[论文解读] Oligarchy as a Phase Transition: The effect of wealth-attained advantage in a Fokker-Planck description of asset exchange

Bruce M. Boghosian, Adrian Devitt‐Lee+3 more|arXiv (Cornell University)|Nov 3, 2015

Complex Systems and Time Series Analysis参考文献 14被引用 4

一句话总结

本文通过引入连续的财富获取优势（WAA）机制，扩展了资产交换的Yard-Sale模型，使更富有的参与者在交易中获得系统性优势。利用非线性Fokker-Planck方程，研究发现当WAA达到临界阈值时，系统经历二阶相变，形成寡头垄断状态，并在凝聚态与非凝聚态财富分布之间出现共存相；同时表明，在临界点处，非寡头财富分布的尾部行为从高斯分布转变为指数衰减。

ABSTRACT

In earlier work, we derived a nonlinear, nonlocal Fokker-Planck equation for the Yard-Sale Model of asset exchange. In the absence of redistribution, we showed that the Gini coefficient is a Lyapunov functional for this model, tending to one in the time-asymptotic limit, corresponding to maximal inequality. When a one-parameter model of redistribution is introduced, we showed that the model admits a steady state similar to Pareto's Law. In this work, we analyze the form of this distribution in greater detail, both analytically and numerically. We find that, while Pareto's Law is approximately valid for low redistribution, it gives way to something like Gibrat's Law at higher redistribution. We also prove that, while this Pareto or Gibrat behavior persists over many orders of magnitude, it ultimately gives way to gaussian decay at extremely large wealth. Following the work of Moukarzel et al., we introduce a bias in favor of the wealthier agent. We derive the corresponding modification to the Fokker-Planck equation, and we show this leads to wealth condensation when the bias exceeds a critical value. Earlier work took the bias to be a discontinuous function of the wealth differential between the two transacting agents, and reported a first-order phase transition to absolute oligarchy. By contrast, in this work we take the bias to be a continuous function of the wealth differential, and consequently we observe a second-order phase transition with a region of coexistence between the oligarch and a distribution of non-oligarchs. We additionally show that the onset of wealth condensation has a reciprocal effect on the character of the non-oligarchical part of the distribution. Specifically, we show that the above-mentioned gaussian decay at extremely large wealth is valid both above and below criticality, but degenerates to exponential decay precisely at criticality.

研究动机与目标

• 建模财富获取优势（WAA）如何在经济交换模型中驱动财富凝聚。
• 利用Fokker-Planck框架，分析不同WAA水平与再分配程度下的稳态财富分布。
• 确定在连续WAA条件下，向寡头垄断的转变是第一阶还是第二阶相变，与先前的不连续模型形成对比。
• 研究财富分布尾部的渐近行为，特别是其在临界点处的变化。
• 确立所得Fokker-Planck方程在财富动力学中的普适性与解析可解性。

提出的方法

• 推导出包含连续WAA的Yard-Sale模型的非线性、积分微分Fokker-Planck方程，将财富交换建模为随机过程。
• 引入一个连续的偏置函数，使财富较多的参与者在交易中获得优势，该函数依赖于交易双方的财富差异。
• 通过尺度变换与标准型变换，将系统简化为依赖于再分配率 $\overline{\tau}$ 和WAA强度 $\overline{\zeta}$ 的双参数解族。
• 对稳态代理密度函数在低、中、高财富值区域进行渐近分析，以表征分布尾部的特征。
• 通过数值求解稳态Fokker-Planck方程，验证理论预测并检测临界行为。
• 采用有限代理数的蒙特卡洛模拟，探究连续极限并确认在临界点处财富占比存在斜率不连续性。

实验结果

研究问题

• RQ1 连续WAA机制是否在Yard-Sale模型中引发向寡头垄断的首阶或二阶相变？
• RQ2 稳态财富分布如何随再分配和WAA强度的变化而演化？
• RQ3 非寡头财富分布尾部的渐近行为是什么？其在财富凝聚临界点处如何变化？
• RQ4 财富凝聚的出现是否引起财富分布尾部形状的定性改变？
• RQ5 具有连续WAA的Fokker-Planck方程是否可普遍应用于基于代理的交换系统中的财富动力学建模？

主要发现

• 模型在临界点 $\overline{\zeta} = \overline{\tau}$ 处表现出向寡头垄断的二阶相变，此时最富有代理所持财富比例出现斜率不连续。
• 在临界点处，非寡头财富分布的尾部从高斯衰减退化为指数衰减，该结论得到渐近分析与数值证据的双重验证。
• 当 $\overline{\zeta} < \overline{\tau}$（亚临界）时，尾部衰减为 $\exp(-w^2/2)$，表现为高斯行为；当 $\overline{\zeta} > \overline{\tau}$（超临界）时，尾部仍保持高斯衰减。
• 在临界点（$\overline{\zeta} = \overline{\tau}$）处，尾部呈指数衰减 $\exp(-3w/2)$，其斜率 $-3/2$ 经理论与数值分析共同确认。
• 有限代理数（64至512）的蒙特卡洛模拟收敛于理论连续极限，且在 $\overline{\zeta}/\overline{\tau} = 1$ 处清晰显示出斜率不连续，证实了二阶相变的存在。
• 稳态基尼系数随 $\overline{\zeta}$ 连续增加，在完全凝聚相中达到 $G=1$，且在临界点处表现出非解析行为，表明相共存现象的存在。