Crises and Physical Phases of a Bipartite Market Model

Nima Dehmamy, Sergey V. Buldyrev, Shlomo Havlin, H. Eugene Stanley, Irena VodenskaarXiv 2016 · DOI · arXiv

Abstract

We analyze the linear response of a market network to shocks based on the bipartite market model we introduced in an earlier paper, which we claimed to be able to identify the time-line of the 2009-2011 Eurozone crisis correctly. We show that this model has three distinct phases that can broadly be categorized as "stable" and "unstable". Based on the interpretation of our behavioral parameters, the stable phase describes periods where investors and traders have confidence in the market (e.g. predict that the market rebounds from a loss). We show that the unstable phase happens when there is a lack of confidence and seems to describe "boom-bust" periods in which changes in prices are exponential. We analytically derive these phases and where the phase transition happens using a mean field approximation of the model. We show that the condition for stability is $\alpha\beta < 1$ with $\alpha$ being the inverse of the "price elasticity" and $\beta$ the "income elasticity of demand", which measures how rash the investors make decisions. We also show that in the mean-field limit this model reduces to the Langevin model by Bouchaud et al. for price returns.

Introduction

Traditional economic models often rely on idealized assumptions that fail to predict real-world market dynamics, a shortcoming made painfully obvious by the 2007-2008 global financial crisis. To better understand systemic risk, cascades, and contagion, researchers need quantitative models that capture the continuous, complex network interactions of market players. This paper addresses the stability conditions of the GIPSI (Group-Impulse Portfolio-Sharing Investment) model, seeking to understand the behavioral conditions that cause a market to absorb a financial shock versus those that trigger a destructive boom-bust cycle.

Methodology

The authors treat the financial market as a physical system, applying an "effective theory" approach to model its linear response to shocks. The market is represented as a bipartite graph where weighted links denote the holdings of an investor in a specific asset.

Bipartite network structure of the GIPSI model, connecting investors (blue squares) to assets (red circles).

The dynamics are governed by coupled, second-order differential equations with time delays, relying on two primary behavioral parameters:

  • $\alpha$ (Inverse Price Elasticity): Measures how much prices change in response to supply and demand.
  • $\beta$ (Income Elasticity of Demand): Measures the "rashness" of investors as they adjust portfolios in response to equity changes.

To analytically solve the system, the authors utilize a mean-field approximation, simplifying the network into a 1-investor vs. 1-asset model. This allows for the use of perturbation theory to identify critical points where the system undergoes phase transitions.

Empirical Grounding

To validate the model, simulations were grounded in empirical data from the 2009-2011 Eurozone crisis. The dataset included sovereign bonds from five European countries (Greece, Italy, Ireland, Portugal, and Spain) as assets, and the network of major financial institutions trading these bonds as investors.

Results: Market Phases and Stability

The research identifies that market stability is dictated by a single metric: the product of the behavioral parameters, $\gamma = \alpha\beta$. A second-order phase transition occurs at exactly $\gamma = 1$, separating the market into three distinct physical phases:

  1. Stable (Oscillatory) Phase ($\gamma < 0$): Characterized by contrarian behavior (e.g., buying when prices drop), negative shocks cause minor oscillations before returning to equilibrium. This behavior acts as a systemic shock absorber.
  2. Stable (Logistic) Phase ($0 < \gamma < 1$): The system absorbs shocks and settles into a new equilibrium. While individual leveraged investors might fail, the systemic shock is contained.
  3. Unstable (Boom/Bust) Phase ($\gamma > 1$): Market confidence is lost. Positive shocks lead to exponential growth (bubbles), while negative shocks lead to exponential decay and widespread systemic failure (crashes).
Phase diagrams showing the sum of final prices and relaxation time as a function of $\alpha$ and $\beta$, illustrating the stability boundary.

In the mean-field limit, this model naturally reduces to the Langevin equation for price returns proposed by Bouchaud in 1998. Notably, the GIPSI model generates the non-linearities required for a crash naturally from its network dynamics, rather than requiring them to be added manually.

Discussion and Limitations

While the model provides a powerful framework for understanding market phases, it involves several simplifications. The bipartite assumption ignores intra-layer connections, such as interbank lending networks, which are critical for real-world contagion. Furthermore, the mean-field approximation assumes uniform leverage and rashness across institutions, a point of contention among economists studying "fire sales." Future work will explore the model's applicability to broader stock markets beyond sovereign debt.

Summary updated

Abstract

We analyze the linear response of a market network to shocks based on the bipartite market model we introduced in an earlier paper, which we claimed to be able to identify the time-line of the 2009-2011 Eurozone crisis correctly. We show that this model has three distinct phases that can broadly be categorized as "stable" and "unstable". Based on the interpretation of our behavioral parameters, the stable phase describes periods where investors and traders have confidence in the market (e.g. predict that the market rebounds from a loss). We show that the unstable phase happens when there is a lack of confidence and seems to describe "boom-bust" periods in which changes in prices are exponential. We analytically derive these phases and where the phase transition happens using a mean field approximation of the model. We show that the condition for stability is $αβ&lt;1$ with $α$ being the inverse of the "price elasticity" and $β$ the "income elasticity of demand", which measures how rash the investors make decisions. We also show that in the mean-field limit this model reduces to the Langevin model by Bouchaud et al. for price returns.
BibTeX
@misc{dehmamy2016crises,
  author = {Nima Dehmamy and Sergey V. Buldyrev and Shlomo Havlin and H. Eugene Stanley and Irena Vodenska},
  title = {{Crises and Physical Phases of a Bipartite Market Model}},
  year = {2016},
  eprint = {1609.05939},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/1609.05939},
}
Figure