关于「原典并读:《管子》轻重、衡数与计然之策」,《选择架构与价格界面》给出的定量结论是什么?

结论:收成冲击不持续(\(\rho = 0\))时,没有常平仓的价格均值为 54.0、标准差 22.8,4,000 年中有 399 年低于 30、490 年高于 80;常平仓把高于 80 的年份降到 0,低于 30 的年份降到 168(库存上限使它在丰年收不完),价格标准差降到 16.9,并且累计盈利 1,744.2——低买高卖本身就是「君得其利」。收成冲击持续(\(\rho = 0.5\))时,连续的歉收会耗尽库存,有 38 年价格突破 80,而这 38 年正是库存用完的年份:区间政策的可信度取决于库存,而不只是政令。

数据口径:选项多一些会卖得更好还是更差?划掉一个原价能把顾客愿意支付的价格抬多高?三档定价中间那一档旁边放一个「不划算」的选项,会不会让中间档卖得更多?本书是教学与研究用的数学模型,不是营销、法律或合规建议。 书中三档价格示例 9.99、49.90、149.00 美元与 5D Library 自己的定价相同,已在利益披露中说明,书中分析不构成对该定价的验证。

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: 0.5, 16.9, 168, 1744.2, 22.8, 30

harvest persistence rho0.0
no granary : mean price54.0, sd 22.8, years < 30: 399, > 80: 490 of 4000
granary 30-80 : mean price52.0, sd 16.9, years < 30: 168, > 80: 0 of 4000
granary net cash +1744.2, years it ran out while price > 800
harvest persistence rho0.5
no granary : mean price55.2, sd 26.7, years < 30: 553, > 80: 611 of 4000
granary 30-80 : mean price52.4, sd 19.3, years < 30: 296, > 80: 38 of 4000
granary net cash +2408.7, years it ran out while price > 8038
'dear at the top turns cheap': mean change next year when this year is in the top
decile of prices (no granary)
rho 0.0-48.3 (top-decile price >= 83.9)
rho 0.5-33.2 (top-decile price >= 90.3)
rho 0.9-26.1 (top-decile price >= 155.4)
random walk in z (2,000 paths of 40 years): mean log change after p >120:
+0.004 (21,968 cases)
查看完整验算程序
# THE AUTHOR'S MODEL, an arithmetic reading of two classical passages (an analogy only).
# Guanzi "Guo xu": goods that are plentiful are cheap, scarce dear; dispersed light,
# concentrated heavy. Shiji "Huozhi liezhuan", Jiran: grain at 20 hurts farmers, at 90
# hurts traders; keep it between 30 and 80. A fictitious grain market: harvest
# H_t = exp(z_t), z_t = rho z_{t-1} + 0.2 e_t; inverse demand p = 50 * Q^(-2) (price
# elasticity 0.5). A granary buys at the floor 30 and sells at the ceiling 80 while its
# stock lasts; quantities that reach the market are Q = H - purchases + sales.
import math, random, statistics

def market(rho, years=4000, band=True, cap=1.5, seed=73):
    rng = random.Random(seed); z, stock, cash, prices, out = 0.0, 0.5, 0.0, [], 0
    for _ in range(years):
        z = rho * z + 0.2 * rng.gauss(0, 1); H = math.exp(z); Q = H
        p = 50 * Q ** -2
        if band and p < 30:                              # buy until p = 30
            buy = min(H - math.sqrt(50 / 30), cap - stock)
            stock += buy; Q = H - buy; cash -= buy * 50 * Q ** -2
        elif band and p > 80:                            # sell until p = 80
            sell = min(math.sqrt(50 / 80) - H, stock)
            stock -= sell; Q = H + sell; cash += sell * 50 * Q ** -2
            out += Q < math.sqrt(50 / 80) - 1e-12
        prices.append(50 * Q ** -2)
    return prices, cash, out

for rho in (0.0, 0.5):
    print(f"harvest persistence rho = {rho}")
    for lab, band in (("no granary", False), ("granary 30-80", True)):
        P, cash, out = market(rho, band=band)
        lo, hi = sum(p < 29.999 for p in P), sum(p > 80.001 for p in P)
        print(f"   {lab:14s}: mean price {statistics.fmean(P):5.1f},"
              f" sd {statistics.stdev(P):5.1f},"
              f" years < 30: {lo:4d}, > 80: {hi:4d} of {len(P)}")
        if band:
            print(f"   {'':14s}  granary net cash {cash:+.1f}, years it ran out while"
                  f" price > 80: {out}")
print("'dear at the top turns cheap': mean change next year when this year is in the top")
print("   decile of prices (no granary)")
for rho in (0.0, 0.5, 0.9):
    P = market(rho, band=False)[0]
    cut = sorted(P)[int(0.9 * len(P))]
    d = [P[t + 1] - P[t] for t in range(len(P) - 1) if P[t] >= cut]
    print(f"   rho {rho}: {statistics.fmean(d):+6.1f} (top-decile price >= {cut:.1f})")
rng = random.Random(1); dd = []
for _ in range(2000):                                    # a random walk: no reversion
    z = [0.0]
    for _ in range(40):
        z.append(z[-1] + 0.2 * rng.gauss(0, 1))
    p = [50 * math.exp(-2 * x) for x in z]
    dd += [math.log(p[t + 1] / p[t]) for t in range(40) if p[t] >= 120]
print("   random walk in z (2,000 paths of 40 years): mean log change after p >= 120:")
print(f"   {statistics.fmean(dd):+.3f} ({len(dd):,} cases)")

唯一数据源与引用

范雅福(Fan, Y.). (2026). 《选择架构、累积前景理论与销售界面价格锚定的计算行为学》第 7 章「原典并读:《管子》轻重、衡数与计然之策」(版本 1.0.0). 5D Library. https://doi.org/10.5281/zenodo.23836260

DOI: 10.5281/zenodo.23836260 · 阅读本章所在专著 · 5DL-B-047 · v1.0.0 · CC-BY-NC-ND-4.0

BibTeX
@book{fan2026047,
  author    = {Fan, Yafu},
  title     = {Choice Architecture, Cumulative Prospect Theory and Computational Behavioural Economics of Pricing Interfaces},
  year      = {2026},
  publisher = {5D Library},
  edition   = {1.0.0},
  doi       = {10.5281/zenodo.23836260},
  url       = {https://doi.org/10.5281/zenodo.23836260}
}

本书其他章节结论