关于「科斯猜想与承诺装置:限量要先把卖家自己绑住」,《限量稀缺与早鸟阶梯定价》给出的定量结论是什么?
结论:科斯猜想在月汐的需求上是一张具体的账单。p20 显示,可以随时再印的卖家,再印越频繁,24 个月的现值毛利越低:从一次性的 RM 121,226 降到每月再印的 RM 96,921,比被相信的限量少 RM 24,341。Bulow 提醒我们区分消耗型与存留型的稀缺,工作坊席位天然被日历约束,编号牌卡则必须另设装置。Stokey 的结论经 p24 验证:事先公布的降价表赚不过一口价,人人同时到场时上升的价格表也赚不过,早鸟阶梯若有用,理由必然在别处。p21 把四件装置加一件可选的补差承诺排成一叠:编号把 RM 289 上的毛利补回 RM 23,518,后面的装置不再改变销量,却把出清价推高到 RM 335.42。p23 说明编号的威慑来自登记簿,登记率 0.4 时多印 20 套几乎一定败露。p22 则把违约的诱惑摊开:销毁印版与补差承诺之后仍剩 RM 3,607,真正让限量可信的,是品牌还有未来。下一章转到早鸟阶梯,看它在什么条件下真的赚钱。
数据口径:「限量」「早鸟」宣称的消费者保护规范只写一般框架,不是法律意见。 书中以标注框区分已发表的证据、可运行的计算、原典与作者的假说,每条假说都写明什么结果会推翻它;原典考辨取《韩非子·外储说左上》的「买椟还珠」,讨论包装溢价与内容价值的分离。
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: 121226, 24, 24341, 289, 96921
A. one sale and never again (continuous model, as in p11):
open edition, one window: RM 236 x 857.7 sets, margin RM 120,874
credible limit 500 with uplift 0.12: clearing RM 340.75, margin RM 121,263
Moontide plan, 500 sets at RM 289: margin RM 96,165
B. no commitment: reprints at cost allowed over 24 months, 0.85 per half year, no uplift
reprint every sales sets by month 24 first sale (RM x sets) PV margin PV / one window
24 months 1 860 236.04 x 860 121,226 1.000
12 months 2 1105 223.92 x 520 109,767 0.905
6 months 4 1265 200.51 x 560 99,552 0.821
3 months 8 1265 213.49 x 405 98,469 0.812
1 months 24 1265 222.54 x 245 96,921 0.800
monthly reprints, first sale RM 222.54 x 245 sets; next five: 218.17, 214.16, 210.46, 206.99, 203.76
monthly reprints, last sale RM 163.56; bucket one-sale check RM 236.04 x 860 sets
C. credible limit minus monthly-reprint PV: RM 24,341 (0.201 of the credible limit)查看完整验算程序
# Chapter 2. p20: The Coase conjecture on Moontide's demand: what happens when the seller may reprint.
# Book-wide WTP model (ASSUMPTION, from p10): 1,440 collectors, lognormal WTP, median RM 260, sigma 0.40.
# Unit cost RM 88, payment fee 0.03, so the margin of a set sold at p is 0.97 p - 88. Resale accounts
# are left out. Part A repeats the continuous one-sale benchmarks of p11 (signal uplift s = 0.12 only
# when the limit is believed). Part B: no commitment. The seller may reprint at cost every m months over
# a 24-month horizon (T = 24 / m sales, per-sale discount d = 0.85 ** (m / 6), i.e. the half-year
# waiting factor 0.85 of p11 spread evenly over the months; an assumption, like every parameter here).
# Collectors are grouped into 288 equal buckets of 5 (value = quantile at the bucket midpoint). Backward
# induction: the marginal buyer of a sale is indifferent between buying now and waiting for the next
# sale the policy reaches; the seller chooses how far down the demand curve each sale goes. No uplift.
import math
from statistics import NormalDist
N, BUCKET, MEDIAN, SIGMA, COST, FEE, S_UP, EDITION = 1440, 5, 260.0, 0.40, 88.0, 0.03, 0.12, 500
B = N // BUCKET
VAL = [MEDIAN * math.exp(SIGMA * NormalDist().inv_cdf(1 - (j + 0.5) / B)) for j in range(B)]
def margin(p):
return p * (1 - FEE) - COST
def surv(v):
return 0.5 * math.erfc((math.log(v) - math.log(MEDIAN)) / SIGMA / math.sqrt(2))
def solve(T, d):
"""Return (PV margin, list of (price, sets) per sale) from the full market."""
w_prev, first_prev, pols = [0.0] * (B + 1), [None] * (B + 1), []
for _ in range(T):
w, pol, first = [0.0] * (B + 1), [(0, None)] * (B + 1), [None] * (B + 1)
for k in range(B + 1):
best, bk, bp = d * w_prev[k], k, None
for k2 in range(k + 1, B + 1):
v, f = VAL[k2 - 1], first_prev[k2]
p = v if f is None else v - d ** (f[0] + 1) * (v - f[1])
g = (k2 - k) * BUCKET * margin(p) + d * w_prev[k2]
if g > best:
best, bk, bp = g, k2, p
w[k], pol[k] = best, (bk, bp)
if bk > k:
first[k] = (0, bp)
elif first_prev[k] is not None:
first[k] = (first_prev[k][0] + 1, first_prev[k][1])
pols.append(pol)
w_prev, first_prev = w, first
path, k = [], 0
for pol in reversed(pols):
k2, p = pol[k]
path.append((p, (k2 - k) * BUCKET))
k = k2
return w_prev[0], path
if __name__ == "__main__":
best = max(range(120, 501), key=lambda p: margin(p) * N * surv(p))
open_m = margin(best) * N * surv(best)
lo, hi = 100.0, 800.0
for _ in range(80):
mid = (lo + hi) / 2
lo, hi = (mid, hi) if N * surv(mid / (1 + S_UP)) > EDITION else (lo, mid)
lim_m, plan_m = EDITION * margin(lo), EDITION * margin(289)
print("A. one sale and never again (continuous model, as in p11):")
print(" open edition, one window: RM {} x {:.1f} sets, margin RM {:,.0f}".format(
best, N * surv(best), open_m))
print(" credible limit {} with uplift {:.2f}: clearing RM {:.2f}, margin RM {:,.0f}".format(
EDITION, S_UP, lo, lim_m))
print(" Moontide plan, {} sets at RM 289: margin RM {:,.0f}".format(EDITION, plan_m))
print("B. no commitment: reprints at cost allowed over 24 months, 0.85 per half year, no uplift")
print(" reprint every sales sets by month 24 first sale (RM x sets) PV margin PV / one window")
rows = {}
for m in (24, 12, 6, 3, 1):
pv, path = solve(24 // m, 0.85 ** (m / 6))
units = sum(u for _, u in path)
rows[m] = (pv, path)
print(" {:5d} months {:5d} {:16d} {:11.2f} x {:6d} {:11,.0f} {:.3f}".format(
m, 24 // m, units, path[0][0], path[0][1], pv, pv / rows[24][0]))
pv1, path1 = rows[1]
sales = [(p, u) for p, u in path1 if u > 0]
print(" monthly reprints, first sale RM {:.2f} x {} sets; next five: {}".format(
sales[0][0], sales[0][1], ", ".join("{:.2f}".format(p) for p, _ in sales[1:6])))
print(" monthly reprints, last sale RM {:.2f}; bucket one-sale check RM {:.2f} x {} sets".format(
sales[-1][0], rows[24][1][0][0], rows[24][1][0][1]))
print("C. credible limit minus monthly-reprint PV: RM {:,.0f} ({:.3f} of the credible limit)".format(
lim_m - pv1, (lim_m - pv1) / lim_m))唯一数据源与引用
范雅福(Fan, Y.). (2026). 《高奢稀缺配额拍卖、早鸟阶梯涨价动力学与限量实体品溢价母典》第 2 章「科斯猜想与承诺装置:限量要先把卖家自己绑住」(版本 0.1.0). 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-136 · v0.1.0 · CC-BY-NC-ND-4.0
BibTeX
@book{fan2026136,
author = {Fan, Yafu},
title = {Luxury Scarcity Allocation Auctions, Early-Bird Price-Ladder Dynamics and Limited-Edition Physical Premiums},
year = {2026},
publisher = {5D Library},
edition = {0.1.0},
note = {DOI pending}
}