5D Library › Shareholder Free-Riding Games, Dynamic Vesting Mechanisms and Defection Early-Warning Models · 中文
Chapter 2. Vesting: What It Repairs and What It Does Not
Dead equity
A common story in start-ups is a co-founder who leaves after a few months and keeps half the company. Wasserman’s study of founders documents how often early splits are agreed quickly and how often they later cause trouble (Wasserman 2012). Vesting with a cliff is the standard answer: shares are earned month by month over four years, and nothing is earned before the end of the first year. p03 compares it with static shares when a co-founder has a one-in-four chance of leaving within two years.
Calculation
# Chapter 2. p03: Dead equity. Two founders split 50 / 50. Founder B
# leaves in month m with chance 0.25 spread evenly over the first 24
# months (else B stays four years). With static shares B keeps 50%
# whatever m is. With four-year vesting and a one-year cliff, B keeps
# nothing before month 12 and m / 48 of the 50% afterwards (the firm
# buys back the rest at cost). A replacement needs 15% of the firm
# for the remaining time. 100,000 draws, seed 93.
import random
def one(rng, vesting):
if rng.random() >= 0.25:
return None
m = rng.uniform(0, 24)
if not vesting:
return 0.50
return 0.0 if m < 12 else 0.50 * m / 48
if __name__ == "__main__":
rng = random.Random(93)
for vesting in (False, True):
out = [one(rng, vesting) for _ in range(100000)]
left = [x for x in out if x is not None]
dead = sum(left) / len(left)
lab = "4-year vesting, 1-year cliff" if vesting else "static 50 / 50"
print(" {}".format(lab))
print(" share of founders who leave early: {:.3f}".format(
len(left) / len(out)))
print(" equity held by a departed founder (mean): {:.3f}".format(
dead))
print(" active owners' share after a 15% hire: {:.3f}".format(
(1 - dead) * 0.85))
print(" leaver in month 11: keeps {:.3f}; month 13: {:.3f}".format(
0.0, 0.50 * 13 / 48))
# static 50 / 50
# share of founders who leave early: 0.251
# equity held by a departed founder (mean): 0.500
# active owners' share after a 15% hire: 0.425
# 4-year vesting, 1-year cliff
# share of founders who leave early: 0.248
# equity held by a departed founder (mean): 0.094
# active owners' share after a 15% hire: 0.770
# leaver in month 11: keeps 0.000; month 13: 0.135With static shares a departed founder holds 0.500 of the firm whatever the month of departure. If a replacement then needs 15% of the firm, the active owners are left with 0.425 of it. With vesting the departed founder holds 0.094 on average, and the active owners keep 0.770 after the same hire. The cliff makes a sharp difference around the end of the first year: a founder who leaves in month 11 keeps 0.000, one who leaves in month 13 keeps 0.135 of the firm.
Retention is not effort
Does vesting also restore effort? p04 separates two things: whether a founder stays, and how hard the founder works while staying. Founder B’s outside offer rises over four years; B leaves when the offer exceeds the value of staying, which under vesting includes the unvested shares B would forfeit.
Calculation
# Chapter 2. p04: What vesting fixes and what it does not. Founder B's
# outside offer grows from 0.5 to 2.0 units a month over 48 months
# (B's own salary is 0.5). B leaves when the offer exceeds the value of
# staying: salary plus the shares B would forfeit by leaving, spread
# over the months left. The firm gains 1 unit of value for each month
# both founders stay. Under static shares B forfeits nothing; under
# vesting B forfeits the unvested part of a stake worth 30 units at
# exit. While B stays, effort is still that of a 50% owner (p01
# efficiency 0.750), so vesting can repair retention but not effort.
def months_stayed(vesting, stake=30.0):
for t in range(48):
offer = 0.5 + 1.5 * t / 47
if vesting:
vested = 0.0 if t < 12 else t / 48
forfeit = stake * (1 - vested)
else:
forfeit = 0.0
stay = 0.5 + forfeit / (48 - t)
if offer > stay:
return t
return 48
if __name__ == "__main__":
print(" rule months B stays joint months / 48")
for lab, v in (("static 50 / 50", False), ("vesting", True)):
m = months_stayed(v)
print(" {:<20} {:14d} {:17.3f}".format(lab, m, m / 48))
for stake in (10.0, 60.0):
m = months_stayed(True, stake)
print(" vesting, stake {:4.0f} {:14d} {:17.3f}".format(
stake, m, m / 48))
s, v = months_stayed(False) / 48, months_stayed(True) / 48
print(" value created, retention x effort: static {:.3f},"
" vesting {:.3f}".format(s * 0.75, v * 0.75))
# rule months B stays joint months / 48
# static 50 / 50 1 0.021
# vesting 20 0.417
# vesting, stake 10 8 0.167
# vesting, stake 60 40 0.833
# value created, retention x effort: static 0.016, vesting 0.312Under static shares B forfeits nothing by leaving and goes in month 1, as soon as the outside offer beats the salary; the two founders work together for 0.021 of the planned time. Under vesting B stays 20 months (0.417 of the time), and the effect depends on what the shares are worth: 8 months if the stake is worth 10, 40 months if it is worth 60. But while B stays, B still owns half of what B produces, so effort is still that of Chapter 1 (efficiency 0.750). Combining the two, the value created rises from 0.016 to 0.312 of the full value. Vesting repairs much of the retention problem and none of the effort problem; the latter needs the tools of Chapters 1 and 4.
Hypothesis · not established
The author’s interpretation (not yet tested): among start-ups that adopt founder vesting, early founder departures become less frequent, but measures of founder effort while in post do not change. What would show it wrong: in a pre-registered comparison of start-ups before and after adopting vesting, departure rates do not fall, or founder effort rises as much as retention does.
Summary
Vesting removes most dead equity and keeps founders longer, and the cliff decides a great deal around month 12. It does not change how hard a founder works while staying; that remains a team problem.
References
Fan, Y. (2026). Shareholder Free-Riding Games, Dynamic Vesting Mechanisms and Defection Early-Warning Models: Moral-Hazard Equity Dilution, Shotgun Buy-Sell Clauses and Behavioral Divergence Signals [股东搭便车博弈、动态股权成熟与背刺早期预警模型] (Version 1.0.0). 5D Library. https://doi.org/10.5281/zenodo.24090243 · Licensed under Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International (CC BY-NC-ND 4.0).