关于「筛选均衡:把 Rothschild 与 Stiglitz 的保险市场搬到快递员的门口」,《货到付款押金与预付筛选》给出的定量结论是什么?
结论:第二章用分数精确求出分离菜单:守信买家的押金为 4.1108,随手下单的买家留在普通货到付款;当随手下单买家的占比不超过 0.1900 时,一份混同合同就能打破菜单。信任赤字是筛选的代价:押金不超过拒收成本、又让守信买家得益,只在一个信任成本窗口里成立。
数据口径:本书把一单的账拆开:示意参数下,签收一单的毛利为 4.5342,拒收一单要付出双程运费、货损与资金占用共 4.3522,盈亏平衡拒收率为 0.4202;在买家自愿付款的保守假设下,追讨要打平需要联系率与付款率之积达到 0.2068,现实中很难做到。 东南亚拒收率没有找到经同行评议的估计,书中参数全部为示意,全部计算由书中程序生成。
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: 0.1900, 4.1108
| phi | D* | D*/K | phi*D* | gain_H@0.25 | min share | cap |
| 0.00 | 5.1522 | 1.184 | 0.0000 | 0.4416 | 0.0000 | D*>K |
| 0.02 | 4.8453 | 1.113 | 0.0969 | 0.3447 | 0.0585 | D*>K |
| 0.05 | 4.4479 | 1.022 | 0.2224 | 0.2192 | 0.1310 | D*>K |
| 0.08 | 4.1108 | 0.945 | 0.3289 | 0.1128 | 0.1900 | ok |
| 0.12 | 3.7335 | 0.858 | 0.4480 | -0.0064 | 0.2533 | ok |
| 0.20 | 3.1544 | 0.725 | 0.6309 | -0.1893 | 0.3455 | ok |
| 0.30 | 2.6421 | 0.607 | 0.7926 | -0.3510 | 0.4222 | ok |
| fair cap D* <= K needs phi >= 0.058047 | ||||||
| committed buyers gain at share 0.25 only if phi < 0.117647 |
查看完整验算程序
# Chapter 2. p23: the trust deficit as the price of screening.
# phi is the expected loss a buyer attaches to each unit paid in
# advance to an unfamiliar seller. It is pure deadweight: an honest
# seller gains nothing from it. For each phi the program recomputes
# the separating deposit D*, the deadweight phi D* borne by a
# committed buyer, his gain over pooled COD at a casual share of 0.25,
# and the casual share below which the menu breaks (as in p22).
# Exact fractions; parameters as in p21, phi illustrative.
from fractions import Fraction as F
P, C, SO, SR = F(20), F(13), F(9, 5), F(9, 5)
RATE, DR, DF, FEE, A = F(3, 25), 21, 10, F(3, 100), F(4, 5)
QH, QL, LAM = F(1, 20), F(7, 20), F(1, 4)
c = C + SO + FEE * P + P * RATE * DF / 365
K = SO + SR + C / 20 + (C + SO) * RATE * DR / 365
def price(q, d):
return c + (q * (K - d) + A) / (1 - q)
def util(q, p, d, phi):
return (1 - q) * (P - p) - (q + phi) * d
def deposit(phi):
d0 = price(QL, 0) - price(QH, 0)
return (1 - QL) * d0 / (QL + phi - (1 - QL) * QH / (1 - QH))
def breaks(lam, phi, ds):
ul = util(QL, price(QL, 0), 0, phi)
uh = util(QH, price(QH, ds), ds, phi)
qb = lam * QL + (1 - lam) * QH
lo, hi = F(0), K
for q, t, strict in ((QL, ul, False), (QH, uh, True)):
a = util(q, price(qb, 0), 0, phi) - t
b = util(q, price(qb, 1), 1, phi) - util(q, price(qb, 0), 0, phi)
if b == 0:
if (a <= 0) if strict else (a < 0):
return False
continue
if b > 0:
lo = max(lo, -a / b)
else:
hi = min(hi, -a / b)
return lo < hi
def threshold(phi, ds):
lo, hi = F(0), F(1)
for _ in range(50):
mid = (lo + hi) / 2
if breaks(mid, phi, ds):
lo = mid
else:
hi = mid
return hi
if __name__ == "__main__":
qb = LAM * QL + (1 - LAM) * QH
print(" phi D* D*/K phi*D* gain_H@0.25 min share cap")
for num in (0, 2, 5, 8, 12, 20, 30):
phi = F(num, 100)
ds = deposit(phi)
uh = util(QH, price(QH, ds), ds, phi)
gain = uh - util(QH, price(qb, 0), 0, phi)
cap = "ok" if ds <= K else "D*>K"
print("{:4.2f} {:7.4f} {:6.3f} {:7.4f} {:11.4f} {:10.4f} {}"
.format(float(phi), float(ds), float(ds / K),
float(phi * ds), float(gain),
float(threshold(phi, ds)), cap))
d0 = price(QL, 0) - price(QH, 0)
lowphi = (1 - QL) * d0 / K - QL + (1 - QL) * QH / (1 - QH)
lo, hi = F(0), F(1, 2)
for _ in range(50):
mid = (lo + hi) / 2
ds = deposit(mid)
g = util(QH, price(QH, ds), ds, mid) - util(QH, price(qb, 0), 0,
mid)
if g > 0:
lo = mid
else:
hi = mid
print("fair cap D* <= K needs phi >= {:.6f}".format(float(lowphi)))
print("committed buyers gain at share 0.25 only if phi < {:.6f}"
.format(float(hi)))唯一数据源与引用
范雅福(Fan, Y.). (2026). 《货到付款拒收博弈与东南亚电商信任赤字、押金阈值与预付折扣激励设计母典》第 2 章「筛选均衡:把 Rothschild 与 Stiglitz 的保险市场搬到快递员的门口」(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-314 · v1.0.0 · CC-BY-NC-ND-4.0
BibTeX
@book{fan2026314,
author = {Fan, Yafu},
title = {Cash-on-Delivery Refusal Games and the Trust Deficit in Southeast Asian E-Commerce: Deposit Thresholds and Prepayment-Discount Incentive Design},
year = {2026},
publisher = {5D Library},
edition = {1.0.0},
note = {DOI pending}
}