priority list (ECHA 2009), as compared with the 1,400 substances mentioned for authorisation in the White Paper (European Commission 2001).

Once a substance is included in Annex XIV and the so-called ‘sunset date’ has passed (Article 58),¹⁵ authorisation must be sought by any manufacturer, importer or downstream user wanting to use the substance or to place it on the market on its own, in preparations (above certain concentrations) or for incorporation into articles (Article 56). Articles containing an Annex XIV substance may be imported to the EU without authorisation.¹⁶ Authorisation is to be granted by the Commission on a case-by-case basis for a specified period and under conditions subject to review, if the risks to health and the environment arising from the listed intrinsic properties of a substance are ‘adequately controlled’ (Article 60, as defined in Annex I). However, the adequate control route is closed for a SVHC that has the properties PBT or vPvB or that is a CMR for which a threshold cannot be determined (Article 60). For such substances, or when control is not adequate, authorisation requires that socioeconomic benefits outweigh risks and that there are no substitutes available (Article 64). These aspects must be considered by a Committee for Risk Assessment and a Committee for Socioeconomic Analysis before the Commission can take a decision on an application, which must include a socio-economic analysis according to Annex XVI.

Concerning substitution, which is mentioned both in general and in relation to authorisations in the REACH Preamble (70, 72–75), a substitution plan is to be included in the application, but only if the applicant identifies a safer alternative (Article 62). There are options for third parties to present alternatives, but even if the ECHA committee considers them suitable, authorisation must still be granted if adequate control is considered to be in place (Article 60). In general, the burden of proof for substitution rests on the regulators, and substitution requirements are more or less restricted to some parts of the authorisation phase (Koch and Ashford 2006).

3.3.4 Restrictions (Title VIII)

The fourth cornerstone in REACH is the possibility of issuing restrictions. Restrictions apply to Community-wide situations when the production, market release or use of a substance (on its own, in preparations or in articles) entails an ‘unacceptable risk’ to the environment or to human health (Article 67–68).¹⁷ ‘Unacceptable risk’ is not explicitly defined, but the provisions state that any restriction ‘decision shall take into account socio-economic impact, including the availability of alternatives’ (Article 68). In contrast to the authorisation title, under which the existence of alternatives might block authorisation, alternatives play the opposite role here, meaning that without alternatives, restrictions may be blocked.

The restriction procedure is complex and includes reviews by the Committee for Risk Assessment (during nine months) and the Committee for Socioeconomic Analysis (during 1 year), the publishing of draft recommendations and decisions and, eventually, a final decision by the Commission via comitology¹⁸ (Articles 69–73). After the decision, the substance and the restrictive conditions placed on its use are included in Annex XVII. For a CMR substance on its own, in mixtures or in other articles, which could be used by consumers, the process is simplified, and the Commission can take action much more easily (Article 68).

The restriction option has been considered to be a ‘safety net’ (European Commission 2007) if other provisions in REACH or action under other laws are deemed to be insufficient. However, the Commission, the ECHA or member states are charged with the responsibility for preparing a restriction decision and must file a dossier clarifying the motives and the most appropriate risk reduction measures, after which a complicated decision-making process follows. The principal aspects here are more or less the same as for substances and decisions under the previous EC chemicals legislation.¹⁹

3.3.5 Other Central Elements of REACH

In addition to the four cornerstones described above, REACH introduces a number of other relatively novel elements. Corresponding to central topics in the regulatory debate, several provisions in REACH promote data sharing during the registration phase, which is of importance for lowering industry’s compliance costs and decreasing animal testing as far as possible (Title III).

Furthermore, REACH includes new or increased demands on the bidirectional flow of data and information in the supply chain (Title IV) and places demands on downstream users (Title V). For instance, REACH includes provisions on safety data sheets for substances on the candidate list (Article 31 and Annex II) and forces suppliers of articles to actively provide information for the safe use of the articles (Article 33). In addition, REACH entitles consumers to, without charge, request information within 45 days on the safe use of articles containing SVHCs in concentrations above 0.1 wt% (Article 33). It remains to be seen to what extent the increased flows of information will impact on the management of chemicals. However, companies that work with environmental management systems and companies located closer to consumers in commodity chains will probably seek more actively to decrease chemical-related risks.

When it comes to Confidential Business Information (CBI), REACH (Title XII) contains several provisions granting rights to companies, including protection of information regarding the full composition of preparations, the precise use and tonnage of substances and links between companies (Article 118).²⁰ On the other hand, Article 119 explicitly lists information that ‘shall be made publicly available’ by electronic means, for instance names and classifications of substances and safety-related data, unless the submitter can justify not doing so. Additional information from companies can also be made available upon request (see further in EDF 2007a).

3.4 REACH and the Precautionary Principle

Some, though not all, parts of REACH have just entered into force. A detailed evaluation of the implementation and the impact in practice can clearly not be made at this point. However, there are already signs indicating that the administrative challenges connected to registration may have been underestimated and that the potential impact of authorisation may have been overestimated. For instance, there were seven substances on ECHA’s first priority list in the authorisation phase – extremely few compared to earlier expectations. However, the following analysis will focus on the REACH system and the legal text, given previous experiences of EU chemicals law, in relation to the precautionary principle as interpreted above.

The first core element, ‘group or worst-case classification’ of unknown substances, could be implemented by interpreting uncertainty as if adverse effects exist until the opposite has been reasonably proven or by classifying on the basis of analyses of similarities between substances, for instance by using structural activity relationships. REACH does not incorporate such a classification concept but acknowledges the use of alternative testing methods.²¹ However, REACH applies the idea of ‘no data, no market’, which is a form of worst-case classification in the sense that a non-registered and, therefore, more or less unknown substance is considered so problematic so as not to be permitted at all. In this way, and by placing the burden of proof on the company seeking registration, REACH provides strong incentives for generating data. However, substances in low quantities and those included in many articles are not included, data requirements are often limited and periods for transition to REACH are often long. Consequently, the registration phase under REACH can hardly be regarded as precautionary.

The second core element, that policies can be based on intrinsic properties, is clearly recognised in REACH. This is most visible in the fundamental requirement for authorisation for substances being vPvB, i.e. substances not necessarily being even suspected to be toxic or otherwise problematic for health or the environment. However, the complex processes for identifying and listing SVHCs for authorisation – combined with exemptions, the risk of very late deadlines and budget restrictions among regulators – may lead to similar implementation problems with REACH as with previous EU laws on risk assessment of existing substances (see also de Sadeleer 2007c). It is difficult to see, for instance, why member states that might wish to stall stricter policies would not be able to do so in the future as well. Furthermore, REACH allows authorisation of even the most troublesome chemicals if adequate control is assumed, albeit such control does not necessarily prevent exposure to humans and the environment. Nevertheless, the burden of proof in the authorisation phase basically rests with chemical companies, which is a more or less unique precautionary measure in an international context.

Turning to preventive measures, the third core element of precaution, REACH offers a multitude of regulatory action possibilities. However, substitution requirements – a cornerstone in precautionary policies – will be the exception rather than the rule under REACH.²² To be sure, possibilities for substitution will block the authorisation of some SVHCs, and some REACH elements promote substitution, in particular the increased flow of and access to information and transparency (Lahl 2007). Nonetheless, for most substances falling under REACH, no substitution requirements are provided at all. Some of the substances that will be authorised may comprise exceptions, however this is quite late in the regulatory chain (Koch and Ashford 2006), and even some of the most hazardous among these may be authorised in spite of existing substitutes, as long as adequately control is assumed. In addition, REACH provides no incentives for a company that applies for authorisation to consider seriously alternatives to what it actually wants to do. Finally, regarding restrictions, the burden of proof still rests with public agencies, with all the well-known implementation problems that are connected to a system of that kind (de Sadeleer 2007c). Thus, on this point, REACH will probably not be much more precautionary than previous legislation.

The fourth core element, the maximin principle, could partly be said to relate to the ‘no data, no market-element’, since this in principle prevents the marketing of a chemical rather than the opposite in case of uncertainty. However, this point refers to data gathering activities and not to decision-making on the basis of uncertain data. On the latter point, no provision in REACH explicitly recognises or implements the maximin principle, not even under the restriction title. On the contrary, traditional cost–benefit analysis plays a central role in the authorisation and restriction procedures.

Finally, REACH clearly reverses the burden of proof when it comes to registration and, to some extent, even during the authorisation phase, even though responsibility for placing a substance on the candidate list rests within the public domain. Concerning evaluation and restrictions, the burden of proof rests with agencies. This challenges the Commission, the ECHA and competent authorities of member states with heavy obligations, and there is an obvious risk for administrative overload, as under previous legislation.

In summary, REACH implements some of the core elements of the precautionary principle but only goes half way at best.

4 The Precautionary Principle in US Chemicals Policy

This section provides an overview of the precautionary principle and chemicals regulation in US policies and describes and analyses the Toxic Substances Control Act in more detail, in particular in relation to the precautionary principle.

4.1 The Precautionary Principle in US Policies

Even though the precautionary principle is not explicitly mentioned in US federal law, fundamental aspects of the principle, such as taking anticipatory action under uncertainty, were reflected in much of the early US environmental legislation, including the 1970 Clean Air Act, the 1972 Clean Water Act and the 1973 Endangered Species Act²³ (Ashford 2007; Applegate 2000). Under the Clean Air Act, the concept ‘precautionary’ was, for instance, explicitly mentioned by courts in cases as early as in 1976 and 1979.²⁴ Concerning US legislation on chemicals substances, the initial ambition was to diverge from the otherwise typical US practice of taking action within the tort system first when damage has occurred and cause–effect relations are more or less proven. By establishing elements of precaution in law, legislators hoped to replace the reactive tort system with one that focused on prevention in the field of health and environment protection (Ashford 2007; Applegate 2000; Wagner 2000).

Over time, though, these elements have been increasingly countered both by amendments of statutes and by the development of other principles and ideas, such as risk acceptability and cost–benefit balancing, and the burden of proof is nearly always placed squarely on the regulator (Applegate 2000). In the field of chemicals policies, protection has been declining since the 1980s due to legislation on use cost–benefit analysis, review activities by the Office of Management and Budget, congressional review and replacement of members of the judiciary (Ashford 2007). In the following, the US legislation on chemicals in general and industrial chemicals in particular will be described and analysed.

4.2 History of Chemicals Law in the US

Chemicals policy in the US goes back quite some time. Today the laws are dispersed over several pieces of legislation. The ‘Consumer Product Safety Act’ regulates the general and chemical safety of thousands of common products, and the related ‘Consumer Product Safety Improvement Act’ of 2008 sets standards for children’s products, including a ban on six phthalates. The ‘Federal Hazardous Substances Act’ requires labelling and authorises the Consumer Product Safety Commission to issue prohibitions when labelling is not considered sufficient for hazardous household products. Under the 1938 ‘Federal Food, Drug and Cosmetics Act’, the US Food and Drug Administration assesses notifications of new substances in materials that come into contact with food, and the law holds industry responsible for the safety of ingredients in cosmetic products. The so-called 1958 ’Delaney Clause’ prohibited any carcinogenic additive in food, regardless of its potency, and was an early example of the application of a precautionary measure, even though the implications of the clause have been downplayed since then.

However, the cornerstone of US chemicals legislation is the Toxic Substances Control Act, which regulates most substances and most of the quantities, in particular industrial chemicals. In the following, the statutes and the more than 30 years of implementation of TSCA will be described and analysed. The investigation is limited to statutes with a general reach; the parts of TSCA that address specific substance groups will not be dealt with.²⁵

4.3 The Toxic Substances Control Act

TSCA was enacted in 1976 and has basically not changed since then. The most important provisions in TSCA are Sections 4, 5, and 6 on testing, premanufacturing clearance and regulation of hazardous substances, respectively. TSCA differentiates between existing and new chemicals (Section 8(b)),²⁶ and it assigns authority to the US Environmental Protection Agency (EPA) to implement the various statutes. TSCA was partly based on the ‘Report on Toxic Substances’ by the US Council on Environmental Quality (CEQ 1971), which stated that lack of data and lack of government control caused problems needed to be managed by regulatory means. CEQ considered previous media-based environmental legislation to be insufficient and wanted a comprehensive life cycle-based approach, focusing on chemicals as such instead of on their presence in different environmental compartments. This view was supported by the Congress, which considered proactive policies to be both safer and cheaper than reactive ones, including the tort system (Applegate 2008). Congress wanted to shift the burden of proof, including placing the responsibility for generating data on manufacturers (see e.g. Section 2(b)(1)), and several legal techniques aimed at providing for this.

4.3.1 Testing of Chemical Substances and Mixtures (Section 4)

According to Section 4(a) of TSCA, the EPA is obliged to require testing of existing or new substances if necessary for filling data gaps in order to be able reasonably to determine if a substance or mixture ‘may present an unreasonable risk of injury to health or the environment [emphasis added]’ or if it will be produced in substantial quantities and may reasonably either enter the environment in such quantities or may expose people in substantial or significant quantities.²⁷ According to the EPA, a Section 4 process can take between 2 and 10 years, and testing has been required for some 200 substances since 1979 (OPPT 2008). There is an ­obvious Catch 22 here: the EPA needs data in order to determine ‘unreasonable risk’ but cannot require this data until such risks have been more or less demonstrated. Even the requirement of demonstrating that there is ‘insufficient data’ (Section 4(a)) can be extremely burdensome in practice EDF (2007a).

For these and other reasons – such as budgetary constraints and political priorities – the EPA is often forced to work on the basis of voluntary agreements with industry and other stakeholders. Perhaps the most well-known example is the ‘High Production Volume Challenge Program’ from 1998, initiated by the EPA, the Environmental Defense Fund, the American Petroleum Institute and the American Chemistry Council. The program focuses on ‘sponsor’ companies agreeing to generate basic hazard data for chemicals produced or imported at a volume of 1 million pounds²⁸ or more per year. Nearly 1,400 chemicals are in the program, with an additional 860 provided by coordinated international efforts (OPPT 2008). The program has been successful in the sense that industry actually accepted the challenge and that much previously unpublished data has been made public. However, obvious delays, incomplete submission of data and low quality data in the program as such (EDF 2007b) as well as non-inclusion of chemicals at lower volumes (GAO 2007) illustrate clear shortcomings in the voluntary approach, in particular if agencies do not have adequate resources.²⁹ It is not even certain that the program will yield data with the quality needed for the EPA to be able to determine if risks are unreasonable or not.

Simultaneously with aiming to finalise the HPV Challenge Program up to 2010, the EPA has initiated a new voluntary program, the Chemical Assessment and Management Program (ChAMP), launched in 2008 in response to the 2007 US–Canadian–Mexico Security and Prosperity Partnership. ChAMP focuses on data on both hazards and risks from some 6,750 chemicals produced or imported at volumes of 25,000 pounds or more per year and can obviously not be evaluated in detail, even though there is a lot to indicate that similar problems may occur within this program as with the HPVC program (EDF 2008).

Yet another initiative is the ‘Voluntary Children’s Chemical Evaluation Program’, which has been running since 2000 as a pilot project and centres on 20 chemicals that children are exposed to.³⁰ In a recent evaluation, the program was commended by industry and criticized by civil society (GAO 2007).

4.3.2 Manufacturing and Processing Notices (Section 5)

Section 5(a) requires premanufacture notification (PMN) to be submitted by an operator at least 90 days before a new substance is produced or imported or before an existing substance is used in what has been determined by an EPA rule to be a ‘significant new use’.³¹ TSCA requires data on, for instance, use, exposure and ­hazards, but only information that is already available, i.e. testing is not required even if key data are missing. The EPA states that 67% of PMNs include no test data and that 85% include no health data (OPPT 2008). The EPA, therefore, uses a number of other assessment options and screening tools, including models for structural activity relationships and voluntary agreements on testing (OPPT 2008). In addition, the EPA has published criteria for the identification of persistent and bioaccumulative substances, which can be used for further action under Section 5 (EDF 2007a).

If a substance ‘may present an unreasonable risk [emphasis added]’ or will be produced in substantial quantities that may reasonably enter the environment or may cause significant or substantial human exposure, then the EPA, pending development of more information needed for evaluating effects, may limit or prohibit the use of the substance (Section 5(e)). If there is a reasonable basis to conclude that use of a substance ‘will present an unreasonable risk [emphasis added]’, the EPA ‘shall’ take action with one or another form of restriction (Section 5(f)). Up until September 2006, this did not happen more than four times (OPPT 2008). Furthermore, TSCA stipulates that the EPA in some cases must publish a statement if action is not taken (Section 5(g)).

These PMN data, combined with regulatory opportunities, enable the EPA to work more effectively than under Section 4. The EPA can send signals on both avoidance of hazardous substances and guidance towards safer chemicals by issuing lists on categories of chemicals of concern and by informal communication and negotiation with submitters (Lowell 2003). Companies have withdrawn PMNs for some 1,700 chemicals once the EPA has indicated that it wanted to go further, and for over 1,300 chemicals, the EPA has placed demands on workplace controls (GAO 2007). However, less than 5% of PMNs go through a full risk assessment (EDF 2007a), and Applegate (2000) refers to the fact that the EPA took no formal action on 98% of PMN fillings in 1995. In addition, the ‘new substances’ correspond to only about 1% of the market.

4.3.3 Regulation of Hazardous Chemical Substances and Mixtures (Section 6)

With regard to existing substances, Section 6 authorises the EPA to regulate single chemicals in the ‘least burdensome’ way by nearly any type of measure in the life cycle, for instance by manufacturing restrictions, labelling or use requirements and regulations on concentration levels. The preconditions are that ‘there is reasonable basis to conclude’ that a substance or a mixture ‘presents or will present an unreasonable risk of injury to health or the environment [emphasis added]’. The level of evidence required for regulatory action is clearly higher here than under Sections 4 and 5.

The central concept of ‘unreasonable risk’ is not defined in TSCA, but guidance can be found in the provision that the EPA can promulgate rules to protect against unreasonable risk ‘after consideration of the effect on the national economy, small business, technological innovation, the environment, and public health’ (Section 6(c)(1)(D)). For the sake of consistency, these factors should be included in the weighing of ‘unreasonable’ under other provisions in TSCA as well, in particular considering that the Congress stated that ‘a determination that a risk … is unreasonable involves balancing the [risk] against the effect of … regulatory action on the availability to society of the benefits of the substance … taking into account the availability of substitutes … and other adverse effects which such proposed action may have on society.’ (H. R. 1976). The EPA (1994) itself stated, in a proposed rule concerning lead fishing sinkers, that ‘The unreasonable risk finding can be characterised as a judgment that the risk of health or environmental injury from the substance/mixture outweighs the burden to society of potential regulations’.

These requirements probably comprise key explanations as to why TSCA has not delivered the protection that was anticipated (Denison 2009; Applegate 2008; Ashford 2007). Since its enactment, only a handful of substances have been banned or restricted under Section 6 (OPPT 2008): certain CFCs (today superseded by air legislation), PCBs (statutory), dioxin in a specific case (superseded), certain uses of metalworking fluids (ban in place), asbestos (basically overturned) and hexavalent chromium (ban in place).

Furthermore, under judicial review, a court ‘shall hold unlawful and set aside such rule if the court finds that the rule is not supported by substantial evidence in the rulemaking record [emphasis added]’ (Section 19(c)), which causes problems in practice for the EPA, for example when attempting to prove that a requirement is the ‘least burdensome’. This became evident after the EPA in 1989 had nearly completely banned asbestos, and some companies filed suit against the EPA by claiming, for instance, that there was a lack of substantial evidence on unreasonable risks. Supporting this claim by ruling that the EPA had not considered all of the evidence, not analysed the least burdensome option, not done a cost-efit analysis for all regulatory options, a US Court of Appeal returned most of the EPA rule for reconsideration (GAO 2007; Applegate 2000).³² Considering the vast knowledge of the problems caused by asbestos and the fact that the EPA had worked on preparing the rule for about a decade, the case clearly illustrates the high level of evidence required. Since then, the EPA has not used Section 6 for restrictions (OPPT 2008).

On the other hand, a number of EPA-initiated voluntary agreements and programs have led to well-needed risk management measures in some important cases, for instance regarding fluorinated substances such as PFOA, detergents and certain brominated flame retardants (see further in EDF 2007a).

4.3.4 Reporting and Retention of Information (Section 8)

Section 8 regulates the division between ‘new’ and existing substances, the latter initially being all chemicals on the market before December 1979 and listed (according to Section 8(b)) in the ‘TSCA Chemical Substance Inventory’. The approximately 62,000 substances that by then were included in the inventory were considered safe from a legal point of view, and they can still be used as long as restrictions under Section 6 are not adopted. This so-called ‘grandfathering’ means that most chemicals, at almost all quantities, are practically unregulated. Since the first list, nearly 21,000 ‘new’ substances have been added in accordance with the PMN process, and the inventory, thus, encompassed nearly 83,000 substances in 2006 (OPPT 2008). As stated by the EPA, companies are requested to keep records of and report to the EPA on a number of basic chemical data (Section 8(a)) regularly included in the inventory.

Finally, a chemical company that obtains information that ‘reasonably supports the conclusion’ of ‘substantial risk of injury to human health or the environment’ caused by a chemical must inform the EPA as soon as possible (Article 8(e)). This function was intended as an early warning system, however as a consequence of a paucity of results, the EPA sent a letter to industry in 1990 about ‘an amnesty program’ for 3 years, giving companies a chance to submit any data they had failed to report, which many companies then did in the coming years (Lowell 2003).

4.3.5 Other Central Elements in TSCA

In addition to the provisions mentioned above, TSCA regulates a number of other issues. For instance, if a risk can be managed satisfactorily by other federal laws, TSCA prevents the EPA from taking action (Section 9(a)), unless it is in the public interest to regulate for protection (Section 9(b)). However, transfers of issues to other laws have been limited.

Section 14 in TSCA prevents information reported to the EPA from being disclosed to the general public if it is considered to involve trade secrets of different kinds; even the identities of companies and chemicals can be kept secret. Exceptions involve information concerning health and safety studies and information that the EPA considers necessary to disclose in order to protect public health or the environment (EDF 2007a). Taking 1990 as an example, chemical identification was claimed to involve confidential business information (CBI) in around 90% of PMNs, but it often turns out that the rate drops quite significantly after the PMN process has been completed or in the rare cases when the EPA challenges a CBI claim (OPPT 2008; EDF 2007a).

4.4 TSCA and the Precautionary Principle

As Applegate (2008) writes, the intentions behind TSCA were to put a legal system in place that was preventive and that placed responsibility for generating information on chemical companies. Despite ‘pockets of precaution’ (Wagner 2000), though, TSCA has not achieved this result. The separation between existing and new chemicals, a substance by substance approach, the omission of certain articles and the strong burden of proof placed on the EPA have led to a situation with an obvious lack of data and severe difficulties in imposing safety measures (Denison 2009; Sachs 2009; Applegate 2008; Ditz 2007; Ashford 2007; GAO 2005).

The failure is most obvious in relation to existing substances, with very little action taken. For new chemicals, the EPA has been somewhat more successful, but this category includes only a minority of the chemicals on the market. As a consequence, the EPA has often considered it necessary to enter into enforceable consentual agreements with industry that are far from being sufficiently effective.

Looking at the first core element of the precautionary principle – ‘group or worst-case classification’ – TSCA can hardly be seen to be stimulating the generation of data. Instead, it rewards the maintenance of unknown chemicals by producing ‘incentives for scientific ignorance’ (Wagner 2000). Besides the fact that the EPA sometimes uses, for instance, structural activity relationships when screening substances, neither policy nor law contains elements of group or worst-case classification. The burden of proof for making a claim about a specific property of a substance rests very strongly on the EPA.

The second precautionary element, basing policies on intrinsic properties, hardly plays any central role in TSCA. On the contrary, the provisions generally place a strong focus on the risk concept, so not even obvious toxicity is necessarily a factor leading to further regulatory action. Persistence and bioaccumulation have an even weaker position in the system and do not automatically play any legal role, even though the EPA uses such parameters, for instance in the screening of PMNs.

When it comes to the third element, preventive measures in the form of substitution or various types of restrictions are very difficult to implement under TSCA. The EPA is not completely free to take action (under e.g. Section 6), even when risks have been more or less completely proven, since a number of other provisions must be followed and since the burden of proof rests completely on the EPA. This is clearly illustrated by the passiveness of the EPA in relation to Section 6 action after the asbestos court case.

The fourth core element, the maximin principle, is not implemented at all. On the contrary, there is a strong reliance in TSCA and in the implementation of the act on formal risk assessments and cost–benefit analysis, which more or less completely blocks complementary approaches, such as decision-making aiming at reducing the likelihood of the worst case becoming a reality.

Finally, summarising the burden of proof in TSCA, it basically always rests on the EPA. Not even if the concept is divided into ‘the burden of persuasion’ and ‘the burden of proving a causal link’ (see further in Ashford 2007) does any component fall on chemical companies.

In summary, with few exceptions, TSCA is not at all in line with the core elements of the precautionary principle.

5 Discussion

The problems with chemicals management have in many ways been the same in most countries, including lack of data on chemicals, distinctions between new and existing chemicals, unconsolidated legislation, ineffective processes for risk assessment, burden of proof being placed on governments, limited prevention and weak incentives for substitution. Consequently, costs and administrative burdens have been high, and environmental and health effects have been severe – facts that are recognised by most parties today (European Commission 2009). Neglect of early warnings has led to several serious miscalculations during the past century (Harremoës et al. 2001).

It is quite clear that traditional risk assessment and risk management approaches are not working sufficiently well in the field of chemicals policy, in particular not in cases of high uncertainty. The traditional approach could hardly deal with the early chemical problems, characterised by evident impacts such as acute effects, and is even less effective in the present situation, with globalised flows of articles that contain hazardous chemicals and the resulting complex chemical cocktail, which may cause diffuse but significant adverse effects on human health and the environment.

The Toxic Substance Control Act is a nearly perfect realisation of the old approaches and is clearly outdated. The EPA has been somewhat successful with stimulating voluntary action, but that is far from sufficient, judging from the results in relation to expectations. Previous EU chemicals legislation was permeated by many of the same problems, and it is obvious that the designers of REACH have learned important lessons from the past.³³

Comparing generation and interpretation of data, REACH potentially³⁴ requires between 22 and 54 tests, depending on the quantity, whereas TSCA requires zero tests for existing substances and potentially 14 – voluntary – tests for higher volumes of new substances (GAO 2007).³⁵ REACH no doubt requires more data and clearly places the responsibility for providing the data on chemicals companies and interprets lack of data as if a chemical is not desirable. Furthermore, data generated, even with respect to intrinsic properties, are more often used for further regulatory action under REACH than under TSCA, for example in the authorisation phase. REACH, however, suffers from the traditional approach in the restriction phase, though not as much as TSCA.

If TSCA is more protective than REACH on any single point at all, it is in relation to the evaluation of new substances, which in the US regularly takes place in a screening process before marketing, whereas marketing in the EU is more or less free once a registration is finalised, i.e. before the evaluation phase starts. TSCA is more protective on this point, but that is the exception that confirms the rule.

At present, the focus in the EU when it comes to chemicals law is on implementation of REACH. If there is political room for improvement, it should in my view focus on the following aspects³⁶:

• Fewer exemptions from REACH of specific categories of chemicals that are not regulated with the same degree of protection as REACH would require.

• Inclusion of chemicals in articles in a more comprehensive manner in REACH.

• Duty to register substances in lower quantities than 1 t per company and year.

• Increased data requirements for registration, in relation to all quantities.

• Extended time periods between registration and market introduction, enabling improved fast screening evaluation, possibly leading to further public management.

• Strict demands on general and early substitution; substitutes should be identified for substances in high quantities or with dangerous properties already in the registration phase.

• Fewer bottlenecks, lower barriers and decisive time limits for the process of identifying substances for the authorisation procedure.

• No authorisation when substitutes exist; statutory deadline on maximum time-limit for review of authorisation.

• Automatic phasing-out over time of CMRs, PTBs, vPvBs and chemicals with other hazardous intrinsic properties.

• Substantially lower burden of proof for public agencies when it comes to decisions on restrictions.

• Increased transparency regarding data provided by industry and agencies.

Whether or not the REACH statutes will change and develop in the coming years is, of course, difficult to say. However, it seems likely that continued administrative problems could reignite the debate, in particular once a new European Parliament and a new European Commission are in place during 2009.

Turning to the US situation, several agencies and scholars have for a long time pointed out shortcomings in TSCA³⁷ without any results. However, it seems plausible that US chemicals legislation could be amended in a fundamental manner in the coming years, due to both the new political landscape and the impact of REACH.³⁸ Considering the strong transatlantic trade relations, which includes exports from the US to the EU of chemical substances on their own and in various articles like electronics (Ackerman et al. 2006), REACH will clearly affect US companies that must comply with provisions concerning imports, thereby paving the way for regulatory reforms in the US (Scott 2009; Sachs 2009; Wirth 2007). At present, the debate over US chemicals policy is underway, and interest in REACH is quite apparent.³⁹

On the federal level, a number of legislative initiatives are up for discussion in the US Congress, some being more inspired by REACH than others.⁴⁰ In May 2008, two identical bills called the ‘Kids-Safe Chemicals Act’ were sent to the Senate and to the House of Representatives.⁴¹ The proposals are inspired by REACH and include a proposal for an additional part under TSCA, ‘Title V – Child Safe Chemicals’ as well as proposals for a number of amendments on existing sections in TSCA. The aim is to ‘reduce exposure of children, workers and consumers to toxic substances’, and the main elements include safety statements for manufacturers (i.e. partly reversed burden of proof), explicit safety standards and priority lists (including for prenatal exposure) and provisions in the direction of ‘no data, no market’, substitution and improved access to information (H. R. 2008; S 2008; see also Scott 2009 and Sachs 2009). A similar development is taking place on the state level, for instance in California, Maine and Washington (Scott 2009; Sachs 2009; Ditz 2007).

It is worth noting here that the initial main arguments against REACH – allegedly high costs and lower competitiveness – could not be successfully substantiated (see e.g. Sachs 2009; Selin 2007; Karlsson 2006) and would probably be even harder to make in the US since the development in the EU, a key export market, has already taken place. With REACH-inspired amendments in TSCA, a more level transatlantic playing field could develop, which would, in turn, have an impact on international chemicals law and chemicals legislation in other countries as well (Sachs 2009; Fisher 2008; Park et al. 2008).⁴²

When it comes to developing US legislation in line with the precautionary principle, regulators must ask whether the best option would not be to toss TSCA into the garbage bin, without recycling the statutes. Developing a legal system based fundamentally on the core elements of the precautionary principle would be preferable, in my view. Proposals pointing in that direction have been developed by, for instance, the Swedish Committee on New Guidelines on Chemicals Policy (CNC 2000).

However, if the choice is made to develop TSCA, then a number of basic amendments are needed, partly in line with an improved version of REACH, as sketched above. In my view, a reform should include the following⁴³:

• Inclusion of criteria for hazardousness, including toxic and non-toxic properties, guiding data requirements and further decisions on preventive measures

• Responsibility placed on companies to submit data, and lower barriers for the EPA to request additional information

• Treatment of existing chemical substances as, in principle, new substances; inclusion of the ‘no data – no market’ principle

• A comprehensive approach regarding chemicals in articles

• A general requirement to substitute hazardous chemicals when alternatives exist

• Duties for the EPA to decide on a span of preventive measures, even in case of uncertainty, with a burden to prove no need for regulation placed on companies

• A general requirement for substituting hazardous chemicals when alternatives exist

• Partial regulation of voluntary initiatives that fail to yield high quality data

• Increased disclosure of data on properties of chemicals

It remains to be seen how the debates in the US and its Congress develop and what the legal implications might be, but if TSCA were to develop gradually in accordance with the suggestions above, it could well mature into a legal system with the same protective functions as REACH, or preferably into something even better.

Evidently, managing uncertainty related to industrial chemicals is not an easy task. So far, the voluntary and regulatory approaches implemented around the world have not succeeded more than marginally. Implementing the core elements of the precautionary principle in EU and US legislation on industrial chemicals, as suggested and elaborated on in this article, would clearly enable better control of chemicals and, thereby, most likely lead to increased safety for human health and the environment.