General Equilibrium and the Welfare Theorems Explained: How All Markets Clear at Once

Understanding general equilibrium — simply put

Most economic analysis looks at one market at a time: the market for coffee, or for rented flats, holding everything else fixed. General equilibrium looks at every market together. A rise in the price of gas changes what electricity costs, which changes the cost of making fertilizer and food, which changes what households have left to spend on restaurants, which changes wages in hospitality. A general equilibrium is a set of prices at which all of these markets clear at the same time, so no buyer or seller anywhere wants to change what they are doing.

How does it actually help?

The simplest case is an exchange economy: two people, two goods, no production. Anna starts with 8 units of good x and 2 of good y; Ben starts with 2 of x and 8 of y. Anna values the two goods equally and Ben prefers y. Their preferences are:

Set the price of y to 1 and call the price of x p. Each person sells their endowment at these prices and buys their preferred bundle. A general equilibrium is the price p at which the total demand for x equals the 10 units available; the market for y then clears automatically. Here that price is p = 6/11 ≈ 0.545: good x is cheaper than y because Ben, who holds little x, does not value it highly. At that price Anna ends up with about 5.83 units of x and 3.18 of y; Ben with 4.17 of x and 6.82 of y. Both are better off than at their starting point.

The final allocation is also Pareto efficient: there is no way to rearrange the 10 units of each good that makes one person better off without making the other worse off. That is the first welfare theorem at work.

Why does this matter?

The welfare theorems are the formal case for markets, and the formal account of when they fail. The first welfare theorem explains why economists start from the presumption that competitive prices allocate resources efficiently. Its assumptions list exactly what can go wrong: externalities such as pollution, market power, missing markets and private information. Each one is a reason for regulation, taxes or public provision. General equilibrium models are also what treasuries, central banks and trade negotiators use to trace a tax, a tariff or an energy shock through the whole economy rather than one market.

What makes the results hold?

Equilibrium exists when preferences and technologies are well behaved and every agent takes prices as given. The first welfare theorem needs little more than complete, competitive markets and consumers who always want slightly more. The second welfare theorem also needs convex preferences and technologies. These assumptions, and what fails when they do not hold, are set out after the worked example.

The takeaway

Partial equilibrium tells you what happens in one market if nothing else moves. General equilibrium tells you what happens when everything moves. The real example below shows that difference in Europe's inflation data, and the formula section sets out the theorems and their proofs.

A real example: how Europe's energy shock spread

In a partial equilibrium view, a jump in energy prices is a story about the energy market. Eurostat's harmonized index of consumer prices (HICP) for the euro area shows the shock did not stay there.

Reading the result

The components peaked in sequence. Energy inflation reached about 42% in mid-2022. Food inflation climbed through the second half of 2022 and was still above 10% in July 2023, because energy is a large input into fertilizer, processing and transport. Non-energy industrial goods followed the same pattern, as producers passed on higher input costs. Services, where wages are the main cost, rose last: services inflation reached 5.6% in July 2023, when energy prices were already 6.1% lower than a year earlier.

That sequence is a general equilibrium process. Energy is an input to almost every other sector, so its price enters every other sector's cost function. Higher prices for food and goods cut households' real incomes, which shifted demand across markets, and wage bargaining eventually passed the shock into the price of labor and labor-intensive services. A model that studied the energy market alone would have predicted a shock that faded as energy prices fell; the data show it moving through the rest of the economy for more than a year afterwards.

The table describes price movements, not a model estimate. Attributing each component's path to the energy shock rather than to supply bottlenecks, the post-pandemic recovery in demand, or monetary policy requires an estimated general equilibrium model, such as the ones central banks use.

The assumptions

Price-taking. Every consumer and firm treats prices as given. If an agent can move prices, it chooses quantities to influence them, and equilibrium outcomes need not be efficient: a monopolist restricts output below the efficient level.

Complete markets. Every good, in every location, date and state of the world, has a market and a price. If some market is missing, such as a market for clean air or for insurance against an uninsurable risk, agents cannot trade the gains that would make the allocation efficient.

No externalities. Each agent's utility or production depends only on its own consumption and inputs. When one firm's pollution enters others' production functions, the polluter does not pay for the damage, so equilibrium output is too high and the first welfare theorem fails.

Local nonsatiation. Near any bundle there is one each consumer prefers. This is the only preference assumption the first welfare theorem needs: it guarantees that any allocation a consumer strictly prefers must cost more at equilibrium prices, which drives the proof below.

Convexity. Preferences and production sets are convex. This is needed for the existence of equilibrium and for the second welfare theorem: without it, some efficient allocations cannot be supported by any set of prices, because no price line separates the better-than set from what the economy can produce.

Symmetric information. Everyone knows the quality of what is traded. When sellers know more than buyers, markets can unravel, which is the subject of information economics.

The formula

The derivations below are optional for anyone who only needs the intuition.

Consider an exchange economy with I consumers and L goods. Consumer i has utility function u_i and endowment ω_i, a vector of quantities. At price vector p, consumer i's income is p·ω_i and their Marshallian demand is x_i(p, p·ω_i), from utility maximization. The aggregate excess demand function is:

A Walrasian (competitive) equilibrium is a price vector p* with z(p*) = 0 (or z(p*) ≤ 0, with a zero price on any good in excess supply), together with the allocation each consumer demands at p*.

Two properties of z(p) follow from individual optimization. It is homogeneous of degree zero: doubling every price doubles every income and leaves demand unchanged, so only relative prices matter and one good can serve as numeraire. And it satisfies Walras's law:

Walras's law means that if L − 1 markets clear, the last one clears too. In the Anna-and-Ben example, clearing the market for x was enough.

Solving the example. With p_y = 1 and p_x = p, Anna's income is 8p + 2 and Ben's is 2p + 8. Cobb-Douglas demands spend a fixed share of income on each good, so the market for x clears when:

At the equilibrium, each consumer's marginal rate of substitution (the rate at which they would swap y for x) equals the price ratio. For Anna it is y_A/x_A = 3.18/5.83 ≈ 0.545; for Ben it is (1/3)(y_B/x_B) = (1/3)(6.82/4.17) ≈ 0.545. Equal marginal rates of substitution is the condition for Pareto efficiency in an exchange economy.

Existence. If each consumer's preferences are continuous, strictly convex and strongly monotone, and endowments are positive, z(p) is continuous on the set of normalized prices and an equilibrium exists. The standard proof applies Brouwer's fixed point theorem to a price-adjustment map that raises the relative price of goods in excess demand; Arrow and Debreu (1954) extended the result to economies with production.

First fundamental theorem of welfare economics. If preferences are locally nonsatiated, every Walrasian equilibrium allocation is Pareto efficient.

The proof is short. Suppose an equilibrium allocation x* at prices p* is not efficient, so some feasible allocation x′ makes everyone at least as well off and someone strictly better off. Anyone strictly better off under x′ must find it unaffordable at p* (otherwise they would have chosen it), and by local nonsatiation anyone indifferent finds it at least as expensive. Adding up across consumers:

But x′ is feasible, so Σ x′_i ≤ Σ ω_i, and since prices are non-negative Σ p*·x′_i ≤ Σ p*·ω_i. The two statements contradict each other, so no such x′ exists and x* is Pareto efficient.

Second fundamental theorem of welfare economics. If preferences are continuous, convex and locally nonsatiated, then any Pareto efficient allocation x̂ can be supported as a competitive equilibrium after a suitable reallocation of endowments (lump-sum transfers). The proof uses the separating hyperplane theorem: convexity ensures a price vector exists that separates the set of allocations everyone weakly prefers to x̂ from the set of feasible allocations, and those prices support x̂ as an equilibrium.

The second theorem separates efficiency from distribution. In principle, a government can achieve any efficient allocation by redistributing wealth in lump sums and letting markets do the rest, without distorting prices. In practice lump-sum transfers based on ability to pay are rarely feasible, so real redistribution uses taxes on income or spending, which change prices and create efficiency costs.

Production. With firms, efficiency requires that every consumer's marginal rate of substitution between two goods equals every firm's marginal rate of transformation between them:

When this isn't the right tool

Equilibrium need not be unique or stable. The Sonnenschein-Mantel-Debreu results show that individual rationality places almost no restriction on the shape of aggregate excess demand beyond continuity, homogeneity and Walras's law. An economy can have several equilibria, and a simple price-adjustment process need not converge to any of them.

Fixing one distortion can make things worse. The theory of the second best (Lipsey and Lancaster, 1956) shows that when one efficiency condition cannot be met, satisfying the others is not necessarily an improvement. Removing a single tax in an economy with many distortions does not guarantee higher welfare.

Real economies violate the assumptions. Externalities, market power, missing insurance markets and private information are common. The welfare theorems are best read as a precise statement of what markets achieve when these problems are absent, and therefore of where policy has a role.

Static models leave out time and money. Applied work uses computable general equilibrium models for trade and tax policy, and dynamic stochastic general equilibrium models for macroeconomic policy, both built on the same market-clearing logic.

Further reading

• Consumer Theory and Duality Explained — internal
• Producer Theory and Cost Functions Explained — internal
• Supply and Demand Explained — internal
• Understanding Inflation — internal
• Eurostat, euro area flash inflation estimate, June 2022 — external reference
• Eurostat, euro area flash inflation estimate, December 2023 — external reference
• Arrow and Debreu (1954), Existence of an Equilibrium for a Competitive Economy, Econometrica — external reference

Inflation figures are from Eurostat's euro area flash estimate releases. The two-person exchange economy is an illustrative teaching example, not observed data.