Working Paper: CEPR ID: DP14273
Authors: Alessandro Iaria; Ao Wang
Abstract: According to the Hicksian criterion, two products are complements if their (compensated)cross-price elasticity is negative. While attractive in theory, the implementation of theHicksian criterion can be hard: computing elasticities requires the estimation of structuralmodels allowing for both complementarity and substitutability. Here, we insteadinvestigate the correlation criterion, whose implementation only requires the comparisonof observed market shares. We show that, in a large class of non-parametric models, thecorrelation criterion satisfies all the axioms by Manzini et al. (2018) and how, in mixedlogit models, it can be used to learn about the Hicksian criterion.
Keywords: Hicksian; complementarity; substitutability; correlation; demand elasticity; demand estimation; market shares
JEL Codes: No JEL codes provided
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| correlation criterion (C10) | products classified as complements (D10) |
| joint market share of products x and y (D26) | products classified as complements (D10) |
| correlation criterion satisfies monotonicity axiom (C10) | products cannot become substitutes with increased joint purchases (D10) |
| individual-level purchase data (D12) | information about Hicksian complementarity and substitutability (D10) |
| covariance between individual-level purchases (D12) | insights into product relationships (L14) |