Tech and Antitrust

Four years ago I wrote Aggregation Theory, which argued that technology companies, uniquely enabled by zero marginal costs, were dominant by virtue of user preference driving suppliers onto their platforms, creating a virtuous cycle. Then, one month later, I predicted that the end state of Aggregation Theory would be increased demands for antitrust action. From Aggregation and the New Regulation:

This last point is key: under Aggregation Theory the winning aggregators have strong winner-take-all characteristics. In other words, they tend towards monopolies. Google is perhaps the best Aggregation Theory example of all — the company modularized individual pages from the publications that housed them even as it became the gateway to said pages for the vast majority of people — and so, given their success, perhaps it shouldn’t be a surprise that the company is under formal investigation by the European Union.

There was a second more subtle point in that article, though:

In other words, the regulation situation for these massive winner-take-all companies is not hopeless, but it has changed: their strength derives from the customer relationships they own, which means quiet backroom deals and straight-up arm wrestling of the Google and Uber varieties are liable to backfire in the face of overwhelming public opinion; it is in shaping that public opinion that the real battle will be fought. And while it’s true that the direct relationship aggregation companies have with their users is an advantage in this fight, the overwhelming power of social media is the new counterweight: it is easier than ever to reach said users with a report or column that resonates deeply. Your average writer or reporter has more (potential) power, not less.

This seems like the best explanation for how we have arrived at the current moment; Reuters reported last week that the U.S. Department of Justice and the Federal Trade Commission were divvying up tech companies for potential antitrust investigations — Google and Apple to the former, and Facebook and Amazon to the latter — a seemingly natural endpoint to what has been a mounting drumbeat for regulatory action against tech.

There’s just one problem: it’s not clear what there is to investigate.

I should state an obligatory caveat: I am not a lawyer or economist, which is relevant given that U.S. antitrust cases are adjudicated in court and largely driven by expert testimony. That reality, though, only underscores the point: any case against these four companies (with possibly one exception, which I will get to momentarily), will be extremely difficult to win.1 To explain why, it is worth examining all four companies with regards to:

  • Whether or not they have a durable monopoly
  • What anticompetitive behavior they are engaging in
  • What remedies are available
  • What will happen in the future with and without regulator intervention

In addition, for comparison’s sake, I will evaluate late 1990’s Microsoft, the last major tech antitrust case in the United States, along the same dimensions.

Durable Monopolies

The FTC defines monopolization as follows:

Courts do not require a literal monopoly before applying rules for single firm conduct; that term is used as shorthand for a firm with significant and durable market power — that is, the long term ability to raise price or exclude competitors. That is how that term is used here: a “monopolist” is a firm with significant and durable market power. Courts look at the firm’s market share, but typically do not find monopoly power if the firm (or a group of firms acting in concert) has less than 50 percent of the sales of a particular product or service within a certain geographic area. Some courts have required much higher percentages. In addition, that leading position must be sustainable over time: if competitive forces or the entry of new firms could discipline the conduct of the leading firm, courts are unlikely to find that the firm has lasting market power.

There are (at least) two major questions that arise from this: how is the relevant market defined, and what does it mean for market power to be sustainable over time?

1990s Microsoft: Microsoft was found to have a monopoly on operating systems for personal computers, and that advantage was found to be durable because of the lock-in created by the network effects between developers using the Windows API and users. Both conclusions were reasonable.

Google: Google certainly has a dominant position in search, but the real question is around durability. Google has long argued that “Competition is only a click away”, which has the welcome benefit of being true.

The European Commission handled this objection by arguing that Google also enjoys network effects:

There are also high barriers to entry in these markets, in part because of network effects: the more consumers use a search engine, the more attractive it becomes to advertisers. The profits generated can then be used to attract even more consumers. Similarly, the data a search engine gathers about consumers can in turn be used to improve results.

This is certainly a much more tenuous lock-in than the Windows API, but I think it is a plausible one.

Apple: There is no company for which the question of market definition matters more than Apple. The company is eager to point out that the iPhone has a minority smartphone share in every market in which it competes; even in the U.S., Apple’s best market, the iPhone has 45% share, less than the 50 percent of sales the FTC suggests as a cut-off.

In Europe, Apple is likely in trouble when it comes to the European Commission’s investigation of Apple about Spotify’s complaints about the App Store. In the Google Android case the European Commission determined that “Google is dominant in the markets for general internet search services, licensable smart mobile operating systems and app stores for the Android mobile operating system.” That last clause leaves room for Apple to be found dominant in app store for the iOS mobile operating system, at which point taking 30% of Spotify’s revenue (or else forbidding Spotify to even link to a web page with a sign-up form) will almost certainly be ruled illegal.

I strongly suspect the Department of Justice will have a much more difficult time convincing a federal court that such a narrow definition is appropriate, but at the same time, I’m not certain that “smartphones” are the correct market definition either. Suggesting that users changing ecosystems is a sufficient antidote to Apple’s behavior is like suggesting that users subject to a hospital monopoly in their city should simply move elsewhere; asking a third party to remedy anticompetitive behavior by incurring massive inconvenience with zero immediate gain is just as problematic as making up market definitions to achieve a desired result.

Facebook: Here again market definitions are very fuzzy. Most people have multiple social media accounts across both Facebook and non-Facebook services, which means any sort of workable market share definition would have to rely on “time-spent” or some other zero-sum metric. Moreover, it’s not clear what is or is not a social network: does iMessage count? What about text messaging generally? What about email?

There certainly is an argument that Google and Facebook are a duopoly when it comes to digital advertising, but it is not as if either has the power to foreclose supply: there is effectively infinite advertising inventory on the Internet, which suggests that Google and Facebook earn more advertising dollars because they are better at advertising, not because they foreclose competition.

Amazon: There really is no plausible argument that Amazon has a monopoly. Yes, the company has around 37% of e-commerce sales, but (1) that is obviously less than 50% and (2) the competition is only a click away! Moreover, it’s not clear why “e-commerce” is the relevant market, and in terms of retail Amazon has low single-digits market share.

Anticompetitive Behavior

But for a few exceptions, everything that follows is moot if the company in question is not found to have a durable monopoly. After all, “anticompetitive behavior” is simply another name for “driving differentiation”, which no one should want to be illegal for any company that is not in a dominant position; it is the potential to make outsized profits that drives innovation.

Still, it is worth examining what, if anything, these companies do that might be considered problematic.

1990s Microsoft: Microsoft was found guilty of illegally bundling Internet Explorer with Windows and unfairly restricting OEMs from shipping computers with alternative browsers (or alternative operating systems). The first objection is particularly interesting in 2019, given that it is unimaginable that any operating system would ship without web browser functionality (which, at a minimum, would obviate an essential distribution channel for 3rd-party software). The second is much more problematic: as I wrote in Where Warren’s Wrong, competition-constraining contracts from dominant players should be viewed with extreme skepticism, as their purpose is almost always to extend dominance, not increase consumer welfare.

Google: Again — and note a developing theme here — Google’s anticompetitive behavior is relatively clear. First, the company consistently favors its own properties in search results, particularly “above-the-fold” — that is, results that are not actually search results but which seek to answer the user’s query directly. A partial list:

  • Google by-and-large removed video segments from competing properties in favor of YouTube videos
  • Google offers local results from Google Maps above search results that tend to favor Yelp, TripAdvisor, etc.
  • Google offers hotel and flight listings above search results that tend to favor Booking, Expedia, etc.
  • Google displays AMP-enabled websites (a Google technology) above search results that are agnostic about how a web page is displayed.
  • Google displays tweets for individuals (thanks to a beneficial relationship with Twitter) above search results that tend to favor LinkedIn, Facebook, etc.

Of these local is probably the most open-and-shut case (although Google’s efforts around travel and hospitality are on the same track): Google Maps results were worse, got better when Google scraped data from competitors (which it stopped doing after an FTC investigation), and now is somewhat competitive by sheer force of exposure to customers defaulting to Google search.

Then, of course, there is Android, where Google leveraged the Play Store to force Android OEMs to feature Search and Chrome, and further forbade said OEMs from shipping any phones with open-source Android alternatives (a la Microsoft). This is one case the European Commission got exactly right.

Apple: As I argued in Antitrust, the App Store, and Apple, Apple is leveraging its position in the smartphone market to earn rents in the market for digital goods:

To put it another way, Apple profits handsomely from having a monopoly on iOS: if you want the Apple software experience, you have no choice but to buy Apple hardware. That is perfectly legitimate. The company, though, is leveraging that monopoly into an adjacent market — the digital content market — and rent-seeking. Apple does nothing to increase the value of Netflix shows or Spotify music or Amazon books or any number of digital services from any number of app providers; they simply skim off 30% because they can.

For this to be illegal does not necessarily require that Apple have a monopoly: tying (i.e. iOS users must use the App Store) is per se illegal in theory, but in practice the Supreme Court has dramatically constricted the definition of tying to include a requirement that the tie-er have market dominance; the Supreme Court also declined to review the Court of Appeals decision in the Microsoft case, which held that courts should use a rule of reason test for software specifically that also considers the benefits of tying, not simply the downsides.

I would certainly argue that the requirement that digital content use Apple’s payment processor (and thus give up 30%) has downsides that outweigh the benefits, but the truth is that this is a case that, under U.S. antitrust law, is harder to make than it was 20 years ago.

Facebook: There are certainly plenty of reasons to be upset with Facebook when it comes to issues of privacy, but the company has not done anything illegal from an antitrust perspective.

I am, to be clear, distinguishing anti-competitive behavior from anti-competitive mergers. I have made the case as to why Facebook’s acquisition of Instagram was so problematic, and this is the area that needs the most urgent attention from anyone who cares about competition. The single best way to maintain a dominant position in a market as dynamic as technology is to use the outsized profits that come from winning in one market to buy the winner in another; it follows, then, that the best way to spur competition in the long run is to force companies to compete with new entrants, not buy them out.

Amazon: Make no mistake, Amazon drives a very hard bargain with its suppliers. Those suppliers, though, have a whole host of alternatives through which to sell their product. Meanwhile, those hard bargains accrue to consumers’ benefit.

Similarly, it is very hard to see why Amazon can’t offer its own branded goods; this practice is widespread in retail, and for good reason: consumers get a better price, not only on the store-branded goods, but also on 3rd-party goods that can be priced more competitively since the retailer is making its margin on its own goods.

In short, more than any company on this list, the arguments against Amazon fall apart on the first point: Amazon simply isn’t a monopoly.


Remedies by definition come last: there has to be something worth remedying! Still, it is interesting to consider what the appropriate remedy for each company would be if they were indeed found to be a monopoly engaged in illegal anticompetitive behavior.

1990s Microsoft: Microsoft was originally ordered to be broken-up, although this remedy was overruled on appeal. The idea was that Windows would better serve all 3rd-party software suppliers if it weren’t incentivized to favor its own offerings. Ultimately, though, the company agreed to open up its API, although critics argued that the specifics simply cemented Windows’ dominance, instead of making it possible to build a Windows alternative that could run 3rd-party Windows applications.

The European Commission went further both in terms of requiring interoperability and also presenting users with choice in terms of both browsers and media players. In both cases 3rd-party competitors actually won in the long run — but they won because they were clearly better (first Firefox and then Chrome, and iTunes).

Google: An effective Google remedy would likely be more about constraining Google behavior than it would be about restructuring Google itself. Google might be forbidden from offering its own results for things like local search, or be forced to feature results from competitors according to an algorithm overseen by a court observer. There would also likely be a large fine.

Apple: The obvious remedy for Apple would be allowing 3rd-party payment processors for apps; frankly, I think this might go too far, as there are real benefits to Apple controlling everything API-related on the iOS platform. I would be satisfied with Apple allowing apps to launch web views for payment processing that is clearly handled on the app’s own webpage.

Alternatively, Apple could be forced to significantly reduces its App Store take rate, but I would prefer that Apple be forced to compete for payment processing business, which would achieve a similar result.

Facebook: Facebook, fascinatingly enough, given its lack of anticompetitive behavior, has the most obvious remedy: break apart Facebook, Instagram, and WhatsApp. I do believe this would be beneficial for competition: Instagram being an independent company would not only add another competitor for digital advertising, but would also make other companies like Snapchat more competitive by virtue of forcing advertisers to diversify. Again, though, this is more about a failure in merger review.

Amazon: Amazon has made anti-competitive acquisitions of its own, like Zappos and Those platforms are gone though, making any sort of breakup unrealistic (this is likely at least one factor in Facebook’s plans to integrate messaging across its platforms — that will make a breakup that much more difficult). And as far as selling its own products goes, not only is that probably not a problem, but there is little evidence 3rd-party sellers are being hurt by Amazon’s policies, and plenty of evidence that they are helped by having access to Amazon’s customers. Moreover, highly differentiated suppliers have found success prioritizing other retailers if Amazon squeezes too hard.

The Future

Ideally, an antitrust action is not simply about punishing bad behavior in the past, but also about ensuring competition going forward. To that end, it is worth considering whether the upheaval that would result from any sort of investigation would actually make a long-term difference.

1990s Microsoft: Here the Microsoft case is particularly pressing. It is my contention that Microsoft failed to compete on the Internet and in mobile because the company was fundamentally unsuited to do so, both in terms of culture and capability.

The implication of this conclusion is that the antitrust case against Microsoft was largely a waste of time: the company would have been surpassed by Google and Apple regardless (and that the company only returned to prominence when it embraced a market that suited its capabilities and transformed its culture).

Many disagree to be sure, arguing that the antitrust case prevented Microsoft from foreclosing Google, although it is never clear as to how Microsoft would have done so (nor any explanation as to why Microsoft failed in mobile, where they were not constrained). A better argument is IBM: the government may have ultimately failed in its antitrust case against the mainframe behemoth, but IBM did voluntarily separate its software sales from its hardware sales, setting the stage for its own disruption; then again, the bigger factor was that IBM simply didn’t care enough about PCs to lock them down effectively.

Google: I wrote that we had reached Peak Google in 2014; clearly I was wrong, at least as far as the company’s results and stock price were concerned, but notably the company is ever more dependent on search advertising. One of my biggest mistakes was underestimating the degree to which Google could monetize mobile, not simply through increased adoption but also stuffing results with ever more ads (which, in the limited viewport of smartphones, are even easier to tap on).

That, though, is also an argument that my mistake was one of timing, not thesis (still a mistake, to be clear). For all of Google’s seeming advantages in machine learning, the company has yet to come up with a true second act in terms of driving revenue and profits (with the notable exception of YouTube, an acquisition).

Frankly, I suspect this is why Google is the most at-risk in this analysis: when a company is growing, it has no need to engage in anti-competitive behavior; it is only when the low-hanging fruit is gone that the risk of leveraging one market into another becomes worth it.

Apple: That analysis applies to Apple as well: the company introduced the “Services Narrative” in the 6S cycle, which in retrospect was when iPhone growth plateaued. Suddenly the rent Apple collected from apps was not simply an added bonus to a thriving iPhone business but a core driver of the company’s stock price.

At the same time, it is not as if iPhones are disappearing: there is still an argument to act for the sake of all of the businesses that will be hurt in the meantime. The same argument applies to Google: just because antitrust action isn’t necessarily causal when it comes to a company being eclipsed doesn’t mean it can’t be an important tool to maintain competition in the meantime.

Facebook: As I noted above, Instagram bought Facebook another five-to-ten years of dominance. That, though, is itself evidence that social networks are not forever. Each generation has its own preferences, and as long as acquisition rules around network-based companies are significantly beefed up, the best solution for Facebook, at least from an antitrust perspective, is simply time.

Amazon: This probably deserves a longer article at some point, but I think there is reason to believe that Amazon’s consumer business has also slowed considerably. The company is pushing more into ads, squeezing its suppliers, and driving customers to 3rd-party merchants with their attendant higher margins (for Amazon). This makes sense: there are certain categories of products that make sense for e-commerce, and Amazon does very well there, but will — and perhaps has — hit a ceiling as far as overall retail share is concerned.

Indeed, a mistake many tech company critics make is assuming that graphs that are up-and-to-the-right continue indefinitely; nearly all of those graphs are S-curves that will flatten out, and it is dangerous making regulatory decisions without some sort of insight as to when that flattening will occur.

Ultimately, when it comes to antitrust actions against tech companies in the U.S., there really isn’t nearly as much there as all of the attendant fervor would suggest. Google is absolutely vulnerable, Apple somewhat less so, and it is very hard to see any sort of case against Facebook or Amazon.

And again, this is probably a trailing indicator: Google and Apple have maximized their gains from their most important products, while Facebook and Amazon (particularly AWS) still have growth potential. I don’t think this alignment is a coincidence.

That is not to say that tech deserves no regulation: questions of privacy, for example, are something else entirely. Nor, for that matter, is antitrust irrelevant in the United States generally: concentration has increased dramatically throughout the economy.

What is driving that concentration matters, though: at the end of the day tech companies are powerful because consumers like them, not because they are the only option. Consumer welfare still matters, both in a court of law and in the court of public opinion.

  1. I will mention the European Commission’s different standards in passing; I addressed the differences between U.S. and European approaches more fully in 2016’s Antitrust and Aggregation