📈 economics

Equilibrium Transmission and Dynamics in Price Networks

This paper proposes a network model demonstrating that equilibrium price transmission and dynamic adjustment are jointly constrained by a shared structural characteristic, revealing that aggregate prices may mask underlying cross-sectional disagreement which can only be detected through a combined analysis of cross-sectional dispersion and time-series dynamics.

Original authors: HyeongWoo Kim

Published 2026-09-29

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: HyeongWoo Kim

Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

In the world of economics, prices are often treated as a single, unified signal. When a market settles, we tend to assume that everyone has finally agreed on what something is worth. This view suggests that if the overall price is stable, the individual opinions of buyers and sellers must have converged into a single, shared truth. However, this assumption overlooks a fundamental reality of human interaction: people often hold different views even when the final number looks the same. In a strategic environment, where people watch each other to decide their own actions, the average price is just a projection of many different minds. It is possible for the collective number to remain steady while the underlying disagreements between individuals remain strong, hidden from view. Understanding how these hidden differences behave is crucial, because the way a group settles on a price today might dictate how that group reacts to new information tomorrow.

A new study by HyeongWoo Kim explores this hidden layer of economic life by building a model of how people form price judgments while watching one another. The research focuses on a specific question: can the way a group settles on a price tell us how that group will adjust if things change later? The author constructs a framework where a fixed network of connections links individuals together. In this network, every person has their own private sense of value, but they also look at the judgments of their neighbors to decide what they should think. The study finds that the structure of these connections does two things at once. First, it determines how much of the original difference in private values survives to become a difference in the final prices. Second, that same structure dictates how quickly or slowly those price differences fade away or grow after a disturbance.

The core discovery is a tight link between the static result and the dynamic process. The researcher shows that the specific way a disagreement is transmitted through the network to reach a steady state is not just a snapshot; it is a blueprint for the future. If you know how a particular type of disagreement is handled when the system is calm, you can mathematically determine how that same type of disagreement will move over time. This means that the equilibrium state contains a structural signature that restricts the possible paths the system can take later. It is not enough to look at the final price and assume the system is stable; one must look at the pattern of differences across the group. If the pattern of differences in the steady state does not match the rules of movement observed in the data, the entire model of how people interact is likely wrong.

This finding challenges the common practice of relying solely on aggregate data, such as a single market price index, to understand economic stability. The paper demonstrates that two different economies could produce the exact same average price and appear identical to an outside observer, yet be completely different underneath. One economy might have deep, persistent disagreements that are hidden by the way the data is averaged, while another might have a true consensus. Because the average price only reveals the shared part of the judgment, it discards the information about the disagreements. Consequently, a stable average price does not guarantee that the underlying disagreements have vanished or that they will behave in a predictable way. The study proves that to truly understand the stability of a market, one must observe the cross-sectional distribution of individual judgments, not just the final number.

The research also clarifies how the speed of adjustment depends on the nature of the disagreement. In the model, different types of disagreements travel through the network at different speeds and with different intensities. Some disagreements might be dampened quickly, while others might linger or even oscillate. The study establishes that the intensity with which a disagreement survives in the steady state is directly tied to how persistent it will be over time. If a specific type of disagreement is transmitted strongly into the final prices, the model predicts it will also be slow to fade away. This creates a testable rule: if researchers can measure how much disagreement exists in a steady state and how fast it moves, they can check if these two observations fit the same underlying rules of human behavior. If they do not fit, it suggests that the assumed rules of interaction are incorrect.

Ultimately, this work provides a new way to look at economic networks. It moves beyond simply asking whether networks make prices more volatile or more stable. Instead, it asks whether the specific pattern of differences we see in a calm market can tell us how that market will react to a shock. The answer is yes, but only if we look at the right data. The study shows that the equilibrium transmission of differences is not a passive outcome; it is an active constraint on the future. By linking the static picture of who disagrees with whom to the dynamic picture of how those disagreements evolve, the research offers a more complete and rigorous way to test theories of how people interact and form expectations. It suggests that the true story of a market is not found in its average price, but in the hidden geometry of its disagreements.

Technical Summary: Equilibrium Transmission and Dynamics in Price Networks

Problem Statement
The paper addresses a fundamental limitation in models of heterogeneous expectations and strategic interaction: the inability of aggregate price data to reveal the underlying cross-sectional structure of disagreement. While aggregate prices are often interpreted as evidence of converged economic judgments, the author argues that in a strategic environment, individual agents may hold systematically different price judgments even when the aggregate price is stable. This occurs because the aggregate price is a projection of the cross-sectional state; variation orthogonal to the aggregation weights (disagreement modes) can remain economically relevant and dynamically distinct while being invisible in the aggregate. The central question is whether the cross-sectional transmission of heterogeneity observed in equilibrium imposes structural restrictions on how that heterogeneity subsequently adjusts over time.

Methodology
The paper develops a network model of strategic price formation involving $n$ agents. The framework is built on three core components:

Network Structure: Agents interact through a fixed, symmetric, row-stochastic matrix $G$ (with zero diagonal), representing strategic substitution. The matrix $G$ is decomposed into eigenmodes, separating a unique consensus mode (eigenvalue $g_1=1$ ) from a set of disagreement modes ( $j \in D$ ).
Fundamental Heterogeneity: Agents possess heterogeneous fundamental benchmarks $\theta$ , representing private valuations or signals.
Behavioral Adjustment: Agents follow a partial-adjustment rule combining persistence in their own judgments ( $r$ ) and strategic response to the expected judgments of others ( $k$ ). Expectations are forward-looking, incorporating an extrapolation parameter ( $\lambda$ ) based on recent changes in others' judgments.

The model derives a second-order vector difference equation for individual judgments. By projecting this system onto the eigenbasis of $G$ , the multidimensional problem is reduced to a collection of independent scalar second-order difference equations, one for each network mode.

Key Contributions and Results

1. The Mode-Specific Equilibrium Transmission Coefficient ( $T_j$ )
The paper defines a key equilibrium object, the transmission coefficient $T_j$ , which measures how fundamental heterogeneity in mode $j$ is transmitted into equilibrium price judgments:
$T_j = \frac{1-r}{1-r + kg_j}$
where $g_j$ is the eigenvalue of the network matrix associated with mode $j$ . This coefficient quantifies the attenuation or amplification of fundamental heterogeneity through strategic interaction. Under uniform aggregation, the aggregate stationary price depends only on the consensus mode ( $T_1$ ), rendering the disagreement modes ( $j \in D$ ) invisible in the aggregate price.

2. The Equilibrium-to-Dynamics Restriction
The primary theoretical contribution is the derivation of a structural restriction linking the static equilibrium object $T_j$ to the dynamic adjustment of the same mode. The dynamic law for mode $j$ is governed by a characteristic polynomial $q_j(\mu)$ . By algebraically eliminating the latent network eigenvalue $g_j$ using the definition of $T_j$ , the paper shows that the characteristic polynomial can be expressed entirely in terms of the equilibrium transmission coefficient and the common behavioral parameters $(r, \lambda)$ :
$q_j(\mu | T_j) = \mu^2 - \left[ r - (1+\lambda)(1-r)\frac{1-T_j}{T_j} \right]\mu - \lambda(1-r)\frac{1-T_j}{T_j}$
This result implies that the equilibrium cross-sectional transmission of heterogeneity structurally constrains the subsequent time-series adjustment of that same mode.

3. Cross-Mode Common-Parameter Restriction
Because the behavioral parameters $(r, \lambda)$ are common across all network modes, the model imposes a joint restriction. The collection of observed equilibrium transmission coefficients $\{T_j\}$ and the corresponding dynamic coefficients $\{a_j, b_j\}$ across all disagreement modes must be jointly consistent with a single pair of parameters $(r, \lambda)$ . This transforms the equilibrium-to-dynamics mapping from a mode-specific algebraic identity into a testable cross-mode structural restriction.

4. Observational Equivalence and Hidden Disagreement
The paper demonstrates that distinct cross-sectional configurations can generate identical stationary aggregate prices. Two economies with the same aggregate fundamental projection and consensus transmission can differ significantly in their disagreement dispersion ( $D^*$ ) and the dynamic persistence of their disagreement modes. Consequently, aggregate price stability does not imply the vanishing of disagreement, nor does it reveal the underlying dynamic adjustment processes.

Significance and Claims
The paper claims its significance lies not in the introduction of heterogeneous beliefs, network structures, or strategic interaction per se—mechanisms already central to existing literature—but in the specific equilibrium-to-dynamics mapping. The distinctive object is the mode-specific equilibrium transmission coefficient, which serves as a bridge between cross-sectional equilibrium information and time-series adjustment.

The framework provides a sharper prediction than models relying solely on aggregate convergence. It posits that the equilibrium transmission of heterogeneity through a network mode restricts the dynamic law of that same mode. This restriction is empirically testable only by combining cross-sectional data (to identify $T_j$ ) with time-series data (to identify dynamic roots), as aggregate price data alone cannot recover the disagreement structure. The paper concludes that equilibrium transmission contains structural information about subsequent dynamic adjustment that is generally invisible in aggregate price behavior, offering a new basis for identifying and testing the joint equilibrium-dynamic structure of strategic price networks.

More like this