Identification of Hedonic Equilibrium and Nonseparable Simultaneous Equations

Victor Chernozhukov, Alfred Galichon, Marc Henry, Brendan Pass

TL;DR

This paper tackles the challenge of identifying consumer preferences and market equilibrium for complex, multi-attribute goods using only market data. It introduces novel "twist condition" and "cyclical monotonicity" shape restrictions, drawing on optimal transport theory, to achieve nonparametric identification of these preferences and market structures, even with unobserved heterogeneity.

What Problem Does This Paper Address?

In many markets, goods are differentiated by multiple attributes (e.g., a car's features, a house's location and size). Consumers have diverse preferences for these attributes, and their choices, along with production technologies, determine market prices and the types of goods traded (a hedonic equilibrium). A key challenge is to understand why consumers value these attributes and how their choices aggregate to form market outcomes, especially when consumers have unobserved characteristics (like taste or income) that influence their decisions. Traditional methods often struggle with multi-attribute goods and unobserved heterogeneity, requiring strong assumptions or multiple markets. This paper aims to develop a framework for nonparametric identification of consumer preferences and market equilibrium from a single market, even when goods have multiple attributes and consumers have multiple dimensions of unobserved heterogeneity.

Methods

The paper builds upon existing work in hedonic modeling and optimal transport theory. The core contribution lies in extending identification strategies from single-attribute goods to multi-attribute goods.

  • Twist Condition: This is a generalization of the "single crossing" condition used in simpler models. It imposes a specific structure on how consumer utility changes with different attribute combinations and unobserved heterogeneity.
  • Cyclical Monotonicity: This property, derived from the equilibrium condition that trades maximize joint surplus, ensures that the matching between consumers and goods is stable and cannot be improved by reallocating partners.
  • Optimal Transport Theory: The authors leverage advanced tools from optimal transport, particularly concepts like "generalized convexity" and "multivariate quantiles," to link these shape restrictions to the identification of preferences and market outcomes.
  • Nonparametric Identification: The goal is to recover the functional forms of preferences and market structures without assuming specific parametric forms (like linear or quadratic utility functions).

Datasets

Not discussed in the paper. The paper focuses on theoretical identification conditions rather than empirical application with specific datasets.

Key Insights and Ideas

  • Beyond Scalar Attributes: The paper demonstrates that the identification strategies developed for single-attribute goods can be generalized to goods with multiple quality dimensions, a significant step for real-world applications.
  • The Power of Shape Restrictions: The "twist condition" and "cyclical monotonicity" are shown to be crucial for disentangling the effects of observable and unobservable characteristics on consumer choices and market outcomes.
  • Optimal Transport as an Identification Tool: The paper highlights the power of optimal transport theory in providing a rigorous mathematical framework for identifying complex economic models, particularly in the presence of unobserved heterogeneity.
  • Unobserved Heterogeneity is Key: The ability to handle multiple dimensions of unobserved heterogeneity is central to the paper's contribution, allowing for more realistic modeling of consumer behavior.

Key Results

The paper establishes nonparametric identification results for:

  • Hedonic Equilibrium Models: Under the proposed shape restrictions (twist condition and cyclical monotonicity), consumer preferences (utility functions) and the equilibrium price schedule can be identified from observed market data, even for goods with multiple attributes and multidimensional unobserved heterogeneity.
  • Nonseparable Simultaneous Equations Models: The framework is extended to identify nonseparable simultaneous equations, a broader class of models that includes hedonic equilibrium as a special case.
  • Absolute Continuity of Traded Qualities: A technical result proving the absolute continuity of the distribution of endogenously traded qualities is derived, which is of independent interest for understanding market dynamics.

Limitations and Future Work

  • Normalization of Unobserved Heterogeneity: The current identification results often require a priori normalization of the distribution of unobserved heterogeneity. The authors suggest that further research is needed to relax this constraint, particularly by using data from multiple markets.
  • Point Identification with Lower Dimensional Heterogeneity: While local identification is achieved under weak assumptions for lower-dimensional unobserved heterogeneity, global identification requires stronger assumptions on the endogenous price function, for which sufficient conditions on primitives are not yet established.
  • Empirical Application: The paper focuses on theoretical identification. Future work would involve applying these identification strategies to real-world data to estimate preferences and market outcomes.
  • Partial Identification with Multiple Markets: The paper briefly discusses partial identification using data from multiple markets but notes that more research is needed to develop point identification conditions in this setting.

Summary updated

Abstract

This paper derives conditions under which preferences and technology are nonparametrically identified in hedonic equilibrium models, where products are differentiated along more than one dimension and agents are characterized by several dimensions of unobserved heterogeneity. With products differentiated along a quality index and agents characterized by scalar unobserved heterogeneity, single crossing conditions on preferences and technology provide identifying restrictions in Ekeland, Heckman and Nesheim (2004) and Heckman, Matzkin and Nesheim (2010). We develop similar shape restrictions in the multi-attribute case. These shape restrictions, which are based on optimal transport theory and generalized convexity, allow us to identify preferences for goods differentiated along multiple dimensions, from the observation of a single market. We thereby derive nonparametric identification results for nonseparable simultaneous equations and multi-attribute hedonic equilibrium models with (possibly) multiple dimensions of unobserved heterogeneity. One of our results is a proof of absolute continuity of the distribution of endogenously traded qualities, which is of independent interest.
BibTeX
@article{chernozhukov2020identification,
  author = {Victor Chernozhukov and Alfred Galichon and Marc Henry and Brendan Pass},
  title = {{Identification of Hedonic Equilibrium and Nonseparable Simultaneous Equations}},
  journal = {Journal of Political Economy},
  year = {2020},
  doi = {10.1086/712447},
}
Figure