What follows is a framework for understanding digital firms, a framework that emerged as I edited an article on Meta Platforms, an article to be posted next week Tuesday. There are three main elements to this framework: firstly, that competition is a discovery process; secondly, that transaction costs define the scope and limits of a firm; and thirdly, that where information is the fundamental attribute of an industry, that industry will tend toward increasing returns. These planks, if you like, are underpinned by a conviction that the economy is a complex organism. The consequence of these elements or planks, is that the Internet Revolution naturally resulted in the emergence of platforms, and aggregators, because it demanded a kind of firm that was not the centre of production, but the centre of a network.
Competition as a Discovery Process
Complexity economics views the economy as a system not necessarily in equilibrium, but rather as one where agents constantly change their actions and strategies in response to the outcomes they mutually create. It holds that computation as well as mathematics is useful in economics, that increasing as well as diminishing returns may be present in an economic situation, and that the economy is not something given and existing, but forms from a constantly developing set of actions, arrangements, and technological innovations. The economy is thus comprised of evolving networks of interacting agents, institutions, and technologies—networks of networks. The macro- level patterns of the economy—growth, innovation, business cycles, market booms and busts, inequality, and carbon emissions—then emerge from these dynamic micro- and meso-level interactions. From the complexity economics perspective, change is largely an endogenous phenomenon, not simply the result of unexplained shocks from outside the system.
“Complexity Economics: An Introduction”, by W. Brian Arthur, Eric D. Beinhocker, & Allison Stange
A consequence of taking a complexity approach to competition is that competition is seen as a multi-level process, in which firms compete and relate with other firms within an industry, who, given the tendency of wealth toward destruction, seek to survive over the long-run and grow in the short run. Within markets, firms maximise their profits and compete for market share by providing sustainable products and services. Competition between firms can also be described in Coasian language as competition between intra-firm and inter-firm organisation, between whether economic activities should be done within the firm, or by the market. Within firms, individuals, units and divisions compete and cooperate to maximise their individual payoffs. Nicolas Petit and Thibault Schrepel call these the macro, meso and micro levels of competition. Competition at the industry level forces changes within firms that result in a firm facing new competitors at the market level. Concretely, by way of example, each of Meta Platforms divisions compete and cooperate over resources, and at the market level, Meta enjoys a monopoly in social media networks, but faces fierce competition at the industry level, where it is part of the Attention Economy. (While preparing this article for publication, I received Ben Thompson’s latest Stratechery blog post, “Meta and Open”, which makes a similar point). The emergence of TikTok at the industry level forced changes in Instagram, by way of Reels, which triggered an evolution from a chronological feed of content surfaced from one’s social network to algorithmically sorted content from the universe of all Instagram users.
Competition exists because firms, and in general, economic agents, have imperfect information. The irreducible complexity of the economy makes prediction a fraught exercise and veils the future with uncertainty. Contrary to the dominant neoclassical economic paradigm, if firms had perfect information, where one knows all the relevant facts to make a decision, as happens under perfect competition, they would not compete, it is the veil of uncertainty thrown over economic activity that makes competition necessary. The greater the uncertainty, the greater the competition. Competition is a learning process in response to ill-defined situations. In fact, in 1975, when Leonid Kantorovich won the Nobel Prize, it was for work that demonstrated that central planning could work given perfect information. Rather than a precondition for perfect competition, perfect information is a precondition for central planning.
With knowledge diffused across the market, management’s primary problem is not the allocation of given capital, but the utilisation of incomplete knowledge. The economy as a whole is an economy of economic agents who respond to ill-defined situations by, in the words of Arthur, “‘making sense’ or recognizing some aspects of them, and choosing their actions, strategies or forecasts accordingly”. Everything is contested and uncertain, from the optimum mix of planning and market transactions, to the best combination and use of resources. The process of gathering, interpreting and making use of knowledge is a constant struggle. In a lecture titled, “Competition as a Process of Discovery”, F.A. Hayek explained that,
…it is salutary to remember that, wherever the use of competition can be rationally justified, it is on the ground that we do not know in advance the facts that determine the actions of competitors. In sports or in examinations, no less than in the award of government contracts or of prizes for poetry, it would clearly be pointless to arrange for competition, if we were certain beforehand who would do best.
“Competition as a Process of Discovery”, by F.A. Hayek
Since Adam Smith wrote The Wealth of Nations, economists have known that prices, like an “invisible hand”, coordinate the actions of economic agents in systems in which knowledge is widely distributed, so that supply adjusts to demand and production to consumption. The price of a thing contains the relevant signals for action, even if few of the market’s actors can fathom the whys and the wherefores. The market is an organism wherein the price mechanism constantly collects, assesses, distils and transmits relevant facts to the market’s actors. By responding to price, competition fosters information discovery and the organisation of the market. The market, a complex adaptive system, achieves spontaneous order, an order created by unwitting firms who are constantly adapting to each other’s behaviour and to changes in the order they create, in a recursive loop.
Uncertainty over existing and possible states, creates costs of identifying partners and opportunities, writing and executing contracts, determining whether production costs are lower within the firm or outside of it, all the while trying to reduce the likelihood of a permanent impairment of capital. Not only does increasing uncertainty lead to increasing competition, but increasing competition increases uncertainty, parallel to which, rising uncertainty increases the economy’s complexity and rising complexity increases uncertainty.
Firms respond to uncertainty through a range of actions from innovation, diversification and exploration, to trying to freeze the market structure through collusion, lock-in, erecting barriers to entry, and other such actions.
Transaction Costs and the Firm
Firms are planned economies. Where in the market, the factors of production are superintended by the price mechanism, within the firm, entrepreneurial planning supersedes it. The triumph of capitalism over the last five hundred or so years has demonstrated the wondrous ability of markets to create unparalleled wealth, and yet, firms exist. Ronald Coase in his paper, “The Nature of the Firm”, sought an explanation for this. “Why do firms exist” is the sort of simple question out of which great truths emerge, and it is a question that was especially important because, as Coase pointed out in his Nobel Prize lecture, “most resources in a modern economic system are employed within firms”, where the allocation of resources depends “on administrative decisions and not directly on the operation of a market”. The answer, he found, lay with transaction costs: firms exist because the price mechanism is not costless, it carries transaction costs such as the cost of discovering the relevant prices of factors of production, the cost of negotiating and concluding contracts for each transaction, and the cost to making long-term forecasts about the deployment of factors of production. These transaction costs are the children of uncertainty; absent uncertainty, firms cannot exist because there is no need to try and submerge these costs down by replacing them with intra-firm costs.
Nevertheless firms cannot keep growing and replacing the market’s transaction costs with its own intra-firm costs. The limits to a firm’s horizontal or vertical integration are the decreasing returns to scale that that firm must experience at some point, until the costs of organising a transaction are equivalent to having that transaction done through the market. There may also be rising supply prices for factors of production such that it is cheaper to produce in a smaller firm than to continue expanding. There are also a class of transactions that are simply too costly for a firm to manage on its own and which are always better managed by the price mechanism. Given the irreducible uncertainty inherent in economic activity, it is also true that firms are liable to engage in value destroying activities. All firms must wrestle with the fact that wealth, in the long run, is far more likely to be destroyed than created.
Coase concluded that,
Other things being equal, therefore, a firm will tend to be larger:
(a) the less the costs of organising and the slower these costs rise with an increase in the transactions organised.
(b) the less likely the entrepreneur is to make mistakes and the smaller the increase in mistakes with an increase in the transactions organised.
(c) the greater the lowering (or the less the rise) in the supply price of factors of production to firms of larger size.
“The Nature of the Firm”, Ronald Coase
With the launch of the transaction cost approach, Coase established transaction costs as the basic unit of analysis of firms, a unit that successfully explained why firms exist, what the limits to their size are, and the sort of market structures that can emerge. Oliver E. Williamson, reflecting on the approach, divined three levels of analysis: one that takes firm size as given and studies how the operating units relate to each other; one that seeks to determine the “efficient boundary” separating firm and market, which is to say, what activities should be conducted within and without the organisation; and one that assess how human assets are used.
By succeeding, firms are able to create value, which is to say that they grow their revenues and earn economic profits. By earning economic profits, defined as a return on invested capital (ROIC) in excess of the opportunity cost, scaled by invested capital, incumbents attract competition from entrants greedy for the economic profits on offer. In Edward Chancellor’s examination of Marathon Asset Management’s capital cycle framework, Capital Returns, he expresses this idea beautifully, saying,
Typically, capital is attracted into high-return businesses and leaves when returns fall below the cost of capital. This process is not static, but cyclical – there is constant flux. The inflow of capital leads to new investment, which over time increases capacity in the sector and eventually pushes down returns. Conversely, when returns are low, capital exits and capacity is reduced; over time, then, profitability recovers. From the perspective of the wider economy, this cycle resembles Schumpeter’s process of “creative destruction” – as the function of the bust, which follows the boom, is to clear away the misallocation of capital that has occurred during the upswing.
Capital Returns, by Edward Chancellor
It is commonly believed that the chief purpose of strategy is to make probabilistic bets to build, defend, and extend incumbent competitive advantages, which, as Bruce C. Greenwald and Judd Kahn show, are synonymous with barriers to entry. Barriers to entry are costs faced by potential entrants but not by incumbents. Without mounting significant barriers to entry, economic profits are vulnerable to being competed away. Yet, it is truer to say that firms seek first to reduce transaction costs, accepting trade-offs between production cost economies, where the market holds certain competitive advantages, and governance cost economies, where internal organisation holds certain competitive advantages. In terms of market structure, this is done through discovery of the optimal configuration of the value chain, either through horizontal or vertical integration of the value chain. Where the goal of vertical integration is reducing the costs of distribution, a company integrates forwards, integrating distribution centres and retailers selling its products. Where the goal of vertical integration is reducing the costs of production, a firm integrates backwards, integrating suppliers. Where the company aims to reduce both costs of production and distribution, it practises a balanced form of vertical integration. Whichever form of integration firms pursue, they are motivated by an economising imperative to reduce transaction costs.
Information, Market Structure, and Digital Firms
The impact of the internet on the economy has been to decouple matter from information and expand the production possibility frontier. New markets, industries and economic sectors have emerged and continue to emerge, as capital and entrepreneurial planning have launched toward technological infrastructure. Firms have and continue to develop new ways of capturing value, and new forms of economic transactions have emerged and continue to emerge. The possibilities flung open by the internet have also created new forms of firms, firm-types which enjoy increasing returns -the tendency of what is advantaged to gain further advantage and what is disadvantaged to to be further disadvantaged-, and which are compelled to enhance consumer welfare. The Internet Revolution is the most consequential economic transformation of the world since Johannes Guttenberg developed modern movable type printing in 1440.
Traditionally, under a monopoly, the monopolist sets the price for all, and, driven by profit-maximisation, tends to seek market prices higher than would be obtained under other market structures, while eroding the quality of goods and services it provides. Profit maximisation leads the monopolist to set a price greater than the marginal cost and which equals marginal revenue. In doing so, the monopolist does not serve consumers who value their goods and services at less than the market price, creating a deadweight loss, which refers to potential gains that are unearned by either the monopolist or the consumers. As such, deadweight loss represents the market’s inefficiency, and society’s loss due to the monopoly structure.
In a world of scarce supply and significant marginal costs, the monopolist’s pursuit of a producer surplus is at odds with the consumer’s pursuit of a surplus, and, given the power of the monopolist, the producer surplus is enlarged and the consumer surplus reduced and prone to decline. Consequently, as Lina M. Khan noted in her note, “Amazon’s Antitrust Paradox”, antitrust law has focused on “the short-term interests of consumers, not producers or the health of the market as a whole; antitrust doctrine views low consumer prices, alone, to be evidence of sound competition.” Of course, it is obvious, as Khan, Thompson and others have observed, that when marginal cost is equal to zero, then all that exists is consumer surplus, that, rather than degradation of services and exploitation of consumers, there is the enhancement of consumer welfare. While Khan is correct in observing that, “gauging real competition in the twenty-first century marketplace—especially in the case of online platforms—requires analysing the underlying structure and dynamics of markets (…) a company’s power and the potential anticompetitive nature of that power cannot be fully understood without looking to the structure of a business and the structural role it plays in markets”, the underlying assumption of the curse of bigness, betrays a widespread failure to appreciate the uniqueness of digital firms.
In 1994, Kenneth J. Arrow realised that information is “almost the exclusive basis for value” in digital firms and digital goods, a fact that has impacted the utility of old theories of value. There and in a 1996 paper, Arrow saw that in the prior thirty years, standard economics had wrought a rich theory of asymmetric information, where one party faces costs to obtain information, assumed to be scarce, that the other party cannot. Yet, where Shannon’s communication theory, and decision theory have developed powerful analyses of information as a random variable, economics then and even now, has a paucity of research on information as a choice variable. Information is an economic good like other commodities, being costly and valuable, however, information is special because it generates increasing returns. Arthur, in his remarkable book, Increasing Returns and Path Dependence in the Economy, began it by saying that,
Conventional economic theory is built on the assumption of diminishing returns. Economic actions engender a negative feedback that leads to a predictable equilibrium for prices and market shares. Such feedback tends to stabilise the economy because any major changes will be offset by the very reactions they generate. The high oil prices of the 1970s encouraged energy conservation and increased oil exploration, precipitating a predictable drop in prices by the early 1980s. According to conventional theory, the equilibrium marks the “best” outcome possible under the circumstances: the most efficient use and allocation of resources.
Increasing Returns and Path Dependence in the Economy, by W. Brian Arthur
Although there are examples of how information leads to increasing returns to scale, such as with Adam Smith’s theory of the benefits of the division of labour, much of standard economics is blind to information, with Arrow pointing to the “analytic difficulties which would follow introducing information as an economic variable”, adding that
Increasing returns can occur for other reasons than information. But with information, constant returns are impossible. Two tons of steel can be used as an input to produce more than one ton of steel in a given productive activity. But repeating a given piece of information adds nothing. On the other hand, the same piece of information can be used over and over again, by the same or a different producer. This means both that the way information enters the production function is different than the way other goods do and that property rights to information take on a different form. These remarks are obvious enough, but their implications are not.
To elaborate the point, the usual logic of the price system depends on constant returns. For conventional inputs, the buyer can buy more or less at a given price (or at least close to it if there are elements of monopoly). But information is different. Technical information needed for production is used once and for all. The same information is used regardless of the scale of production. Hence, there is an extreme form of increasing returns.
“Technical Information and Industrial Structure”, by Kenneth J. Arrow
Information as the central economic good of a market results in markets defined by positive, rather than negative feedback, and this has clear effects on the kinds of structures and dynamics that emerge. Arthur gave the following example:
The history of the videocassette recorder furnishes a simple example of positive feedback. The VCR market started out with two competing formats selling at about the same price: VHS and Beta. Each format could realise increasing returns as its market share increased: large numbers of VHS recorders would encourage video outlets to stock more pre-recorded tapes in VHS format, thereby enhancing the value of owning a VHS recorder and leading more people to buy one. (The same would, of course, be true for Beta-format players.) In this way, a small gain in market share would improve the competitive position of one system and help it further increase its lead.
Such a market is initially unstable. Both systems were introduced at about the same time and so began with roughly equal market shares; those shares fluctuated early on because of external circumstance, “luck,” and corporate manoeuvring. Increasing returns on early gains eventually tilted the competition toward VHS: it accumulated enough of an advantage to take virtually the entire VCR market. Yet it would have been impossible at the outset of the competition to say which system would win, which of the two possible equilibria would be selected. Furthermore, if the claim that Beta was technically superior is true, then the market’s choice did not represent the best economic outcome.
Increasing Returns and Path Dependence in the Economy, by W. Brian Arthur
Although some parts of the resource-based economy are subject to increasing returns, resource-based economies are largely defined by diminishing returns. It is in the knowledge economy where increasing returns are definitive. Importantly, whereas negative feedback loops exert a gravitational pull toward the initial equilibrium, and hinder the emergence of a leader, positive feedback loops, push the market away from the initial equilibrium, fostering competition, technological change, innovation, and new business models, and, in the long run, may reduce uncertainty through winner-take-all effects, market tipping and path dependence. The nature of an industry’s returns, then, determines whether its structure is monopolistic, a monopoly, or oligopolistic. In the presence of increasing returns, monopolies and oligopolies are the natural outcomes, because, not only do firms gain in efficiencies as they grow larger, but consumers gain in utility the more they consume. By dint of merit or accident, consumption agglomerates around the happy few firms who are anointed winners.
Indeed, winners may emerge for random, contingent reasons that benefit inferior products or services, such as QWERTY’s victory over Dvorak Simplified keyboards, even as lock-in may emerge because a product or service is superior, such as Apple’s suite of integrated products. This randomness makes it hard to predict winners and losers and the knock-on effects of such success. Did the authors of the Section 230 of the 1996 Communications Decency Act foresee the rise of online “town squares” such as Twitter, Facebook, and Instagram, or that such town squares would become battlegrounds over what is true, and whether false or even harmful statements should be struck down? In 2007, Facebook launched Facebook Platform, with Zuckerberg asserting that, “Right now, social networks are closed platforms. And today, we’re going to end that”, and yet, rather than develop a platform that could go toe-to-toe with Apple and Google, the whole effort was a gigantic, missed opportunity. It was mobile, a technology outside Facebook’s control, that turned Facebook into such a juggernaut that in his first earnings call in 2012, Zuckerberg could claim that, “Mobile is a huge opportunity for Facebook. Our goal is to connect everyone in the world”.
Information also implies that the owner can allow others to use it without ceding ownership, and without destroying that information. Given that reproduction of information is cheaper than its production, intellectual property rights exist to create artificial scarcities of information to motivate firms to acquire information. Nevertheless, information is diffused, through labour mobility, through markets, publication, and informal, interpersonal contacts. Information’s tendency to cheap or even costless diffusion makes it difficult to turn it into a property, information wants, as it were, to be free, it is a fugitive resource, overlapping firms, in part because of the limited mobility of a firm’s workers, resulting in “an increasing tension between legal relations and fundamental economic determinants”, to quote Arrow.
Platforms and Aggregators
Consider the following thought experiment:
In a village of one thousand people, there is one print newspaper, in which two journalists work. Computers, mobile phones, and the internet have not yet been invented. In that village, aside from those journalists, there are three hundred people who would, if given the opportunity, write for the rest of the village to read. In this village, it is only possible to publish through the newspaper, and so, only these two happy journalists write and publish anything. The costs of competing with the newspaper are prohibitive. In fact, the newspaper only exists thanks to the largesse of a wealthy retiree who settled there. As time passes, a gale of creative destruction passes over the village, ushering in computers, mobile phones, and the internet. Suddenly, anyone can publish! Bad poems, misjudged and angry articles, and thoughtful, perhaps even moving content. The cost is nothing, so near zero one might as well call it zero. It costs nothing to post, nothing to share. Suddenly, our newspaper of two journalists is faced with three hundred competing writers. Even if the majority of the content produced by these intrepid three hundred is bad, the law of large numbers tells us that this group will produce more quality content than the newspapers. The newspaper cannot compete because its costs of competition go up: the only way to compete is by hiring three hundred writers! The market has no costs to compete, the newspaper’s costs mount. Supposing this newspaper was led by a young man named Mark Zuckmayer, who realised that there was more value being created outside the firm compared to within, and that the costs of production were cheaper outside the firm compared to within, surely he would say, “The Internet Revolution forces me to push production outside the firm, to the market, where it is cheaper to produce. The trick is to capture that value!” This shape-shifting business, a unicorn among firms, would transform itself from a centre of production, to a kind of coordinator of market activity. This is the story of the Internet Revolution. It is not just that transaction costs tumbled toward zero, it is that it necessitated a new type of firm, a firm that was not a centre of production, but the centre of a network.
While the pre-internet era was defined by scarce supply, abundant consumers and users, and significant transaction costs, the Internet Revolution ushered in a special sort of multi-sided platform company, who not only matched supply and demand providing each with network benefits in traditional ways, the way credit cards match cardholders with merchants, or the yellow pages matched advertisers and consumers, but who boasted abundant supply, abundant consumers and users, and, crucially, enjoyed zero transaction costs. The key “event” of the Internet Revolution, the end of transaction costs, and its corollary, mounting transaction costs for legacy incumbents, turned a millenia old business model into the most profitable business model in the history of the world. For such platform companies or “matchmakers” demand is the appropriate lens of analysis. In a world of no transaction costs, serving the world from Day One is achievable, and the battle is over scalability, which rests upon a business’ ability to deliver the best available user experience. Enhancing consumer welfare is the inevitable and singular goal of the business.
Alongside these platforms, were what Thompson describes as an “aggregator”: a firm that has all of the following qualities: a direct relationship with its users; zero marginal costs for serving them; and a demand-driven multi-sided network with decreasing acquisition costs. He calls them “aggregators” because they “aggregate modularized suppliers — which they often don’t pay for — to consumers/users with whom they have an exclusive relationship at scale”. In his discussion of “Aggregator Theory”, Thompson describes how in the pre-internet era, value creation depended upon establishing horizontal dominance, as a monopoly or as part of an oligopoly, in one of the three links in the value chain, supply, distribution, and consumers and users; or, by integrating backwards into supply to offer a vertical solution. For printed newspapers, for instance, advertisers were modularised, and supply, in the form of content they created, was integrated with advertisements, which they delivered to their readers within some locale. Vertical solutions largely depended on controlling distribution and leveraging relationships with suppliers. In the internet era, where distribution is free, transaction costs are near-zero, and the addressable market is the entire globe, value creation depends upon controlling and scaling the user relationship, and supply is both commoditised and gargantuan. The more users there are, the greater the value users enjoy, and the more attractive the aggregator is for suppliers, creating a virtuous cycle. The network’s winner-take-all-effects is the prize aggregators compete for, and to win it, companies do not try to control scarce resources, but to control demand for them.
Meta, the subject of my next blog post, is not merely an aggregator, it is a “super aggregator”, sharing that title with just one other company: Alphabet. As super aggregators, not only do Alphabet and Meta have near-zero transaction costs for serving their end users, they also enjoy near-zero transaction costs with respect to both suppliers and advertisers. For Meta, its suppliers are its users, who provide content freely, for which Meta has exclusive access. Where publishers in the pre-internet era integrated content and advertisements, as an aggregator, it has modularized advertisements, integrating its ad inventory and profile data in order to programmatically deliver finely targeted ads through an advertising network that directly match advertisers and adverts to their most probable customers. By commoditising and modularising pre-internet publishers’ integration of advertisers and content, Meta has been able to use its integration of advertisers and adverts to earn attractive economic profits. As Clayton Christensen said in The Innovator’s Solution, as he developed what he first called the “law of conservation of attractive profits”, and later, the law of conservation of modularity:
Formally, the law of conservation of attractive profits states that in the value chain there is a requisite juxtaposition of modular and interdependent architectures, and of reciprocal processes of commoditization and de-commoditization, that exists in order to optimise the performance of what is not good enough. The law states that when modularity and commoditization cause attractive profits to disappear at one stage in the value chain, the opportunity to earn attractive profits with proprietary products will usually emerge at an adjacent stage.
The Innovator’s Solution, by Clayton Christensen