There is a growing debate around the possible existence of a kill zone around tech titans. This is an area where venture capitalists will not finance start-ups because of fear of both exclusionary conduct and aggressive acquisition strategies by technology incumbents.
This paper, available here, draws upon existing literature and antitrust agencies’ work, as well as data from the venture capital industry, to argue for the need to investigate the existence and magnitude of the kill zone, as well as its possible causes.
Section II looks at evidence from the venture capital industry.
Venture capital consists of equity investments in companies with innovative ideas characterised by both high growth potential and high risk of failure. Venture capitalists invest across different stages of the life cycle of start-up companies. Recent years have seen larger and later-stage deals, with funds being funnelled to fewer companies, many of which are large enough to be valued at over USD 1 billion, together with a reduction in the number of early/seed stage deals. In 2018, follow-on rounds amounted to nearly 12 times the capital invested in first-time financings, a 15-year high.
Recent studies have indicated that the existence of a kill zone could be behind the decline of venture capital investments, and thus of new entry, in areas where the major digital platforms are active or make large acquisitions. Research has found that first-time financings contracted drastically, compared to similar technology sectors, in those categories where Amazon, Facebook and Google are active. In effect, the amount of investments in these technology markets have declined as they have increased in other sectors. There is also evidence of a reduction of c. 45% in investment in start-ups by venture capital firms after major acquisitions by Facebook and Google in the same space.
When the target company succeeds – which very often does not happen – venture capitalists aim to exit their investment at the highest possible returns through an IPO, if the company goes public, or a sale through M&A, if it remains private. In recent years, the number of US IPOs have consistently decreased. In this context, M&A plays an increasingly prominent role, accounting for the large majority of venture-backed exits. Digital incumbents are particularly active acquirers of technology start-ups. Google, Amazon, Facebook, Apple and Microsoft (GAFAM) have made 44.9 deals on average in each of the last 10 years.
Section III describes possible causes for the creation of kill zones.
The evidence described in the previous section is consistent with the existence of reduced first-time venture-backed funding in markets dominated by digital incumbents. However, it is unclear what the reasons behind this may be. These developments could stem from changing requirements of start-ups themselves as their technological and commercial needs evolve, such as the widespread adoption of ‘blitzscaling’ strategies where start-ups enter a digital niche. Moreover, changes have taken place also in the investment industry landscape through an expansion of the types of capital provided, which now include non-traditional investors and sovereign wealth funds, which may impact VC investment strategies.
Section IV discusses why competition enforcement might be needed.
Digital incumbents seem to face very little threat of entry. Some recent studies and antitrust agency reports suggest that digital markets are becoming progressively less dynamic. Key players in the digital industry remained the same over the last two technology waves, staying dominant through the shift to mobile and the rise of artificial intelligence, without significant impact on market share or profit margins – which are such as to attract entry were the markets to be contestable. Radical, disruptive innovation would seem necessary to challenge the digital incumbents’ core businesses. Instead, at present incumbents allegedly have both the ability and the incentive to foreclose promising potential competitors. Their position allows them to collect large amounts of data and to identify emerging trends early on and to react to them, whether by adopting aggressive exclusionary practices to protect their core market or by pre-emptive acquisitions of innovative start-ups at generous multiples.
As a policy objective, antitrust agencies should ensure that a potential new entrant with a superior technology may grow as an independent company in the kill zone or in adjacent markets. In that regard, antitrust agencies have been in recent years more active in relation to conduct cases, especially in Europe. As regards merger control, however, many acquisitions by digital incumbents fell below the merger control thresholds, while those that were reviewed may have been mistakenly cleared.
In the light of this, a number of possible solutions have been advanced. First, the reversal of the burden of proof in relation to the acquisition of start-ups in specific circumstances has been advanced. This would introduce a rebuttable presumption of anticompetitiveness if the incumbent acquirer does not provide evidence of absence of such effects or of efficiencies. Second, the development of theories of harm better suited to digital markets should be pursued. The author notes that theories of harm based on actual or potential competition, or on innovation, are unlikely to apply to the acquisition of start-ups by digital companies on their own. However, it may be appropriate to apply conglomerate theories of harm when the merging parties operate in the same ‘technological space’ or ‘users’ space’. Third, in order to avoid the difficulties with defining counterfactuals for inherently uncertain outcomes in dynamic markets, it has been suggested that the standard of review should allow one to look at a balance of harms, instead of just at the likelihood of anticompetitive effects occurring. Fourth, to address the ambiguous effects of more stringent merger control, it has been suggested that one should recognise the negative dynamic effects of the current de facto laissez-faire regime on ex ante incentives to innovate and the importance of maintaining market contestability.
Section V concludes.
While there is convergence on the need to do more as regards acquisitions by digital platforms, there is far less agreement on how to go about it. On the substantive front, it will be necessary to adopt new theories of harm, to adapt antitrust tools to changing market conditions, and better to incorporate ex ante incentives to innovate in merger assessments. These developments will need to be tested in the courts, and may create a risk of false positives along the way. Other alternatives are procedural – e.g. allowing for some mergers to be reviewed, and the target divested, ex post. Finally, and complementarily to antitrust intervention, there may be a need to resort to ex ante regulation, especially to promote market contestability
This paper takes a very interesting perspective on the digital economy. By looking at it from the perspective of venture capital, it effectively adopts the view of the industry, which is always informative. When coupled with the analysis of recent trends in VC practice, and how they reflect on market developments, the relevance of this paper should be obvious. As much was recognised by one of the most respected technology analysts in the world – and a leading tech investor and venture capitalist to boot – who described the paper as recommended reading because ‘[while] I don’t agree with a lot of this, [-] it’s logical, coherent and properly argued (…) and worth reading to understand how people with the power to make laws are thinking about this.’
I think Andrea – who is a good friend of mine – would be surprised to hear that he has the power to make laws. Were he to have that ability, I would have suggested that he subject the paper to some editing, in particular to streamline the argument and clarify its structure.
For example, I know that the concept of ‘kill zone’ emerged from the venture capital / technological industry itself, so it is obvious that there is a link between the two. However, this link – and in particular the link between reduced investment, or certain investment patterns, and the creation of a kill zone, and why this matters for competition law – could have been made clearer. As it is, the paper does not provide any actual evidence of a kill zone, which would seem an important basis for the interventions proposed in section IV. Further, it is unclear to me why venture firms would not invest in companies in the kill zone, given that the dominant mode of venture capital exit is by purchase – i.e. how is the reduction in venture capital investment related to the existence of a kill zone, when placing a start-up in this zone would seem to provide the best monetisation option? As noted by Motta and Peitz and in other papers I reviewed earlier, the existence of a “kill zone” cannot not be explained by the possibility of a start up being acquired, but rather by the threat that the incumbent may engage in exclusionary practices to prevent competition from taking place. This also seems to be the reason why the paper above by Lemley and McCreary attacks the VC funding model as bad for competition. Likewise, it is not clear how the concept of ‘kill zone’ relates to increases in late-stage investments, or developments in early-stage investments.
As a result of this, it was unclear to me from this paper how venture capital investments matter for competition law, other than by providing some additional evidence of increased market concentration and reduced contestability in digital platforms’ core markets. This matters not only as a matter of the internal logic of the paper, but also because answering these questions is necessary to identify the problem that the policy reactions are supposed to solve.
The author provides a very interesting overview of recent proposals to address these problems, while making it clear that he thinks that more work is needed. I tend to agree. In particular, I think that part of that work must seek better to delineate the competitive problems that kill zones / digital incumbency create.
Nowhere is this clearer than in the discussion of the various proposals concerning the reversal of the burden of proof. The sheer number of different proposals concerning the scope of the presumption of harm for digital mergers is an indication of the lack of agreement about what exactly we are concerned about. For example, this paper identifies the following proposals regarding when the presumption of harm could apply:
- to acquisitions by already dominant platforms or ecosystems, in a context characterised by strong network effects possibly reinforced by data-driven feedback loops, of early-stage small, but successful, innovative start-ups with a quickly growing user base and significant competitive potential, especially if there is a systematic pattern of such acquisitions (EU Expert Report);
- where there exists a cannibalisation effect between the target’s and the acquirer’s products or services and there is a risk of the acquirer discontinuing the potential innovation, the acquirer should have the burden of proof to demonstrate that it will not discontinue the target’s innovation (Borreau and De Streel);
- whenever the acquirer is super dominant (Valletti);
- when the agency can demonstrate a reasonable prospect of harm in a digital market (Fletcher).
There are also other proposals (e.g. see the paper reviewed above), but this is an illustrative sample. Each proposal presupposes a different theory of harm, and maybe even different types of harm to competition – and this despite not a single acquisition by GAFAM firms having been prohibited thus far. In other words, each proposal addresses a different concern, even if these differences are smoothed over by all proposals targeting digital platforms. This links to the point I made in previous reviews: other than a generic fear that we are missing problematic digital acquisitions, or that cumulative acquisitions are a threat to market contestability where digital markets have tipped, what is the exact concern we are trying to address here? And why is it that we cannot address this concern with the tools we currently have? Once we answer these questions, it will be easier to devise an effective response.