关于「风险仓切分:切分数、尾部指数与刺破公司面纱」,《责任图渗流与担保尾部破产》给出的定量结论是什么?

结论:第五章说明风险仓切分的效果取决于尾部:切成八个仓,轻尾时破产概率从 0.0102 升到 0.0355,重尾时从 0.0489 降到 0.0027;七条仓间担保串成的链使它升到 0.0943。

数据口径:书中的资产负债表、损失分布与法律参数均为示意;全部计算由书中程序生成;本书不是个人化的法律、税务或财务建议,各法域规则不同,书中也不提供任何规避合法债务或逃避执行的方法。 书中以标注框区分已发表的证据、可运行的 Python 计算(17 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: 0.0027, 0.0102, 0.0355, 0.0489

P(ruin) and mean buckets failed per year, by number of buckets k
loss lawk = 1k = 2k = 4k = 8
exponential0.01020.0100.08230.0840.05620.3670.03551.081
lognormal0.06050.0610.16970.1780.06740.4420.01600.959
Pareto 1.50.04890.0490.11140.1150.02330.2560.00270.545
查看完整验算程序
# Chapter 5. p51: splitting eight business lines into k risk buckets
# (separate entities). Each line loses X a year, mean 0.5, independent.
# Equity C = 8 and going-concern value F = 8 are split evenly over the
# buckets. A bucket whose loss exceeds its equity fails: the owner loses
# its equity and its going-concern value; otherwise the owner bears the
# loss. Ruin = the owner loses more than half of C + F. No guarantees and
# no piercing here (see p52). Seeded Monte Carlo; illustrative inputs.
import math
import random

LINES, C, F, PATHS = 8, 8.0, 8.0, 100000
KS = (1, 2, 4, 8)
rng = random.Random(20261112)


def draw(kind):
    u = 1.0 - rng.random()
    if kind == "exponential":
        return -0.5 * math.log(u)
    if kind == "lognormal":
        return 0.5 * math.exp(rng.gauss(-0.72, 1.2))
    return (0.5 / 3) * u ** (-1 / 1.5)        # Pareto, tail index 1.5


print("P(ruin) and mean buckets failed per year, by number of buckets k")
print("{:12s}".format("loss law") +
      "".join("{:>15s}".format("k = {}".format(k)) for k in KS))
for kind in ("exponential", "lognormal", "Pareto 1.5"):
    ruin = [0] * len(KS)
    fails = [0] * len(KS)
    for _ in range(PATHS):
        xs = [draw(kind) for _ in range(LINES)]
        for i, k in enumerate(KS):
            per = LINES // k
            cap, fr = C / k, F / k
            lost = []
            for b in range(k):
                lb = math.fsum(xs[b * per:(b + 1) * per])
                if lb > cap:
                    lost.append(cap + fr)
                    fails[i] += 1
                else:
                    lost.append(lb)
            if math.fsum(lost) > (C + F) / 2:
                ruin[i] += 1
    print("{:12s}".format(kind) + "".join(
        "{:8.4f} {:6.3f}".format(r / PATHS, f / PATHS)
        for r, f in zip(ruin, fails)))

唯一数据源与引用

范雅福(Fan, Y.). (2026). 《多子公司责任隔离图渗流、个人连带担保尾部破产放大与经营资产风险仓切分》第 5 章「风险仓切分:切分数、尾部指数与刺破公司面纱」(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-177 · v1.0.0 · CC-BY-NC-ND-4.0

BibTeX
@book{fan2026177,
  author    = {Fan, Yafu},
  title     = {Liability Percolation across Subsidiary Graphs, Tail-Ruin Amplification from Personal Guarantees and Risk-Bucket Partitioning of Operating Assets},
  year      = {2026},
  publisher = {5D Library},
  edition   = {1.0.0},
  note      = {DOI pending}
}

本书其他章节结论