Chillin'Competition

[In the course of the coming weeks we will be publishing a number of guests posts on different topics (feel free to send ideas our way!). We start this series of guest posts with a contribution from Daniel Beard QC and Jack Williams on discrimination concerns underlying new regulatory initiatives]

Language is both beguiling and dangerous.  Being an author would be no fun if it were otherwise.  But being an author and being a regulator are very different jobs.  Sometimes compelling words and rhetoric can distract from the regulatory job at hand. The political force of a cri de coeur risks losing sight of the raison d’être.

The mood of regulators and governments across the EU is that, in relation to digital technology and the operation of competition law, something must be done.  All sorts of interesting suggestions are being made, from new governmental inquiry units and changes in merger scrutiny to whole new competition tools. 

Amongst the myriad reports and studies one thing comes out loud and clear: discrimination is bad. It is particularly bad where the discrimination is by a dominant undertaking in a digital industry characterised by high entry costs, low marginal costs, and network effects. Indeed, it is so obviously bad that a new ex ante regulatory regime is needed so we can make sure it never happens in the first place.

The level of support for this idea is striking. Perhaps it comes from the language itself: who is going to say that preventing “discrimination” is wrong?  After all, when we first think about the notion of discrimination, we do not think about digital service self-preferencing; we think, for example, about powerful struggles for freedom from race or gender inequality.  Legal battles to protect people from discrimination on the basis of the colour of their skin or their sex, religion, physical ability or sexual orientation have been long and hard fought. It is easy, and obviously right, instinctively to think discrimination wrong when that is our starting point.

Yet, transferring the concept of discrimination away from recognised status cases involving individuals, the analysis becomes problematic and the effective use of ex ante tools even harder to gauge. There are three types of issue: substantive difficulty; procedural complexity; and fairness.

Substantively, discrimination analysis can confuse as much as enlighten. If you want to change the way that electricity transmission pricing works, can you treat low carbon, renewables and traditional generators in the same way? The electricity they produce is the same.  The way they do it is not.  Does that mean that you say all the generators are similar? If they are comparably similar, is there a justification for different treatment?  The test in such economic cases is whether you are treating like cases differently (or different cases the same way) without justification.

The problem is that the same essential considerations are used for assessing whether the cases are alike and whether there is a justification for differential treatment. What is really being asked is whether the particular benefits of renewable generators mean that arrangements for charging them differently from, for example, gas fired plant is reasonable overall.  It is an assessment of priorities and objectives within the sector. Labelling the issue as discrimination does not itself assist the substantive analysis of whether the changes properly pursue the legitimate aims of the sectoral regulation. Indeed, using the language of discrimination in economic contexts may ultimately distract from the key policy issues which a measure is trying to address.

It might well be because discrimination assessments outside recognised status cases are hard that until recently we had not seen as many cases as we might otherwise expect applying Articles 101(c) or 102(c) TFEU.  Broadly, those provisions prevent agreements or provisions by dominant entities applying different conditions to equivalent transactions – stopping discrimination.  Aside from various music collecting societies cases these apparently very powerful tools have not historically been wielded frequently.  Plainly things have changed more recently with cases such as MEO, Google Shoppingor Royal Mail v Ofcom.  Yet none of those cases is an advert for the clarity and simplicity of using discrimination as a key tool.

Ex ante non-discrimination provisions are applied to regulate utility industries.  Where a legacy incumbent inherits a system monopoly such that they have significant market power, non-discrimination requirements are a standard feature of the regulatory regimes. Those requirements often focus on criteria and terms for access to established networks: what information has to be provided for a connection? How are connections to be made and paid for?

However, there is a real difference in assessing discrimination – and justification – where networks have built through investment and competition (rather than bequeathed), where the nature of the services being provided by the dominant entity are changing, and where the criteria for assessment have to consider different business models and types of demand.  The substantive difficulties involved in applying discrimination outside status cases are all the greater in dynamic rather than mature industries. And that is in addition to setting the thresholds for what sorts of entity are going to be subject to the ex ante tools in the first place. 

In addition to these substantive concerns about using discrimination as a key tool, there is the second problem of a new regulatory process introducing complexity and inertia. Experience shows that building a novel regulatory infrastructure can itself create another layer of definitional disputes and uncertainties. 

Sponsoring lawyers and compliance departments to be involved in extensive exchanges with regulators is an excellent means of shifting resources from technology developers to professional services providers.  But it is far from clear that social utility and consumer welfare is enhanced by such a process.  It can increase the risk of undue intervention by regulators. It can encourage undue conservatism on the part of operators.  In a fast-moving industry a complex regulatory process can be a real impediment to innovation. 

Such a regime may offer particular protection to a cautious undertaking.  After all, if it can reach agreement with a regulator on an ex ante approach, the prospects of ex post fines under existing rules must diminish. Affecting the dynamism and risk taking of large companies can damage innovation.   It cannot be assumed that the best and fastest development always comes from start-ups or scrappy mavericks.   Does having a prior regulatory clearance mechanism mean better outcomes for consumers or simply a slower digital industry in Europe? 

Then there is a third issue: fairness. It is perhaps not going to cause people to rally to the barricades (or blogs) where the new process is directed primarily at large multinational firms, but the rule of law should not pick and choose. 

If companies are required to pre-justify conduct, that is effectively a shift in the burden of proof: they are being required to prove what they are doing is acceptable rather than a regulator having to show the contrary later.   There is a reason the burden is placed on regulators.  It is not just to make their life hard.  It is because it is generally accepted to be unfair to assume that someone is doing wrong unless they can show otherwise. The use of presumptions against people is something to be cautious about.  The fact that they would be used in the complex and confusing territory of discrimination assessments might make that all the more concerning. 

Whatever course is to be taken on ex ante regulation it needs to recognise that big themes and buzz words do not equal good law. We must not get carried away with an easy reliance on prejudicial language.  Perhaps we need to be more discriminating on discrimination in order to decide what new tools we really need.