There is a card game called Hearts, which I learned as a kid. It exists in various versions and under different names, but they are all broadly the same. Players must follow suit and the highest card wins each trick. The goal of the game is to avoid winning tricks that contain hearts, and each heart you collect earns a penalty point. We played a version where ending up with the Queen of Spades earned you a disastrous 13 extra penalty points (almost certainly causing you to come last). The fun of the game is in trying to screw over your opponents – dumping your hearts, or even better the Queen, into tricks they will win.

But here’s the catch. Competing on the merits is all well and good, but the game also allows for a strategy called “shooting the moon”. To shoot the moon a player must collect all the hearts and the Queen. Pulling this off is like grabbing the golden snitch – it’s an automatic win. But miss even one heart card and you fail (and fail badly!).

More on this later.

We now know there is a divergence of views between the European Commission and the CMA on whether behavioural remedies work in non-horizontal mergers. The Commission has, in line with its approach in several similar past cases, accepted commitments from Microsoft and allowed its acquisition of the gaming company Activision. The remedies guarantee the licensing of Activision games to rival cloud gaming services. The EC said:

“These commitments fully address the competition concerns identified by the Commission and represent a significant improvement for cloud game streaming compared to the current situation.”

Meanwhile the CMA, in its decision to block the same deal, has articulated a message it has been trailing for a while: these sort of commitments don’t work because they replace the organic and dynamic process of competitive rivalry with a stilted and inflexible regulatory solution. The CMA’s rebuttal to the EC (in a rather spicy twitter thread) was clear:

“they would replace a free, open and competitive market with one subject to ongoing regulation of the game Microsoft sells, the platforms to which it sells them, and the conditions of sale”

In other words, “Yuck”, says the CMA. Given me market forces over regulatory supervision any day.

The CMA’s position has a noble free-and-open markets philosophy behind it. Let competition do its work, rather than bend over backwards to permit a deal which needs lots of regulatory cosmetic surgery to make it work. On Saturday CMA Chairman Marcus Bokkerink wrote an op-ed in the Telegraph newspaper making the same point “for the CMA the starting point is prevention, not regulation” – with a nod to their decision to prevent the supermarket merger between Sainsbury’s and Asda.

Who’s got it right? Can behavioural access-type remedies work to preserve competition in cases like this? After all, it is worth noting that several of the new regulatory tools in the DMCC bill currently before UK parliament aim to safeguard digital competition in exactly this way, through opening up access to rivals.

Why do behavioural remedies work to prevent foreclosure?

Here is how I would describe the logic behind the CMA’s position:

– I have shown you have the ability and incentive to do a bad thing

– You are committing not to do the bad thing

– I have looked at your commitment and agree it stops you doing some bad things

– But there are other things it might not stop you from doing, and some of those might be bad – so the commitment isn’t good enough

The EC sees it the same way at steps 1 and 2. But then it diverges, forming a view that the commitments not only prevent the bad things from happening, but that they will make things even better.

In my view, the EC has economics on its side in this disagreement.

Foreclosure strategies all rest on a cost benefit calculus: whether I gain more from foreclosing rivals than it costs me to harm them. That’s why there is both an ability and an incentive limb to the legal test. Many firms have the ability to harm their rivals without having the incentive to do so. Change the cost benefit calculus and you change the incentives. Change incentives and you fundamentally change the situation.

This is what a (good) behavioural remedy in a foreclosure case does. It changes the incentive calculus from one where foreclosure makes sense to one where it does not. Once the incentive to foreclose has been eliminated, the situation has fundamentally changed.

The critical point to understand (and where the CMA has it wrong in its analysis) is that a behavioural remedy doesn’t need to be 100% comprehensive in order to be 100% effective. This is because even an imperfect remedy can be more than sufficient to bring about a fundamental change in the incentive to foreclose (thereby entirely eliminating the foreclosure concern). It’s like approving a vaccine that is 90% effective – it might still fundamentally change the calculus in defeating the spread of a virus.

Suppose I’ve run my SLC analysis and found that the benefits of foreclosure outweigh any costs by 2:1 – a very clear cut case. So when I shut of sales to my rivals, for every £1 I lose upstream, I expect to gain extra £2 in profits downstream.

Now suppose I put a remedy in place that covers 50% of the market – so half of my rivals are protected and the other half remain vulnerable to foreclosure.  What happens?

– For the group of rivals who have protection from the remedy, nothing happens – I continue to sell to them, and they continue to compete downstream. No gains and no losses to me.

– What about the group of rivals who have no protection? I still have the ability to foreclose them. Do I still have the incentive? My losses form shutting off sales to them are the same as before. But the gains have changed. I can’t count on 2:1 returns anymore – because even if I succeed in foreclosure, their customers now have a choice between diverting to me, or diverting to other rivals who are protected by the remedy.

If you take a simply example where I have a 30% market share downstream, and where customers divert in proportion to market shares, then the foreclosure calculus turns negative. For every £1 lost from foreclosure, I would only expect to get £0.92 back downstream. That’s because more than half of the foreclosure gains now get diverted to my rivals.

So even with this extreme example (very strong foreclosure incentives, and a remedy with very weak coverage) you still bring about a fundamental change in the incentive to foreclose. Once the incentive is gone, as a regulator I don’t need to worry about the terms you will set in situations outside the scope of the commitments – because you will be incentivised to reach a deal on commercial terms. The remedy has partial coverage, but it works to protect everyone.

This logic is familiar to anyone who works in telecoms regulation, where wholesale access remedies are rarely – if ever – designed to be comprehensive and have full market coverage. Instead they function as a “regulatory backstop” that put incumbents in a position where they are incentivised to do commercial deals, because they know foreclosure isn’t a feasible option.

When do behavioural remedies work to prevent foreclosure?

One can take this thinking a step further, and look at the ‘critical’ level of market coverage needed for a behavioural remedy to reverse foreclosure incentives. The required coverage of the remedy fundamentally depends on two factors:

– How strong the pre-remedy incentive to foreclose is

– How strong the market position of the downstream merging party is

Both of these factors are measurable (and will already have been measured in the standard SLC analysis). The table below shows how this critical ‘coverage’ level for a behavioural remedy changes according to:

– The ratio of gains to losses from foreclosure

– The downstream market share of the merged firm

It shows that one needs both very strong foreclosure incentives and a very strong downstream position before anything like full coverage of a remedy is needed.

Critical level of coverage for a behavioural remedy under different scenarios:

Now it might not be that remedies lack 100% coverage simply because they apply to some rivals and not to others. It might not be that clear cut: instead it could be that the remedy applies to all rivals but that it loses complete coverage in other ways – e.g. because of changes to the relevant product that make the commitments less than fully watertight. Whatever the exact mechanism, I would argue that the same broad conceptual framework can apply – testing whether the efficacy of the remedy is sufficient to clearly undermine the commercial incentive to foreclose.

Shooting the moon

A foreclosure strategy turns out to be a bit like shooting the moon. Instead of competing on the merits you go for the alternative strategy of engineering an out right win. But to succeed you generally need the strategy to work 100% – in the language of Pokémon “you gotta catch ‘em all”. If you miss one or two of the hearts – or even worse if you are left missing the Queen of Spades – your situation flips quickly from an outright win to a very expensive loss. And once you realise some of the hearts are gone, the right strategy isn’t to carry on trying to shoot the moon – the losses will just get worse in the end. Once that option is lost, it makes sense to do an immediate 180 and go back to playing the game normally.

This is why a good behavioural remedy doesn’t necessarily fit the description given by the CMA as something where free market outcomes are replaced by fixed regulatory terms. If the coverage of the remedy is sufficient, it should flip incentives in a way that means market outcomes are back to being dictated by competitive forces, not by the regulatory backstop.

More importantly, sufficient coverage in this context doesn’t need to be 100%. And so the CMA’s focus on its absolute comprehensiveness isn’t right. If were in the business of appealing that decision, I would certainly look to argue that there is a flaw in the reasoning: because the CMA should have gone back and revisited whether its incentive analysis remained valid in a post-remedy world. Rather than only focusing on whether the remedy addressed every conceivable channel of ability.