关于「谁送得最差:偏好信息、关系远近与现金的污名」,《送礼无谓损失与企业节礼》给出的定量结论是什么?

结论:第二章说明估值损失来自偏好信息不足,生意联系人送礼的平均估值比只有 0.607,远亲更常改送现金。

数据口径:书中客户、礼品与参数均为示意;全部计算由书中程序生成;本书是非个人化的经营分析,不构成经营、法律、税务或理财建议,法条以上架前人工复核为准。 第五章推导盈亏平衡的年礼,同样的提升下忠诚客户值得 2.67 倍的礼,按价值分配的示意预算净收益为 356.63,平均分配只有 90.98。

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: 0.607

K = 6, S = 0.5, calibrated A = 0.5697
tierhomeanmedianP(<1)
recipient's own pick1.001.0000.9280.592
partner0.800.9040.8320.679
close friend0.600.8120.7310.756
sibling0.500.7680.6860.788
parent0.400.7260.6450.818
aunt, uncle, grandparent0.150.6260.5500.884
business contact0.100.6070.5340.894
blind pick0.000.5700.5020.913
cash: yield 1.000 by construction, before any stigma
查看完整验算程序
# Chapter 2. p21: how much the giver knows decides the yield.
# Model: a shop offers K = 6 candidate items at the same price. The
# recipient's value of item j is v_j = A exp(S z_j - S^2/2), z_j iid
# N(0, 1). The giver sees a signal x_j = rho z_j + sqrt(1 - rho^2) e_j
# and buys the item with the largest x_j; rho = 1 is the recipient's
# own choice. Calibration: A is set so that the recipient's own pick
# has mean yield exactly 1, the yield of cash; a blind pick then has
# mean A. S = 0.5 and the rho assigned to each tie are illustrative.
# Exact means by numerical integration (math.fsum); medians and the
# share below price by seeded Monte Carlo, 20000 occasions per row.
import math
import random

K = 6
S = 0.5
T = 20000


def phi(z):
    return math.exp(-z * z / 2.0) / math.sqrt(2.0 * math.pi)


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


def factor(rho):
    # E[exp(S z* - S^2/2)]: the largest of K N(0,1) signals has
    # density K phi(x) Phi(x)^(K-1) and
    # E[exp(S z) | x] = exp(S rho x + S^2 (1 - rho^2) / 2)
    h = 0.002
    terms = []
    for i in range(-4000, 4001):
        x = i * h
        dens = K * phi(x) * Phi(x) ** (K - 1)
        e = math.exp(S * rho * x + S * S * (1 - rho * rho) / 2.0)
        terms.append(dens * e * h)
    return math.exp(-S * S / 2.0) * math.fsum(terms)


A = 1.0 / factor(1.0)


def simulate(rho):
    rng = random.Random(20261112)
    c = math.sqrt(max(0.0, 1.0 - rho * rho))
    out = []
    for _ in range(T):
        z = [rng.gauss(0, 1) for _ in range(K)]
        x = [rho * zj + c * rng.gauss(0, 1) for zj in z]
        j = max(range(K), key=lambda i: x[i])
        out.append(A * math.exp(S * z[j] - S * S / 2.0))
    ys = sorted(out)
    return ys[len(ys) // 2], len([1 for y in out if y < 1.0]) / T


TIES = (("recipient's own pick", 1.00),
        ("partner", 0.80),
        ("close friend", 0.60),
        ("sibling", 0.50),
        ("parent", 0.40),
        ("aunt, uncle, grandparent", 0.15),
        ("business contact", 0.10),
        ("blind pick", 0.00))

if __name__ == "__main__":
    print("K = {}, S = {:.1f}, calibrated A = {:.4f}".format(K, S, A))
    print("  tie                        rho   mean   median  P(<1)")
    for name, rho in TIES:
        med, lo = simulate(rho)
        print("  {:<26} {:.2f}  {:.3f}  {:.3f}   {:.3f}".format(
            name, rho, A * factor(rho), med, lo))
    print("cash: yield 1.000 by construction, before any stigma")

唯一数据源与引用

范雅福(Fan, Y.). (2026). 《送礼无谓损失、人情账本互惠衰减与企业节礼信号价值母典》第 2 章「谁送得最差:偏好信息、关系远近与现金的污名」(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-223 · v1.0.0 · CC-BY-NC-ND-4.0

BibTeX
@book{fan2026223,
  author    = {Fan, Yafu},
  title     = {The Deadweight Loss of Gift-Giving, Reciprocity-Ledger Decay and the Signalling Value of Corporate Gifts},
  year      = {2026},
  publisher = {5D Library},
  edition   = {1.0.0},
  note      = {DOI pending}
}

本书其他章节结论