Is Economics a Failure?

Alexander Rosenberg’s Blunt Instrument is a provocation, by intention. Its central claim starts with the recognition that economics systematically fails to predict real-world outcomes. But then Rosenberg goes on to argue that economic theory, in its classic rational-choice and equilibrium form, is structurally incapable of becoming a genuinely predictive science. Economists have spent more than a century elaborating progressively more elegant—and, to be honest, more interesting—mathematical models. The problem is that economists can only predict the past, especially the recent past, using these models. The inability to make any sort of useful predictions about the future is a problem, both practically and as a matter of scientific method.

Yet the book is not an outsider’s uninformed attack, because Rosenberg knows economists’ methods and approach. (Candor requires that I disclose that Rosenberg is a longtime friend, and someone with whom I co-teach classes in the “PPE” program he helped found. His Routledge book, The Philosophy of Social Science, is now in its fifth edition, and has nearly 2,500 citations). Blunt Instrument’s claim is more interesting and constructive: despite the manifest failures of economics and the limits even on its successes, economic theory is indispensable. Economics gave rise to game theory and mechanism design, the tools we need for designing institutions that can cope with strategic, self-interested human behavior. It is more appropriate, in Rosenberg’s view, to think of economics not as a science of prediction, but rather a practical instrument of institutional diagnosis and self-defense.

The book’s construction reflects that thesis. Rosenberg begins with why non-specialists should care about economic theory, then moves on to rational choice theory, to Adam Smith, to the mathematization of economics, money and macroeconomics, profit, and finally to game theory, financial markets, and institutional design. The table of contents alone shows that this is not a narrow methodological tract; it is a guided tour of the conceptual foundations of modern economics, culminating in the claim that forecasting is not what economic theory is meant to do. Instead, economics is the only substantive tool available for developing even a limited understanding of rules, incentives, and decision environments. In Rosenberg’s own public discussion of the book, he says bluntly that economics is “not a science” in the strong predictive sense, but “a tool that we use for coping with one another’s behavior,” with game theory being the indispensable part for institutional design.

That’s the first major strength of the book: clarity of purpose. Rosenberg is succeeding in doing something few writers even attempt: explaining to non-economists what economic theory is for. The MIT Press description emphasizes that the book “draws back the curtain” on the equations and graphs of micro and macro models, and that seems exactly right. Rosenberg is a philosopher of science, and he brings that training to bear in a way that forces readers to ask a basic but neglected question: Why do rational-choice assumptions persist if they are descriptively false? Too much writing on economics alternates between technocratic boosterism and theatrical anti-market caricature.

The second major strength is the book’s positive turn toward game theory and mechanism design. These are means of analyzing strategic interactions and designing rules that channel self-interest toward workable outcomes. The relevant solution concept is not correct prediction but trial and error. Still, having a theoretical scaffolding within which to conduct trials and see what happens is essential. In his discussion of the book, Rosenberg highlights auctions, hospital matching, school assignment, kidney exchange, and antitrust as examples of domains where theory matters because it helps build incentive-compatible institutions. This gives the book a constructive spine. Rather than ending with “economics is broken,” Rosenberg ends with “economics is useful for a different reason than economists often say.” That is a serious and worthwhile thesis.

The Problem

My criticism of the book is more fundamental and likely unfair, because it is more valid as a critique of the models than as a book about those models. I was for much of my professional life persuaded of the value of the standard neoclassical model, the existence proofs for a competitive economy, and the worth of the two “welfare theorems.” The advantage of the equilibrium theory is that it reveals the usefulness of market processes, because generating prices reveals information about the relative scarcity of resources, and if exchange is allowed to take place, then “utility is maximized,” in the language of the modern scholastics.

Further, and no less important, having a benchmark marking the equilibrium ideal gives state actors valuable information about how actual market processes fall short, often well short, of the ideal. That means that the state can then target these “market failures,” and policy can be optimized.

But I now realize that this entire model is intellectually bankrupt, and the dirigiste support for equilibrium theory and the welfare theorems is deeply instrumental, even disingenuous. There are two reasons this is true.

First, equilibrium analysis is nonsense. I have spent much of the last year writing and recording eleven podcast episodes on the Scottish Enlightenment and the writings of Adam Smith. The total is nearly 250,000 words, more than 13 hours. Smith’s central argument is that the system of natural liberty, combined with the development of a population capable of self-command in service of justice and propriety, will result in the growth of “opulence,” his word for prosperity. The primary driver of prosperity is the increasing returns that accrue to division of labor. In technical parlance, this implies sharply increasing returns to scale over much of the production range for a wide variety of commodities. But the vaunted theorems that demonstrate the existence of equilibrium explicitly rule out sharply increasing returns.

What that means is that speculations based on equilibrium theories are the equivalent of the medieval “school men” investigating hypothetical questions of how many angels can dance on the head of a pin. Who cares? National economies and international trade systems are almost never “in equilibrium,” and if they happen on temporary stability, the duration of the equilibrium and the direction of change of the nearly immediate disruption are unknowable.

What is strange is that the conclusion is widely known and, in fact, uncontroversial. The skepticism of equilibrium theory shown by Austrian economists is well established. But no less an “anti-Austrian” than John Maynard Keynes “was indeed anti-equilibrium,” as E. Roy Weintraub put it in his 1975 History of Political Economy paper, “‘Uncertainty’ and the Keynesian Revolution.” Nick Kaldor went further, in his 1972 Economic Journal paper, “The Irrelevance of Equilibrium Economics,” arguing that the Keynesian approach is essentially a negation of the Walras-Arrow-Debreu project of demonstrating that general equilibrium exists and has desirable properties, especially in labor markets.

The two central Keynesian claims, in fact, are (a) investment and production are highly uncertain, to the point of being incapable of generating accurate pricing, and (b) markets, and especially labor markets, can exhibit stability over considerable periods without in any way “clearing,” or showing a tendency toward price adjustments that would eliminate excess demand or excess supply. Stability is not equilibrium, at least not equilibrium in the sense implied by the models that Rosenberg wants us to admire. It is one thing to say that the method of economics, using game theory, is useful; Rosenberg wants us to protect the results of that method, the models themselves. But those models are, at best, useless.

Second, in spite of the nonsense, dirigistes need equilibrium analysis to justify state control of the economy. If the model that assumes equilibrium has nothing to do with actual economies' function and change, why do “we need it anyway,” as Rosenberg’s subtitle suggests? I said above that the models are useless, at best; what are they at worst?

The answer is that economic models are explicitly designed as a scaffolding to justify government control and direction of the economy. The idea that private decentralized forces would be allowed to direct resources is anathema to Team Progressive. The general equilibrium model is based on a set of assumptions, often called the “perfect competition” assumptions. These include perfect information, no market power to set prices, no public goods, and no significant economies of scale. As economist John Ledyard famously noted, these assumptions are actually the list of “market failures,” each of which justifies a type of extensive and intrusive government control of aspects of the economy.

No remotely realistic model of commerce would countenance such nonsensical assumptions, of course. So the “need” for the nonsense model is to justify the nonsense assumptions to legitimate state control, which was the unspoken objective from the outset. Progressives are quite willing to accept an economy that performs far below its potential if this cost can be exchanged for the ability to use regulation and tax policy to achieve their policy objectives. A recent example helps illustrate this.

The “price takers” assumption means that no firm can be large relative to the market. The advantage of this assumption in the model is that it ensures competition, which disciplines market power and encourages low prices and high quality. But the only industries that can approximate this structure are commodities markets, with homogenous products—corn, pork bellies, rolled steel, etc.

The advantage of the received economic equilibrium theory is that it can be confected to imply that any move toward approximating “perfect competition” improves the function of the market. The recent resurgence of the nonsensical neo-Brandeisian school of antitrust takes this opportunity for control and direction fully to heart, holding that the main concern is the number of firms, not their viability.

The problem is that focusing solely on the number of firms reduces competition in the world outside the toy model. Exactly this issue was raised in the merger proposal between Spirit Airlines and JetBlue, first proposed in April 2022. Each company was faltering on its own, and the combined company could realize synergies in scheduling, administration, and connections that would allow it to lower prices.

But the merger proposal was rejected because it would reduce the number of airlines, giving fliers fewer choices and reducing “competition.” The Biden antitrust mandarins raised alarms that there would be one fewer line quoted on Orbitz or Expedia for the air routes that both companies flew. Several states also filed lawsuits challenging the merger on the grounds that it would diminish competition and harm travelers.

Both Spirit and JetBlue were left to reconsider their strategies separately; in November 2024, Spirit filed for bankruptcy. JetBlue continues to operate at a loss, planning additional cost cuts, flight reductions, and parked aircraft.

The very idea of “perfect competition” and “price-taking” in the airline industry is risible. What matters is having several firms large enough to be viable. Forcing the industry closer to the idealized notion of blackboard models used to mislead undergraduates is a scientifically unsupportable approach from the outset. What’s important to understand is that prices in an oligopolistic industry, with firms large enough to take advantage of scale economies, will actually be lower than the “competitive” prices in an atomistic, inefficiently structured industry that happens to conform to the imaginary world of economic models.

It is important to note, however, that my criticisms come from an outsider, someone who was catechized as a traditional economist but who has been an apostate for decades. Rosenberg is writing within the tradition that treats the model of perfect competition as given and desirable. Rosenberg notes that this model does not perform well descriptively or predictively, but it is nonetheless worth salvaging for the value of its insights into policy problems, particularly into institutions and mechanism design. That’s a valuable contribution, as far as it goes. But the critique must go deeper than Rosenberg seems willing to accept.

Michael Munger is Professor of Political Science and Economics at Duke University. His research focuses on the relations between political and commercial institutions.