The Rational Employee

The call

My boss asked how I was doing first. That was the tell.

We had a standing one-on-one, a video call, and he never opened with me. He opened with the roadmap, or a partner who was being difficult, or a number he wanted me to look at. This time he asked about my weekend and waited for the answer. I knew before he said it. He had brought me into the company two years earlier. He had, a few weeks before this, given me a raise I hadn’t asked for. And now he was telling me he was leaving.

What I felt was not what I expected to feel. It wasn’t fear, though there was reason for it. It wasn’t betrayal. It was sadness, for him. He had given a large piece of his life to this company. He had arrived early, when the equity grants were generous and the story was clean, and he had stayed through the part where the story got complicated. This was not the ending he had signed up for. On the screen he looked lighter than I had seen him in months.

A few weeks later, once he was out, he told me something he could not have said while he was still inside. He had started doubting nine months earlier. Not the company’s survival, exactly. Its chances of paying the people who had built it. Nine months is a long time to go to work every day on a belief you no longer hold. I wanted to know what had kept him there that long, and what had finally let him go. The answer, it turned out, had almost nothing to do with him.

Relief or grief

Here is a question. If you were laid off tomorrow, with three months’ severance, what would you feel in the first ten seconds?

Not what you would feel by the afternoon, when the mortgage math kicks in. The first ten seconds, before the rational mind arrives. People split cleanly. Some feel grief: they like it there, they would miss the people, the loss is real. Some feel annoyance, the way you would at a flat tire. And a surprisingly large group feels something they are embarrassed to name. Relief. Lightness. The sense of a decision that has finally been made, by someone else, so that they no longer have to make it.

If you are in the third group, this essay is about you, and the news is better than you think. You are not weak, or disloyal, or bad at your job. You are behaving exactly the way a well-designed compensation plan is built to make you behave. The plan is working. It is working on you.

That is the argument. Startups retain people with equity that vests over time, and the mechanism is so effective that it holds onto the people who should leave at least as tightly as the people who should stay. Understanding why requires a small amount of economics, a smaller amount of psychology, and one uncomfortable look in the mirror. I’ll supply the first two. The mirror is below.

Five kinds of paper

I have been paid in every flavor of equity there is. I have worked at a company whose shares traded on a real exchange, where the paper was money and everyone knew it. I was an early employee at a financial company that went public while I was there, and I sat through the lockup watching a number on a screen become real. I have held stock in a company quoted on the pink sheets, where you could technically sell and nobody would buy. At one point I turned down an option grant there, because accepting it would have triggered a tax bill larger than any price the shares were plausibly going to reach. Think about that: equity so illiquid that the rational move was to refuse it. I have worked at a private company where the shares were worth whatever the last investor said they were worth. And I am at one now, a venture-backed financial technology company, two years in, half vested, with a strike price and a preferred price and a 409A valuation that do not agree with each other.

None of this makes me an economist. It makes me a specimen. I have been the believer, the one doing the math at night, the one holding a lottery ticket, and I have watched the people around me cycle through the same roles. The pattern is not personal. It is structural, and once you see the structure you cannot unsee it.

Part I: What the handcuffs are for

A vesting schedule is not a reward. It is a contract designed to solve a problem that economists call the principal-agent problem, and it is worth being precise about what the problem is.

The principal (the company, really its investors) wants an employee (the agent) to work hard on something whose payoff is years away and whose outcome the employee can barely influence. The company cannot watch the employee every minute. It cannot measure effort directly. So it has to design a payment that makes the employee want to do the right thing on their own. Bengt Holmström won a Nobel Prize for working out how to do this. His 1979 paper on moral hazard shows that when you can’t observe effort, the best you can do is tie pay to outcomes, and accept that the employee will bear risk they would rather not bear. His later work with Paul Milgrom on multitask incentives explains why you can’t incentivize everything: pay for one thing and people neglect the rest.

Equity does two jobs in this framework. It ties pay to the outcome, so the employee cares about the company’s value. And it ties pay to time, so the employee stays long enough for their effort to matter. The second job is the one people forget. A four-year vest with a one-year cliff is not primarily about alignment. It is a commitment device: a way for the company to make leaving expensive without having to negotiate about it every quarter. Thomas Schelling, in The Strategy of Conflict, showed that binding yourself (burning a bridge, posting a bond) can be a source of power. A vesting schedule binds the employee, and the employee agrees to it on day one because on day one it costs nothing.

Here is the first thing to notice. The commitment device has an expiry date. It is strongest in year one, when almost everything is unvested, and weakest in year four, when almost nothing is. The company’s hold on you is a decaying asset. Everyone knows this and almost nobody plans for it, because the decay is slow and the moment it reaches zero never announces itself.

The second thing to notice is subtler. At a company of any size, your effort does not move the share price. You could work twice as hard or half as hard and the valuation at the next round would be the same to three decimal places. So the alignment job, the one the economists care about, is doing very little for most employees. What the equity is actually doing is the retention job. It is a bond you post against your own departure, denominated in a currency whose value you are constantly re-estimating.

That last phrase is the whole essay. The currency is belief.

Part II: The four corners

If belief is the currency, then every employee at a private company can be placed on a map with two axes. One axis is how much you have left to vest. The other is how much you believe the common stock will ever be worth anything.

That second axis needs a definition, because the usual one is wrong. People ask “will the company succeed?” The question that matters is narrower: will an exit be large enough to pay the common shareholders after the investors have been paid first? Venture-backed companies carry a liquidation preference: investors get their money back, sometimes with a multiple, before anyone holding common stock sees a dollar. At many companies the gap between the preferred price and the common price (the 409A valuation, the number the IRS requires for setting option strikes) is two or three to one. Your options are a bet that the company clears that stack. A company can survive, even grow, and never clear it.

Four corners, then.

The Believer has lots left to vest and believes. This is the employee the plan was designed for. The handcuffs are on, and they don’t chafe, because the thing inside them looks valuable. Believers are the most exposed people on the map, because belief is the input they control least and it is usually the first thing to move.

The Handcuffed has lots left to vest and doubts. They have started running the numbers. They know what the preference stack looks like. They stay because forfeiting the unvested half feels like a real loss, and because a refresh grant might arrive. This is the uncomfortable middle, and most people in it do not admit they are there.

The Vested Veteran has little left to vest and doubts. Their shares are theirs whether they stay or go. The company has no lever left except a refresh grant it will almost certainly size incorrectly. Every rational model says they leave at the next natural transition, and when they do, everyone is surprised except the other veterans.

The Lifer has little left to vest and still believes. Founders live here, and a few early hires. Nothing financial is holding them and nothing financial could make them leave. Companies make a mistake when they look at their Lifers and assume loyalty is the general case.

And in the middle, where most people actually are, sits the Lottery Ticket Holder: half vested, belief somewhere around a coin flip, treating the equity as a scratch-off stapled to a paycheck. Not cynical. Just accounting.

My boss was a Vested Veteran by the time he told me. I am a Lottery Ticket Holder, two years in, half through, with a strike near the 409A and a preferred price nearly three times higher. We were in different corners of the same map, which is why the same news landed on us so differently.

Part III: Why the smartest people leave first

Three ideas from economics explain why departures at a private company come in clusters, why they are led by the people who know the most, and why the company can do almost nothing about it.

The first is costly signaling. Michael Spence’s 1973 paper on job-market signaling made a simple point: a signal is only credible if it is expensive to send. Anyone can say the company is in trouble. A senior person who walks away from unvested equity is paying to say it. That is why one departure from the top moves everyone else’s beliefs more than a hundred all-hands slides. The information content is in the forfeiture.

The second is the information cascade. Bikhchandani, Hirshleifer and Welch showed in 1992 how rational people, each watching the actions of those before them, can abandon their own private information and follow the crowd. At a startup the mechanism is visceral. You had a private estimate of the company’s odds. Someone you respect leaves. You update. Your updated estimate leaks into a conversation with a peer, who updates. Within a quarter, a belief that nobody held individually has become the consensus, and the exits cluster around the next natural transition, a reorg, a new operating model, a layoff, because a transition is a Schelling point: a moment when leaving carries less reputational cost because everyone understands why you might.

The third is adverse selection, George Akerlof’s market for lemons. The company cannot observe your private belief. It can only offer retention grants and watch who takes them. The people most eager to accept a refresh are the ones who still believe. The doubters, the ones the company most needs to keep, take the outside offer quietly. Any refresh program retains exactly the population it was going to keep anyway, and the company learns this only in the exit interviews.

Put the three together and you get the dynamic my boss lived through. His own belief dropped nine months before he left; that is private information. The equity that might have held him had already vested; the bond had expired. A refresh sized to his tenure and level was far smaller than a fresh four-year grant from a company whose preference stack he had not yet learned to fear. The reorg gave him a moment when leaving looked like a reasonable response to events rather than a verdict on the company. He left. Then another early employee left on the same transition, for what I assume were the same reasons. And every person who watched them go, including me, updated.

Could the company have stopped it? Only by doing something it could not afford and could not target: paying fully vested senior people a grant sized to their outside option, which is to say a second founding grant, and paying it specifically to the ones who had stopped believing, whom it could not identify. The rational firm accepts the churn and plans around it. Most do not plan. They are surprised, every time, by the people who knew the most.

Part IV: The part the economists skip

Everything so far treats the decision as money. It isn’t, and everyone who has ever stayed at a job too long knows it isn’t. The honest version of the equation has four terms the equity model leaves out.

Liking it there. Sherwin Rosen’s theory of compensating differentials says that jobs pay partly in cash and partly in conditions, and that workers trade one for the other. A pleasant job pays less in money because it pays more in everything else. The reverse is also true. If you like the people and the work is interesting and your boss leaves you alone, that is income. It does not appear on the pay stub, and an outside offer has to clear it along with the equity.

Dreading the search. John McCall’s job search model introduced the idea of a reservation wage: the offer below which a person won’t bother looking. Searching costs time, energy, and a certain amount of self-respect. Ambitious people have a low reservation offer; they take the recruiter’s call to find out what they’re worth. Comfortable people have a high one; the search sounds exhausting, and this is fine. Same equity math, opposite behavior.

Starting over. Switching costs rise with age and family. A new boss, new politics, a reputation to rebuild. And a detail the brochures leave out: at most companies, leaving starts a 90-day clock to exercise your vested options, which means writing a check for the strike price and possibly a tax bill, to hold shares you still can’t sell. The one place illiquidity still bites the employee is on the way out the door.

Fear of loss. Daniel Kahneman and Amos Tversky’s prospect theory established that people feel a loss roughly twice as strongly as an equivalent gain. Unvested equity is a loss if you leave and a gain if you stay. So it weighs double, whatever the spreadsheet says. This is not a flaw in the vesting schedule. It is the vesting schedule’s secret weapon, and it works hardest on the people whose spreadsheet already says go.

These four terms don’t add a new axis to the map. They move the line. The boundary between “stay” and “go” is not fixed at the midpoint of belief or vesting; it slides. For someone who loves the place it slides far enough that almost nobody leaves. For someone who dreads it, it slides the other way, until even a large unvested stake can’t justify staying, and yet people stay anyway, because of the fourth term. Take your Monday score from the top of this piece and apply it to the map. Watch which corner you end up in without touching the money.

Part V: The Hostage

There is a fifth type, and it is the one the layoff question was designed to find.

The Hostage has lots left to vest and dreads going to work. The stress is too high, or the new boss is impossible, or there is a feeling of having been slighted that will not go away no matter how many good quarters follow. The equity math might say stay. Loss aversion says stay, twice as loudly. And so the Hostage stays, and cannot explain, even to themselves, why they keep signing up to be disappointed.

The mechanism has a name, and it comes from a pigeon. B. F. Skinner and Charles Ferster showed in Schedules of Reinforcement that the way to produce the most persistent behavior is not to reward it every time. It is to reward it unpredictably. A pigeon that gets a pellet on every tenth peck learns to peck ten times. A pigeon that gets a pellet at random intervals pecks forever, because the next one might be the one. Casinos are built on this. So, without meaning to be, is the modern startup.

A company that disappointed you reliably would lose you. You would adjust and go. A company that disappoints you unpredictably keeps you: the surprise raise, the boss who pulls you aside to say you matter, the quarter where the numbers finally turn. Each one resets the clock. The Hostage is not irrational. The Hostage is responding correctly to a reinforcement schedule that happens to be the most addictive kind there is.

This is why a layoff can feel like relief. The company has paid you severance to stop paying the loss. It has resolved a decision you were not able to make, and it has done so in a way that lets you tell everyone, including yourself, that it was not your choice. People in this state are almost always past the rational exit point. They have been waiting, sometimes for years, for someone else to say the word.

If you want a diagnostic, it isn’t the spreadsheet. It’s the ten seconds.

Part VI: Employee number three

The other person who left on that reorg was employee number three. Nine years. My boss’s boss. I want to be precise about him, because the easy version of this story would make him a cautionary tale, and he isn’t one.

He is not someone I could see myself becoming. We are very different people, with different styles and different instincts, and in most of the ways that get measured he is simply better than me: smarter, more committed, with a level of energy I could not match on my best week. He is younger than I am, with more upside and harder skills. If I had started sacrificing everything, with total focus, twenty years ago, I might have become something like him, in a less visible place. After an early stretch where we didn’t understand each other, we arrived at a kind of mutual respect. He came to see value in the things I’m actually good at: humor, being a good teammate, not needing credit, creativity, diplomacy. I don’t think he believes I’m at his level. Maybe I’m not.

And yet, when it ended, the perception around him was that he hadn’t been successful enough. I find that brutal, inaccurate, and unfair, and the disappointment in the building was thick. I had an hour with him after the news, after the decision, when there was nothing left to negotiate. I didn’t offer analysis. I just stayed in the empathy, because the thing he was living through has no analytical fix: it is hard to watch something you loved and built turn away from you and from what you thought you were contributing. Every employee at every startup already knows, in the abstract, that tying your self-worth to a job is a mistake. I had learned it myself, more than once. But I had learned it the way you learn by falling off a two-story building and breaking a leg. This was a skyscraper.

Here is what the theory in this essay says about him, and where it falls short. Nine years is nearly the whole life of a venture-backed company. He had vested everything, long ago. He had watched the preferred price climb past the common price and keep climbing. He had seen more reorgs than anyone. By the four-corner model he was a Vested Veteran who should have left years earlier, and the fact that he didn’t is the thing I keep turning over. What held him was not the equity. It was the other terms, the ones from Part IV: the people, the thing he had built, the identity of being the person who was there at the start. Those held for a long time. The reorg didn’t change his equity math. It changed the other terms, all at once, and the bond that had quietly expired turned out to be the only thing left holding him there. The model predicted the exit. It said nothing about what the exit would cost him, because it has no term for what a decade of yourself is worth.

I want to be careful here, because the easy reading of this essay is “leave.” That is not the argument, and the reason I know the model is incomplete is that I have broken every rule in it and been paid for it.

On the morning the company I had helped build went public, I called in sick. I was not rational-sick. I was sick to my stomach with how betrayed I felt. The thing I had invented had become the engine of the company’s earnings, nine figures over its lifetime, and the celebration was happening without me. No trip to New York. No bell. Three layers of management stood on that podium taking credit for a thing I had built, and I was already on the far side of the door, pushed aside before the party started. If you had asked me the layoff question that morning, I’d have answered relief without hesitating.

And then I did something no rational employee does. I took the proceeds of two property refinances and bought the company’s stock in the open market, at roughly ten times the price it had listed at, far above the price of my own grant. Everything in this essay says that is the behavior of a Hostage: self-worth tied to the job, loss aversion, the sunk cost of years. But it wasn’t belief in the company that made me buy. It was belief in the thing I had built, which I understood better than the people ringing the bell did. I had tied my self-worth to the invention and my money to the company. Those are not the same bet, and I did not know it yet.

I stopped believing at roughly the same time I decided to leave, which is exactly what the four-corner model predicts, and I sold everything at about seventeen times the listing price. The stock peaked a couple of percent higher. Some months later it went to zero. Not down. Zero. I was not rational. I was lucky. Psychologists call the mistake I am tempted to make here outcome bias, Baron and Hershey’s term for judging a decision by how it turned out rather than by what was known when it was made. Nassim Taleb built a whole book around the same error. Shift the timing by a year and this is a story about a man who put two refinances into a company that had already shown him the door, and lost it all.

So the point is not that rational people leave. It is that a rational person knows there are three separate bets on the table and refuses to let them blur. The job is one bet, paid in cash and in conditions, and it can go bad on a Tuesday. The equity is a second bet, on whether an exit clears the preference stack, and it should be sized like any other speculative position, which is to say not with refinance proceeds. And the thing you built is not a financial instrument at all. It is the one part of the story that is actually yours, and it is worth exactly what it was worth the morning they rang the bell without you. The mistake, the one the vesting schedule is engineered to produce, is to let the second bet carry the weight of the third.

The rational employee

What does a rational employee actually do? Not leave, necessarily. Three things.

They define belief correctly. Not “will the company survive” but “will an exit clear the preference stack and pay the common.” They know the preferred price, the 409A, and their strike, and they know that the gap between the first two is the honest measure of how far away the money is.

They watch the clock. The company’s hold on them is a decaying asset, strongest after the cliff and near zero by the end of the original grant. They know roughly where they are on that curve, and they treat a refresh conversation not as a favor to ask but as a data point to collect: the company’s answer reveals how it sees its own odds, and how it sees them.

And they take the ten-second test, honestly, once a year. If a layoff would feel like relief, the equity is not the reason they are staying. Loss aversion is. The handcuffs are doing their job on someone who was not supposed to need them.

My boss took nine months to act on something he already knew. He is not unusual. He is the median. The plan was working on him, right up until the moment it had nothing left to work with, and then a reorg gave him permission. The only thing I would change is the nine months.

I’m two years in, half vested, holding a lottery ticket priced as honestly as I know how. I like the people. I am not done. I know which of those terms is doing the work, and I check. That’s all the rationality there is.

Further reading

The economics

• Holmström, B. (1979). Moral Hazard and Observability. Bell Journal of Economics. Why pay has to be tied to outcomes when effort can’t be seen.
• Holmström, B. and Milgrom, P. (1991). Multitask Principal-Agent Analyses. Journal of Law, Economics, and Organization. Why you can’t incentivize everything.
• Schelling, T. (1960). The Strategy of Conflict. Commitment devices and why binding yourself is power.
• Spence, M. (1973). Job Market Signaling. Quarterly Journal of Economics. A signal is credible only when it costs something.
• Bikhchandani, S., Hirshleifer, D. and Welch, I. (1992). A Theory of Fads, Fashion, Custom, and Cultural Change as Informational Cascades. Journal of Political Economy. Why departures cluster.
• Akerlof, G. (1970). The Market for Lemons. Quarterly Journal of Economics. Why retention grants reach the wrong people.
• Rosen, S. (1986). The Theory of Equalizing Differences. Handbook of Labor Economics. Liking your job is income.
• McCall, J. (1970). Economics of Information and Job Search. Quarterly Journal of Economics. Reservation wages and why some people never look.

The psychology

• Kahneman, D. and Tversky, A. (1979). Prospect Theory: An Analysis of Decision under Risk. Econometrica. Losses weigh roughly twice as much as gains.
• Ferster, C. and Skinner, B. F. (1957). Schedules of Reinforcement. Why unpredictable rewards produce the most persistent behavior.
• Baron, J. and Hershey, J. (1988). Outcome Bias in Decision Evaluation. Journal of Personality and Social Psychology. Judging a decision by its result instead of by what was known.
• Taleb, N. N. (2001). Fooled by Randomness. Why lucky people write confident books.

The plumbing

• IRS, Section 409A and nonqualified deferred compensation. Why private-company common stock gets a separate valuation from preferred.

Links go to the original papers or publisher pages; several sit behind journal paywalls. Open versions of most can be found by title.