5D Library › Slow-Crash Liquidity Reserves, Real-Asset Option Valuation and Intergenerational Wealth Preservation · 中文

Chapter 4. Putting a Cash Reserve to Work in Stages, and Where Distress Discounts Come From

Four rules for spending a reserve

With cash set aside for a slow crash, how should it be spent? p07 compares four rules on the paths of Chapter 1: everything at once when the index is 20% below its peak; a quarter each at falls of 20%, 30%, 40% and 50%; equal monthly amounts from month 6 to month 30; and a quarter each in months 24, 27, 30 and 33. Cash not used stays cash.

Calculation

# Chapter 4. p07: Spending a cash reserve in a slow crash. The reserve is
# put into the market under four rules, on the 2,000 paths of p02
# (seed 88): all at once when the index is 20% below its peak; four
# equal parts at 20, 30, 40 and 50% below the peak; equal monthly parts
# from month 6 to month 30; four equal parts in months 24, 27, 30 and
# 33. Cash not yet used stays cash (multiple 1). Reported: the value
# multiple of the whole reserve at month 60, and the share of paths
# where part of the reserve was never used.
import importlib.util
import random

def load(name):
    spec = importlib.util.spec_from_file_location(name, name + ".py")
    m = importlib.util.module_from_spec(spec)
    spec.loader.exec_module(m)
    return m

p02 = load("p02_slow_crash")

def by_drawdown(p, levels):
    value, used, peak, k = 0.0, 0, p[0], 0
    for t in range(61):
        peak = max(peak, p[t])
        while k < len(levels) and p[t] <= (1 - levels[k]) * peak:
            value += p[60] / p[t] / len(levels)
            used, k = used + 1, k + 1
    return value + (len(levels) - used) / len(levels), used < len(levels)

def by_month(p, months):
    return sum(p[60] / p[m] for m in months) / len(months), False

if __name__ == "__main__":
    rng = random.Random(88)
    paths = [p02.path(rng) for _ in range(2000)]
    rules = (("all at -20%", lambda p: by_drawdown(p, (0.2,))),
             ("parts at -20 to -50%",
              lambda p: by_drawdown(p, (0.2, 0.3, 0.4, 0.5))),
             ("monthly, months 6-30",
              lambda p: by_month(p, range(6, 31))),
             ("months 24, 27, 30, 33",
              lambda p: by_month(p, (24, 27, 30, 33))))
    q = lambda v, a: sorted(v)[int(a * (len(v) - 1))]
    print("   rule                    median   10th pct   part unused")
    for name, f in rules:
        res = [f(p) for p in paths]
        v = [r[0] for r in res]
        unused = sum(1 for r in res if r[1]) / 2000
        print("   {:<22}  {:6.2f}   {:8.2f}   {:11.3f}".format(
            name, q(v, 0.5), q(v, 0.1), unused))
#    rule                    median   10th pct   part unused
#    all at -20%               0.93       0.79         0.000
#    parts at -20 to -50%      1.16       1.00         0.152
#    monthly, months 6-30      1.07       0.93         0.000
#    months 24, 27, 30, 33     1.33       1.14         0.000

Spending everything at once does worst, with a median value multiple of only 0.93 at month 60: a 20% fall is still far from the bottom. Staging by the size of the fall gives 1.16, but 0.152 of paths never fall 50%, so the last part of the cash is never used. Equal monthly amounts give 1.07. Following a timetable with four parts in months 24 to 33 does best, a median of 1.33, and 1.14 even in the worst tenth of cases. In this model, watching the calendar is more useful than watching the fall, because forced sales and the low both cluster in the second half of the decline; but this depends on how long the model’s slow crash lasts, and Chapter 1 showed that a real decline may take only seventeen months.

Where distress discounts come from

Why can deep discounts be had in a crisis? Shleifer and Vishny pointed out that the best buyers of an asset are usually firms in the same industry, and in a crisis those firms are themselves short of cash; with fewer buyers able to bid, the seller must accept a lower offer from an outside buyer (Shleifer and Vishny 1992). p08 writes this as an auction.

Calculation

# Chapter 4. p08: Why fire-sale discounts are deepest when everyone is
# short of cash (Shleifer and Vishny). An asset is worth 1 to an owner
# who can run it. n potential buyers from the same industry each value
# it between 0.7 and 1.0 (uniform), but each is short of cash with
# chance c. The seller takes the second-highest bid among buyers with
# cash (an English auction); with fewer than two such buyers she gets
# only the outside offer of 0.5, from a buyer who cannot run the asset.
# 20,000 sales per case, seed 88.
import random

def price(rng, n, c):
    bids = sorted((0.7 + 0.3 * rng.random() for _ in range(n)
                   if rng.random() >= c), reverse=True)
    if len(bids) >= 2:
        return bids[1]
    return 0.5

if __name__ == "__main__":
    rng = random.Random(88)
    print("   buyers   discount, normal (c 0.2)   in a crisis (c 0.8)")
    for n in (2, 4, 6, 10):
        d = []
        for c in (0.2, 0.8):
            d.append(1 - sum(price(rng, n, c) for _ in range(20000)) / 20000)
        print("   {:6d}   {:24.3f}   {:19.3f}".format(n, *d))
#    buyers   discount, normal (c 0.2)   in a crisis (c 0.8)
#         2                      0.307                 0.487
#         4                      0.155                 0.443
#         6                      0.107                 0.391
#        10                      0.068                 0.290

In normal times (each potential buyer short of cash with chance 0.2), with four potential buyers the discount is only 0.155; in a crisis (chance 0.8), with the same four potential buyers the discount is 0.443. Even with ten potential buyers, the crisis discount is still 0.290. The “30% to 50% off” of Chapter 3 is, in this model, exactly the crisis case; it is not the buyer’s negotiating skill but the result of whether the buyer still has cash.

Note

Discipline in spending a reserve: write down the rule for spending in advance (by timetable or by size of fall), and require that whatever the market does, no more than half the reserve is spent before month 24; base offers for distressed assets on “how many buyers in the same industry still have cash”, not on how far prices have fallen.

Summary

Spending the whole reserve at a 20% fall does worst in a slow crash; spending in stages in the second half of the decline does best, but only if the decline really is long enough. The deep discounts of a crisis come from industry buyers being widely short of cash, and holding cash is itself bargaining power.

References

Shleifer, Andrei, and Robert W. Vishny. 1992. “Liquidation Values and Debt Capacity: A Market Equilibrium Approach.” The Journal of Finance 47 (4): 1343 to 1366.

Fan, Y. (2026). Slow-Crash Liquidity Reserves, Real-Asset Option Valuation and Intergenerational Wealth Preservation: Thirty-Month Drawdown Kinetics, Distressed Resource Acquisition and Ergodic Dynasty Trusts [慢崩盘流动性储备、实物资产期权估值与跨代财富保全动力学] (Version 1.0.0). 5D Library. https://doi.org/10.5281/zenodo.24090231 · Licensed under Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International (CC BY-NC-ND 4.0).