General equilibrium theory

• 1. Introduction

  ›

  General equilibrium theory asks whether every market in an economy can clear at once, with supply matching demand everywhere at the same time. In a 1979 article, economist Nicholas Georgescu-Roegen took aim at the field. He wrote that some of its work amounted to "just plain mathematical exercises, not only without any economic substance but also without any mathematical value." As an example, he pointed to a paper that assumed more traders existed than there are points in the set of real numbers. That complaint captures a tension running through this entire field. Economists spent decades building a rigorous mathematical account of how prices coordinate an entire economy. Others have asked whether that account describes anything a real economy actually does. What follows traces how that account was built, what it can prove, and where its own architects found it could not answer its own questions.

• 2. Walras's Founding Model

  ›

  French economist Léon Walras published Elements of Pure Economics in 1874. It was the first attempt in neoclassical economics to model prices across an entire economy at once. The book built up through a series of models, starting with two commodities and gradually adding many commodities, production, growth, and money. Some economists have argued the project fell short, and that Walras's later models in that series are inconsistent.

  Walras's model was built for the long run. In it, a capital good's price stays the same whether it serves as an input or an output. Every industry also earns the same rate of profit in this model. That structure conflicted with treating capital goods quantities as fixed. When Walras later introduced capital goods into his models, he simply took their quantities as given, in arbitrary ratios. Kenneth Arrow and Gérard Debreu instead kept the initial quantities of capital goods as given. But they built a short-run model, where capital prices and the rate of interest could vary over time.

  Walras set an agenda that most 20th-century economists in this field went on to follow, including whether equilibria are unique and stable. In his own book, Walras's Lesson 7 fails to establish uniqueness, stability, or even the existence of an equilibrium. He also proposed a dynamic process for reaching general equilibrium, which he called the tâtonnement, or groping process.

  In the tâtonnement process, an imagined auctioneer announces prices, and agents state how much of each good they would offer or wish to buy. No transactions or production happen at these disequilibrium prices. Instead, prices fall for goods with excess supply and rise for goods with excess demand. Walras could never give a definitive answer to whether this process would actually terminate in equilibrium.

  That gap between the model and a working market economy would resurface decades later, in a dispute over the shape of an ordinary supply curve.

• 3. The Sraffa Challenge

  ›

  Piero Sraffa showed, in the late 1920s, that economists following Marshall could not explain why a consumer good's supply curve sloped upward. That demonstration pulled Anglo-American economists toward general equilibrium through the 1930s. Marshall's own theory of supply and demand fixes the price of one good while holding all others constant. It works only when a shift in demand for that good does not also shift its own supply curve.

  If an industry uses only a small share of some factor of production, a modest rise in its output will not bid up that factor's price. The industry's supply curve then stays essentially flat. If an industry instead relies heavily on that factor, growing its output raises costs and tilts the supply curve upward. Sraffa pointed out that such a factor is often shared with substitute goods. A shift in demand for the original good then ripples into the supply of its substitutes, and back into its own supply curve again. General equilibrium theory exists to trace exactly this kind of interaction between markets.

  Continental European economists advanced the field further during the 1930s. Walras had defended general equilibrium mainly by counting equations against variables, an approach that breaks down for non-linear systems. It can also yield meaningless negative prices or quantities. Replacing some equations with inequalities, and applying more rigorous mathematics, strengthened general equilibrium modeling going forward.

  That firmer mathematical footing set up a full reformulation of the theory during the 1950s.

• 4. The Arrow-Debreu-McKenzie Model

  ›

  Kenneth Arrow, Gérard Debreu, and Lionel W. McKenzie jointly built the modern conception of general equilibrium during the 1950s. Debreu laid the model out in his 1959 book Theory of Value as a fully axiomatic system. He followed the mathematical style associated with the group known as Nicolas Bourbaki. In that approach, terms like goods and prices are not pinned down by the axioms themselves. That openness leaves room for several different readings of what the model actually describes.

  One reading distinguishes commodities by where they are delivered, turning the model into a spatial account of trade between locations. A second reading distinguishes commodities by when they are delivered. Under this reading, all markets clear once, at some initial moment, and agents trade contracts specifying a good and its delivery date. Forward markets exist for every good at every date, but no markets exist at any point in the future.

  A third reading lets a contract specify the state of nature on which delivery depends. Debreu described this in Theory of Value, writing that such a contract yields "a theory of risk free from any probability concept."

  These three readings can combine into a single model, where goods are identified by delivery date, delivery place, and delivery circumstance all at once. The result is a complete price for something as specific as one ton of Winter red wheat. That contract might specify delivery on the 3rd of January in Minneapolis, contingent on a hurricane in Florida during December. Supporters treat this complete-markets version as a useful simplified guide, even though critics note it sits far from how any real economy works.

  Some recent research instead studies incomplete markets. These describe an intertemporal economy that faces uncertainty, without contracts detailed enough to let agents fully allocate consumption and resources through time. Such economies generally still have an equilibrium. That outcome, however, may no longer be Pareto optimal, since consumers lack a way to tie prices to the relevant marginal rate of substitution. Under some conditions, the economy can still be constrained Pareto optimal, meaning a central authority using the same limited contracts could not improve on the outcome. What would be needed instead is a full set of possible contracts. That gap points to inefficiency rooted in underdeveloped financial institutions or credit constraints facing some members of the public.

  None of this yet asks whether such an equilibrium treats every consumer fairly or efficiently.

• 5. The Welfare Theorems

  ›

  The First Fundamental Theorem of Welfare Economics states that market equilibria are Pareto efficient. That means no reallocation of goods could make one consumer better off without making another worse off. In a pure exchange economy, this result holds as long as preferences are locally nonsatiated. It also holds for economies with production, regardless of the shape of the production function. The theorem implicitly assumes complete markets and perfect information, so in an economy with externalities, equilibria can turn out not to be efficient after all.

  This theorem matters because of what it implies about blame. Under its assumptions, any market equilibrium is efficient by definition. So when an equilibrium does turn out inefficient, the fault lies with some specific market failure, not with markets in general.

  The Second Fundamental Theorem goes further, stating that any Pareto efficient allocation can be supported as an equilibrium under some set of prices. Not every efficient allocation will already be part of one on its own. In practice, this means reaching a chosen efficient outcome only requires redistributing agents' initial endowments and then leaving the market alone. That suggests efficiency and equity can be treated as separate concerns, rather than as a trade-off. The conditions behind this theorem are stronger than the first. Consumer preferences and production sets now must be convex. One description compares convexity to saying that "the average of two equally good bundles is better than either of the two bundles."

  Neither theorem promises that such an equilibrium exists in the first place, or that only one of them could exist.

• 6. Existence, Uniqueness, Stability

  ›

  To guarantee an equilibrium exists at all, it is enough for consumer preferences to be strictly convex. With enough consumers in the economy, that requirement can be relaxed. Convex feasible production sets can also guarantee existence, though this is less plausible, since convexity rules out economies of scale. Proofs of existence traditionally rely on fixed-point theorems. These include the Brouwer fixed-point theorem for functions, and the more general Kakutani fixed-point theorem for set-valued functions. Lionel McKenzie, along with Kenneth Arrow and Gérard Debreu, first supplied that proof.

  Uzawa later proved the reverse relationship, deriving Brouwer's fixed-point theorem directly from Walras's law. Following that result, many mathematical economists have judged proving existence to be a deeper achievement than proving either welfare theorem. A separate method, called global analysis, proves existence using Sard's lemma and the Baire category theorem. Gérard Debreu and Stephen Smale pioneered that approach.

  In 1969, economist Starr applied the Shapley-Folkman-Starr theorem to prove that an approximate equilibrium exists even without convex preferences. The Shapley-Folkman-Starr results measure the distance between such an approximate equilibrium and the equilibrium of a fully convexified economy. That bound applies once the number of agents exceeds the number of goods being traded. According to economist Guesnerie, these results were afterward much exploited in the theoretical literature. Nonconvexities in preferences, he noted, do not undo standard results like Debreu's theory of value. This holds so long as the consumption side of the economy is large enough. Guesnerie added, in a footnote, that working out these results in general form was one of the major achievements of postwar economic theory.

  The Sonnenschein-Mantel-Debreu theorem, proven during the 1970s, showed that an aggregate excess demand function inherits only a few properties from individual demand functions. These are continuity, homogeneity of degree zero, Walras' law, and a specific boundary behavior when prices approach zero. Any function meeting just those conditions could represent the excess demand of an economy full of rational, utility-maximizing individuals, which makes uniqueness far from automatic. Under mild assumptions, though, the number of equilibria will be finite and odd. Either the revealed preference property or the gross substitute property, applied to the whole economy, is enough to guarantee the equilibrium is unique.

  A regular economy is one where equilibria are finite in number, and therefore locally unique. That lets economists apply comparative statics, as long as shocks to the system stay modest. Debreu offered a reassuring finding: most economies qualify as regular. Economist Michael Mandler challenged that claim in 1999. He showed that when production is modeled through fixed coefficient processes, rather than continuously differentiable ones, agents such as firms drive endowments over time toward a continuum of equilibria. Mandler warned that this indeterminacy is not just a technical nuisance. It undermines the price-taking assumption of competitive models. Small manipulations of factor supplies can sharply raise a factor's price, discouraging factor owners from treating prices as fixed.

  Stability asks a related question: after some shock disturbs prices that once cleared every market, does the economy converge back to the same outcome. If more than one equilibrium exists, some of those equilibria will necessarily be unstable. An unstable equilibrium disrupted by a shock will settle into a different set of prices and allocations. Whether convergence happens at all depends not just on how many equilibria exist, but on the specific process guiding price changes, such as the Walrasian auction.

  These proofs all assume that the information every agent needs is somehow available from the start, an assumption one economist attacked directly in 1945.

• 7. Hayek's Knowledge Problem

  ›

  In a 1945 essay called The Use of Knowledge in Society, economist Friedrich Hayek challenged a core assumption behind general equilibrium theory. Hayek argued that economic knowledge is scattered across countless individuals. Much of it exists only in tacit, situation-specific forms that resist being gathered in one place. That posed a problem for any model, Walrasian or centrally planned, that assumes complete information could be collected centrally.

  Hayek proposed instead that market prices work as decentralized signals. These signals compress scattered knowledge about preferences, resources, and opportunities into simple summary statistics. Those price signals, he argued, coordinate an entire economy's decisions without requiring anyone to hold centralized knowledge or give centralized direction. His essay predates the full Arrow-Debreu formalization of 1954. Later economists have still read it as a critique of perfect-information equilibrium models. They also read it as an account of how real markets coordinate activity despite widespread ignorance and uncertainty.

  That emphasis on ongoing discovery, rather than a fixed equilibrium state, is one thread in a wider set of objections other economists have raised against the theory.

• 8. Critics And Computation

  ›

  Some economists have argued that a model built around the tâtonnement process describes a centrally planned economy, not a decentralized market one. Economist Frank Hahn pushed the question further. He asked what has come to be called Hahn's problem: can anyone construct an equilibrium where money actually has value? Hahn investigated whether general equilibrium models could include money in some essential way. The Arrow-Debreu-McKenzie framework itself, despite using an arbitrary numéraire, does not include money at all.

  Hahn also defended general equilibrium modeling on different grounds, describing its value as a negative function. General equilibrium models show what an economy would need to look like for an unregulated market to reach Pareto efficiency. That benchmark still matters even when real economies never fully meet it.

  Until the 1970s, general equilibrium analysis stayed purely theoretical. Growing computing power and new input-output tables then let economists try to model entire national or global economies numerically. Herbert Scarf pioneered applied general equilibrium models in 1967, offering the first method to solve the Arrow-Debreu system numerically. John Shoven and John Whalley, two of Scarf's students at Yale, implemented that method in 1972 and 1973. These applied models were popular through the 1970s. They fell out of favor in the 1980s, since they could not deliver precise solutions and cost too much to compute.

  Computable general equilibrium models replaced their applied predecessors in the mid-1980s. They delivered faster, larger results for a whole economy, and became the preferred tool of governments and the World Bank. The terms 'AGE' and 'CGE' are still used interchangeably in the literature. But no Scarf-style applied model has been built since the mid-1980s. Current CGE work instead rests on static, simultaneously solved macro balancing equations drawn from the standard Keynesian macro model, rather than on Arrow-Debreu theory directly.

  Keynesian and Post-Keynesian economists have rejected general equilibrium theory outright, along with much of neoclassical economics generally. They argue that real economies are never actually in equilibrium, and that reaching one, if it happened, could be slow and painful. Economist Robert Clower pushed for a reformulation toward disequilibrium analysis. He argued that monetary exchange fundamentally changes how an economy should be represented, compared with a barter system. New classical macroeconomics took the opposite path, building its macroeconomic theory directly on general equilibrium foundations. Its best-known model, real business-cycle theory, treats business cycles as driven by changes in the real economy, not by markets failing to reach potential output.

  Within socialist economics, the book Anti-Equilibrium mounted a sustained critique of general equilibrium theory. It drew on János Kornai's experience with the failures of Communist central planning. Economists Michael Albert and Robin Hahnel later built their Parecon model on that same critical foundation. A separate strand, new structural economics, instead extends the theory through a matrix-form computable general equilibrium model built on John von Neumann's general equilibrium model. That structural model can be computed using the R package GE, and can incorporate taxes, money, and endogenous production functions and institutions.

  When production functions or economic institutions become variables inside that structural model, the resulting equilibrium takes on its own name: structural equilibrium.