Abstract
The paper proposes that competition authorities adopt an outcome-based approach to deal with anticompetitive mergers for which the parties provide a solidly based innovation defense. Such an approach would shift to the merging parties part of the risk of approval due to the stochastic nature of the innovation process. It would also discourage exaggerated innovation claims, and encourage firms to preserve competition while pursuing innovation. This “innovation clawback” mechanism could reduce competition authorities’ reluctance to approve potentially pro-innovative mergers. The proposal draws on analogous mechanisms in environmental law, biodiversity protection, pharmaceutical reimbursement and EU R&D State aid. The proposed mechanism would therefore make merger control more forward-looking and innovation-friendly while preserving its fundamental objective of protecting consumer welfare.1
1. The uncertainty of innovation: who should bear the risk of failure?
Aside from the difficulty of predicting whether an anticompetitive merger could lead to innovation, competition authorities are faced with a second difficulty which is the stochastic nature of the innovation process.
One of the reasons competition authorities are suspicious of innovation gain claims in merger control may be that whereas they can be reasonably certain of the risks to other dimensions of competition that the mergers they examine entail, innovation claims (even backed by an analysis of innovation capabilities) do not guarantee that innovation will take place. They thus run the risk of a false negative, and a reputational risk, if they clear anticompetitive mergers which do not achieve claimed innovation gains. As a result, to protect themselves against this reputational risk2 they may discount claims of dynamic efficiencies as being “speculative” even if the merging firms have put forth a strong case.
There can be three reasons for which a merger accepted on the basis of a successful innovation defense may fail to deliver innovation. The first is that the merged firms, which had the capability to increase innovation and devoted all of their efforts to the innovation after the merger, were unlucky and failed to come up with an innovation outcome. The second is that the competition authority made a mistake in its assessment that the merging firms would have the capability to achieve innovation. The third reason may be that the merged firms had the capability to innovate (as indicated by the analysis undertaken at the time of their merger) but not the incentives: instead of leveraging innovation capabilities, they chose to enjoy the monopolistic rents from the restriction in competition that their merger entailed rather than devoting sufficient resources to promoting competition on innovation.
The first risk is unfortunately impossible to address, though merging firms should not be penalized for lack of luck. As far as the second risk is concerned, the work undertaken on the assessment of the capability of mergers to increase competition on innovation,3 which needs to be pursued, may usefully contribute to reducing the second risk: the risk of a faulty evaluation of the capabilities of the merging firms and their competitors to innovate post-merger.
The remaining question is whether merger control can address the third risk by changing its operating practices to eliminate, or at least reduce, the risk aversion of the competition authority while keeping the incentive of the merging firms to limit themselves to credible claims of efficiencies when they notify their merger and to do their best to achieve the claimed innovation at the lowest cost to society, after the merger has been authorized.
Economists have in the past tried to address this issue. The Oxera study,4 published in 2026, proposes an adaptation of the consumer welfare standard when a competition authority faces the stochastic nature of some of the relevant competitive processes (typical of R&D projects). It suggests that a competition authority should then “capture the impact of a merger by reference to the change in Expected Consumer Welfare (i.e. ΔECW), defined as the difference in Expected Consumer Welfare between the merger scenario and the counterfactual”. Technically, the Expected Consumer Welfare “captures the level of consumer welfare across multiple possible relevant outcomes, weighted by the probability of each outcome”. However, they acknowledge that this solution would raise practical difficulties and that the competition authority would have difficulty placing specific values on the probabilities and associated outcomes of specific competitive events. The Oxera study therefore suggests that “reliance on ECW should be primarily seen as an analytical benchmark to evaluate the impact of a merger in situations where the future evolution of important competitive variables is uncertain” but not a practical solution to the problem at hand.
One of the weaknesses of the current system of merger control is that once a merger has been authorized, in most cases, for both legal and practical reasons, the merger cannot be undone. Given the fact that merging firms and the competition authority controlling the merger know that undoing a consummated merger is nearly impossible, the merging firms theoretically have an ex-ante incentive to make unjustified efficiencies or innovation claims to attempt to ensure their merger will be approved. In practice, however, this hardly ever happens because the law sets strict conditions that discourage the formulation of efficiency or innovation claims. Furthermore, once the merger has been approved the merging firms have an incentive to benefit to the maximum from the restriction to competition that their merger allows and will not face any constraint if they devote fewer efforts to innovating than anticipated.
Today, there is a movement to revise the EU merger guidelines to signal interest in the efficiency or innovation benefits either as the rationale for the deal or through a defense. Even under that reform, competition authorities may have an incentive to distrust the efficiency claims provided by the merger parties and may try to limit the risk of approving anticompetitive mergers likely to lead to innovation by conditioning their approval on commitments on the amount of resources to be devoted toward the achievement of the claimed efficiencies, but they tend to be skeptical about the usefulness of such commitments5.
2. The proposed clawback mechanism
One possible means to overcome these difficulties could be to adopt a mechanism such that, if the competition authority finds, at the end of an agreed-upon period, that the claimed innovation has failed to materialize, the affected consumers could be compensated by the merged firms for the loss of consumer surplus they suffered through the restriction on competition associated with the merger, as estimated by the competition authority, without the victims having to prove a fault on the part of the Commission or of the merging firms. Alternatively, if compensating affected consumers is impractical, the firm could be required to pay to the State or to a Special Fund the value of the lost consumer surplus.
The merging firms would have no obligation to compensate consumers in two circumstances. First, if they succeeded in innovating as planned within the period. Second, if they failed to innovate within the period after their merger, but kept competition at its pre-merger level during that period (for example, by refraining from closing down one line of R&D undertaken by the target because it would exercise a negative externality on the line of R&D of the acquiring firm or if they refrained from increasing their prices post-merger).
If such a mechanism did exist, the risk associated with erroneously allowing an anticompetitive merger on the basis of an innovation defense would be partly shifted from the competition authority to the merging parties.
The merging firms would then face two sets of strategic choices.
First, unless they were fairly confident that they could and would deliver on the innovation, they would have an incentive, after the approval of the merger, to limit or eliminate the restriction to competition on the relevant market rendered possible by their merger (since they would be exposed to having to reimburse the consumer surplus loss on the market and not merely the consumer surplus losses associated with their sales) if the innovation did not materialize. This means that they would have an incentive to not increase their prices above the pre-merger level or to not degrade the quality of their offering compared to what it was pre-merger or to not reduce their output or to not eliminate or redirect a line of R&D because it competes with another similar line.
If, on the other hand, the merging firms were confident that they would deliver on the innovation, they could take the risk of benefiting from the restriction on competition that their merger made possible. Since the competition authority would have authorized the merger only if the consumer surplus benefit associated with the innovation was greater than the reduction in consumer welfare associated with the restriction of competition which the merger could entail, the net consumer surplus would be increased.
Second, when agreeing on the period for the development of their innovation, the merging firms would also be faced with a strategic decision. The longer the period they would propose to come up with the innovation, the higher the probability that they would succeed but the higher the cost for them if they benefited from the restriction to competition during this period and failed to come up with the innovation.
Such a mechanism would require the merging firms to be precise with respect to the innovation claimed as a result of the merger. But the precise delimitation of the expected innovation is, in any case, necessary for the innovation defense irrespective of whether the suggested mechanism exists. Indeed, to decide whether an anticompetitive merger should be allowed on the basis of an innovation defense, the competition authority would verify that the innovation claimed would meet the three criteria which apply to an efficiency defense6: the innovation must be beneficial for consumers (consumers must be no worse off after factoring the positive effects of the merger on innovation and the consumer welfare loss due to the restriction to competition which it entails), must be merger specific and must be verifiable (thanks to an analysis of the capacities of the merging firms). This means that only precise innovation claims can be accepted as a defense against an otherwise anti-competitive merger (for example, the development by the merging parties of a vaccine or a new treatment for a known disease). Thus, the assessment by the competition authority at the end of the agreed-on period of whether the innovation was developed in accordance with the initial claims of the merging parties should not raise a great deal of difficulty.
At the end of the period agreed upon by the parties and the competition authority, the competition authority would have to assess the decrease in competition associated with the merger if the merging firms failed to develop an innovation. Such an assessment requires the skills which are the core skills of the competition authority. Furthermore, presumably, the competition authority in its initial decision would have detailed the theory of harm associated with the merger, and in doing so would have indicated which elements would characterize a reduction in competition (for example, the fact of merging two independent and competing research projects).
The competition authority would have to undertake an investigation at the end of the period agreed upon to determine whether the promised innovation has been delivered and, if it has not been delivered, whether the possible anticompetitive effects of the merger which it had discussed in its decision authorizing the merger, have materialized and what is the loss of consumer welfare during the period agreed-upon. It would thus have to answer classical questions of competition for a competition authority (questions for which it would have set the metrics in its initial decision) and would not have to deal with complex scientific questions (such as, for example, why the firms failed to innovate according to their initial claims).
3. Addressing potential concerns and implementation challenges
Critics on the business side could argue that in a number of cases they cannot precisely anticipate the innovation that they will come up with if they merge. It is possible that a line of research could lead to different possible results but in such a case, the firms can in all likelihood map the possible expected outcomes of their innovation efforts. The competition authority is able to take these expected results into consideration to assess whether the merger is likely to increase the consumer welfare. If the merging firms argue that there is no way for them to know what innovation they will develop if they merge, it seems difficult for them to argue that their anticompetitive merger is nevertheless likely to lead to innovation which will compensate the lost consumer welfare due to the merger. In that case, their merger should not be allowed.
Critics on the competition authority side might argue that this procedure would require them to open an investigation of the reduction in consumer welfare associated with the restriction of competition in cases where it is found that the promised innovations did not materialize. This is true but four observations are in order: First, it is only in the case in which a competition authority has accepted the innovation defense of an otherwise anticompetitive merger and that the promised innovation did not materialize that it would have to investigate the reduction of consumer welfare due to the restriction of competition (and not in all cases for which the competition authority would have accepted an innovation defense). Second, if such a mechanism were implemented, authorities would trade more work for preserving their reputation as protector of consumer welfare and for more relevance in merger control. The real question is thus whether competition authorities would rather have less work and be considered less relevant when it comes to merger control (particularly in high-tech industries). Third, the specification of the theory of harm and benefits in the original decision authorizing the merger will serve as the framework for examining if the innovation has materialized and, if not, whether the competition concerns have led to a loss of consumer welfare. It is thus possible for the competition authority, if its decisions are clearly reasoned, to simplify the scope of the investigations at the end of the agreed-on period. Fourth, it is generally agreed among competition authorities that doing ex post analysis of merger decisions is one of the best ways to increase the quality of such decisions over time.7 The investigations suggested here, which will only occur if innovation has not materialized, go in the direction of what competition authorities agree they should do.
Another concern could be voiced. Forcing merging firms which have claimed that they would innovate and failed to do so within the agreed-on period to compensate society for the loss of consumer welfare that their merger entailed does not cover the whole social cost associated with the anticompetitive merger (and is thus insufficiently deterrent). Indeed, the concentration realized through the merger continues after the period agreed upon for the development of the innovation and additional welfare loss may be due to the merger. This is a technicality which can be taken care of in the mechanism which does not require that the penalty for not innovating and restricting competition should be strictly limited to 100% of the consumer welfare loss during the period agreed upon. As we shall see below, similar mechanisms in other areas of European law allow for a sliding scale of compensation of the undue advantage.
Still another concern regards the question of whether such a mechanism could be adopted without requiring a (highly unlikely) revision of the merger control regulation. This question would deserve further analysis but it would seem that commitment decisions where the authorization of the merger would be conditioned on the commitment of the merging firms to compensate the lost consumer surplus if they fail to innovate within a period agreed-on with the Commission could be a way to implement the underlying principle of the mechanism without amending the merger regulation.
Finally, to allay the skepticism expressed by the European Commission with respect to the investment commitments proposed in the Draghi report, in the mechanism we suggest there is no such investment commitment required from the merging firms. If the firms fail to innovate, for whatever reason, and have restricted competition, they would have to compensate the lost consumer surplus that occurred during the negotiated period. Such a condition would not require continual monitoring and enforcement since it is the success or the failure to innovate at the end of the agreed-on period which would determine whether the merging firms have an obligation and not the means the merging firms have devoted to achieving the result.
A mechanism of the kind proposed would decrease the incentive of merging firms to make unsubstantiated efficiency claims at the time of the merger. It would also decrease the risk aversion of the competition authority and therefore the risk that pro-innovation mergers might be blocked and it would compensate consumers affected by a restriction to competition if a merger failed to deliver on innovation.
4. Precedents for outcome-based accountability
This mechanism goes in the direction of one proposal of the Draghi report which states8: “Competition authorities need to be more forward-looking and agile. For example, since innovation in the tech sector is rapid, merger evaluations in this sector must assess how the proposed concentration will affect future innovation potential, despite its uncertainty.”
Such a mechanism would be consistent with an important evolution of the scope of civil liability. Indeed, whereas originally the idea was that civil liability should have a function of deterrence and therefore that the obligation to compensate for damages should be allocated to a law violator, legislators (particularly in civil law countries) have increasingly frequently considered that professional producers of goods and services should bear the risk of their activities regardless of any proven “fault” caused by damages to their customers or to third parties. The doctrine of No-Fault Liability for Lawful Acts (responsabilité sans faute pour acte régulier) has become more widespread.9
In line with this evolution, in the EU and its member states (or former member state, in the case of the United Kingdom), there are various examples of situations where a firm benefiting from an administrative authorization does not commit a violation, acts in good faith, but nevertheless has to compensate parties or to pay a sum to the authorizing body because the expected outcome from the authorization did not materialize. Such legal mechanisms exist in laws or regulations dealing with infrastructure construction or the environment, sectors in which it is often impossible to undo the work which led to the damage.
For example, under the German Federal Nature Conservation Act10, European “No Net Loss” biodiversity principles are enforced through an outcome-based lens.
The law provides that if a development project is likely to cause significant damage to the environment or to the landscape and that a part of this environmental damage cannot be avoided or compensated within a reasonable period, the project can only be allowed to proceed if the competent authority has determined that the nature-conservation interests do not prevail over the competing interests. But, in such a case, the developer must pay to the state through the Ersatzzahlung (substitute payment) mechanism the anticipated cost of the unmitigated ecological damage with the money earmarked for other nature-conservation and landscape-management measures.11
A related mechanism is used in the United Kingdom. The 1990 Town and Country Planning Act (TCPA) was amended by the Environment Act 2021. Following this amendment, Schedule 7A of the TCPA12 provides that grants of planning permission in England must be subject to a condition to secure a biodiversity gain objective of 10%, which must be legally maintained for at least 30 years. This is an absolute statutory objective. If this condition is not met, the Local Planning Authority can take enforcement action against the developer for breach of a planning condition. To rectify it, the developer will be forced to pay to restore the site, buy offsite “biodiversity units,” or, as a last resort, purchase statutory credits from the government.
In 2004, the European Union adopted a new regime for the prevention and remediation of environmental damage by adopting Directive 2004/35/EC “on environmental liability with regard to the prevention and remedying of environmental damage”.
This innovative legislation established, for the first time in the EU, a comprehensive liability regime for damages to the environment, based on the ‘polluter-pays’ principle by making those having caused environmental damage liable for remediation of the damage.
The competent authorities, which are public bodies designated by the Member State to implement and enforce the Directive, determine and assess the extent of environmental damage and agree, with those responsible for the potential damage, on the preventive or remediation measures to be taken. Operators carrying out dangerous activities listed in an annex of the Directive fall under strict liability (no need to prove fault13). If full remediation is delayed, compensatory remediation measures should be taken for the interim loss of natural resources and services.
In the health sector, Italy has become a leading user of Managed Entry Agreements (MEAs) to manage the clinical and financial uncertainty associated with high-cost novel therapies. These agreements aim to provide patient access to innovative drugs while ensuring the financial sustainability of the Servizio Sanitario Nazionale (SSN). Outcome-based MEAs directly link reimbursement to a drug’s real-world performance. The most prominent type is payment-by-result, where the Italian Medicines Agency (AIFA) tracks individual patient outcomes against pre-agreed clinical endpoints. If a treatment fails to deliver the expected benefits, the manufacturer refunds the cost of the drug, transferring the financial risk from the SSN to the company14.
EU State aid law (Articles 107–109 TFEU) is also relevant to our discussion.
EU State aid law already uses ex ante approval combined with ex post financial adjustment where the future outcome differs from the assumptions about performance on which regulatory approval was based, particularly in relation to innovative investment and R&D.15
For example, in the case of repayable advances for R&D&I, which are one category of State aid, the Communication from the Commission16 provides:
“85 (…) the repayable advance (…) may exceed the applicable maximum aid intensities by 10 percentage points, provided that the following conditions are fulfilled:
(a) in case of a successful outcome, the measure must provide that the advance is to be repaid with an interest rate not less than the discount rate resulting from the application of the Communication from the Commission on the revision of the method for setting the reference and discount rates;
(b) in case of a success exceeding the outcome defined as successful, the Member State concerned should request payments beyond repayment of the advance amount including interest according to the applicable discount rate;
(c) in case the project fails, the advance does not have to be fully repaid. In case of partial success, the repayment should be proportional to the degree of success achieved”.
Over and beyond the case previously mentioned, the underlying rationale for EU State aid control is that public subsidies given to private companies distort the market and cause collective economic damage to competitors. The EU will permit this damage if the aid is tied to a strict plan that delivers a larger societal benefit—such as saving regional infrastructure or generating long-term employment. The failure of an authorized aid measure to generate the expected positive effects does not, in itself, lead to the retroactive withdrawal of the Commission’s authorization or to an obligation to reimburse the aid. However, where the beneficiary fails to comply with binding conditions attached to the authorization, the Commission may investigate whether the aid has been misused (Article 20 of Regulation (EU) 2015/1589). If the aid is ultimately found incompatible with the internal market, the Commission may adopt a negative decision and order its recovery, together with interest, in accordance with Article 16.
Conclusion
These examples are not always perfect matches for the suggested mechanism discussed in this paper concerning mergers and innovation. Indeed, in some of the examples quoted above, either the Commission finds that a practice was illegal (in the case of State aid) or the compensation that firms have to pay for not achieving a social benefit is based on an estimate of the value of the social benefit which was not realized. In the case of mergers and innovation, what is proposed is not to withdraw the authorization given to the merger but to hold the merging firms liable for having restricted competition during their unsuccessful search for innovation (and not for having failed to innovate).
But what these examples have in common with our suggested mechanism is that the firms which have secured an administrative authorization to undertake a risky activity are exposed to the risk of having to pay for the loss they impose (on their customers, on third parties or to competition in the case of State aid) if they fail to deliver on the expected benefit which justified the authorization. As argued previously, this is likely to make them more realistic about the benefits they claim, more reluctant to restrict competition, and more incentivized to live up to the expectation they have created once they have secured the authorization.