Mike Walker: “The UK Facebook/Giphy Case: Taking Dynamic Competition Seriously”

The Network Law Review is pleased to present you with a Dynamic Competition Initiative (“DCI”) symposium. Co-sponsored by UC Berkeley, EUI, and Vrije Universiteit Amsterdam’s ALTI, the DCI seeks to develop and advance innovation-based dynamic competition theories, tools, and policy processes adapted to the nature and pace of innovation in the 21st century. The symposium features guest speakers and panelists from DCI’s first annual conference held in April 2023. This contribution is signed by Mike Walker, Chief Economic Adviser at the Competition and Markets Authority.

***

1. Introduction

There are some frequently mentioned statistics about the number of acquisitions carried out by the large digital platforms and, in particular, about the lack of merger control intervention. For instance, the Furman report cited the statistic of more than 400 acquisitions undertaken by Google, Apple, Facebook, Amazon and Microsoft between 2008 and 2018 and not a single one blocked or seriously remedied.¹ The Lear report into mergers in digital markets made a similar comment with respect to nearly 300 acquisitions by Google, Amazon and Facebook over the same ten year period.² It also highlighted that many of these acquisitions were of very young firms whose long term competitive significance was very hard for competition authorities to assess.³ Valletti⁴ tells us to focus on just two numbers: 1,000 and zero. The former is the number of acquisitions by GAFAM over the 20 year period to 2021. The latter was the number of these mergers that have been blocked by merger control authorities worldwide. It is clear that there is no evidence of merger control authorities making Type 2 errors and blocking pro-competitive mergers. However, it is rather less clear that there is no evidence of under-enforcement and Type 1 errors of allowing bad mergers. It is, for instance, hard to see the Google-Doubleclick merger as anything other than a bad merger that has allowed Google to become dominant across the adtech stack, to the detriment of users.

This context explains why the UK Competition and Markets Authority’s (CMA) decision to block the Facebook/Giphy⁵ merger was so controversial. There had been no previous blocks of GAFAM mergers. This article looks at whether the howls of outrage around the decision were justified. Was it as bad a decision as some suggest (“sophomorically” so according to some)? Or can it teach us something about analysing dynamic competition under significant time and data constraints?

2. Basic facts of the case

Facebook announced the completed acquisition of Giphy on 15 May 2020 for around $315m. Facebook is known as a social media firm, but fundamentally it is an advertising firm. In the UK, it has a very strong position in the market for display advertising. According to the CMA’s 2020 Digital Advertising Market Study, it had a market share of more than 50%.⁶a It faced one other significant competitor, Google, although their offerings were significantly differentiated as Google focused primarily on open display advertising, whilst Facebook focused primarily on advertising on its own sites. At the time, Facebook’s global revenues were more than $80bn.

Giphy provides an online database and search engine that allows users to search and share GIFs on social media and messaging platforms. GIFs are short soundless looping videos that you can add to messages and so on (e.g. to a WhatsApp message). Giphy also provides access to “stickers”, which are animated images on a transparent background over which an image or text can be added. GIFs increase user engagement on social media and so are used by social media platforms to help drive advertising revenue. At the time of the acquisition, Giphy was the leading provider of GIFs. It accounted for two-thirds of GIF searches and it had built a very large user base – in the UK alone, over a billion GIF searches were run by users each month using Giphy’s library. High levels of usage are clearly an initial requirement for any monetisation strategy.

Although Giphy provided GIFs for free to users, including providing free access to its GIF library to social media platforms, Giphy had started monetising GIFs in the US through a novel form of digital advertising, “paid alignment”. This allowed brands to pay to align their GIFs to popular search terms or pin them to Giphy’s trending feed. Whilst the model was still nascent, paid alignment was used by a number of leading consumer brands, was growing in revenues and number of advertisers, and allowed for the monetising of messaging with advertising. Internal Giphy documents indicated that Giphy was planning to expand paid alignment internationally, including in the UK. Facebook ended paid alignment after it acquired Giphy.

3. What were the CMA’s concerns?

Stepping back briefly from this particular case, it is important to understand the wider policy context within which the decision was taken. The CMA has been very clear in recent years that it puts great store on innovation. Whilst static economic efficiency is obviously good for short term consumer welfare, it is innovation that drives long run growth and hence long run consumer benefits. The CMA’s view is that it is competition that drives innovation, not entrenched monopolies. Entrenched monopolies have an interest in maintaining their positions against disruptive innovation and we think that the best way to encourage disruptive innovation is to encourage competition. We also think that the threat of innovation from upstarts is likely to encourage incumbent firms themselves to seek to innovate. In short, there is value in the process of competition over innovation, even if the probability that the particular competitor will innovate successfully or disruptively is low.

Turning to the specifics of the Facebook/Giphy case, we thought that there was value to having Giphy as a competitor to Facebook’s display advertising. Giphy was clearly a small competitor, but it had already built a very large user base (a precondition to being able to compete in digital advertising markets), had a growing US advertising business and and had plans for rolling it out in the UK (including in collaboration with other platforms).⁷ It might well ultimately not have succeeded in its business plan, but it might have and even if it did not, its existence gave Facebook an incentive to respond. We also had concerns that post-merger Facebook might disadvantage its rivals in social media by restricting their access to Giphy’s library: GIFs are a popular feature on social media platforms, increasing user engagement and thereby advertising revenue. There were no efficiencies plausibly claimed for the merger.

Facebook appealed the CMA decision to the Competition Appeals Tribunal (CAT), the specialist competition court in the UK, but was unable to overturn the decision to block the merger.

4. What does the case highlight?

This case highlights a number of issues around how merger control should be applied in dynamic markets. A number of these were covered in the CMA’s revision in 2021 of their Merger Assessment Guidelines.⁸

First, this case should not be taken as indicating a general aversion on the part of competition authorities to mergers involving GAFAM or, more widely, dynamic industries. These industries produce great products that have considerably increased consumer welfare in recent years. The Lear report highlighted that the vast majority of mergers involving Amazon, Google and Facebook were between complementary products and were likely to be pro-competitive.⁹ However, there is a corollary to this recognition: mergers that restrict dynamic competition may well be more harmful to long run consumer welfare than mergers that just restrict short run static competition.

Second, the fact that you produce great products is not a “get out of jail free” card when it comes to merger control. It is quite possible that a firm might benefit from monopoly rents in addition to Schumpeterian and Ricardian rents. Indeed, it is quite possible that within tech, initial Schumpeterian rents lead to Ricardian rents and then monopoly rents, as the combination of Schumpeterian and Ricardian rents gives the firm an entrenched position of market power.

Third, uncertainty is also not a “get out of jail free” card. This is a change from traditional merger control, where authorities have tended to allow mergers where they think there is a significant degree of uncertainty over the likely effect of the merger. In my view, this is simply bad policy. The standard for merger control in the UK (and also traditionally at European level) is a balance of probabilities standard. If the CMA thinks the probability that the merger will create a substantial lessening of competition is more than 50%, then it blocks it or requires remedies. If it thinks that the probability is less than 50%, then it allows it. There is no scope within that framework for deciding to allow a merger because it is all very uncertain. Allowing a merger requires a judgement that the merger will harm competition with a less than 50% probability. Clearing a merger does not allow a difficult judgement about probabilities to be avoided. However, historically that is how merger control authorities have often acted. They have tended to act as if clearing a merger in some way avoided them having to make a difficult judgement.

There is an important point here about Type I and Type II errors. There is considerable focus on avoiding Type II errors: blocking a merger that is pro-competitive. We are often told that we should focus more on avoiding Type II errors than on avoiding Type I errors. The standard argument here is that market forces will correct Type II errors over time, but not Type I errors. There are two points to be made here. First, the balance of probabilities framework does not allow this trade-off to be made. Second, it is far from clear that the argument around Type I and Type II errors is even correct, particularly with respect to digital platforms. In a world of significant network effects and tipping, it may well be that the market is not able to repair the damage done by a bad merger, even in the long run. In such a world, Type II errors can be very costly.¹⁰ The ongoing harm caused by the Google-Doubleclick merger is a good example of this.

Fourth, it is not the case that in blocking the merger the CMA was predicting the future and deciding that Giphy was likely to be a long-term strong competitor to Facebook in digital advertising. It might be, but that is not what the decision was based on. What the CMA was claiming is that the process of competition over innovation between Facebook and Giphy was likely to yield benefits to consumers.¹¹ It would spur both parties to invest and seek to innovate. This dynamic competition would have been lost had the merger been allowed. The CMA distinguishes between future competition and dynamic competition. Future competition can be harmed by an incumbent firm buying a known likely entrant and thus reducing the degree of competition in the future. This is much more focused on knowing the likely outcome that the merger will stop. Dynamic competition is much more about the loss of competition between existing actual or likely competitors who are competing by investing to innovate for the future. There may be no or little product competition between the firms now, but there may be in the future depending on the success of the investments. A merger may reduce the incentive of the incumbent to invest as it removes the possibility of the acquired firm innovating and threatening the incumbent’s future revenues.

Fifth, incentives are evidence. We are frequently told that we do not have the evidence to draw the conclusions we want to. For instance, where are the internal documents detailing what investment the incumbent would undertake in response to competition from the target firm? Well, maybe they do not exist. Maybe they do but have not been submitted. But where there are clear incentives for a firm to respond to a competitive threat, it seems reasonable to draw conclusions on that basis. It is not as if this is a new approach to merger control. Much of static merger control is based on the incentives of firms (e.g. standard Bertrand differentiated products competition measures; vertical arithmetic; etc.).

Sixth, Facebook was willing to pay a substantial amount of money for Giphy ($315m). It is reasonable to require that statements about the purpose of the merger, and about the potential competitive outcome of the merger, are consistent with that valuation. This is a point that Lear made when criticising the analysis carried out by the UK competition authority in the Facebook/Instagram case.¹² It is also discussed by Kuhn (2021), who notes that those tech mergers that ex post look to have been bad mergers all had very high outlier ratios of purchase price to current revenues.¹³

Seventh, when we are considering concerns around whether a merger might harm dynamic competition, the focus of the analysis is likely to be more on the capabilities of the firms involved, and of other firms, than on the specific products that firms currently produce. Whilst this analysis can often be very challenging within the relatively short timescales of merger control analysis, it is clearly a sensible approach when we are trying to understand the implications of dynamic competition between firms.¹⁴

5. Conclusions

When it blocked the Facebook/Giphy merger the CMA was not claiming that Giphy would definitely be a commercial success, or that it would become a very significant competitor to Facebook, or that it would fare better under an alternative owner. But what it was doing was arguing that the process of competition is an important driver of innovation in dynamic industries and that incumbent firms should not have carte blanche to be allowed to eliminate that competition.

Citation: Mike Walker, “The UK Facebook/Giphy Case: Taking Dynamic Competition Seriously”, Network Law Review, Summer 2023.

Related Posts