关于「原典并读:「九年之蓄」与常平」,《流动性体制、尾部相依与杠铃配置》给出的定量结论是什么?
结论:第七章与《礼记·王制》「三年耕,必有一年之食;九年耕,必有三年之食。以三十年之通,虽有凶旱水溢,民无菜色」、《墨子·七患》所引《周书》与《汉书·食货志》的常平仓并读,用程序显示:若粮食每年损耗超过 1/27(约 3.7%),「九年之蓄」永远达不到。
数据口径:危机来临时,资产会一起下跌吗?储备与安全垫要留多少才够?本书是教学与研究用的数学模型,不是投资建议。 作者要求证明「90% 短债 + 10% 期权」的杠铃配置随机占优;书中证明它不可能被股票占优,且在期权作废的概率高于股票跌破保底的概率时(本书模型即如此)也不可能一阶占优于股票,并在跳跃扩散模型的四种期权定价下显示它对股票、对 90/10 线性组合都不构成一阶或二阶占优,期权偏贵时反而被 90/10 线性组合二阶占优;CPPI 在乘数 6 时有 6.4% 的年份跌破保底线。
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: 27, 3.7
| years of food in store (deterministic surplus 1/3 a year) | |
| spoilage a year after 3 after 9 after 30 ceiling | |
| 0% | 1.00 3.00 10.00 none |
| 2% | 0.98 2.77 7.58 16.7 |
| 5% | 0.95 2.47 5.24 6.7 |
| 10% | 0.90 2.04 3.19 3.3 |
| the ceiling falls below 9 years when spoilage exceeds 1/27 | 3.7%, |
| and below 3 years when it exceeds 1/9 | 11.1% |
| random harvests, 30 years: years with less than full rations | |
| policy spoilage P(any) mean years stock at year 30 | |
| eat the harvest | 0% 98.3% 3.88 0.00 |
| store | 0% 23.9% 0.31 7.71 |
| store | 5% 25.7% 0.34 4.01 |
| store | 10% 27.9% 0.39 2.43 |
| storage with 5 % spoilage: in 87% of the spans with a short year, | |
| the first one comes in years 1-3, before the stock has been built |
查看完整验算程序
# The arithmetic of the "Royal Regulations" (Wang zhi): "in three years of tilling there
# must be one year's food; in nine years, three years' food". In units of one year's
# consumption c = 1, that needs an average harvest of 4/3 (a surplus of 1/3 a year).
# (1) Deterministic stock with spoilage d: S(n+1) = (1 - d) S(n) + 1/3, ceiling (1/3)/d.
# (2) ILLUSTRATIVE random harvests: lognormal (sd 0.15) around 4/3 with a 10 % chance of a
# disaster year at 40 % of normal; consumption is cut only when harvest + stock < 1.
# 20,000 simulated 30-year spans per policy.
import math, random
def stock_after(n, d, s=0.0):
for _ in range(n):
s = (1 - d) * s + 1 / 3
return s
def span(rng, d, policy, years=30):
s, short, firsts = 0.0, 0, None
for y in range(years):
h = 4 / 3 * math.exp(rng.gauss(-0.01125, 0.15))
h *= 0.4 if rng.random() < 0.1 else 1
if policy == "eat the harvest":
avail, s = min(h, 4 / 3), 0.0 # no storage: surplus is spent
else:
avail = h + (1 - d) * s
eat = min(1.0, avail)
if eat < 0.999:
short += 1; firsts = y + 1 if firsts is None else firsts
if policy != "eat the harvest":
s = avail - eat
return short, firsts, s
if __name__ == "__main__":
print("years of food in store (deterministic surplus 1/3 a year)")
print(" spoilage a year after 3 after 9 after 30 ceiling")
for d in (0.0, 0.02, 0.05, 0.10):
cap = "none" if d == 0 else f"{1 / 3 / d:.1f}"
print(f" {d:6.0%} {stock_after(3, d):5.2f}"
f" {stock_after(9, d):5.2f}"
f" {stock_after(30, d):5.2f} {cap}")
print(f" the ceiling falls below 9 years when spoilage exceeds 1/27 ="
f" {1 / 27:.1%},")
print(f" and below 3 years when it exceeds 1/9 = {1 / 9:.1%}")
rng = random.Random(54)
print("random harvests, 30 years: years with less than full rations")
print(" policy spoilage P(any) mean years stock at year 30")
for policy, d in (("eat the harvest", 0.0), ("store", 0.0), ("store", 0.05),
("store", 0.10)):
runs = [span(rng, d, policy) for _ in range(20_000)]
anyp = sum(r[0] > 0 for r in runs) / len(runs)
mean_short = sum(r[0] for r in runs) / len(runs)
print(f" {policy:18s} {d:5.0%} {anyp:6.1%} {mean_short:5.2f}"
f" {sum(r[2] for r in runs) / len(runs):5.2f}")
early = [span(rng, 0.05, "store")[1] for _ in range(20_000)]
hit = [y for y in early if y]
share = sum(y <= 3 for y in hit) / len(hit)
print(f" storage with 5 % spoilage: in {share:.0%} of the spans with a short year,")
print(" the first one comes in years 1-3, before the stock has been built")唯一数据源与引用
范雅福(Fan, Y.). (2026). 《宏观流动性体制转换、极值尾部相依与杠铃资产配置》第 7 章「原典并读:「九年之蓄」与常平」(版本 1.0.0). 5D Library. https://doi.org/10.5281/zenodo.23828049
DOI: 10.5281/zenodo.23828049 · 阅读本章所在专著 · 5DL-B-040 · v1.0.0 · CC-BY-NC-ND-4.0
BibTeX
@book{fan2026040,
author = {Fan, Yafu},
title = {Macro-Liquidity Regime Switching, Extreme Tail Dependence and Barbell Asset Allocation},
year = {2026},
publisher = {5D Library},
edition = {1.0.0},
doi = {10.5281/zenodo.23828049},
url = {https://doi.org/10.5281/zenodo.23828049}
}