Maximum vs minimum harmonization: what to expect from the institutional and legal battles in the EU on gene editing technologies.

New plant-breeding technologies (NPBTs), including gene editing, are widely used and drive the development of new crops. However, these new technologies are disputed, creating uncertainty in how their application for agricultural and food uses will be regulated. While in North America regulatory systems respond with a differentiated approach to NPBTs, the Court of Justice of the European Union (EU) has in effect made most if not all NPBT subject to the same regulatory regime as genetically modified organisms (GMOs). This paper discusses from a law and economics point of view different options that are available for the EU's multi-level legal order. Using an ex-ante regulation versus ex-post liability framework allows the economic implications of different options to be addressed. The results show that under current conditions, some options are more expensive than others. The least costly option encompasses regulating new crops derived from NPBTs similar to those used in 'conventional' breeding. The current regulatory situation in the EU, namely making the use of NPBTs subject to the same conditions as GMOs, is the most costly option.

INTRODUCTION

Scientific progress has resulted in new methods in plant breeding. Concerns have been expressed by scientists and others that those methods can be abused or create unwanted outcomes and hence applications should be closely monitored. These concerns, in combination with the political economy related to the rents being generated and lost by the new methods, result in regulatory policies governing the introduction and use of products developed.1 The appropriate regulatory policies are widely discussed in the USA, the European Union (EU) and other countries.2 While a set of regulatory policies has been developed to govern the approval and use of so‐called genetically modified organisms (GMOs) developed by using transgenic methods, progress did not stop. More precise and cost‐efficient methods have been developed, raising questions about their regulation.3, 4 In the literature they have been summarised under the term ‘new plant breeding technologies’ (NPBTs).5 Examples include CRISPR‐Cas 9, TALEN, zinc fingerprinting, oligo‐directed mutagenesis (ODM), and more. Examples of food products that have received approval for the US market include non‐browning apples and mushrooms.3

Many of the NPBTs introduce mutations to change the genetic information of an organism. Mutation‐inducing processes, mutageneses, have a long history in plant breeding. Under EU law, most products produced using mutagenesis are considered GMOs by definition, but are explicitly exempted from GMO regulation. A debate exists regarding whether mutagenesis exemption, which dates from 2001, only applies to those processes available until 2001 or if the exemption also includes future processes. The Court of Justice of the European Union (CJEU) recently ruled that this exemption does not apply to NPBTs, making most, if not all, of them subject to the same regulatory regime as GMOs.6 This all‐or‐nothing approach has triggered discussion regarding the fitness for purpose of the current EU's GMO regulatory regime to meet the challenges of the 21st century.6 In the literature an agreement has emerged that some sort of reform of the regulatory system is warranted.

Subsequently we will discuss and assess four options for reform, taking the possibility to regulate at different levels of the EU multi‐level system with different intensity as a starting point. Each option will have implications for the benefits and costs and hence the incentives for private and public sector investments in further developing and applying NPBTs. The regulatory outcome can have important economic implications. The additional costs to comply with regulatory safety standards for GMOs have been assessed to be about between US$6 and 14 million,7 but as Smyth et al. 8 show they vary widely between different jurisdictions.

This has implications for the development of natural products for pest management. Gene‐editing technologies are an essential part of developing these technologies as the other papers in this special issue show. Depending on how gene‐editing technologies are regulated in the EU, access to natural products for pest management is more or less restrictive.

The literature on the economic implications of regulatory policies with respect to new technologies is diverse.9 First, it is foremost a legal question, whether or not a new technology requires additional regulation beyond the current level: which of the current regulatory policies apply and do they sufficiently well address potential risks? These legal questions are not independent of economic considerations. Regulatory policies can be described by ex‐ante regulatory standards and ex‐post liability rules. In the case of NPBTs, for example, an investor has to comply with regulatory standards prior to placing a product on the market and faces possible ex‐post liability in the case of non‐compliance and/or damage. Many authors have shown that a combination of ex‐ante regulatory standards and ex‐post liability rules is often superior to using only one or the other.10, 11 Nevertheless, one has to acknowledge that avoiding ex‐post liability is almost impossible. Even if not being held liable by law, an investor may observe negative effects on reputation and related economic consequences.

In this contribution we develop a real option model based on the ex‐ante regulatory standards and ex‐post liability literature to assess NPBT regulatory policies on private sector incentives for investment in NPBTs. We divide the investment into three phases: the research phase, the approval phase and the market phase. Each phase will be more or less strongly affected by ex‐ante regulatory policies. All three phases are characterised by uncertainty about the time length of each phase. Additional uncertainty is added by the probability of ex‐post liability. After the product has reached the market.

This set‐up allows us to compare different regulatory policies and their implications on the incentives for investment. While in the literature ex‐ante regulatory standards have been mainly modelled as one phase, the differentiation into three phases allows for a more detailed regulatory policy analysis. We are in particular able to show the strong effect of the costs of regulatory standards and ex‐post liability rules under uncertainty and irreversibility on the decision to invest. While this is to be expected, as the strong effect of sunk costs on investments is well known from the literature, our model is richer in the details, allowing a more detailed policy analysis.

Further, our approach enables us to illustrate the trade‐offs of different policy options and the factors driving those. Finally, the model is developed for and motivated by the case of investments in NPBTs. However, it is generic enough to provide additional insights into trade‐offs between ex‐ante regulations and ex‐post liability. We are able to show the different outcomes of minimum versus maximum harmonization regarding the regulation of NPBTs, where minimum harmonization refers to decisions being made at EU Member State level and the EU only provides some general guidelines, and maximum harmonization refers to the decisions being made at the EU level and Member States have to comply.

We proceed by first developing the generic model for our analysis before we assess in more detail the legal environment governing the approval of NPBTs in the EU. The assessment allows us to identify four policy options at EU level that are assessed after the real option model has been introduced. The four options discussed are derived from the assessment at EU level, but they not only apply to the EU. The policy options are generic and apply to other cases as well that we discuss before we conclude.

ECONOMIC MODEL ASSESSING INVESTMENTS IN NPBTS

NPBTs can be regulated at several levels of the EU. One needs first to acknowledge that the regulation does not start from scratch. The Treaties of the European Union set out a general framework, and secondary legislation may cover specific cases. As pointed out in more detail below, a regulatory framework at secondary level can follow different harmonization methods, in particular choosing between minimum and maximum harmonization or a combination of both. In the case of NPBTs this results in the four basic options, which have different economic implications.12 The economic implications follow from differences in the costs and benefits companies face under different regulatory policies. In general, the lower the fixed costs, the larger the possibility for companies entering the sector and the more competition about new ideas will emerge. The underlying assumption is that the more firms get involved the better this will be ceteris paribus for the economy under consideration.

In general, the objective of a firm considering investing in developing new crops using NPBTs can be modelled as maximizing the real option value of the investment. There will be a trade‐off between the ex‐ante regulation a firm has to follow and the ex‐post ‘liability’ costs they may face. We start with the general model and later compare alternative institutional arrangements and the possible trade‐offs they include.

Let F denote the value of the option to invest. The firm has to invest into research, R, for developing the new technology. These investments are paid at the start of the research phase. In addition, there are annual research costs r , with t indicating time. The time needed to complete research is not known but expectations exist. The time length for this research will be denoted by the random variable κ. κ follows an exponential failure function with and , where h denotes the failure rate. At κ 1 an application for approval will be submitted. The submission for approval includes approval costs, A, that are considered to be sunk, and some annual reversible costs, a . The time length for approval is not known but expectations exist and it is denoted by random variable κ . At κ 2 the product will be approved for market entry generating a benefits stream, B, expressed in net‐present‐value terms, B 0, at time κ 2. Firms may face ex‐post tort liability and/or reputation costs, θ, if damages linked to the product introduced occur, again modelled as being random and denoted by κ 3.

The expected value of the investment can be written as follows:

This provides the following solution assuming r and a are constant:

Equation (2) is the expected value of immediate investment. This expectation may change over time. As can be seen from Eqn (2), if the expectations about the length of the approval process reduce and/or the fixed approval costs are reduced and/or the benefits, B, increase the expected value of the investment increases, while, if the opposite happens, the expected value decreases. This is not a negligible possibility regarding the debates around NPBTs.

For keeping the model simple, two future possibilities are considered, one in which the future looks bright, B is high, B , and increases with probability q to and one in which the future looks less bright and decreases with probability 1 – q to , . For keeping the model economical relevant the following is assumed:

Assumption: .

Solving the model for a one unit of time, t = 1, delay, such as 1 year, provides the following solution, with subscript p for postponement:

The objective of the firm assuming profit maximization and abstracting from potential issues related to competition among firms is as follows:

Equation (4) allows the threshold for immediate investment versus postponement to be identified by taking the difference between Eqns 2, 3. Immediate investment is economical if:

Lemma 1: If NPBTs are banned, , it is always economical to postpone investment in research and development of NPBTs.

Proof: By definition and the right‐hand‐side of Eqn (5) always positive.

If all the costs are normalized to 1, the weighing factor or hurdle rate for the costs can be summarized as:

Equation 6 shows this factor is clearly larger than 1. Table 1 shows weighing factors for different parameter values. The second row shows hurdle rates for changes in expected values for κ 1, holding κ 2 constant at E[κ 2] = 10, while the fourth row shows the hurdle rates for different expected values of κ 2, holding κ 1 constant at E[κ 1] = 10. The expected values range between 1 and 10 years. These are reasonable numbers. Smart et al. 13 estimated the time for approval of GMOs to be about 6.7 years on average while Fredericks and Wesseler14 estimated the approval length for microbial biological control agents to be about 4.7 years. The reported hurdles are substantially larger than one stressing the importance of regulatory policies on investment. The hurdle rates decrease with a decrease in the expected values of κ 1 (κ 2). The last row shows the hurdle rates for zero approval costs. This observation results in the following Lemma:

Table 1

Hurdle rates for different parameter values

E(κ1)	10	5	2.5	1
Hurdle rate	14.59	10.80	8.91	7.78
E(κ2)	10	5	2.5	1
Hurdle rate	14.59	10.70	8.76	7.59
Hurdle rate zero approval costs	8.66	4.88	2.99	1.86

The hurdle rates are calculated applying Eqn (6). Other parameter values are fixed at μ = 0.04, q = 0.5, if not otherwise.

Lemma 2. A unit increase in all costs requires substantially more than one unit of additional benefits B for justifying immediate investment.

Proof: By definition and Eqn (6) is larger than one.

The implications of the model will be discussed in the context of the debate about approval of NPBTs in the EU.

LEGAL ENVIRONMENT IN THE EU

The regulation of NPBTs has been a long‐contested field in the EU. In view of this uncertainty, several Member States (most prominently Sweden) had introduced specific regulations such as approval requirements. At EU level (which this article focuses on) there are no specific regulations covering the area of NPBTs. In view of the non‐existence of a special legal regime to cover products derived from NPBTs, which involve mutagenesis, NPBTs are according to the recent ruling of the CJEU subject to the existing regulatory regime for GMOs. Various acts exist which cover the regulation of GMOs in the EU (see for a summary Wesseler and Kalaitzandonakes15), out of which Directive 2001/18/EC16 on the deliberate release into the environment (hereinafter Directive) and Regulation 1829/200317 on genetically modified food and feed (hereinafter Regulation) stand out. Both established, among other tools, an authorization requirement for GMOs which are released into the environment and for those to be used in food and feed. Products derived from ‘mutagenesis’, however, are according to Annex I B Directive exempted from the scope of the Directive. Mainly based on the formulation of Recital 17 of the Directive, which stated that ‘(t)his Directive should not apply to organisms obtained through certain techniques of genetic modification which have conventionally been used in a number of applications and have a long safety record’,17 the CJEU held, adopting the evaluation of the risk assessment of the referring national court, that such a long safety record does not exist for NPBTs. Hence, the ‘mutagenesis’ exemption does not apply to NPBTs.

Previously, Advocate General Bobek (AG)18 had proposed a different interpretation, which would have given Member States more leeway in their legislation to decide how to regulate NPBTs. After the final judgment, the AG's proposal now may be used as a blueprint for possible legislative options to come. We will first illustrate the different options proposed by the AG before we assess their impact from an economic point of view.

AG Bobek interpreted the ‘mutagenesis’ exemption as covering NPBTs. As the AG rightly emphasizes, however, this does not necessarily mean that NPBTs will remain unregulated. Rather, he invites the EU to provide clearer and more detailed regulation on NPBTs if it wishes to do so. Alternatively, Member States may regulate the use of NPBTs. The exemption, which is also cleared by the AG, is only a minimum harmonization provision (Opinion of AG Bobek, C‐528/16, ECLI:EU:C:2018:20, para 129). Otherwise, he would see possible tensions with the precautionary principle (Opinion of AG Bobek, C‐528/16, ECLI:EU:C:2018:20, para 122) and the general tendency in EU GMO regulation to give Member States more leeway to decide if and what kind of GMOs they would like to introduce to their market.19 Hence, while the majority of NPBTs remain unregulated in view of the EU Directive, Member States can initiate legislation, which would then bring these kinds of laws under the scrutiny of EU primary law, in particular the freedom of goods in Art. 34 TFEU. If foods and feed are involved, the regime of food and feed law would also be applicable.4

Taking the options derived from the AG's Opinion as a starting point, this would open up four regulative options:4

The default option: NPBTs are regulated at EU level.

EU institutions in secondary legislation frame Member State laws for NPBTs.

NPBTs will be in the future exempted from GMO law and regulations for ‘conventional’ seeds at Member State level apply.

Member States take action and regulate these NPBTs under their own laws with effect for their territory.

Option 1 and 3 can be realised within a maximum harmonization strategy, while option 2 and 4 would typically be realized by a minimum harmonization strategy (see on these strategies in the literature Gerner‐Beuerle).20

ASSESSING THE DIFFERENT OPTIONS

The results of the ex‐ante regulatory standards and ex‐post liability can be used to compare the different policy options discussed above.

Option 1: NPBTs are regulated at EU level

Under this option the GMO Directive will be applied, and NPBTs are not subject to the mutagenesis exemption. Here, the steps for the approval of GMOs would apply. Decisions on the technologies would need to be reached following the comitology procedure of the EU. This includes voting by the Member States in the relevant committees where decisions need to be reached by qualified majority rule. Again, looking back at past experiences on the approval of GMOs, reaching a qualified majority for or against a technology to be included under the mutagenesis exemption is highly unlikely. Finally, the European Commission will decide, and they can be expected to follow the advice from the European Food Safety Authority (EFSA). While in the end a decision will be reached, the process will be highly time consuming. This will increase the waiting time for plant breeders and the factor (1 −  in Eqn (6) will become larger and the incentives for delaying investments increase.

Option 2: EU institutions in secondary legislation frame Member State laws for NPBTs

In this case the EU would develop a general legislative framework to be voted on by the EU Parliament and EU Council. Member States would have the freedom to decide how to specifically apply the framework. This would be similar to the regulatory techniques used in the EU's General Food Law.22 An example to what can be expected for the case of NPBTs are the regulations Member States use for coexistence policies of GMOs. Some Member States use very stringent coexistence policies that come close to a cultivation ban while others apply coexistence policies that have almost no effect on cultivation decisions by farmers (see Beckmann et al. 23, 24 for a detailed discussion). Looking at Eqn (6) the implications of Option 2 are that the term (1 −  will increase due to the time delay a decision to be made about the appropriate regulatory policy. Depending on the effort the EU institutions involved will use to reach an agreement, the time span can be relatively short, say about 2–3 years as, for example, when the Directive was developed in the late 1990s through the early 2000s. Considering that at that time a shared view among Member States was that GMO regulatory policies were needed15 and that today views are more diverse, it is reasonable to expect that the timeframe for developing the regulatory policy will be longer than 3 years. The possible result is that after an agreement at EU level has been reached, approval and compliance with coexistence rules in some Member States will be less demanding (costly) than in other. This reduces A and a .

Option 3: NPBTs will in the future be exempted from GMO law and regulations for ‘conventional’ seeds at Member State level apply

Unlike the default option in the EU, this situation is similar to the one in the USA,3 where NPTS are treated in a similar way to ‘conventional’ crops that are not under the scrutiny of the Directive. Firms will have to register new varieties under the seed law in the different Member States, similar to the GMO‐developed oilseed rape and sunflower in France. In this case B will be large, and κ 1 and κ 2 will be low as will A and a .

The research costs R and r will also be lower as requirements for field trials such as fencing and other compliance costs required for field trials of GMOs25 will not apply. Using Eqn (6) this simplifies for A = 0 and a  = 0 to . The bottom row of Table 1 shows the effect of zero approval costs on the hurdle rate. The hurdle rate substantially decreases.

Option 4: Member States take action and regulate these NPBTs under their own laws with effect for their territories

In this case the NPBTs are considered to be GMOs but are exempted under the mutagenesis clause. Member States can implement additional regulations. As the view among Member States differs26 it is reasonable to expect that some Member States will ban or more heavily regulate their cultivation. They will not have the ability to establish barriers for trade in products derived from the cultivation of NPBTs. In comparison to Option 1, the approval costs A and a are expected to increase as several applications need to be prepared, while B will be reduced as the technology cannot be used as widely. The voting results of Member States at the Standing Committee and the Appeal Committee on GMOs can be used as a proxy to determine which Member States might look more favourably at NPBTs. Using the results of Smart et al. 26 it is reasonable to assume that countries like Austria, Bulgaria, France, Germany, Hungary, Italy, Luxembourg, Malta, Poland and Romania will prevent cultivations, while countries like Czech Republic, Denmark, Portugal, Slovakia, Spain, Sweden, the Netherlands and the UK are more likely to allow cultivation following their country rules for ‘conventional’ seeds. This solution may result in regulatory competition and over time harmonization of regulations is expected.27

A view beyond the EU

The debate about NPBTs is not limited to the EU. The USA and other countries are also discussing the regulation of NPBTs. The USA announced in May, 2018 that the ‘USDA does not regulate or have any plans to regulate plants that could otherwise have been developed through traditional breeding techniques as long as they are not plant pests or developed using plant pests’.28 This at first‐hand may look as a very lenient regulatory policy. A closer look at NPBTs and regulatory policies discloses that many products will still be regulated in a similar way to GMOs for reasons such as a combination of different events and plant‐breeding techniques.3 The announcement, while providing regulatory certainty, does not seem to change the approval costs.

The fact sheet approach used for the approval of crops in Argentina has been appreciated in the literature4 for its simplicity. The additional approval costs are close to zero for NPBTs, reducing the cost factor of Eqn (6). In Canada, products developed by NPBTs fall under the Novel Food Law, which follows a product‐based approach. The costs can be expected to be low, as many food products derived from NPBTs will not be different from other products.

The differences in regulatory policies observed around the world can be considered in the context of this paper as a minimum harmonization policy. Countries have the freedom to choose their own regulatory policies as long as those policies are in line with international legal obligations such as, for most, as disciplined by WTO law. There is an important difference between the EU and the world market. In the EU, regulations at Member State level would only affect cultivation but not extra‐EU trade. At the world market level, this is currently not the case. At EU level products produced by NPBTs that have not received approval for import and processing are baned and a zero threshold level for food products applies. A similar policy applies for imports into the USA depending on the specific trait if they have not received approval by the US Department of Agriculture (USDA). The difference in regulatory policies at the international level can be expected to increase friction in international trade. Even so, one may argue that commodities or food products derived from the application of NPBTs cannot be differentiated from ‘conventional’ commodities and food products, but this has not prevented countries from imposing import restrictions. In many cases it is just enough to know that NPBTs have been used to impose import restrictions. Looking at Eqn (5), this reduces B 0 and increases B , providing strong incentives for delaying investment. Further, the regulatory uncertainties at international level increase the ex‐post liability costs for the private sector. Reaching international agreements on trade in NPBTs can have a strong effect on the incentives to invest in applying the technologies as an agreement reached will increase B 0 and reduce B .

CONCLUSIONS

The regulatory options at EU level have an effect on the incentives of the private sector to invest in NPBTs. Options based on minimum harmonization provide stronger incentives than those based on maximum harmonization. Minimum harmonization is expected to reduce research and approval costs, including approval length. While minimum harmonization may not provide access to the whole EU market for cultivation of crops, maximum harmonization will not do this either. Countries against the cultivation of crops based on NPBTs have other means for de facto banning cultivation.

The results of the model show that it is difficult to derive a solution with a common hurdle rate for all cost items. This requires a careful interpretation of the results. Nevertheless, the basic result that uncertainty and irreversibility have a strong effect on postponing investment in NPBTs holds. Interestingly, the marginal effect of a delay by 1 or 2 years is not that strong as a change in the time length for research and approval. At the time of deciding about investment into NPBTs the marginal effect of ex‐post liability after market entry has only a small effect on the incentives for investment, while ex‐post liability can be used as a substitute for approval costs. Moving in this direction will have a strong effect on incentives for investment. This not only applies at EU level but also at international level. A liability system at international level that compensates approval costs will further strengthen incentives to invest in NPBTs.

We have illustrated the implications of regulatory policies, providing numerical examples. The model predicts that there are gains from regulatory harmonization at EU as well as international level. They should result in an increase in investments in NPBTs. Yet, in the short term – the next 5 years – we do not expect this to happen. Even if stakeholders such as the scientific community, policy makers, civil society groups and others would start lobbying now, time will pass before agreements will be reached. Nevertheless, our results illustrate that lobbying for harmonization at international level can boost investment in natural products for pest management. Considering the challenges humanity faces due to climate change and the resulting need for developing a portfolio of adaptation strategies such lobbying activities are a worthwhile effort.

The model we present only assesses incentives for immediate or postponed investment in the application of NPBTs by the private sector. The model can be enriched by adding details about the benefits generated, including competition issues, but also international market access. The model we present provides the framework. At this point in time it is not possible to test this for the specific case of NPBTs as they are just entering the market and we leave this for future research.