Nobody Exercises Options in a Market With No Exits

An employee stock option is a right to buy shares at a fixed price. It turns into money only when there is a way to sell those shares. In a market that produces roughly two dozen exits in a year, against a funded base counted in the hundreds at the top and the thousands below it, most grants never reach a sale. So the honest answer to the question every senior hire is quietly asking, is this option real pay, is usually no. Not unless the founder has built a mechanism that pays without waiting for an exit that will not arrive.

This is a decision for two people at once. The founder is deciding what to put in an offer letter and whether the equity line is compensation or decoration. The hire is deciding whether to accept a number they cannot bank, or to price it at close to zero and negotiate the cash instead. Both are working off the same broken assumption: that an option is a smaller, later version of salary. It is not. You are not paying them in equity. You are paying them in a maybe.

The exit-scarcity base rate underneath this is by now well established, and I am going to take it as given rather than re-derive it. Africa produces on the order of twenty to twenty-six startup exits a year, overwhelmingly trade sales, with public listings close to absent and most acquisition values never disclosed. Even in a down market the forecast from inside the ecosystem is more of the same: “we will continue to see few exits.”⁷ Hold that number still. The work here is what it does to a pay packet.

Why three lenses, and why not the obvious one

The reflex is to reach straight for option pricing: an option is worth more when the underlying is volatile and the horizon is long. True, and already covered elsewhere in this series. Reached for alone it also flatters the grant, because volatility makes even a long shot look valuable on paper. That is the wrong comfort for a founder writing an offer and a worse one for the person accepting it.

So I am routing this to three models that fail in different directions. The first is a communication game: cheap talk. It asks whether the grant carries any information at all, or whether it is a promise that costs the founder nothing to make. The second is the luck-skill continuum. It asks how much of the payoff the employee can influence, and answers that the deciding event is mostly luck they do not control. The third is base rates. It supplies the reference class that turns the first two from theory into a number. One equilibrium lens, one random-outcome lens, one cycle-regime lens. Where three different kinds of error agree, you can act.

The behavioural story, that people overweight a vivid jackpot, is real and it is doing quiet work under all three. I have folded it into the levers rather than shipping it as a fourth card. It changes what the employee hopes. It does not change what the grant is worth.

The framework: promise, luck, and the base rate under both

1. The signal: a grant that costs nothing to make says nothing

Treat the offer as a message. The founder says, in effect, this equity will be worth real money one day. In game theory a message only carries information when it is costly to send, or when the sender is bound to it. A message that costs nothing and binds no one is cheap talk, and a rational listener discounts it to zero.

Now look at what the standard option grant actually costs the founder to promise. Almost nothing. The dilution only bites in the good state that rarely comes. The strike price is set at fair value, so the option is worth nothing the day it is granted and the employee still has to find cash to buy in later.⁴ The 90-day post-termination window means that if the employee leaves before an exit, the option is usually forfeited outright.¹ A promise this easy to make and this easy to claw back is the textbook definition of cheap talk. Which is why the sharpest hires treat the equity line as noise and negotiate on cash.

The repair is to make the grant expensive to the founder and hard to revoke. An extended exercise window, a cashless-exercise path, a contractual right to sell in the next round: each of these is a cost the founder now carries in more states than just the jackpot. That cost is exactly what makes the promise believable.

Assumes the employee is rational and discounts any claim the founder can make for free and unwind at will.

Fits because a strike-priced, 90-day-window option costs the founder almost nothing to grant and is cheap to reclaim.

Breaks when the founder is bound by binding commitments the employee can see and enforce, at which point the message stops being cheap.

Counteracts the belief that stating a large future value is the same as paying it.

May reinforce the base-rate lens, since a costless promise and a rare payoff point at the same discount.

2. The luck: the payoff turns on an event the holder cannot move

Grant the option is real and vested. Its value still hangs on one thing: a liquidity event. An acquisition, a rare listing, a secondary sale inside a later round. Ask how much of that event the employee controls, and the answer is very little.

Mauboussin’s luck-skill continuum is the tool here. Some outcomes reward practice and effort. Others are dominated by luck, and for those “we must think of skill in terms of a process,” because the result itself is not yours to command.⁶ Whether a bank or a telco decides to buy, whether the IPO window is open, whether a later investor funds a secondary, sits far out on the luck end. The engineer who ships flawlessly for four years does not thereby summon an acquirer. Their diligence changes the product. It barely touches the exit.

So an option ties a large slice of someone’s pay to a lottery whose draw they cannot influence, in a market where the draw almost never happens. That is a poor instrument for rewarding contribution, because it severs reward from the thing the employee actually does. Cash does not have this defect. Nor does a phantom or share-appreciation scheme that pays out on a revenue trigger the team can move.

Assumes the liquidity event is dominated by luck and market timing, not by any individual employee’s effort.

Fits because acquirer appetite, listing windows and secondary demand are exogenous to a hire’s daily work.

Breaks when the employee is senior enough that their work genuinely shifts the odds of a sale, which is a small number of people.

Counteracts the pitch that working harder will make the equity pay.

May reinforce the case for tying reward to an operating metric the team controls instead.

3. The base rate: price the grant at the outcome that usually happens

The first two lenses need a number, and the reference class supplies it. The right comparison is not “startups” in the abstract. It is companies whose realistic end state is a modest trade sale or no sale at all, whose shares have no public market, and whose employees can only cash out at a liquidity event that most firms never reach.⁴

For that class the base rate is brutal. Two dozen exits a year across the continent, trade sales dominant, listings rare, values mostly undisclosed, which is itself a tell that they are small.⁷ Set the expected value of a random grant against that distribution and it rounds toward zero, not because any single company is doomed but because the pooled odds of reaching a sale are low and the odds of a large one are lower still. The vivid exceptions, the ones everyone can name, are exceptions. The base of the distribution is a company that is acquired quietly for a figure nobody prints, or that simply keeps operating, privately, indefinitely.

Assumes your company’s exit odds resemble the market’s, absent hard private evidence that they do not.

Fits because exit counts, the trade-sale share and the wall of undisclosed values are all measured and stable.

Breaks when the founder holds real, specific evidence of a heavy-tailed outcome, in which case the grant is worth more than the base rate says.

Counteracts pricing the option off the jackpot rather than off the common case.

May reinforce the cheap-talk read, since a rare payoff makes a free promise even easier to discount.

The combined read

The three converge on one sentence. The standard grant is a promise that costs the founder nothing, paying off on an event the employee cannot influence, in a market where that event almost never occurs. Fix any one of those and the equity improves. Fix all three and it becomes real compensation. Leave all three and you are diluting your cap table to hand people a certificate they cannot spend.

Fixing the grant, from the free redraft to the funded buy-back

Ordered by what it costs the founder, least first. Each is reversible until the one below it.

• Say the real expected value out loud. Free, and it is the single most credible thing a founder can do. Tell the hire what the option is worth at a realistic small exit and at zero, not only at the dream. Honesty here is itself a costly signal, because most founders will not do it.
• Extend the exercise window. Move the post-termination window from 90 days to five, seven or ten years, so a departing employee does not forfeit vested equity they cannot yet afford to buy. Coinbase went from 90 days to seven years for anyone who stays two years, and a long public list of companies has followed.^{2, 3} This costs the founder future dilution, which is the point: it is no longer cheap talk.
• Build a cashless-exercise path. Let the employee exercise without finding the strike and the tax bill in cash, because in this market the combined cost “can run into six figures” and most people simply walk away from the options rather than write that cheque.¹
• Write a secondary right into the plan. Contract that employees may sell a slice of vested shares in the next priced round. Secondaries are already how African liquidity mostly happens, so this is the realistic exit, formalised.⁹
• Or convert the budget to cash, or to a phantom scheme. If you cannot fund any of the above, do not spend dilution on a grant nobody can bank. Pay the cash, or run a share-appreciation plan that pays out on a revenue trigger the team controls.^{5, 8}

What to settle before the offer, on both sides of the table

Two portfolios, sequenced by reversibility. Do the cheap dominant moves now, buy the cheap insurance, and pre-commit the irreversible calls to a trigger you name in advance.

If you are the founder writing the offer.

• Do now (T+3 to T+14). Model the option’s value at a realistic small exit and at zero, and put both numbers in the offer conversation. Then price the role on cash as if the equity were worth little, and let the equity be the upside it actually is. Reversible, and it costs you nothing but candour.
• Hedge (by T+14). Extend the exercise window before your next hire, not after your first good leaver forfeits a stake and tells everyone. A window change applied plan-wide is cheap insurance against the retention damage of the 90-day cliff.¹
• Defer and trigger (T+28 and beyond). Commit now: the first time you raise a priced round, you carve out a secondary allowance for employees who have vested. Write the trigger down while you can still see why it matters, because in the round itself the pressure will be to give that room to investors instead.⁹

If you are the one being offered options.

• Do now (T+3 to T+14). Price the grant at the base rate, which is near zero, and negotiate the cash on that basis. Ask three questions before you sign: what is the strike, what is the post-termination window, and is there any secondary or cashless path. The answers tell you whether this is pay or a poster.
• Hedge (by T+14). If the equity is a real part of why you are joining, ask for the exercise window in writing and check your own tax exposure. In Kenya the proposed rule would tax option gains within 30 days of the shares vesting, “regardless of whether those shares can be sold.” That is tax on a promise, not on profit, and it can land before any liquidity exists.⁸
• Defer and trigger (T+28 and beyond). Do not count the equity as wealth. Pre-commit the trigger for treating it as real: the day a secondary or an exit is actually on the table, you re-run the math and act then. Until that day, plan your life on the cash.

What usually happens to an unfixed pool

Build the comparison from companies shaped like yours, not flagged like yours. The class that matters is private firms with no share market, a 90-day forfeiture window, a strike the employee must fund, and a realistic outcome of a quiet trade sale or none. For that class the pattern is consistent and unkind. The pool is granted with real intent. It vests. People leave, as people do, and the 90-day window quietly voids most of what they earned because they cannot afford to exercise.¹ The company keeps operating, privately, for years past any fund’s patience. The equity line that was supposed to retain people retains no one, because everyone senior enough to do the math has already discounted it to zero and priced their loyalty in cash.

Adjust for your present state. If your market genuinely produces frequent large exits, the tail fattens and the grant is worth more; watch the exit count and the share of disclosed, sizeable deals for that shift. Until both climb, assume the quiet outcome. And if your ownership rules or tax regime tax the option before any sale, as Kenya has proposed, the grant is worse than neutral for the employee: it is a bill with no matching cash.⁸

Now take out what would have happened anyway. Some of these pools would have failed to motivate regardless, because the company was never going to reach an exit and no structure changes that. Count only the loss the unfixed grant itself caused: the dilution spent, the hires who left over cash you could have paid, and the trust burned when a vested stake evaporated on a 90-day clock. That residue is the real cost of treating an option as pay when it is not.

Flag that rewrites this. If a real secondary market forms, where employees can reliably sell vested shares between rounds at a fair price, the whole argument softens. The option stops being a bet on a rare exit and becomes a semi-liquid asset. African VCs are already turning to secondaries for their own liquidity, which is the first sign of that market forming.⁹ If it matures, revisit every grant in this piece.

Where these three lenses go dark

The ensemble prices the instrument. It cannot price the meaning. For some people, holding equity changes how they work: they behave like owners, and that shift can be worth more than the expected cash value of the option, even when the option itself is unlikely to pay. The models see a discounted lottery ticket. They do not see the psychology of ownership, and that psychology is sometimes the entire point of the grant.

They are also blind to the founder who is genuinely building the rare exit. If you hold specific, private evidence that your outcome is heavy-tailed, the base rate is the wrong reference class for you, and honest equity can be the best pay you offer. The lenses cannot tell that founder from the many who only believe they are that founder, because the belief looks identical from outside.

So here is the decision that survives the doubt. Before you put an equity number in front of anyone, or accept one, write down what the option pays at a realistic small exit and at zero. If the honest figure is close to nothing, then either fix the mechanism, extend the window, fund the exercise, guarantee a secondary, or stop calling it compensation and pay the cash. Do not spend your dilution, or someone else’s years, on a currency neither of you can spend.

Sources and notes

1. Farrell Fritz (via Mondaq), “Designing A Smarter Post-Termination Option Exercise Window: Lessons From Coinbase, Pinterest And Quora.” Verified: body states departing employees “typically have just 90 days to exercise their vested stock options or lose them forever,” that they forfeit equity “if they cannot afford the exercise price and tax bill,” and that the combined cost “can run into six figures. Many simply cannot write that check, and so they walk away from options they genuinely earned.” mondaq.com
2. Coinbase, “Improving Equity Compensation at Coinbase.” Verified: body states Coinbase “extended the post-termination option exercise window from the standard 90 days to seven years for new employees who join and stay at least two years.” coinbase.com
3. Zach Holman, “extended-exercise-windows” (GitHub), a maintained list of startups offering exercise windows past the 90-day norm. Verified: body lists Coinbase and Pinterest at 7 years and Amplitude, Asana, Segment, Supply and others at 10 years, several conditioned on two years of employment. github.com/holman/extended-exercise-windows
4. TechCabal, Muktar Oladunmade, “Next Wave: ESOPs and the future of employee ownership” (9 June 2025). Verified: body states that “because private company shares are illiquid” ESOPs use vesting, that “in privately held startups, employees usually can’t cash out their stock until a liquidity event because there’s no public market for the shares,” that until then “stock options remain a deferred reward,” and that strike prices must be set at or above fair value and communicated clearly. techcabal.com
5. TechCabal, “Can equity incentivise African tech startup teams?” (29 May 2023). Verified: body states it is “essential to strike a balance between equity grants and cash compensation,” offering competitive salaries alongside equity, and that meaningful value is created for founders and team “post a liquidity event.” techcabal.com
6. Farnam Street, “Untangling Skill and Luck,” on Michael Mauboussin’s The Success Equation. Verified: body sets out the luck-skill continuum and quotes that where “luck is rampant, we must think of skill in terms of a process,” because the outcome is not directly controllable. fs.blog
7. TechCrunch, “From InstaDeep to Paystack: Here are Africa’s biggest startup exits and how much they raised” (1 September 2024). Verified: body catalogues the small set of notable African exits and quotes the forward view that “we will continue to see few exits (IPOs),” citing a frozen IPO market and reduced attractiveness to buyers. Cited here for a different finding than the exit-count sources used in the companion pieces on fund size and liquidation preference. techcrunch.com
8. TechCabal, “Kenya to scrap tax breaks on startup employee stock options in 2025 Finance Bill” (5 June 2025). Verified: body states the Finance Bill 2025 would remove the deferral provision so workers “pay income tax within 30 days of receiving shares, regardless of whether those shares can be sold,” and that in early-stage startups “the change amounts to taxing promise, not profit.” techcabal.com
9. TechCabal, “Exclusive: African VCs are turning to secondary markets for liquidity” (12 January 2024). Verified: body reports that VCs are using secondary sales for liquidity, gives the example of a firm returning its first fund after selling part of its Moniepoint stake in a secondary, and notes such secondaries are often bought at a discount. techcabal.com

Note on the exit base rate. The scarcity figure this piece rests on, roughly twenty to twenty-six African exits a year, trade-sale dominated, near-absent listings, most values undisclosed, is treated as established rather than re-derived, and is drawn on for a different finding than the companion pieces that set it out in detail. The argument here does not turn on the precise count. It turns on the mechanism: a costless promise, a luck-dominated payoff, and a reference class in which the payoff almost never lands.