关于「准备金挤兑:Diamond–Dybvig 合同与全局博弈阈值」,《交子、格雷欣与稳定币挤兑》给出的定量结论是什么?

结论:第四章说明流动性保险与挤兑脆弱性同源,最优存款合同在噪声信号下的挤兑概率为 0.6248。

数据口径:书中参数为示意;全部计算由书中程序生成;本书不是个人化的理财建议,不推荐任何币种、稳定币或平台,法规事项以上架前人工复核为准。 书中以标注框区分已发表的证据、可运行的 Python 计算(18 段程序)、原典与作者的假说。

English version

对照数据表

下表数字由本章 Python 程序原样打印,可重新运行复现。 程序只用 Python 标准库与结论吻合的打印值: 0.6248

r = 0.8; theta ~ normal(2.5, 0.5)
promise c1buffer btheta*P(run)
1.00000.01.6470.0440
1.00000.11.5250.0256
1.00000.31.3550.0110
1.05000.01.9380.1306
1.05000.11.8060.0825
1.05000.31.6560.0458
1.10000.02.2520.3101
1.10000.12.1080.2163
1.10000.31.9840.1511
1.15960.02.6590.6248
1.15960.12.4990.4992
1.15960.32.4150.4321
查看完整验算程序
# Chapter 4. p42: a unique run threshold once signals are noisy
# Global-game version of p41 (after Morris and Shin 1998 and Goldstein
# and Pauzner 2005). The long asset's date-2 return theta is uncertain.
# With small private noise, a patient depositor at the threshold signal
# is indifferent when the date-1 withdrawal share f is uniform on
# [LAM, 1] (Laplace belief): the mean of stay(f) - withdraw(f) is zero.
# The bank promises c1 at date 1 and holds LIQ = LAM * c1 + b in cash,
# b an extra buffer. theta is drawn from a normal with mean 2.5 and sd
# 0.5 to price the run risk. c1 = 1.1596 is the p41 optimum for gam = 2;
# the other parameters are illustrative.
import math

LAM, R_LIQ = 0.25, 0.8
MEAN, SD, N = 2.5, 0.5, 2000
PROMISES = [1.0, 1.05, 1.10, 1.1596]


def gap(f, theta, liq, r, c1):
    """stay minus withdraw for a patient depositor."""
    long_ = 1.0 - liq
    need = f * c1
    if need <= liq:
        stay = (liq - need + long_ * theta) / (1.0 - f)
        return stay - c1
    sold = (need - liq) / r
    if sold <= long_:
        return theta * (long_ - sold) / (1.0 - f) - c1
    return 0.0 - (liq + r * long_) / f


def mean_gap(theta, liq, r, c1):
    h = (1.0 - LAM) / N
    pts = [gap(LAM + (i + 0.5) * h, theta, liq, r, c1)
           for i in range(N)]
    return math.fsum(pts) / N


def threshold(liq, r, c1):
    lo, hi = 0.5, 8.0
    for _ in range(50):
        mid = 0.5 * (lo + hi)
        if mean_gap(mid, liq, r, c1) < 0.0:
            lo = mid
        else:
            hi = mid
    return 0.5 * (lo + hi)


def phi(z):
    return 0.5 * (1.0 + math.erf(z / math.sqrt(2.0)))


print("r = {}; theta ~ normal({}, {})".format(R_LIQ, MEAN, SD))
print("promise c1  buffer b  theta*   P(run)")
for c1 in PROMISES:
    for b in (0.0, 0.1, 0.3):
        t = threshold(LAM * c1 + b, R_LIQ, c1)
        print("{:10.4f} {:9.1f} {:7.3f} {:8.4f}".format(
            c1, b, t, phi((t - MEAN) / SD)))

唯一数据源与引用

范雅福(Fan, Y.). (2026). 《交子超发与格雷欣定律、浮动兑换下的提尔斯逆转与稳定币挤兑临界母典》第 4 章「准备金挤兑:Diamond–Dybvig 合同与全局博弈阈值」(版本 1.0.0). 5D Library. DOI 待发布

DOI 待发布 · 专著页正式发布后上线(本站暂无专著页) · 5DL-B-221 · v1.0.0 · CC-BY-NC-ND-4.0

BibTeX
@book{fan2026221,
  author    = {Fan, Yafu},
  title     = {Jiaozi Over-Issue and Gresham's Law, Thiers' Reversal under Floating Exchange and Stablecoin Run Thresholds},
  year      = {2026},
  publisher = {5D Library},
  edition   = {1.0.0},
  note      = {DOI pending}
}

本书其他章节结论