A German retail chain ran an experiment in 2018 that it never meant to publish as a warning. Apprentices in 232 stores, most of them hired straight out of school, were told they could earn a bonus point for every month they worked without a single day of unplanned absence. Three points converted into sixty euros loaded onto the employee card they used to buy groceries in the company's own stores, up to two hundred and forty euros across the year, which came to more than a quarter of a typical monthly apprentice salary. A second group was offered extra vacation days on the same schedule instead of cash, and a control group was offered neither.1

Absence went up. In the group paid cash it rose by roughly half, which the researchers worked out to more than five additional days absent per person per year. The vacation-day group showed no systematic reduction either, though the researchers were careful to say that arm was too small to draw a firm conclusion from.1 The firm had bought the opposite of what it went shopping for.

What the Money Announced

The tempting explanation is gaming. People bank their absences inside a month they have already written off and stay clean the rest of the time. The data closes that door. Among first-year apprentices the cash bonus reduced the probability of actually earning a bonus point in a given month by more than eight percentage points, so whatever they were doing, they were not doing it to maximize the payout.1

The survey the researchers ran afterward points somewhere less comfortable. Apprentices in the cash group reported feeling less guilty about being absent when they were not sick, and felt less bound by their contract to always turn up. On a combined measure of what the authors called the intrinsic costs of absence, the cash group sat nearly half a standard deviation below the control group.1 Nothing had changed about what the company could actually do to an apprentice who failed to show. What changed was what an absence meant.

Pay for something people were already doing out of obligation and you have told them the obligation was negotiable. The price is the message.

I have put a compressed version of this to leadership groups for years, usually as structure eats culture, because the reward system beats the values statement whenever the two disagree. What the retail experiment adds is that the reward system also revises what people understood the value to have been in the first place. It reaches backward.

The Damage Landed on Whoever Joined Last

The average conceals a split worth staring at. Apprentices in their first year carried essentially the whole effect, with an estimate close to double the overall figure. For the second-year group the combined estimate came out with a negative sign, and among the most tenured slice of that cohort absence actually fell.1 The people who had been there long enough to absorb what the job expected of them read the bonus as a nice extra. The people who had not yet worked that out read it as instruction.

Which puts your exposure in a specific place rather than spread evenly across the population you are incentivizing. It concentrates in whoever arrived most recently, and in an organization that has been hiring, that is most of the junior bench.

The Money Stopped and the Behavior Kept Going

The bonus ran for twelve months. The researchers kept watching for two quarters after it ended, and for the first-year cohort the effect remained similar in magnitude to what it had been while the money was still on the table.1 Those apprentices had not been renting a belief for the duration of the scheme. They formed one, and it stayed after the thing that formed it was withdrawn.

Why This Matters More Now Than It Did in 2018

Deloitte told its U.S. tax staff in early 2025 that presence at a Deloitte office or client site would be considered in performance evaluations, with an expectation of two to three days a week.2 Robert Half, a recruitment firm whose business runs on how employers compete for people, surveyed five hundred hiring managers across Australia in October 2025 and reported that 69 percent said bonus structures had been changed to reflect office presence, while the firm's own director cautioned that this is unlikely to be a formal or consistently applied policy anywhere.3

The mechanism the German study exposed is now being run at scale, on a behavior most organizations already carried by expectation rather than by payment. That is the exact condition under which the backfire showed up. Employees give a bolted-on correction the same reading, which is the argument behind why installing a structure to make someone accountable produces compliance rather than ownership, and I am not going to relitigate it here. The reverse problem is at least as common, and I looked at how rarely executives visibly reward the behavior a change was supposed to create earlier this summer.

The Question to Ask Before You Attach the Money

The authors close with advice that costs nothing to follow. Examine the norm that already exists before introducing an incentive for the behavior. So the diagnostic is one question, asked honestly, before the compensation committee gets involved. Is this thing currently carried by expectation, or by nothing at all? If people already do it because they understand it as part of the job, attaching money announces that their understanding was wrong, and the announcement outlives the budget line.

Most compensation design gets debated on whether the amount is large enough to move behavior. That is the second question. The first one is what the existence of the payment tells a workforce about the thing being paid for, and nobody in the room is usually assigned to answer it. Who is asking it in yours?