The person who wants the job most isn't your safest promotion. In a controlled test, they were the worse one.

A role is opening up — a shift lead, a plant supervisor, a second-in-command — and one person on staff has made it plain, repeatedly, that they want it. They stay late. They speak first in meetings. When you mention the opening, they are already telling you they are ready. It is a reasonable instinct to read that wanting as evidence of fitness. A study in one of economics' most selective journals says that instinct is, on average, wrong.

The paper is "How Do You Identify a Good Manager?" by eight economists — Ben Weidmann (UCL and LSE), Joseph Vecci (University of Gothenburg), Farah Said (Lahore University of Management Sciences), Sonia Bhalotra (University of Warwick), Achyuta Adhvaryu (UC San Diego), Anant Nyshadham (University of Michigan), Jorge Tamayo (Harvard Business School), and David Deming (Harvard Kennedy School). It ran in The Quarterly Journal of Economics online in January 2026 and has been circulating in leadership commentary again this summer, which is how it crossed my desk. I read the authors' working-paper draft, posted through the ROCKWOOL Foundation Berlin's discussion paper series — the final typeset article sits behind a paywall I could not get past, and the two may differ slightly.

The heart of it is a lab experiment, and the authors are upfront that a lab is not a workplace. They recruited 555 people through the University of Essex's economics lab — paid £35 on average — and sorted them into 728 three-person teams across four rounds, each team one manager and two workers. Every manager was rotated through several different teams, which let the researchers isolate what that specific person added, separate from how capable their workers happened to be. Before teams were built, each participant rated how badly they wanted to be the manager, on a scale of one to ten. Half the sessions then handed the role to whoever wanted it most. The other half assigned it by lottery.

Teams with a self-selected manager performed about 0.1 standard deviations worse than teams with a randomly assigned one — a gap the authors compare to swapping in a manager with a full standard deviation lower score on a test of fluid intelligence. Part of the reason, they find, is overconfidence: self-promoted managers rated their own people skills higher than randomly assigned managers did, yet scored worse, as a group, on a standard test of reading others' emotions.

People who want to be in charge are significantly more likely to overrate their performance compared to their objective contributions.Weidmann, Vecci, Said, Bhalotra, Adhvaryu, Nyshadham, Tamayo, and Deming, working-paper draft of the study published in The Quarterly Journal of Economics

A lab result like that would be easy to file away as curiosity. The authors went looking for it outside the lab, twice. Participants whose lab performance marked them as strong managers were, per their LinkedIn histories, more likely to have since been promoted in real jobs. More useful to you: the authors also measured the same skills in 225 store managers at a retail chain of 500 grocery stores in South America, using sales and staffing records from February 2020 to May 2022. Managers there rotate between stores on the company's own schedule, which let the authors compare one manager's effect on sales across different stores. The single skill that best predicted lab performance — economic decision-making, not personality or people skills — also best predicted store sales in the field. A one standard deviation improvement in manager quality raised a store's annual sales by an estimated $4.1 million, a 25 percent increase.

The authors are direct about what their method leaves out, in their own words: "our approach does not capture other important managerial skills, especially those associated with long-term, repeated interactions, such as relationship management and conflict resolution." That is their caution, stated in their own discussion section. The rest of what I'd flag is mine: the lab pool was mostly not existing managers, recruited at one UK university; the field data comes from a single retail chain in one country. Neither guarantees your next hire will play out the same way.

Here is why this lands differently for you than for a company with an HR department running assessment centers. A large employer has already built the check this paper describes — skills tests, structured interviews, calibration panels. At a $5–25 million company, there usually isn't one. The person asking loudest for the job is often also the person already acting the part, and it is easy to mistake that performance of readiness for the decision-making skill that, in this data, is what actually moved sales.

The part the research doesn't say, but that I will say plainly: the eagerness that gets someone promoted in a company like yours is often the same eagerness that makes them hard to say no to. It is rarely dishonesty. It is that the person pushing hardest for the role is rarely the one volunteering the case against themselves — and in a company like yours, you may be the only one checking it.

Before you finalize your next promotion, give the leading candidate one real decision to make under the conditions a manager actually faces — split a fixed set of hours, people, or budget across competing jobs, with a deadline attached — rather than deciding on an interview or how long they have been asking.

If you want a cleaner read on how you decide under pressure before you hand someone a team, that is what the Leadership Profile is built to surface.