Dixit–Stiglitz Model

Dixit–Stiglitz Model is the canonical general-equilibrium model of monopolistic competition and endogenous product diversity, formulated by Avinash K. Dixit and Joseph E. Stiglitz in their landmark 1977 American Economic Review article, "Monopolistic Competition and Optimum Product Diversity." For more than four decades after Edward Chamberlin (1933) and Joan Robinson (1933) introduced the concept of monopolistic competition—an industry structure combining downward-sloping firm-level demand curves with free entry that drives economic profits to zero—economists lacked a tractable general-equilibrium formalization capable of handling internal economies of scale without descending into intractable strategic oligopoly interactions. By introducing symmetric constant-elasticity-of-substitution (CES) preferences over a continuum (or large discrete set) of differentiated varieties, Dixit and Stiglitz created the foundational analytical engine driving modern New Trade Theory, New Economic Geography, endogenous growth theory, and New Keynesian macroeconomics.

In the canonical symmetric specification of the Dixit–Stiglitz model, a representative consumer allocates expenditure E across a continuum of differentiated varieties indexed by ω ∈ Ω, where the measure of available varieties is N = |Ω|. The consumer's utility function (or real consumption aggregator Q) takes the constant-elasticity-of-substitution form:

U = Q = [ ∫_ω∈Ω c(ω)^{(σ − 1) / σ} dω ]^{σ / (σ − 1)} = [ ∫₀^N c(ω)^ρ dω ]^{1 / ρ}

where c(ω) ≥ 0 denotes the quantity consumed of variety ω, ρ = (σ − 1) / σ ∈ (0, 1) governs product substitutability, and σ = 1 / (1 − ρ) > 1 is the constant elasticity of substitution between any two varieties. Because σ > 1 (so 0 < ρ < 1), no single variety is essential (c(ω) = 0 is feasible), and varieties are gross substitutes. If σ → ∞ (ρ → 1), varieties become perfect substitutes and the market collapses to homogeneous perfect competition.

Given individual variety prices p(ω), the consumer maximizes Q subject to the budget constraint ∫₀^N p(ω)c(ω) dω ≤ E. Forming the Lagrangian and taking the ratio of first-order conditions for any two varieties ω and ω′ yields (c(ω) / c(ω′))^−1/σ = p(ω) / p(ω′). Integrating across the budget constraint yields the isoelastic Marshallian demand function for variety ω:

c(ω) = Q [ p(ω) / P ]^−σ = (E / P) [ p(ω) / P ]^−σ = E P^{σ − 1} p(ω)^−σ

where P is the dual Dixit–Stiglitz ideal price index (the minimum expenditure required to purchase one unit of the composite consumption bundle Q, so that P Q = E):

P = [ ∫₀^N p(ω)^{1 − σ} dω ]^{1 / (1 − σ)}

A defining structural feature of the Dixit–Stiglitz aggregator is love of variety (or increasing returns to product diversity in consumption). Suppose all N available varieties are priced identically at p(ω) = p and the consumer spends a fixed budget E equally across all N varieties, so c(ω) = E / (Np). Substituting these symmetric quantities into the utility function yields indirect utility and the ideal price index:

U = Q = N^{σ / (σ − 1)} (E / Np) = N^{1 / (σ − 1)} (E / p) and P = N^{−1 / (σ − 1)} p

Because σ > 1, the exponent 1 / (σ − 1) is positive: holding total physical expenditure E / p constant, expanding the number of available varieties N increases consumer welfare U and lowers the true cost-of-living price index P. The love-of-variety elasticity 1 / (σ − 1) is larger when varieties are more differentiated (lower σ) and vanishes as varieties become perfect substitutes (σ → ∞).

On the production side, each differentiated variety ω is produced by a single specialized monopoly firm because fixed setup costs make it unprofitable for two firms to replicate the same brand. Labor L is the sole factor of production, supplied inelastically at a competitive wage normalized to unity (w = 1). Producing q(ω) units of variety ω requires a fixed overhead labor input f > 0 and a constant marginal labor requirement c > 0:

l(ω) = f + c q(ω)

Because average cost AC(q) = f / q + c declines monotonically with output q, every firm operates under internal economies of scale (increasing returns to scale at the firm level).

In a continuum of varieties (or under Chamberlinian "large-group" competition), each individual firm ω is atomistic relative to the aggregate economy and therefore treats total expenditure E and the aggregate price index P as unaffected by its own price p(ω). Facing the isoelastic demand curve q(ω) = E P^{σ − 1} p(ω)^−σ with price elasticity of demand −∂ ln q / ∂ ln p = σ > 1, firm ω maximizes profit π(ω) = p(ω)q(ω) − c q(ω) − f. The first-order condition equates marginal revenue MR = p(1 − 1/σ) to marginal cost MC = c, yielding the Dixit–Stiglitz constant-markup pricing rule:

p(ω) = p = [ σ / (σ − 1) ] c = c / ρ

Substituting this optimal price back into the firm's profit function expresses operating profit as a constant fraction 1 / σ of total firm revenue r = pq:

π = (p − c)q − f = (c q) / (σ − 1) − f = (pq) / σ − f

Under free entry and exit, new firms enter to introduce new varieties until net economic profit is driven to zero (π = 0). Setting (c q*) / (σ − 1) − f = 0 pins down the unique equilibrium firm scale q*:

q* = [ (σ − 1) f ] / c

Consequently, the labor employed by each active firm is l* = f + c q* = σf. Imposing full employment of the aggregate labor endowment L across N symmetric firms (N l* = L) determines the equilibrium number of varieties N*:

N* = L / l* = L / (σ f)

Remarkably, in the baseline CES Dixit–Stiglitz model, increasing the size of the economy L expands the equilibrium number of varieties N* = L / (σf) one-for-one while leaving individual firm output q* = (σ − 1)f / c and markup σ / (σ − 1) completely invariant to market size.

A central theoretical question addressed by Dixit and Stiglitz (1977) is whether a decentralized monopolistically competitive market produces too many varieties at too small a scale (Chamberlin's "excess capacity theorem") or too few varieties relative to the social Pareto efficiency benchmark.

When a new firm introduces a marginal variety in a decentralized market, two opposing welfare distortions operate:

1. Non-Appropriability of Consumer Surplus (Under-Entry Distortion): Because a single-price firm cannot perfectly price-discriminate, its revenue pq is smaller than the gross consumer utility gain contributed by the new variety. Left alone, this consumer-surplus effect biases the market toward producing too few varieties.
2. Business-Stealing / Profit-Destruction Externality (Over-Entry Distortion): When a new firm enters, it shifts demand away from all existing varieties, reducing their sales and profits (p − c)Δq. Because the entrant ignores the profit destruction inflicted on incumbent rivals, this business-stealing effect biases the market toward producing too many varieties.

Dixit and Stiglitz proved a notable razor's-edge theorem for the standard CES utility specification: when labor supply L is inelastic and all sectors have the same CES elasticity σ, the non-appropriability distortion and the business-stealing externality precisely cancel each other out. Specifically, both a first-best social planner (who prices at marginal cost p = c and finances fixed setup costs Nf via lump-sum taxation) and a second-best social planner (who must allow every firm to break even without lump-sum subsidies) choose the identical variety count N* = L / (σf) and firm scale q* = (σ − 1)f / c as the decentralized market equilibrium.

When preferences depart from constant-elasticity CES—generalized to additive variable-elasticity utility U = G(∫₀^N v(c(ω)) dω)—this cancellation breaks down. Depending on the curvature of the sub-utility function v(c) relative to its elasticity c v′(c) / v(c), the decentralized market may generate either insufficient or excessive variety relative to the social optimum.

The analytical tractability of the Dixit–Stiglitz framework transformed four major fields of economic theory from the late 1970s onward:

• New Trade Theory and Heterogeneous-Firm Trade (Krugman, 1979, 1980; Melitz, 2003): Paul Krugman (1980) embedded the Dixit–Stiglitz model into a two-country trading world to explain why industrialized nations with identical factor endowments engage in large-scale intra-industry trade (two-way trade in differentiated goods). Opening from autarky to free trade between two countries of size L doubles the integrated market size to 2L, allowing consumers in both nations to access 2N* varieties and raising real wages w / P = (ρ / c) N^1/(σ−1) purely through love-of-variety gains, even without Ricardian or Stolper–Samuelson comparative advantage. Marc Melitz (2003) later combined Dixit–Stiglitz demand with heterogeneous firm productivities in the Melitz Model to explain exporter selection and intra-industry reallocations, while providing structural foundations for the gravity model of trade.
• New Economic Geography (Krugman, 1991): Combining Dixit–Stiglitz increasing returns and iceberg transport costs τ > 1 with inter-regional labor mobility generates the home-market effect and endogenous core–periphery agglomeration: workers migrate toward the larger market to enjoy a lower Dixit–Stiglitz price index P, while firms locate in the larger market to save transport costs, creating self-reinforcing urban and industrial clusters.
• Endogenous Growth Theory (Romer, 1990): Paul Romer's horizontal-innovation growth model replaces consumer varieties with differentiated intermediate capital goods in a Dixit–Stiglitz production aggregator Y = L_Y^1−α ∫₀^A x(ω)^α dω. As R&D expands the measure of intermediate blueprints A, division of labor deepens and total factor productivity grows endogenously (A^1−α).
• New Keynesian Macroeconomics (Blanchard & Kiyotaki, 1987): Modern Dynamic Stochastic General Equilibrium (DSGE) models require firms to be price-setters rather than price-takers so that nominal price rigidities—such as Calvo pricing and Rotemberg pricing—can exist. The Dixit–Stiglitz monopolistic competition aggregator provides the universal microfoundation for price-setting firms and the New Keynesian Phillips Curve.

Why do firm size and markups remain constant when market size L increases in the baseline CES Dixit–Stiglitz model?

Under CES preferences, the price elasticity of demand σ is constant and completely independent of the number of competing varieties N or the level of consumption c(ω). Because markup p / c = σ / (σ − 1) depends only on σ, the break-even output required to cover the fixed cost f (q* = (σ − 1)f / c) is fixed by technology and preference parameters alone; any increase in market size L is absorbed by the entry of new varieties N* = L / (σf).

How did Krugman (1979) allow firm size to expand with trade if the CES Dixit–Stiglitz model keeps firm scale constant?

In Krugman's 1979 Journal of International Economics paper, preferences take an additive variable-elasticity form where the elasticity of demand increases as per-capita consumption of a variety falls. When trade opens and the number of varieties rises, individual consumption per variety drops, raising demand elasticity, compressing price-cost markups, and forcing each surviving firm to produce at a larger scale q* to cover its fixed costs (the "pro-competitive effect" of trade). In his 1980 American Economic Review paper, Krugman adopted the standard CES Dixit–Stiglitz specification to obtain closed-form solutions with iceberg transport costs.

Why is it necessary to assume σ > 1 in the Dixit–Stiglitz model?

If the elasticity of substitution were less than or equal to unity (σ ≤ 1), the price elasticity of demand facing each monopolistically competitive firm would be inelastic or unit-elastic. A profit-maximizing monopolist facing σ ≤ 1 would find that raising its price p(ω) toward infinity continuously increases or maintains revenue while lowering production costs, so no finite profit-maximizing price or interior equilibrium could exist.

1. Dixit, A. K., & Stiglitz, J. E. (1977). Monopolistic competition and optimum product diversity. The American Economic Review, 67(3), 297–308. https://doi.org/10.7916/D81J97SC
2. Krugman, P. R. (1980). Scale economies, product differentiation, and the pattern of trade. The American Economic Review, 70(5), 950–959. https://doi.org/10.7916/D8B85GCF
3. Krugman, P. R. (1991). Increasing returns and economic geography. Journal of Political Economy, 99(3), 483–499. https://doi.org/10.1086/261763
4. Romer, P. M. (1990). Endogenous technological change. Journal of Political Economy, 98(5, Part 2), S71–S102. https://doi.org/10.1086/261725
5. Blanchard, O. J., & Kiyotaki, N. (1987). Monopolistic competition and the effects of aggregate demand. The American Economic Review, 77(4), 647–666. https://doi.org/10.3386/w1770