Market allocations under conflation of goods

We study competitive equilibria in exchange economies when a continuum of goods is conflated into a finite set of commodities. The design of conflation choices affects the allocation of scarce resources among agents, by constraining trading opportunities and shifting competitive pressures. We demonstrate the consequences on relative prices, trading positions, and welfare.

💡 Research Summary

The paper investigates how the design of a classification that aggregates a continuum of goods into a finite set of tradable commodities—referred to as “conflation”—shapes competitive equilibria in exchange economies. Starting from the Arrow‑Debreu framework, the authors formalize a measurable space of goods I equipped with a probability measure ω. A classification π is a finite partition of I; each cell C∈π is interpreted as a single commodity obtained by mixing all goods in C. Agents have well‑behaved preferences U_i defined on the infinite‑dimensional space L¹⁺(ω) (concave, monotone, continuous). When a classification is imposed, these preferences are restricted to the subspace of simple functions measurable with respect to π, yielding finite‑dimensional utilities V_i(π,·) that depend only on the quantities of the π‑commodities.

The model assumes n agents, each with a proportional claim κ_i on the total endowment (∑κ_i=1), which makes the setting equivalent to a Fisher market. An allocation is feasible if the total amount of each commodity does not exceed its original supply ω(C). A competitive equilibrium for a given π consists of a price vector p∈ℝ_+^{|π|} and an allocation (x_i) such that each agent maximizes V_i(π,·) subject to its budget and any strictly better bundle would violate the budget constraint.

Two central theoretical contributions are presented. First, the authors endow the set Π(≤k) of all classifications with at most k intervals with a Hausdorff‑type pseudo‑metric d_ω that measures the maximal symmetric difference in ω‑measure between the σ‑algebras generated by two partitions. They prove that (Π(≤k), d_ω) is compact and that the correspondence W mapping each π to its set of competitive equilibria is compact‑valued and upper‑hemicontinuous. Consequently, small changes in the classification (e.g., slight adjustments of quality cut‑offs) lead to only small changes in equilibrium prices, allocations, and agents’ utilities. This continuity result enables comparative statics: the welfare impact of refining a classification can be bounded arbitrarily tightly by choosing a sufficiently close π′, regardless of how many new commodities are introduced.

Second, the paper examines Pareto efficiency across different classifications. Within a fixed π, any competitive allocation is Pareto‑optimal (the First Welfare Theorem). However, when comparing allocations generated under distinct classifications π and ρ, a competitive allocation under π need not be Pareto‑optimal relative to the feasible set defined by ρ. The authors illustrate that a finer classification can generate a new equilibrium that makes every agent strictly better off, or conversely, that a coarser classification can improve welfare for some agents at the expense of others. Thus, the classification itself functions as a non‑excludable, non‑rival public good that creates externalities not priced by the market.

To make the abstract results concrete, the authors discuss three families of primitive utilities. If U_i is linear, it can be represented by a measure ν_i absolutely continuous with respect to ω, and the induced finite‑dimensional utility becomes V_i(π,x)=∑{C∈π} x_C ω(C) ν_i(C). For Cobb‑Douglas utilities (U_i(b)=exp∫log b dν_i) the induced utility is multiplicative: V_i(π,x)=∏{C∈π} (x_C ω(C))^{ν_i(C)}. For CES utilities (U_i(b)=(∫ b^ρ dν_i)^{1/ρ}) the induced utility takes the familiar CES form. In each case, standard existence and uniqueness results for competitive equilibria apply. Moreover, the authors show that any primitive utility of the separable form U_i(b)=∫ u_i(t,b(t)) dt, with u_i(t,·) concave, increasing, and satisfying a mild curvature condition, guarantees Gross Substitutability of the induced finite‑dimensional demands, thereby ensuring equilibrium existence and uniqueness up to normalization.

The paper’s policy implications are noteworthy. Designing standards—such as quality grades for agricultural products, protected designations of origin for wine, or classification of financial securities—amounts to choosing a π. While intuition might suggest that finer classifications (more tradable commodities) always improve market efficiency, the analysis demonstrates that this is not guaranteed; additional trading opportunities can generate equilibria that are Pareto‑inferior or even reduce overall welfare. Conversely, a coarser classification may sometimes yield higher aggregate utility if it better aligns agents’ marginal rates of substitution with the available price structure. Because the classification acts as a public good, policymakers may need to internalize its externalities through subsidies, taxes, or cost‑sharing mechanisms that reflect the social value of finer or coarser aggregation.

In summary, the paper provides a rigorous framework for studying how the institutional choice of commodity definition shapes competitive outcomes. It establishes continuity of equilibria with respect to classification, highlights the possibility of cross‑classification Pareto improvements, and offers concrete examples that connect abstract theory to real‑world market design problems. The results extend the classical general equilibrium literature by foregrounding the often‑overlooked role of product definition as a strategic economic variable.