5D Library › 体制转换分数凯利、最大回撤约束与生命周期杠杆配置 · English

第六章 生命周期资产负债表:人力资本、早期杠杆与下滑路径

人力资本是一项资产

年轻的上班族金融财富很少,却拥有一项不出现在任何账单上的大资产:未来收入的现值。默顿的组合规则适用于总财富,而不只是金融部分 (Merton 1969);如果收入大体像债券,那么总财富所需的股票就必须全部放在那个很小的金融账户里 (Bodie et al. 1992)。这就是「生命周期投资」主张早年使用最高二比一杠杆的理由 (Ayres and Nalebuff 2010)。p11 为一位示意的上班族按凯利的 0.3 倍建立资产负债表。

推导

# Chapter 6. p11: A lifecycle glide path from the human-capital balance
# sheet. Real income 50 (thousand) at age 25, growing 1.5% a year to 55,
# then flat; retirement at 65; 15% of income saved; financial wealth F
# starts at 10. Human capital H = present value of remaining income at
# 0.03 a year. Total equity wanted = k * Kelly * (F + H) minus the part
# of H that already behaves like equity (0.2 of H), with k = 0.3 and
# Kelly 1.953 (p01). Leverage on F is that amount / F, kept between 0
# and 2. F grows at the expected return: 0.02 plus 0.05 per unit of
# equity, less 0.02 on borrowed money.
K, KELLY, BETA_H, CAP = 0.3, 1.953, 0.2, 2.0

def income(age):
    return 50 * 1.015 ** (min(age, 55) - 25) if age < 65 else 0.0

def human(age):
    return sum(income(a) / 1.03 ** (a - age) for a in range(age, 65))

if __name__ == "__main__":
    f, first_below = 10.0, None
    print("   age  income  human cap.  financial  wanted lev  used lev")
    for age in range(25, 66):
        h = human(age)
        want = (K * KELLY * (f + h) - BETA_H * h) / f
        lev = min(max(want, 0.0), CAP)
        if lev < 1 and first_below is None:
            first_below = age
        if age % 5 == 0:
            print("   {:3d}  {:6.1f}  {:10,.0f}  {:9,.0f}  {:10.2f}  {:8.2f}"
                  .format(age, income(age), h, f, want, lev))
        ret = 0.02 + lev * 0.05 - max(lev - 1, 0.0) * 0.02
        f = f * (1 + ret) + 0.15 * income(age)
    print("   first age with leverage below 1: {}".format(first_below))
#    age  income  human cap.  financial  wanted lev  used lev
#     25    50.0       1,505         10       58.67      2.00
#     30    53.9       1,464         63        9.53      2.00
#     35    58.0       1,393        152        4.11      2.00
#     40    62.5       1,289        300        2.24      2.00
#     45    67.3       1,142        528        1.42      1.42
#     50    72.5         945        827        1.03      1.03
#     55    78.2         687      1,204        0.81      0.81
#     60    78.2         369      1,657        0.67      0.67
#     65     0.0           0      2,197        0.59      0.59
#    first age with leverage below 1: 51

25 岁时人力资本是 1,505(千),金融财富是 10,所需杠杆是 58.67;以 2 为上限时,这位上班族直到约 40 岁都停在上限。所需杠杆在 45 岁降到 1.42,50 岁降到 1.03,首次低于 1 的年龄是 51 岁;到 65 岁,已经纯粹是金融资产的资产负债表上,股票占比是 0.59。下滑路径不是经验法则,而是人力资本的缩小。

p11 的资产负债表上按年龄分列的所需杠杆与实际杠杆;阴影带是二比一的上限。

当市场与工作一起出问题

资产负债表的论证假定收入不会随市场下跌。p12 放宽这一点:熊市月份里,当月储蓄有 0.3 的机会落空;市场是第一章的两体制市场,借款利率每年 0.04。

推导

# Chapter 6. p12: Four lifecycle rules from age 25 to 65 in the two-
# regime market of p02 (monthly, started from its long-run mix). Saving
# is 15% of the income of p11, paid monthly; in a bear month the saving
# is lost with chance 0.3 (layoffs come with bad markets). Cash earns
# 0.02 a year, borrowing costs 0.04. Rules: all equity; target date
# (0.9 equity at 25 falling evenly to 0.4 at 65); the p11 balance-sheet
# rule capped at 1 (no borrowing); the same capped at 2. 4,000 lives,
# seed 98.
import random
from math import sqrt

R12 = sqrt(12)
REG = ((0.12 / 12, 0.13 / R12), (-0.25 / 12, 0.28 / R12))
STAY, RF, RB = (0.98, 0.92), 0.02 / 12, 0.04 / 12

def income(age):
    return 50 * 1.015 ** (min(age, 55) - 25) if age < 65 else 0.0

HUMAN = [sum(income(a) / 1.03 ** (a - age) for a in range(age, 65))
         for age in range(66)]

def lev(rule, age, f):
    if rule == "all equity":
        return 1.0
    if rule == "target date":
        return 0.9 - 0.5 * (age - 25) / 40
    h = HUMAN[age]
    want = (0.3 * 1.953 * (f + h) - 0.2 * h) / max(f, 1e-9)
    return min(max(want, 0.0), 1.0 if rule == "balance cap 1" else 2.0)

RULES = ("all equity", "target date", "balance cap 1", "balance cap 2")

if __name__ == "__main__":
    rng = random.Random(98)
    end = {r: [] for r in RULES}
    early = {r: 0 for r in RULES}
    for _ in range(4000):
        s = 0 if rng.random() < 0.8 else 1
        st = {r: [10.0, 10.0, 0.0] for r in RULES}
        for m in range(480):
            age = 25 + m // 12
            if rng.random() > STAY[s]:
                s = 1 - s
            x = rng.gauss(*REG[s])
            save = 0.15 * income(age) / 12
            if s == 1 and rng.random() < 0.3:
                save = 0.0
            for r in RULES:
                v = st[r]
                L = lev(r, age, v[0])
                ret = RF + L * x - max(L - 1, 0.0) * (RB - RF)
                v[0] = max(v[0] * (1 + ret), 0.0) + save
                v[1] = max(v[1], v[0])
                if age < 40:
                    v[2] = max(v[2], 1 - v[0] / v[1])
        for r in RULES:
            end[r].append(st[r][0])
            early[r] += st[r][2] >= 0.5
    print("   rule            median at 65  10th pct  90th pct  "
          "DD>=50% by 40")
    for r in RULES:
        v = sorted(end[r])
        print("   {:<14}  {:12,.0f}  {:8,.0f}  {:8,.0f}  {:13.3f}".format(
            r, v[2000], v[400], v[3600], early[r] / 4000))
#    rule            median at 65  10th pct  90th pct  DD>=50% by 40
#    all equity             1,477       351     5,871          0.285
#    target date            1,063       479     2,338          0.118
#    balance cap 1          1,421       402     4,330          0.285
#    balance cap 2          1,725       308     6,568          0.783

上限为 2 的资产负债表规则在 65 岁的财富中位数最高(1,725,全部持股是 1,477,目标日期路径是 1,063),但第十百分位数最低(308,对 351 与 479),并且 0.783 的人生在 40 岁前出现腰斩以上的回撤,全部持股是 0.285,目标日期是 0.118。上限为 1 时,同一规则的中位数是 1,421,第十百分位数 402。早期杠杆抬高中位数的方式,是让相当一部分人把糟糕的第一个十年变成失去的十年;值不值得,取决于投资者能否在 35 岁、工作不稳的时候熬过一次腰斩。

假说 · 尚未验证

作者的诠释(尚未验证): 早年使用二比一杠杆的投资者,在第一次腰斩以上的回撤之后会减少或放弃杠杆,因此实际得到的好处小于假定规则被一直遵守的模拟结果。什么会证明它错: 在持有杠杆股票仓位的年轻投资者账户数据中,经历腰斩以上回撤十二个月后仍保持杠杆的比例,与没有经历这种回撤的可比账户一样高。

小结

在计入人力资本的资产负债表上,年轻上班族的金融账户应持有比其规模所显示的更多的股票,而退休前的下滑路径来自人力资本的缩小。当裁员与熊市同时到来,二比一的早期杠杆抬高退休财富的中位数,却使 40 岁前的腰斩成为常见情形。

参考文献

Ayres, Ian, and Barry Nalebuff. 2010. Lifecycle Investing: A New, Safe, and Audacious Way to Improve the Performance of Your Retirement Portfolio. Basic Books.
Bodie, Zvi, Robert C. Merton, and William F. Samuelson. 1992. “Labor Supply Flexibility and Portfolio Choice in a Life Cycle Model.” Journal of Economic Dynamics and Control 16: 427 to 449.
Merton, Robert C. 1969. “Lifetime Portfolio Selection under Uncertainty: The Continuous-Time Case.” The Review of Economics and Statistics 51 (3): 247 to 257.

Fan, Y. (2026). Regime-Switching Fractional Kelly, Maximum-Drawdown Constraints and Lifecycle Leverage Allocation: Cyclical Drift Estimation Errors, Ruin-Bounded Bet Sizing and Human-Capital Amortization [体制转换分数凯利、最大回撤约束与生命周期杠杆配置] (Version 1.0.0). 5D Library. https://doi.org/10.5281/zenodo.24090256 · Licensed under Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International (CC BY-NC-ND 4.0).