Antitrust authorities are increasingly interested in pricing algorithms

A few months ago, we considered whether pricing algorithms might be the European Commission’s next antitrust target (here). The Commission had warned companies about the risks of using algorithms to collude, and indicated that pricing software formed part of its investigation into consumer electronics.

Since then, the Commission has published its Final Report on the E-commerce Sector Inquiry (“Final Report”, see our commentary here). This specifically identifies the wide-scale use of pricing software as something that may raise competition concerns, noting that the “availability of pricing information may trigger automatised price coordination” (13).  The Final Report also claims that comments from retailers point to manufacturers making use of retail price maintenance.  It suggests that pricing software may make it easier for manufacturers to retaliate against retailers that deviate from desired pricing, which may even limit the incentive for retailers to deviate from pricing recommendations in the first place.  The Final Report also considers that pricing algorithms could facilitate or strengthen collusion between retailers – the algorithms make it easier for retailers to detect any deviations from prices implemented under a collusive agreement (33).

At the start of this month the Commission announced an investigation into whether the clothing manufacturer and retailer Guess illegally restricts retailers from selling cross-border within the EU.  It seems likely that this investigation arose out of information the Commission received through the E-commerce Sector Inquiry. Bearing in mind the Commission’s comments on pricing algorithms in the Final Report, there is real potential for subsequent investigations into their use.   

OECD Roundtable

Further insights into the Commission’s current thinking can be drawn from the OECD Roundtable currently taking place on ‘Algorithms and collusion’ (see here).  The Commission's contribution paper identifies the potential harmful effects that pricing algorithms may have in both vertical and horizontal contexts, namely by facilitating collusion and making collusion easier to enforce.  It expands upon the concerns identified in the Final Report, offering by way of example a scenario where retailer A is adhering to retail price maintenance, and retailer B is monitoring and matching retailer A’s prices using pricing software (16).  This is said to show how artificially high prices caused by retail price maintenance can easily spread to other ‘innocent’ retailers through the use of pricing software.  

The paper notes that where firms are using algorithms to engage in explicit collusion, it is clear that the firms are still liable for their behaviour.  It suggests that to a large extent, pricing algorithms can be analysed under traditional EU competition law.  However, it also spends some time discussing the issue of tacit collusion, where there is no anti-competitive agreement involved and therefore the conduct of non-dominant companies acting independently falls outside the EU competition law framework.  Does algorithmic pricing make tacit collusion more pervasive and more effective?  If so, how should competition authorities respond? 

As the paper recognises, this is an area of on-going debate.  Nevertheless, it considers potential options such as whether the market itself may correct ‘algorithm-enabled tacit collusion’ through the development of ‘consumer algorithms’ that could track prices and even identify ‘maverick’ sellers not engaging in algorithmic pricing that consumers could purchase from.  It also explores whether changes to the law on tacit collusion might be effective, or whether the interpretation of ‘communication’ should be expanded in order to bring algorithm-enabled price matching within the scope of Article 101 TFEU.

The UK CMA's contribution to the Roundtable has less of a focus on pricing algorithms. It identifies a few potential theories of harm that may apply to the use of algorithms more broadly:

  • Facilitating the implementation or maintenance of a collusive agreement.
  • Facilitating behavioural discrimination (e.g. price discrimination where consumers are set individual prices based on algorithmic assessment of the highest price that consumer is likely to pay).
  • Reinforcing dominance or raising barriers to entry (relating to the typical requirement for large volumes of data to make an algorithm effective).
The CMA recognises that algorithms can give rise to many consumer benefits. However, it also notes that they could lead to competitive or consumer harm in novel, untested ways, and that it is challenging to detect and understand the exact effect of complex algorithms, particularly given that they rapidly evolve (whether through constant refinement from developers or via ‘self-learning’).

Despite this, the CMA appears to be more comfortable than the Commission in its ability to combat this sort of issue. It notes that “the flexible, principles-based UK competition law framework has to date shown itself able to accommodate technological change, and to be capable of flexible and effective use to tackle a wide range of novel competition harms” (43). It plans to invest in in-house technological expertise and new digital forensic tools in response to the challenges posed by the use of algorithms.

Conclusion

These two OECD contribution papers were prepared for a Roundtable discussion – they do not reflect the Commission’s or the CMA’s official stance on pricing algorithms.  However, they provide an interesting insight: these authorities are clearly paying a good deal of attention to the potential competition issues raised by algorithms.  Coupled with the increasing enforcement activity by the Commission in the e-commerce sector, pricing algorithms continue to be one of the trending topics in competition law in Europe. 

Unwired Planet v Huawei: a new FRAND injunction

Mr Justice Birss has once again broken new legal ground by granting what he has termed a ‘FRAND injunction in Unwired Planet v Huawei.

As a reminder, in April Mr Justice Birss handed down the first UK court decision determining a FRAND royalty rate (see here). A post-judgment hearing took place in May to establish whether or not Huawei should be subject to an injunction in the UK and the issue of permission to appeal.

The FRAND injunction

At the post-judgment hearing, Huawei had argued that the judge should not grant an injunction. As Huawei intended to appeal the decision, it said that it could not enter into the FRAND licence agreement at this stage, in case the Court of Appeal determined that different FRAND terms were appropriate. Huawei claimed that to grant an injunction now would effectively be punishing it for exercising its right to appeal. It also noted that if an injunction was granted, it would last until 2028 (when the patent found valid and infringed in the first patent trial expired), despite the FRAND licence agreement expiring in 2020. Therefore, Huawei would be forced to negotiate a new licence from an extremely weak position – it would automatically be injuncted if terms could not be agreed. 

Huawei requested that the judge accept undertakings in lieu of granting an injunction, offering to: (a) enter into the licence following its appeal, and (b) to comply with the terms of the licence as if it was in effect (including paying royalties) until its appeal was finished.

Mr Justice Birss essentially took the view that the offer of undertakings now was too little, too late. He decided that an injunction should be granted. However, he recognised the risk that this might affect negotiations or disputes about the terms of the licence in later years. To resolve this, he granted a new kind of injunction, which he called a “FRAND injunction”. This would be like a normal injunction, but with the following extra features:

  • A proviso that it would cease to have effect when the defendant enters into a FRAND licence; and
  • Express liberty to return to court to decide whether the injunction should take effect again at the end of the FRAND licence (if it ends before the relevant patents expire, or ceases to have effect for any other reason).
The injunction is to be stayed pending the result of the appeal, on terms that provide for appropriate royalty payments from Huawei to Unwired Planet in the meantime.

Permission to appeal 

Mr Justice Birss granted Huawei permission to appeal on three main issues:

  1. The global licence: including: (i) whether more than one set of terms can be FRAND, (ii) whether a UK only licence was FRAND, (iii) whether the court is able to determine FRAND terms, including rates, for territories other than the UK, and (iv) whether it is appropriate to grant an injunction excluding Huawei from the UK market unless it took a global licence.
  2. Hard-edged non-discrimination: Huawei have permission to appeal the finding that a distortion of competition is required for the non-discrimination aspect of FRAND to apply, but not whether or not there was a distortion of competition in this case.
  3. Huawei v ZTE (Article 102 TFEU): regarding the judge’s findings on abuse of dominance and injunctive relief.
This permits a fairly wide-ranging appeal, especially as regards the competition law elements of the latter two issues. The trial judgment appeared to downplay the importance of competition law in FRAND issues (see here for more information); the appeal may enable a renewed focus on it.

The FRAND licence 

In his main trial judgment, Mr Justice Birss settled the terms of the FRAND licence to be entered into by Unwired Planet and Huawei. This latest judgment annexes a copy of the final form of that licence. Given the shroud of secrecy that usually surrounds such patent licence agreements, this is a unique insight, reflective of the judge’s desire throughout the case to ensure as much transparency as possible.

Transparency is likely to be helpful as the law in this area continues to develop. With the advent of new technologies developed for 5G and the Internet of Things, new companies may need to enter the FRAND licensing field for the first time. Without being able to draw upon any previous experience of negotiating licences in this area, they will be at a disadvantage in negotiations. 

If other judgments follow Mr Justice Birss’ lead and annex copies of any FRAND agreement determined by the court, these would provide useful points of reference for negotiating parties. It might also reduce the requirements for third party disclosure (a costly, time-consuming exercise) in any subsequent litigation. Otherwise, such disclosure will be essential in FRAND cases involving relatively new entrants to the market – they are unlikely to have many licence agreements that can be used by the judge as comparables as part of the process for determining a FRAND rate.

Conclusion

Yet again, Mr Justice Birss provides plenty of food for thought. Assuming that Huawei does go ahead with its appeal, it will be fascinating to see how the Court of Appeal responds to these issues.

Pat Treacy and Matt Hunt



Transparency may undermine online competition: Commission’s Final Report on the E-commerce Sector Inquiry

On 10 May 2017 the European Commission published its Final Report on the E-commerce Sector Inquiry, together with accompanying Q&As, and, for those who want something rather longer, a Staff Working Document

The inquiry, launched over 2 years ago, and part of the wider Commission Digital Single Market Strategy (see our earlier comment here) has gathered evidence from nearly 1,900 companies connected with the online sale of consumer goods and digital content.
The Report’s main findings

  • Price transparency has increased through online trade, allowing consumers instantaneously to compare product and price information and switch from online to offline. The Commission acknowledges that this has created a significant ‘free riding’ issue, with consumers using the pre-sales services of ‘brick and mortar’ shops before purchasing products online. 

  • Increased price transparency has also resulted in greater price competition both online and offline.  It has allowed companies to monitor prices more easily, and the use of price-tracking software may facilitate resale price maintenance and strengthen collusion between retailers.

  • Manufacturers have reacted to these developments by seeking to increase their control of distribution networks though their own online retail channels, an increased use of ‘selective distribution’ arrangements (where manufacturers set the criteria that retailers must meet to become part of the distribution system) and the introduction of contractual restrictions to control online distribution.
How about changes to competition policy? 

The Report does not advocate any significant changes to European competition policy, but rather confirms the status quo. The key point of interest are as follows: 

  • Selective distribution – whilst the Commission has not recommended any review of the Vertical Block Exemption Regulation (‘VBER’) ahead of its scheduled expiry in 2022, the Commission notes that the use of selective systems aimed at excluding pure online retailers, for example by requiring retailers to operate at least one ‘brick and mortar’ shop, is only permissible where justified (for example in respect of complex or quality goods or to protect suitable brand image).

  • Pricing restrictions – dual pricing (i.e. differential pricing depending on whether sales are made online or through a bricks and mortar outlet) will generally be considered a ‘hardcore’ (or object) restriction of competition when applied to one and the same retailer, although it is capable of individual exemption under Article 101(3) TFEU, for example if the obligation is indispensable to address free-riding by offline stores.  

  • Restrictions on the use of marketplaces – the Report finds that an absolute ban on the use of an online marketplace should not be considered a hardcore restriction, although the Commission notes that a reference for a preliminary ruling is pending before the CJEU (C-230/16 - Coty Germany v Parfümerie Akzente).

  • Geo-blocking – a re-emphasis of the existing position on territorial and customer restrictions – active sales restrictions are allowed, whereas passive sales restrictions are generally unlawful. Within a selective distribution system, neither active nor passive sales to end users may be restricted. The Commission also make clear that companies are free to make their own unilateral decisions on where they choose to trade.

  • Content licensing – the significance of copyright licensing in digital content markets is noted, as is the potential concern that licensing terms may suppress innovative business practices.  

  • Big Data – possible competition concerns are identified relating to data collection and usage. In particular, the exchange of competitively sensitive data (e.g. in relation to prices and sales) may lead to competition problems where the same players are in direct competition, for example between online marketplaces and manufacturers with their own shop.  
What happens next?

The Commission has identified the need for more competition enforcement investigations, particularly in relation to restrictions of cross-border trade.  It is expected that more investigations will be opened in addition to those already in play in respect of holiday bookings, consumer electronics and online video games. In a more novel approach, the Commission’s press release also name-checks a number of retailers (in particular in fashion) who have already reformed their business practices “on their own initiative”.
  
The Commission also highlights the need for a consistent application of the EU competition rules across national competition authorities.  It remains to be seen whether the Commission will seek to use its enforcement investigations to address inconsistencies such as those evident in the more interventionist stance of some national authorities (e.g. the Bundeskartellamt) in respect of issues such as pricing restrictions.

The Commission steps into the excessive pricing arena

An announcement by the European Commission last week resolves an open question about its view on the recent spate of pharma sector excessive pricing cases that have been seen in Italy and the UK.  The Commission has now confirmed that, following dawn raids across four member states in February, it has opened an investigation into Aspen Pharma for suspected breach of Article 102.  The concern is that Aspen’s pricing practices in relation to off patent drugs containing five ingredients used for treating cancer has led to unjustified price increases.  This overlaps with the Italian Market Competition Authority’s decision in October 2016 to fine Aspen €5.2 million for unfair price increases, which covered four of the five ingredients now under investigation (see here for our report).

This is the Commission’s first excessive pricing case in the pharma sector, following a trend set by the national competition authorities (including in the UK and US – although given that US antitrust does not apply to pure pricing issues, the US cases have tended to focus on another form of abuse (e.g. here) which led to the excessive prices).  

The EU NCAs have been well placed to deal with such conduct, as pharma markets are national in scope, and subject to significant regional differences resulting from the different formation of public health services.  However, setting the benchmark price is a difficult – and controversial – aspect in any investigation (see here for our thoughts on the CMA’s recent approach) and it will be of interest to see how the Commission tackles the issue, as its approach may well be followed by other NCAs.  

Bearing these points in mind, we will be particularly keen to see how the Commission deals with the following two points: 

  1. The definition of the relevant market and whether a company is dominant – before assessing whether its prices are excessive, a pharma company must first know whether it is dominant.  With diverging approaches on product market definition (definitions have been drawn from therapeutic / molecular / dosage level and from regulatory guidance), it can be difficult to make an assessment of dominance.  In Aspen’s case, it has found itself  one of few companies willing to manufacture low volume generic drugs, and despite low barriers to entry, no other companies have entered the market to exert a form of price control on Aspen.  It has perhaps therefore become dominant as a result of market failures.    
      
  2. How the Commission determines an acceptable level of profit (i.e., what is the meaning of ‘excessive’?).  While under patent protection, pharmaceutical product prices are generally constrained in some way (e.g. through profit caps under the UK PPRS), but in theory, profits could be competed upwards following patent expiry, even if overall prices decline (this is a key part of the argument raised by Pfizer and Flynn in their appeals of the CMA infringement decision).  The recent opinion of AG Wahl considering unfair prices (albeit in a copyright licence context) concluded that there is no single method of determining the benchmark, and acknowledges that there is a high risk of error, but a price should only be excessive if it is significantly and persistently above whatever benchmark is determined.  Whilst AG Wahl was unable to point to any guaranteed failsafe methods of analysis, he stated that an authority should only intervene when there is no doubt that an abuse has been committed. 

The investigation does signal that the Commission is keen to address the fairness of pricing in the pharmaceutical industry, but as with all such investigations, its approach should not be one of a price regulator.  Indeed, the Commission is at pains to point out that it is looking at a case where the price increases were extremely significant (100s of percent uplift).  It may reveal that a case-by-case approach is not appropriate, and the real issue is regulatory failure that will need to be corrected by legislation (such as that currently before the UK Parliament). 

Patent licensing: 5G & the Internet of Things

The next telecommunications standard, 5G, and the nascent Internet of Things (IoT), promise a world of high-speed interconnectivity. We’re already accustomed to people talking to smart devices to ask them to play music, or to order a taxi or takeaway. The technology for truly smart homes and even smart cities is following closely (see here for examples of how the IoT is already being used).

Standardisation will be essential to maximising the benefits of both the IoT and 5G. After all, there’s no point in having a smart thermostat that can be adjusted remotely if you can’t connect to it whilst out of the house because your phone is made by a different manufacturer. Standardisation will ensure that 5G/IoT devices and systems can connect and work together.

There will be thousands of patents essential to the operation of the standards developed. As the IoT grows and 5G is rolled out, the issue of how these patents are licensed will become increasingly important. Standard essential patents (SEPs) have to be licensed on a FRAND basis, but determining a FRAND royalty rate is a challenging task.

The Commission’s Roadmap

Following two studies on SEPs published at the end of last year (see here), on 10 April 2017, the Commission released a Roadmap that sets out its plan to publish a ‘Communication on Standard Essential Patents for a European digitalised economy’ later this year, possibly as early as May or June. The Communication (which will not have the binding force of a Directive or Regulation) is intended to complement the Commission’s Digital Single Market project, and to help work towards the goal of having 5G rolled out across the EU by 2025. 

The Roadmap identifies three main issues that the Commission will seek to address in its Communication to ensure a balanced, flexible framework for SEP licensing:

  • Opaque information about SEP exposure: the lack of effective tools for potential licensees too identify which patents they need to take licences for in order to implement relevant standardised technologies.
  • Unclear valuation of patented technologies: the difficulties in assessing the value of new technology (for both licensors and licensees), including the lack of any widely accepted methodology.
  • Risks of uncertainty in enforcement: the general framework provided by the CJEU in Huawei v ZTE is a starting point for agreeing a FRAND royalty rate, but it does not provide complete guidance. There are many technical issues that aren’t addressed, such as how portfolio licensing, related damages claims, and ADR mechanisms should be dealt with. (NB: some of these issues have been recently addressed in the UK case Unwired Planet v Huawei, see our initial thoughts on that case here.)

It is unclear if the Communication will address other issues with standardisation and SEP licensing, such as over-declaration, hold-up, hold-out, the appropriate royalty base (a particularly difficult challenge in the diverse world of the IoT) or whether a total aggregate royalty burden is appropriate.

Use-based licensing?

The Roadmap does not offer any specific details as to how the Commission intends to solve the issues it identifies. However, according to MLex (here – subscription required), an outline of the Commission’s Communication seen in February suggested that the Commission was considering a licensing model that would enable licensors to offer licensees different royalty rates depending on how the relevant technology is used. This could extend to allowing licensors to refuse to offer a licence if the final use of the technology cannot be identified and tracked.

This suggestion has caused concern amongst a number of companies. ACT | The App Association, an organisation sponsored by Apple, Facebook, Microsoft, PayPal & others which represents more than 5,000 small technology firms, has written to the Commission claiming that such a licensing model poses a substantial threat to innovation. It argues that in order for suppliers to obtain licences under this model, they might be required to monitor their customers’ business practices and potentially to charge different customers different prices depending on how they use the technology. It also claims that this model could appropriate the value created by new innovators. If a company succeeds in developing a new use for a particular sensor by incorporating it into a health app for example, it might find itself being charged higher royalties for this new use of the sensor.   

We have reported before on how new licensing models may be required to make best use of the IoT. Qualcomm, Ericsson and Royal KPN (among others) have backed a new licensing platform called Avanci. This is designed to remove the need for lengthy bi-lateral licence negotiations by making flat rate FRAND licences available for particular patent portfolios. However, it is now over six months since Avanci launched and there have been no reports yet of it successfully concluding any licence agreements.

The proper royalty base for patent licences has been a controversial topic for a number of years now. There has been considerable debate over whether licences should be based upon the price of the smallest component in which the patent is implemented, or the final price of the end product. This issue has been addressed by the courts on occasion, particularly in the US (see our post on CSIRO v Cisco here for example).

The new Communication is intended to provide best practice guidance on SEP licensing. If the Commission does opt for a use-based licensing model, this would be a controversial choice. However, whatever the Commission decides, given the number of conflicting interests and amounts of money at state, it is unlikely to satisfy everyone. 

Unwired Planet v Huawei: Is FRAND now a competition law free zone? Not so fast…

It has been two weeks since Mr Justice Birss handed down his latest judgment in Unwired Planet v Huawei (see here for a summary), which is almost long enough to get to grips with the 150 or so pages. There has already been a huge amount of discussion as to what this judgment means in practice and we have even overheard some suggest that, when it comes to FRAND in the future, we can simply ignore competition law altogether. This week we were invited by our friends at the renowned IP law blog, IPKat, to have our say on this. You can check out our thoughts on the IPKat blog here.

Unwired Planet v Huawei: UK High Court determines FRAND licence rate

Mr Justice Birss has just handed down the first decision by a UK court on the ever controversial topic of what constitutes a FRAND royalty rate.  At well over 150 pages, the judgment covers a lot of ground: a lot of ink is likely to be spilled about it over the coming weeks and months.  From what we’ve seen so far, the judge has not been afraid to make findings that will have a considerable impact on licensing negotiations in the TMT sector. 

We’ve summarised the headline conclusions below, but also keep an eye out for future posts in which we’ll analyse some of the judge’s findings and reasoning in more detail.

Background

In March 2014, Unwired Planet (“UP”) sued Huawei, Samsung and Google for the infringement of 6 of its UK patents.  Five of these were standard essential patents (“SEPs”) that UP had acquired from Ericsson.  They related to various telecommunications standards (2G GSM, 3G UMTS, and 4G LTE) for mobile phone technology. 

Five technical trials, numbered A-E, were listed on the validity and infringement of the patents at issue.  These were to be followed by a non-technical trial on competition law and FRAND issues.  UP’s patents were found valid and infringed in both trial A and trial C, but two were held invalid for obviousness in trial B. Trials D and E were then stayed, and as Google and Samsung had settled with UP during the proceedings, this just left Huawei and UP involved in the 7 week non-technical trial, for which judgment has just been given. 

Judgment

There’s a lot to unpack in this judgment, but here is a short list of what we think are the most important findings:

General principles:
  • There is only one set of FRAND terms in a given set of circumstances.  Note the contrast between this and the comments of the Hague District Court in the Netherlands in Archos v Philips (here, in Dutch) which seem to interpret the CJEU decision in Huawei v ZTE as meaning that there can be a range of FRAND rates.
  • Injunctive relief is available if an implementer refuses to take a FRAND licence determined by the court. Mr Justice Birss indicated that an injunction would be granted against Huawei at a post-judgment hearing in a few weeks’ time (although presumably Huawei can avoid this by now taking a licence on the terms set by the Judge).
  • UP is entitled to damages dating back to 1 January 2013 at the determined major markets FRAND rate applied to UK sales. 
  • What constitutes a FRAND rate does not vary depending on the size of the licensee.
  • For a portfolio like UP’s and for an implementer like Huawei, a FRAND licence is worldwide.
  • It’s still legitimate to make offers higher or lower than FRAND if they do not disrupt or prejudice negotiations.
Abuse of dominance:
  • UP did not abuse its dominant position by issuing proceedings for an injunction prematurely (it began the litigation without complying with the Huawei v ZTE framework).
Calculating the FRAND rate:
  • A FRAND royalty rate can be determined by making appropriate adjustments to a ‘benchmark rate’ primarily based upon the SEP holder’s portfolio. 
  • In the alternative, if a UK-only portfolio licence was appropriate, an uplift of 100% on the benchmark rates would be required.
  • Counting patents is the only practical approach for assessing the value of sizeable patent portfolios, although it may be possible to identify a patent as an exceptional ‘keystone’ invention.
  • Comparable, freely negotiated licences can be used as to determine a FRAND rate.
The FRAND rates as determined:


Other FRAND terms:
  • The Judge goes into some details as to the terms which will be FRAND in the licence between Unwired Planet and Huawei – much of which will be worth reading for licensors and licensees in this field.  Of particular note is the royalty base for infrastructure (excluding services). 
Other remedies:
  • Damages are compensatory and are pegged to the FRAND rate.
Comment

There have been near continual disputes between the major players in the TMT field over the last decade or so.  The meaning of FRAND has been strategically important in a large number of cases.  However, many of these companies are very effective negotiators.  In the vast majority of cases, they are able to agree licences without resorting to litigation.  Where proceedings are initiated, the parties are usually able to settle long before a judgment is reached, particularly given the time and expense required to take a FRAND case all the way to trial.  (Such expense is, however, usually dwarfed by the value of the licence – many licences in this field are valued in $billions.) 

The scarcity of judicial opinion in this area means this is a rare opportunity to see how a respected UK judge has approached a number of the unresolved questions regarding FRAND. 

A number of significant questions remain unanswered however, and we will be exploring these in future blog posts.  There’s also the matter of the upcoming post-judgment hearing in a few weeks’ time, which will establish whether or not Huawei will actually be subject to an injunction in the UK, and of course the chance that either party might wish to appeal.  All in all, there’s plenty of interest to talk about, plenty of advice to be given to clients, and the FRAND debate will undoubtedly continue on.

EU Commission’s Microsoft / LinkedIn Decision – watershed for competition and data?

On 6 December 2016, the European Commission approved the acquisition of LinkedIn by Microsoft, conditional on compliance with a series of commitments.  The full text of the decision has recently been published, affording some useful insight into the Commission’s reasoning.

The merger is one of a number of high profile technology cases in which data is the key asset. Cases such as this are challenging the Commission’s relaxed attitude to the potential effects on competition of deals involving significant volumes of data (for example, the Commission’s 2014 clearance decision of Facebook’s acquisition of WhatsApp – now the subject of an investigation into whether Facebook provided misleading information in the context of that merger review).  

Similarly, in the LinkedIn / Microsoft decision, the Commission’s assessment was that the post-merger combination of data (such as the individual career information, contact details and professional contacts of users) did not raise competition concerns.

The Commission identified two potential concerns: 

  1. The combination of data may increase the merged entity’s market power in the data market or increase barriers to entry / expansion for competitors who need this data in order to compete – forcing them to collect a larger dataset in order to compete with the merged entity; and 
  2. Even if the datasets are not combined, the companies may have been competing pre-merger on the basis of the data they control and that this competition could be eliminated by the merger. 
These concerns were dismissed by the Commission on a number of grounds, the most interesting being that the combination of their respective datasets is unlikely to result in raising the barriers to entry / expansion for other players as there will continue to be large amounts of internet user data available for advertising purposes which are not within Microsoft’s exclusive control.

The Commission’s approach contrasts with that of some commentators (and indeed some of the Commission’s own non-merger enforcement activities) which have highlighted the potential for platforms to gain an unassailable advantage over competitors in relation to data. 

Concerns of data ‘tipping points’ were among the reasons why French and German competition authorities have published a joint paper on data and competition law. 

Germany has amended its domestic competition law to increase the legal tools available to prevent market dominance and abuses in relation to data. These changes will come in to force later this year and include: 

  1. controversially) amending the German merger thresholds to require notification of deals involving innovative companies (like start-ups) with a transaction value of EUR 400 million; and
  2. introducing specific criteria for reviewing market power in (digital) multi-sided markets, for example allowing the Bundeskartellamt (BKA) to consider: concentration tendencies; the role of big data; economies of scale; user behaviour; and the possibilities to switch a platform.
The additional merger threshold is intended to allow the BKA to review mergers in which the transaction value is high but the parties’ turnover in Germany is below the existing EUR 25 million threshold; for example, when Facebook’s acquisition of WhatsApp for USD 22 billion was not notifiable in Germany (although it was reviewed by the Commission). 

France and Germany’s robust approach to competition concerns in relation to data is in contrast with the less interventionist position in the UK. This is demonstrated by recent UK government report on digital platforms which found that, “In many sectors, e.g. search engines or social networks, firm behaviour and survey evidence suggests that in the event of even a modest hike in costs users would expect to find an alternative and cease using the service. It is difficult to reconcile this behaviour, and this finding, with the sense that there is an important “moat” which prevents users switching to alternative services over time. Any moat that does exist only seems to be enough to keep them in one place if the platform continues to be free and improve its service over time.

Given the moves towards ex ante regulation of data in France and Germany, and given the ex post investigation into Facebook/WhatsApp, it remains to be seen whether future merger investigations will take a similarly permissive approach.

Will pricing algorithms be the European Commission’s next antitrust target?

There has been considerable debate over the last year or so about the potential anti-competitive impacts of pricing algorithms. They could lead to discriminatory pricing, for example a company quoting different prices to different people based on an algorithmic analysis of their personal data, or cases of collusion, for example companies using algorithms to automatically fix prices. 

In a recent speech, Commissioner Vestager sounded a clear warning against the latter example: “companies can’t escape responsibility for collusion by hiding behind a computer program”. She also indicated that the use of pricing software forms part of the issues being investigated in the Commission’s new investigation into price-fixing in consumer electronics.

However, as pricing algorithms increase in complexity and sophistication, and their use becomes more prevalent, it will not be easy for competition authorities to establish where the use of such algorithms equates to actionable infringements of competition law.

How might pricing algorithms be used?

A new book by Professors Ariel Ezrachi and Maurice Stucke have identified four scenarios in which pricing algorithms may promote anti-competitive collusion (see here; also developed in more detail in their book Virtual Competition). 

The first is where firms collude as in a traditional cartel, but use computers to manage or implement the cartel more effectively, or to monitor compliance, for example by utilising real-time data analysis. Competition authorities have already investigated this kind of subject-matter – for example, the CMA issued an infringement decision last year against two companies that agreed to use algorithms to fix prices for the sale of posters and frames on Amazon (see here).

The second example is a hub-and-spoke scenario whereby one pricing algorithm may be used to determine prices charged by numerous users. Evaluating this sort of issue is a current challenge for competition authorities. Last year, in Eturas, the CJEU held that travel agents participating in a platform that implemented a discount cap could be liable if they knew about the anti-competitive agreement and failed to distance themselves from it (see here).  An ongoing case in the US (Meyer v Kalanick) is examining Uber’s ‘surge’ pricing algorithm, which increases the price of an Uber journey as demand increases.  The claimants allege that this constitutes an implied horizontal price-fixing agreement.  

The examples seen so far involve relatively straightforward cases of the use of algorithms as an aid or means to fix prices (although the Uber example arguably involves only unilateral conduct, rather than collusion).  However, Ezrachi and Stucke’s final two scenarios move into more uncertain territory – what if there is no express collusion by the companies? 

In the third scenario, each firm independently adopts an algorithm that continually monitors and adjusts prices according to market data. Although this can lead – effectively – to tacit collusion, particularly in oligopolistic markets (those with a small number of sellers), there is no agreement between companies that could form the basis of an investigation.  However, there can evidently be an anti-competitive effect: if an online retailer can track the prices used by another online retailer for common products, and immediately adjust its own prices to match any discounts, it can prevent the second online retailer from gaining a reputation for lower prices. The incentive for either retailer to lower its prices is removed.  On the other hand, examples from the analogue world suggests that this kind of market review can be used to ensure lower prices for consumers, at least for now (think supermarkets’ price match promises…).

In the fourth scenario, machine learning and the increasing sophistication of algorithms expand tacit collusion beyond oligopolistic markets, making it very difficult even to detect when it’s happening.

The latter two examples pose obvious difficulties for competition authorities. If they do consider such actions to be anti-competitive, how would they prove the requisite intention to co-ordinate prices?

How will competition authorities react?

As discussed above, competition authorities have already undertaken investigations against companies using pricing algorithms in collusion. We have previously noted the CMA’s interest in developing digital tools to aid its investigations (here). It seems certain that such tools will be necessary as these algorithms become more sophisticated and harder to detect.

The actions of non-dominant companies in using pricing algorithms whilst acting independently do not fall within the current competition law framework, even if such use ultimately results in higher prices for consumers. Commissioner Vestager has accepted that “what matters is how these algorithms are actually used”. This sensibly suggests that for now the Commission’s focus will remain on the more clear-cut cases of collusion. Anything else is arguably a matter for policy and regulation rather than enforcement by competition authorities.

However, Commissioner Vestager also stated that “pricing algorithms need to be built in a way that doesn’t allow them to collude”, suggesting that they needed to be designed in a way that will oblige them to reject offers of collusion. It is unclear whether this means Commissioner Vestager intends to target the use of pricing algorithms more generally, or simply to drive home that the competition rules apply equally where collusion is achieved algorithmically.

The fourth scenario, where machine learning algorithms tacitly collude to fix prices, does sound speculative. However, recent developments such as Carnegie Mellon’s Liberatus beating four of the world’s best professional poker players (here) and Google Deep Mind’s AlphaGo victory against Lee Sedol (here) indicate that it might not be too far from becoming reality in the near future.

Collusion in the online economy – new competition law traps for the unwary?

We reported last year on the Eturas decision, in which the Court of Justice ruled that technical measures applied on an online platform gave rise to a potentially anti-competitive agreement.  The Lithuanian Court which had referred the matter to the CJEU then went on to consider liability, based on the participants’ knowledge of the relevant facts (for a review of this decision, see here).

But the risks posed by agreements over platform T&Cs are not the only thing for companies to be aware of.

The European Commission is now carrying out active enforcement in relation to geo-blocking, which can be achieved primarily through technical measures.  The Steam video games investigation is looking in particular at whether anti-piracy measures have an anti-competitive effect. 

Meanwhile, the CMA last autumn issued a statement noting another practice potentially raising antitrust concerns.  This concerned agreements restricting the use of paid online search advertising (e.g. through use of Google AdWords).  The CMA suggested that restrictions on bidding for particular ad terms, or on negative matching (identifying terms for which ads should not be shown) may infringe the competition rules.  It appears that the CMA sees this in terms of potential effects on competition, rather than as a new form of object restriction, with the CMA stating that the practices are particularly likely to be problematic “where one or more similar agreements include parties that collectively represent a material share of the relevant markets and, in the context of brand bidding restrictions, as a result of negative matching obligations in relation to brand terms which an advertiser would not negatively match but for the agreement”.   It should therefore not be assumed that such a provision would in fact be restrictive of competition – but it is something which bears watching.  Indeed, the CMA is not the only competition authority to have lighted on this issue – a similar point is under investigation in the United States, where the FTC accuses 1-800 Contacts of “orchestrating a web of anticompetitive agreements with rival online contact lens sellers that suppress competition in certain online search advertising auctions”.

In conjunction with this statement, the CMA also announced a market study into digital comparison tools; it described the study as an opportunity to explore the nature of competition between price comparison websites and their relationship with service providers.  This may lead to further issues in this area; in the meantime, judgment in the Coty case, which considers contractual prohibitions on the use of certain online sales channels, such as price comparison websites, is due from the CJEU in the near future.

And then there’s the risk of good ol’-fashioned collusion, with a modern twist.  One thing that comes to mind is the new attention on the significance of privacy conditions for consumers.  Now that these are recognised as a parameter of competition (see here, for example), is there a risk that exchanging information about planned changes to privacy conditions / other online trading T&Cs, or actually agreeing a common strategy for these could amount to a breach of Article 101 or its national equivalents?   Or that an agreement between separate companies to adopt a common practice on such terms (in particular if it results in less protection for consumers) could amount to active collusion?  These are open questions for now, but companies should remember that – while benchmarking is often sensible – they should ultimately take their own decisions, and keep their own counsel, about such matters.

Coincidentally, the consumer arm of the CMA has just closed an investigation into the online terms and conditions of cloud service providers following changes agreed by a number of companies.  The closure statement notes that “the CMA remains interested in unfair terms and conditions, particularly in the digital economy”.  It should not be assumed that this interest is limited only to the parts of the CMA responsible for enforcement of consumer laws…