The Uncovered Option: Why Every Function Counts the Same Number Differently

A buyer receives an email in spring with his name in it, because he pushed through a supplier switch that saves the group six hundred thousand euros this fiscal year, and that number is real, it sits in the system, it is attributed to him, and it will follow him all the way into his next performance review. Two years later a line in another plant stands still for nine days because that very cheaper supplier failed to deliver a batch, and the damage exceeds the saving several times over, but it appears in no email with a name on it, it spreads across production, logistics and sales, it is booked as operational noise, and nobody would think to net it against the celebrated saving from back then. Between these two events there is no moral failure and no particular short-sightedness on the part of this one buyer, but something more unsettling, namely a bookkeeping that sees the one immediately and by name and the other never.

The convenient explanation for such patterns reaches for personality. Procurement is just frugal to the point of pain, marketing spends other people's money, legal is structurally timid, and so on down the whole org chart, as if departments were little national characters with innate temperaments. That is comforting, because it can be fixed with other people, and it is almost always wrong. Move the same person from procurement into marketing and after six months she will spend like marketing, not because her character has changed but because she now sits on a different curve, and that curve reckons with her long before she reckons with the curve.

The Same Number, Counted in Opposite Directions

It helps to picture a function's position not as a task but as a bet with a particular shape. Procurement holds a bet whose gain is thin, certain and immediately visible and whose loss is thick, rare and far away, and a bet of this shape is called convex in the language of finance, because the holder collects the upside while the downside seeps away somewhere into the system. Every euro saved is booked, celebrated and attributed to a name, while the destroyed value, the fragility of the supply chain, the lost negotiating position, the default risk, arrives later, spreads across departments and belongs to nobody. Whoever sits on such a curve need not be a miser to behave like one, it is entirely enough that he responds rationally to what the measurement system shows him and to what it hides from him.

Marketing holds the mirror-image bet, and its shape is concave. Every euro spent is immediately and by name attributable, it stands in the budget, it has a face, while the return arrives late, remains disputed in its attribution and in the end is usually booked as a market success of the sales team, which after all closed the deal. A function whose costs are charged to it in full and whose returns are credited to someone else appears in the system as a pure cost item, and a pure cost item is at every budget round the most obvious candidate for the red pen, which fully explains why the marketing budget is the first to fall in a crisis even though nobody seriously claims that marketing harms the business. One notch more sharply concave sit legal and compliance, who never draw a gain from any enabled deal and a visible loss from every mistake, which is why their only rational strategy is to prevent as much as possible, and why it is unfair to blame them for exactly that.

Why the Bookkeeping Lies in One Direction

The mechanism that produces these curvatures is unspectacular and effective for exactly that reason, and it has to do with time. Costs become visible at the same moment as the decision, the invoice is there, the number is fixed, it can be booked, whereas value destruction almost always arrives later, surfaces somewhere else and has no unambiguous originator anyone could hold to account. A measurement system that follows this non-simultaneity, and every real measurement system does, does not merely measure a little imprecisely, it measures in one systematic direction, because it captures one side of the ledger at once and the other side only when nobody draws the connection anymore. This time asymmetry of attributability is the actual engine of the whole thing, and it is not a bug you fix with better software but a property of how time and responsibility hang together in organisations.

You can catch it with a small test. Ask any function what its greatest achievement of the past year was, and you get a number that materialised at the same time as a decision, the saving, the reduced cycle time, the count of cases closed. Ask about the largest damage it avoided, and the room goes quiet, not because there was none, but because avoided damage by definition never occurred and therefore sits in no system that only books what happened. What did not happen has no author, and what has no author counts, in an organisation, for about as much as a rumour.

The CEO Would Have Taken All Twenty-Three

The sharpest illustration of this comes from Richard Thaler, who in his book Misbehaving describes how he put twenty-three division managers of one company to a simple choice. Each of them was asked whether he would take on a project with a fifty percent chance of a two million gain and a fifty percent chance of a one million loss, a project, that is, with a clear expected value of plus half a million, and only three of the twenty-three said yes. The CEO, who sat beside them and listened, answered the same question by saying that of course he wanted all twenty-three projects, because he sees the portfolio in which the good and bad outcomes average out to an expected plus of eleven and a half million, while each individual manager holds not the portfolio but the one case on which, in case of doubt, his career hangs. The manager who declines is neither stupid nor unduly timid, he simply reckons correctly on his own curve, and his curve is concave, because the loss can cost him the job and the gain brings him a good number and a pat on the shoulder.

The theoretically cleanest version of this pattern, though, is not Thaler's anecdote but a paper by Bengt Holmström and Paul Milgrom from 1991, the Multitask Principal-Agent Analyses in the Journal of Law, Economics, and Organization. Their core result can be stated in a sentence that sounds harmless on first hearing and shakes half of management practice on the second: as soon as a person has several tasks, one of which is well measurable and another poorly, and both are served by the same incentive system, effort flows into the measurable task and the unmeasurable one starves. Not because people considered the unmeasurable task unimportant, but because a high-resolution incentive system on a partial metric pulls attention precisely to where it is rewarded and casts everything else into the shade, which is the reason it is sometimes wiser to build no strong incentive system at all than a strong one on the wrong half.

The Most Convenient Objection in the World

Here one has to be honest, because there is an objection to this whole argument that is strong and must not be waved away. Procurement savings are real, they do not sit in the system for no reason, and the claim that behind every saving lurks a later, invisible damage can be used by any department that does not want to justify its own costs, which makes it the most convenient excuse in the world. Every overpaid unit could say its returns simply arrive later and invisibly, and it would be immunised against any cut. An argument that saves every cost position saves none, and whoever takes payoff curvature seriously has to take this objection seriously first, otherwise he has merely built a rhetorical tool with which any inertia can be dressed up.

The answer to it is not to dispute the savings, but to dispute the one-sidedness of the bookkeeping. A function that books its gains and leaves its losses to the corridor is not a savings program but an uncovered option at the company's expense, a position, that is, that pockets the upside and bills the premium for the downside to someone else. The difference between a real saving and an uncovered option is not whether something was saved, but whether the same place that gets credited with the gain also stands liable for the loss that arrives later. As long as it does not, the pretty number in the system is not an achievement but a sold put whose buyer does not yet know he has bought it, and the price of that put only surfaces when the line stands still.

Where HR Sits in This Calculation

For everyone who works in HR, the transfer is almost too exact to still pass as an analogy. HR bears the cost of every hire, every measure and every program in full and immediately visible, it stands in the budget, it has a face, and HR is practically never allowed to book the value of a good retention, an avoided mishire, a culture that drives nobody away, because a person who stays and does good work produces no metric, only the absence of a problem. And at every mishire HR is named, while the quiet success stays anonymous, which yields a deeply concave curve and fully explains why recruiting optimizes for time-to-fill and cost-per-hire and not for fit. Those are, after all, the only figures that can be evidenced at the same time as the decision and without an attribution dispute, while the fit that actually matters becomes visible only after years and by then has no clear author left to praise.

So whoever accuses an HR department of hiring too fast and too cheaply instead of too well is essentially accusing it of responding to the only curve anyone has built for it, which is about as helpful as accusing a compass of pointing north. The interesting question is never why a function behaves the way its curve prescribes, but why the curve is cut the way it is and who could re-cut it without simply producing a new, equally one-sided curve.

What Would Have to Change, and What Stays Open

The obvious response, that one should measure avoided damage and long-term value too, is correct and still leads straight into the next trap, because a softly defined notion of value quickly degenerates into a mushy innovation budget that justifies everything and disciplines nothing. Between a bookkeeping that sees only hard, immediate costs and one that books every vague hope for the future as value lies the actual design space, and it is narrower and harder than the advice literature admits. Three things stay honestly open. How you re-cut attribution so that the same place stands liable for both the saving and the follow-on damage, without it turning into an unaffordable bureaucracy. Whether judgement at the portfolio level rather than the single-case level, as Thaler's CEO performed it intuitively, is even enforceable in a real corporate hierarchy with its individual, career-anxious human beings. And whether the option value of an exploration decision can be quantified so that a controller accepts it, for which the real-options literature around Avinash Dixit and Robert Pindyck, Investment under Uncertainty, supplies the vocabulary, the value of waiting, the cost of irreversibility, but not automatically the result, because a vocabulary is not yet a booking.

What can be said before these questions are solved, though, is the diagnostic part. When someone in the next meeting explains that one department is too cautious or another too wasteful, it is worth setting personality aside for a moment and asking instead what shape its payoff has, where its gains are booked and where its losses disappear to. Most of the time it turns out that the supposed character flaw is a perfectly correct answer to a crookedly cut curve, and that the person you want to reprimand is doing the only thing the system permits.