关于「规则与相机抉择:最优计划为什么会自我推翻」,《时间不一致与反掀桌质押》给出的定量结论是什么?

结论:第一章用基德兰德与普雷斯科特的玩具模型说明,背约那一期最诱人,而可信度一旦丧失,每期都要多付代价;冷酷触发下贴现因子高于 0.333 才能靠声誉守住规则。

数据口径:书中参数为示意;全部计算由书中程序生成;本书不是法律意见,也不是个人化财务或经营建议,所有例子都是合成情境,价格一律为无币种示意单位。 书中以标注框区分已发表的证据、可运行的 Python 计算(22 段程序)、原典与作者的假说。

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: 0.333

k lam pi_disc L_disc L_rule L_cheat gap tempt
10.52.0001.5000.5000.167 1.0000.333
11.01.0001.0000.5000.250 0.5000.250
12.00.5000.7500.5000.333 0.2500.167
14.00.2500.6250.5000.400 0.1250.100
20.54.0006.0002.0000.667 4.0001.333
21.02.0004.0002.0001.000 2.0001.000
22.01.0003.0002.0001.333 1.0000.667
24.00.5002.5002.0001.600 0.5000.400
check k=1 lam=1: gap = k^2/(2 lam) = 1/2 ; temptation = k^2/(2(1+lam)) = 1/4
ordering L_cheat < L_rule < L_disc holds: True
查看完整验算程序
# Chapter 1. p11: rules versus discretion in a Kydland-Prescott toy model
# Loss L = lam * pi^2 / 2 + (y - k)^2 / 2 with y = pi - pi_e.
# k is the planner's wish to push output above its natural level,
# lam the weight on inflation. Discretion: the planner optimises after
# expectations are set; rational expectations then force pi = k / lam.
# Rule: pi = 0 is announced and kept. Cheat: the rule is announced,
# believed (pi_e = 0) and then broken once. All values illustrative.
from fractions import Fraction as F


def discretion(k, lam):
    pi = k / lam
    return pi, lam * pi * pi / 2 + k * k / 2


def rule(k, lam):
    return F(0), k * k / 2


def cheat(k, lam):
    pi = k / (1 + lam)
    y = pi
    return pi, lam * pi * pi / 2 + (y - k) ** 2 / 2


print("k    lam   pi_disc  L_disc  L_rule  L_cheat  gap   tempt")
for k in (F(1), F(2)):
    for lam in (F(1, 2), F(1), F(2), F(4)):
        pd, ld = discretion(k, lam)
        _, lr = rule(k, lam)
        _, lc = cheat(k, lam)
        fmt = "{:<4} {:<5} {:>7.3f} {:>7.3f} {:>7.3f} {:>8.3f} {:>5.3f}"
        print((fmt + " {:>6.3f}").format(
            str(k), float(lam), float(pd), float(ld), float(lr),
            float(lc), float(ld - lr), float(lr - lc)))
k, lam = F(1), F(1)
pd, ld = discretion(k, lam)
_, lr = rule(k, lam)
_, lc = cheat(k, lam)
print("check k=1 lam=1: gap = k^2/(2 lam) =", ld - lr,
      "; temptation = k^2/(2(1+lam)) =", lr - lc)
print("ordering L_cheat < L_rule < L_disc holds:", lc < lr < ld)

唯一数据源与引用

范雅福(Fan, Y.). (2026). 《动态时间不一致博弈、收益私有化亏损转嫁与反掀桌质押契约母典》第 1 章「规则与相机抉择:最优计划为什么会自我推翻」(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-422 · v1.0.0 · CC-BY-NC-ND-4.0

BibTeX
@book{fan2026422,
  author    = {Fan, Yafu},
  title     = {Dynamic Time-Inconsistency, Asymmetric Freeroll Reneging and Collateralised Anti-Hold-Up Contracts},
  year      = {2026},
  publisher = {5D Library},
  edition   = {1.0.0},
  note      = {DOI pending}
}

本书其他章节结论