There are many people wanting to unroll past tech acquisitions. Others certainly would advocate doing it differently if there was another chance. For instance, Ben Thompson called Facebook being allowed to purchase Instagram “the greatest regulatory failure of the last decade.” I don’t really see that but that is another matter.
The problem, at the moment, is that there is a lack of clear economic thinking on this topic. Thanks to my colleague Kevin Bryan and his co-author Erik Hovenkamp, that ends today. In a new paper, they show that, amongst regulatory options, in many cases, compulsory licensing is much better than blocking acquisitions.
Let me outline their very straightforward argument. They imagine an industry where there is a leader (who leads by virtue of having better quality products than others) and laggards. Innovation in this industry comes from startups who produce components that potentially can improve the quality of any firm’s products. However, the start-ups can’t bring the products to market themselves (or to do so must acquire an incumbent).
If there is no regulation, what happens? Bryan and Hovenkamp show that the leading incumbent will always purchase the startup not so much because it will increase the quality of their own products but, instead, because if they don’t do so, the startup will be acquired by another firm who will now sell a product that is closer to the leaders and so will compete more intensively. The leader pays out the big bucks to the startup to prevent all that from happening. And, moreover, even though it could improve the product quality of others, the leader has no incentive to license the component to them.
The real cost of this comes because of how this process changes the incentives of startups. Imagine that startups can produce products that ‘work more’ for specific firms. They can produce a product that improves the quality of the leader by more than it does those of laggards or vice versa. If that choice is otherwise free, society would benefit if the startup helped the laggards more as this increases the intensity of competition. However, because these innovations are just components, that isn’t enough to get the laggard to pay more than the leader. (This would change if the innovation was such that if the laggard got it, it would become the leader — this is something that is a direct implication of an earlier result from myself and Scott Stern).
So what happens if you ban the leader from acquiring the startup? Some potentially good things. First, it is acquired by the laggard which means more competition and also a more favourable direction of innovative effort (towards the laggard than the leader). Second, even better, the laggard has an incentive to license the component to the leader which improves further the direction of innovative effort. But there is a bad thing: the startup earns a lower return which may cause it not to innovate in the first place.
This, however, can be resolved by an acquisition under a condition that the leader is forced to license the startup innovation to the laggards. In this case, the profits of the startup end up depending upon the impact on overall value in the industry and so the direction of innovative effort is ‘more’ neutral and because the innovation is diffused, it potentially has a favourable impact on competition. Indeed, this policy has more bite precisely where the incentives of the leader to acquire the firm would otherwise have resulted in maximal competitive harm.
How does this help us with regard to Facebook and Instagram? The answer is: not much. This is because Instagram isn’t a component that can be used by incumbent social networks but is an alternative platform. Thus, the argument that Instagram should be blocked is based on the notion that Instagram could potentially have led the industry. In that case, the antitrust issues would be distinct from those discussed in this paper.
On the other hand, Google’s acquisition of Waze falls squarely within the scope of the paper. That acquisition gave Google important traffic data that would assist it in competing with other navigation software. Moreover, that data could easily be licensed to others and so the remedies that Bryan and Hovenkamp have in mind could have been applied. The open question, which I believe can be answered but hasn’t yet, is whether this argument could be used to be precipitate a breakup following an acquisition. I have to say, the Google-Waze issue does present an intriguing possibility here. I note that it is on Elizabeth Warren’s list. Maybe she should turn attention to compulsory licensing as a more robust policy alternative.