PicoBlog

This question of what constitutes a natural monopoly is more important today than you may realize. It affects you if you are involved with consumption or production of electricity or of technology platforms, and that's all of us! I'd like to pick up the thread of natural monopoly theory that I started to spin a couple of weeks ago, motivated by Manuela Mosca's excellent article on the intellectual history of natural monopoly theory, apply this question to electricity, and drop some hints about natural monopoly theory and technology platforms but leave that topic for another day.

Technological change enabled the telecommunications industry to evolve away from being a natural monopoly, and the electricity industry is in the process of doing the same. Some people have been arguing that tech companies should be regulated as utilities, using a natural monopoly argument (they're incorrect, again a topic for another time ...). Whether you are a consumer, an industry professional in these industries, a regulator or policy maker, or a policy expert, knowing what does and does not constitute natural monopoly will be a useful tool for thinking about the policy implications of dynamic innovation.

As I mentioned before, Mosca does a great job of illustrating how the late 19th and early 20th century development of natural monopoly theory was historically contingent, intertwined with technological and economic changes. The late 19th century saw dramatic innovations in technologies that could operate at large scale, and indeed reduced average cost by operating at larger scale. Natural monopoly theory was embedded in a social and cultural context of the American Progressive Era, with farmers complaints about the transport rates railroads charged them, concerns about "big business" and the trust-busting impetus that emerged and culminated in several different strands of new competition policy that took the form of regulation and antitrust.

The intellectual history I explored in the first part of this series highlighted the natural monopoly concept's evolution up to about 1948. This period is the height of original work in neoclassical economics, carrying forward the foundational synthesis work of Alfred Marshall in the 1880s-1890s, through A.C. Pigou's codification and formalization of welfare economics in the 1920s. 1948 is a landmark year, when Paul Samuelson's economics textbook is first published. Samuelson famously picked up the ball from Marshall, Edgeworth, Pigou and others to embed the modern mathematical, formal approach to economic theory that would go on to define economic theory for the rest of the 20th century.

Samuelson's textbook characterizes natural monopoly as "decreasing average cost over the relevant range of demand", crystallizing the focus on economies of scale. As Mosca observes,

In the first edition of Samuelson's textbook (Samuelson 1948), and until the end of the 1970s, natural monopoly was considered to occur in cases where economies of scale exist over the entire range of market demand. In this literature, scale economies are attributed to high fixed costs and low or zero variable costs. Excess capacity is a characteristic of the natural monopoly here described, because the firm produces below its minimum efficient scale ... Scale economies are also considered a barrier to entry: due to the monopoly power derived from them, natural monopoly is seen as a market failure, and government intervention is required (in the forms of nationalization, regulation, or antitrust). Natural monopoly is more likely to occur where the market is small: a growth of market demand can shift the demand function to the right, and this could eliminate the monopoly. Moreover, as natural monopoly depends upon the technology employed, technological innovations can open it to competition and, again, eliminate it.

Here we can pick up my previous criticism, most notably how static the model is – for a given production technology and a given market demand, a natural monopoly exists if the average cost curve is downward sloping until it intersects the demand curve. But in reality, production technologies are way more complicated and change in ways that can affect lots of elements of production cost, including average cost. And in a flourishing, thriving, dynamic economy, demand for electricity increases, shifts out as more entrepreneurs create more ways for us to benefit from using electricity, including the devices on which I'm writing this and you're reading it.

The most important criticisms of this static approach, criticisms that are highly relevant to the 21st century regulatory and technology issues in electricity and tech industries, come in the late 1970s in the form of contestability theory. Primarily the work of William Baumol, John Panzar, and Robert Willig (BPW 1982), contestability theory was simultaneously a deeper analysis of the components of production costs and an attempt to reintroduce some dynamics into this overly static model [full disclosure: I went to graduate school to work with John Panzar and he served on my dissertation committee, although my Ph.D. work ultimately went in a different direction]. Their work changed the definition of natural monopoly.

On the components of production costs, BPW point out that it matters a lot whether the fixed costs are sunk costs. Once incurred, can those assets be repurposed or sold to another firm to produce something else? For the electric system, the capital costs, power plants and poles and wires etc., those physical assets are pretty much sunk costs (and here's one place where the argument that tech firms are natural monopolies falls very short). If those costs aren't sunk, a competitor could come in, spend that capital and incur those fixed costs, and if they fail and have to exit they can sell the assets to someone else and/or use them to produce something else. That threat of potential entry disciplines the incumbent firm's pricing and output decisions. Sunk costs, not fixed costs, are the source of the technology-based entry barrier.

Another question is whether the firm produces one product or multiple products with that set of assets. Traditionally in electricity, firms use those assets to produce multiple products. A vertically-integrated utility produces power, transportation, and various grid services (voltage regulation, frequency regulation, etc.), and power quality can vary in ways that amount to different products. If it's cheaper for that integrated firm to produce multiple products than it is for stand-alone firms to produce each one separately (C(A+B) < C(A) + C(B)), then that firm enjoys economies of scope. That means that it's not enough to say economies of scale => natural monopoly – to define natural monopoly you have to identify both economies of scale and scope. Economies of scope can arise out of economies of scale, so they are related but different.

One of the most important ideas that BPW introduce is the concept of subadditivity, which combines economies of scale and scope. Subadditivity requires both that (1) production by a single firm costs less and (2) the monopoly is sustainable because entry is not profitable. This second condition is where contestability reintroduces dynamics into the natural monopoly model, by requiring the analyst to ask whether entry could be profitable and whether therefore the market could be potentially competitive. Their work also shows that economies of scale are not necessary for natural monopoly. This definition is very different from the early 20th century definition of economies of scale as creating an entry barrier, because it's not economies of scale that create the entry barrier, it's the fact that the fixed costs are sunk and those assets aren't redeployable. Again Mosca summarizes the importance of this work clearly:

Not only does this theory demonstrate that scale economies do not help to define natural monopoly properly, but also that they alone may not constitute a barrier to entry: to deter entry, economies of scale must be associated with sunk costs. The policy implication of the theory of contestable markets is that ‘even in the absence of anything resembling pure competition, market forces are potentially a guardian of the general welfare far more powerful and effective than is commonly supposed’ (Baumol et al. 1982: 347). In other words, monopoly has no damaging effects in contestable markets, and there is no need for government intervention.

That BPW quote reminded me of a comment I've heard frequently from Pat Wood, former chair of the Public Utility Commission of Texas and the Federal Energy Regulatory Commission and now CEO of Hunt Energy Network: “Competition on its worst day is better than I ever could be as a regulator on my best day” (quoted in this very good recent RTO Insider article). It also reminded me of this old video I did talking about monopoly and goverment regulation. This implication that market forces can be a dynamic guardian of the general welfare even in the absence of "perfect" competition is a powerful insight that should inform our thinking about innovation and future regulatory and business models.

Technological innovation changes whether or not a market is a natural monopoly by changing the cost structure, by changing economies of scale, economies of scope, and whether or not the fixed costs are sunk costs. Think about how the invention of the combined-cycle gas turbine generator in the 1980s changed the cost structure of generation, reducing some of the economies of scale associated with large coal-fired, hydro, and nuclear generation and paving the way to competitive wholesale power markets. Other innovations change economies of scope, and most importantly, other innovations change what used to be sunk costs into recoverable fixed costs. BPW's analysis gives us the analytical tools to think more deeply about the nature of production costs and their implications for whether or not an industry continues to be a natural monopoly that should be subject to regulation.

One way that digital innovation can reduce sunk costs is through modular device and network architecture. Modular systems have standardized components, so that if one component fails it doesn't take down the whole system, you just replace that component with an identical device. Think of a server farm, for example. The electricity industry has a long history of one-off bespoke design, which will of course amplify the sunk cost nature of the investments. In contrast, digital networks evolved quickly into modular systems with standardized components and standard, mutually agreed-upon interfaces (like the USB interface, or TCP-IP in networking). As the electric system digitizes, it is likely to become more modular, but unless there are other uses for those production assets, they may still be sunk costs. And this is where my skepticism arises about whether tech platforms are natural monopolies. To make that argument stick you have to argue that the network effects they create are a form of sunk cost.

Reference: Baumol, William, John Panzar, and Robert Willig. 1982. Contestable Markets and the Theory of Industry Structure. New York: Harcourt Brace Jovanovich.

ncG1vNJzZmijnqTErbHDoJypqp%2BXuaa5jaysm6uklrCsesKopGioX5qwsLrOpqCcZZakwq%2BwwK2gqKajYruiwNSrmKVlnaS7sLzOpbBmbZVl