关于「破产边界:准备金 / 负债比、沉睡依赖与二维压力图」,《积分负债与沉睡精算》给出的定量结论是什么?
结论:第六章画出破产边界:两年内透支概率不超过百分之一需要 0.492 的准备金比例,而一封到期提醒就会使依赖沉睡的计划少赚 0.121 的利润。
数据口径:「忽略沉睡会错估负债」须分方向:完全忽略沉睡在示意数据中把负债多记 0.473,而用未成熟数据直接计数则把每分递延额少记 0.272,后者更危险。 书中参数为示意;全部计算由书中程序生成;本书是非个人化会计、税务或法律建议,以执业会计师意见为准。
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: 0.492
| paths 4000 | months 24 | full cost 90000 |
| rho | reserve | P(ruin) |
| 0.00 | 0 | 0.9050 |
| 0.05 | 4500 | 0.5315 |
| 0.10 | 9000 | 0.3870 |
| 0.20 | 18000 | 0.3058 |
| 0.30 | 27000 | 0.1403 |
| 0.50 | 45000 | 0.0088 |
| rho for 1 in 100 ruin: 0.492 |
查看完整验算程序
# Chapter 6. p61: probability that the redemption reserve runs dry.
# Illustrative Monte Carlo, 4000 paths of 24 months, fixed seed.
# Expected-to-redeem points start at 15000000 and 750000 are added a
# month. Monthly hazard 0.05 times lognormal noise (log-sd 0.3); with
# chance 0.02 a month a shock starts that multiplies the hazard by 4
# for 2 months. Cash cost c = 0.006 per point; operations put 4600 a
# month toward redemptions. A path is ruined if reserve cash goes
# below zero. The same paths are used for every reserve ratio rho
# (reserve over the full fulfilment cost 90000).
import math
import random
PATHS, MONTHS, OUT0, ADD, H, SD = 4000, 24, 15000000, 750000, 0.05, 0.3
P_SHOCK, M_SHOCK, C, INFLOW = 0.02, 4.0, 0.006, 4600.0
def net_paths(rng):
paths = []
for _ in range(PATHS):
out, left, net = OUT0, 0, []
for _ in range(MONTHS):
if left == 0 and rng.random() < P_SHOCK:
left = 2
m = M_SHOCK if left > 0 else 1.0
left = max(0, left - 1)
h = min(1.0, H * m * rng.lognormvariate(-SD * SD / 2, SD))
red = out * h
out = out - red + ADD
net.append(INFLOW - C * red)
paths.append(net)
return paths
def lowest(net):
run = [math.fsum(net[:k + 1]) for k in range(len(net))]
return min(run)
if __name__ == "__main__":
lows = [lowest(n) for n in net_paths(random.Random(20261113))]
full = C * OUT0
print("paths {} months {} full cost {:.0f}".format(
PATHS, MONTHS, full))
print(" rho reserve P(ruin)")
for rho in (0.0, 0.05, 0.10, 0.20, 0.30, 0.50):
ruined = len([1 for x in lows if rho * full + x < 0])
print("{:6.2f} {:8.0f} {:8.4f}".format(rho, rho * full,
ruined / PATHS))
lows.sort()
print("rho for 1 in 100 ruin: {:.3f}".format(-lows[PATHS // 100] / full))唯一数据源与引用
范雅福(Fan, Y.). (2026). 《会员积分负债与沉睡率精算、积分发行兑付率与递延收入的破产边界母典》第 6 章「破产边界:准备金 / 负债比、沉睡依赖与二维压力图」(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-315 · v1.0.0 · CC-BY-NC-ND-4.0
BibTeX
@book{fan2026315,
author = {Fan, Yafu},
title = {Loyalty-Point Liabilities and Breakage Actuarial Science: Issuance, Redemption Rates and the Insolvency Boundary of Deferred Revenue},
year = {2026},
publisher = {5D Library},
edition = {1.0.0},
note = {DOI pending}
}