Big Tech Platforms and Schumpeter’s Creative Destruction

Schumpeter’s indirect entry theory fits the average tendencies of competition in digital industries. When the model is added to standard assumptions and suspicions, a refined policy emerges. The prescription is neither less nor more Government intervention, but a different one.

In antitrust policy circles, Joseph Schumpeter’s “creative destruction” theory is often mentioned but seldom listened to. The Schumpeterian neglect is understandable, but unfortunate. [Understandable, because the historical record of the economy involves mostly incremental/imitative entrants rather than the radical breakthroughs associated with creative destruction (Teece, 2018).]

My recent book Big Tech and the Digital Economy: The Moligopoly Scenario sheds light on how Schumpeter’s insights give us a useful model of “indirect entry” that fits the tendencies of competition in digital industries. Schumpeter considered that the main source of economic competition is wrought by product differentiation, complements, or new combinations, not by imitation, replication, and substitution of existing products. He colorfully stressed that this is as powerful “as bombardment is in forcing a door” (Teece, 2020).

The issue matters because the theory of government regulation of industry operates on a different assumption. The main source of competition that keeps markets free and competitive comes from entry of substitute products. And a suspicion of anticompetitive conduct by incumbents or of structural market failure arises when direct entry is not observed. The recent US antitrust charges against Google bring a good example. The engine of the case against Google brought by the Department of Justice is that by purpose or effect, Google’s payments to Apple and other original equipment manufacturers (OEMs) for pre-installation restricted the entry of rival search engines like DuckDuckGo.

But when the Schumpeterian model of “indirect entry” is added to standard assumptions and suspicions, a refined policy emerges. The prescription is neither less nor more Government intervention, but a different one. To see this, let us consider first models of indirect entry (1) and why they suit digital industries (2), explore high-level implications for policymaking (3), and discuss concrete illustrations by looking at two ongoing cases against big tech firms (4).

1. Indirect entry

Tim Bresnahan and Shane Greenstein have documented that the competitive story of the computer industry is one of indirect entry. Every decade or so, the gravitational center of competition shifted: from mainframes (1960s), to minicomputers (1960-1970s), microcomputers (1970-1980s), and servers (1980-1990s).

Three key insights emerge from the work of Bresnahan and Greenstein: One, platform shifts in the computer industry occurred by indirect entry. Two, new firms, not dominant incumbents, won platform shifts. Three, the shifts occurred sequentially, leading to platform replacement every 10 years or so.

Given its proximity to the computer industry, it is appealing to examine the digital industry under the Bresnahan and Greenstein model of indirect entry. What do we see?

A process of indirect entry characterizes the digital industry from its early days to the present. Search engines leapfrogged portals and browsers as entry points to the web. Mobile telephony cannibalized desktop computing. Social networking redefined personal and professional communications. In a way, each of today’s dominant consumer-facing platforms emerged by indirect entry.

To be sure, one main difference between the digital industry and the computer industry is that several platform shifts occurred simultaneously, not subsequently. As if all innovation arrivals came at once, search, e-commerce, and social network platforms emerged in parallel.

Also, and this likely holds true for each of today’s platforms, we can see that platform shifts are still not complete, as new firms keep competing for the dominant design by indirect entry. Take social networks. Facebook attacked MySpace with a curated network of college students. Instagram then introduced architectural innovation by adding sophisticated photo capabilities to Facebook. And Tiktok recently complemented Instagram by adding creative music, dancing, and lip-sync functionality.

2. So What?

The usual theoretical paper or government report tends to look for product and consumer similarities to declare a market competitive. When evidence of similarity lacks, inferences of insufficient rivalry, and in turn market failure, are a click away. The Furman report, commissioned by the UK government, argues that “persistent” and problematic dominance characterize Facebook, Google, and Apple because they have a large market share in their lines of business for a long time.

This mindset reflects the well-known economist bias towards measuring competition by reference to market share, concentration, and insufficient direct entry (Sidak and Teece, 2009).

But the average tendency of competition in digital industries is one of indirect entry. Firms do not enter by imitation, replication, or substitution of incumbent firms’ capabilities. Android entered mobile telephony without a handset device; Shopify entered e-commerce without a market place; TikTok entered social networks without a social graph; and Zoom entered video conferencing without network effects.

To be sure, one might counter-argue that these anecdotal examples are compatible with a theory whereby indirect entry is the byproduct of anticompetitive entry deterrence by monopoly platforms. Industry history, however, refutes this idea. Firms as aggressive, sophisticated, and profitable as Microsoft almost systematically failed when they tried to take on platform incumbents by direct competition. Direct entry is a losing playbook, and for a good reason. In industries with increasing returns on both the demand and the supply side, a monopoly market share is efficient (Ducci, 2020).

It might be the case that some politicians, public officers, and economists expect the digital economy to demonstrate higher levels of entry. But if this is the case, it is reasonable to ask (i) why higher threshold levels of entry are socially desirable in digital industries, compared to other industries; and (ii) for the evidence showing that observed entry levels are lower than expectations. Short of this, a sentiment that double standards apply to digital industries is not unreasonable. And a looming suspicion of vindictive, not rational regulation is hard to repress.

3. Policy Neutrality towards Big Tech Entry?

If indirect entry is more the rule than the exception, what should antitrust and regulatory policy concretely do?

To start, it should promote and protect firms that participate in the game of indirect entry. What about incumbent monopoly platforms? Should an antitrust or regulatory system discriminate towards entry depending on whether the entrant is a dominant incumbent or not?

Again, Bresnahan and Greenstein shed some light on this issue. Their work suggests that platform shifts seldom benefit the incumbent, soothing recent concerns of pervasive anticompetitive leveraging in digital markets. Admittedly, digital platforms’ ability to leverage data across markets is a novel source of competitive advantage, as noted in recent expert reports, including the Stigler Committee report on digital platforms. But no corresponding investigation has been offered of organizational, governance, and managerial inefficiencies arising from corporate diversification. In another study, Bresnahan, Greenstein, and Henderson showed that managerial inefficiencies and diversification costs kicked in when the boundaries of the firm widened, offsetting economies of scale and scope in production. At best, the evidence currently available suggests an ambiguous relationship between incumbency and expansion in adjacent markets.

What can we glean from this? One, a regime of policy neutrality should prevail towards indirect entry by new and incumbent firms. Forms of conduct associated with leveraging do not deserve broad presumptions of liability, per se prohibition rules, or line of business restrictions. Recall, in passing, that line of business restrictions trap incumbents behind exit barriers, and incentivize them to ramp up efforts towards rent extraction in monopoly markets.

Two, an indirect entry-minded agency or expert should be uncompromising towards leveraging conduct or transactions by an incumbent firm that by purpose or effect discontinues, extinguishes, or underdevelops the indirect entry business. This form of defensive leveraging was at the heart of the US Microsoft case in the 2000s.

4. What Indirect Entry hints at: The Google, Epic v Apple and Amazon Cases

Three ongoing antitrust cases give insights on the policy power of an indirect entry framework. Let us start with the DOJ complaint against Google. The case exemplifies how obliviousness to indirect entry leads to the development of theories of harm based on wrongheaded models of competition. The complaint focuses on the idea that Google paid for exclusionary pre-installation of its search engine on a variety of communications devices, and that the size of its payments, in particular to Apple, denote the strategic value of monopoly, or the cost of exclusion of competing search engines.

The story of pre-installation payments as a cost of exclusion is perplexing. On (legitimate) demand from Apple and other OEMs, Google, as well as many other firms, pays “Traffic Acquisition Costs” (TACs) to be a default on Apple’s products. (( Google and Apple entered into a licensing contract to embed search in Safari years before the iPhone.)) As public data shows, Google has relentlessly tried to decrease the share of TACs paid to Apple and other firms. What does indirect entry bring here? The evidence needed to see the flaw in the TACs as a cost of exclusion theory. Google would rationally prefer to pay zero for pre-installation.

The development of Google’s own mobile phone ecosystem Android brings this lost fact into broad daylight. Android is a long-term indirect entry strategy to cut down on TACs paid to Apple and other OEMs. Vertical integration and industry cooperation here manifest a Coasian choice to avoid transactional licensing costs. The upshot is this: In forcing the door of the smartphone OS market, Google might plausibly have adopted disproportionate restraints of competition. But it is not all clear if there is anything inherently exclusionary in the Apple payment. Instead, these transfers might just denote legacy contracts, and market power over competitive bottlenecks (the iPhone ecosystem) in a growing industry.

Next, consider the Epic v Apple case. The issue here is a revenue-sharing dispute over the 30 percent cut that Apple takes on in-app payments. Epic claims that Apple can do this because it unlawfully ties access to App Store distribution to exclusive use of iOS in-app payment service. To be sure, Apple started to charge a 30 percent transaction fee when it launched the iPhone 3G and app store in 2007. At the time, it presumably lacked market power. Today, little has changed. Apple keeps charging 30 percent. On Twitter, Randy Picker suggested that this feature of the case raises doctrinal and normative obstacles to a finding of liability because a distinctive act of monopolization is and should be required to apply antitrust law.

Indirect entry, however, allows us to see a potential case of unlawful monopoly maintenance. To start, a consideration of indirect entry suggests that what we are seeing here is exactly the specific class of market power that economic theory is chiefly concerned with. That is market power that cannot be dissipated by competitive forces in the short to mid-term. In spite of Google’s indirect entry, Apple’s fees have hardly budged (just yesterday, Apple announced a reduced cut of 15 percent for small developers though).

With the monopoly box ticked, what remains to be evaluated is whether Apple has unlawfully maintained its market power. The difficult issue here is whether contractual terms and technical policies that prevent platform bypass—such as Apple’s disputed ties—remain justified once the tipping point is crossed. There is no easy answer. Incentives matter. At the same time, a market that has crossed the tipping point operates in something equivalent to “lucky demand conditions,” meaning that the size of output that the firm serves no longer depends on its degree of effort, in turn creating one more example of what Jean Tirole has called “undeserved” market power (Tirole, 2015). An empirical study of Apple’s actual degree of effort at the divisional App Store level would help cast light on whether bypass restrictions remain legitimate.

Last, the current allegations against Amazon’s discriminatorily using third party merchants’ data to self-preference its private labels invite a further distinction between cheap indirect entry and real indirect entry. The question for an investigating economist or Government agent faced with an indirect entry move like Amazon’s is not whether it is efficient–it is unlikely to be because it undermines merchant adhesion to the ecosystem – but whether the firm is seeking to seize a quick profit–John Hicks called this firm type a “snatcher”—or whether it “is interested in building a steady business”—Hicks talked of a “sticker” (Hicks, 1954). The underlying normative consideration is that policymakers should credit real indirect entry moves, not cheap ones. Again, this would allow drawing a line between a dominant firm’s behavior that is more “fox” than “shepherd.”

Conclusion

In digital markets, indirect entry considerations can enrich the stationary, substitute centric, and share-based paradigm that informs policymaking. The market definition and market power assessment in digital markets has become hopelessly blind to indirect entry. A sticky conjecture is that network externalities, big data, and behavioral biases almost mechanically facilitate monopolization outcomes. Indirect entry arguments can play a role at other stages of the antitrust and regulatory process. For example, indirect entry models can help to prioritize cases, formulate theories of harm and justifications, build counterfactual hypotheses, or fashion remedies. At a time where experts try to reinvent a new antitrust, the Schumpeterian leap long called for by innovation, dynamic capabilities, and business theorists appears warranted.