You're sitting in a talent review meeting. The conversation turns to who should lead the new team. Someone mentions Marcus. He's confident, outgoing, always the first to volunteer. He lists "team leadership" and "people management" on his LinkedIn profile. He told his skip-level he's ready for the next step.
Everyone nods. Marcus wants it. Marcus looks the part.
According to a rigorous new study, Marcus is statistically likely to underperform a person you picked at random.
A team of researchers from Harvard, the Institute for Fiscal Studies, and five other universities ran a pre-registered experiment with 555 participants to answer a deceptively simple question: how do you identify a good manager? They randomly assigned managers to multiple teams, controlled for individual skill, and measured what actually drove team performance. Then they validated those findings against real-world promotion data from LinkedIn and sales results from a multi-billion-dollar retail chain.
Above is the working paper "How Do You Identify a Good Manager?" from Ben Weidmann, David Deming, and colleagues at Harvard, IFS, and several partner universities. Below are my takeaways on what this means for how L&D professionals design leadership pipelines, manager development programs, and talent assessment strategies.
TL;DR
- People who volunteer for management roles perform worse than people assigned at random, largely due to overconfidence about their social skills.
- The single best predictor of management success is economic decision-making skill, not personality, emotional intelligence, or demographic traits.
- Promoting your top individual performers into management (the Peter Principle) produces worse managers than random selection by skill.
- Good managers succeed through monitoring and motivation more than task allocation, which flips the priority of most development programs.
- Widening your candidate pool and screening on skills instead of self-nomination could improve management quality by 0.7 standard deviations.
1. The people most eager to manage are often the least equipped to do it well
Think about how most organizations fill management roles. Someone raises their hand. They express interest during a development conversation. They tell HR they're ready. In most companies, that willingness is treated as a signal of readiness.
This study tested that assumption directly. Half the managers in the experiment were chosen because they wanted the role most. The other half were picked by lottery. Teams led by self-promoted managers performed about 0.1 standard deviations worse than teams with randomly assigned managers. To put that in perspective, having a self-promoted manager dragged team performance down by roughly the same amount as assigning a manager with significantly lower cognitive ability.
The researchers found a specific reason for this gap. Self-promoted managers were considerably more overconfident about their abilities, rating themselves much higher than their objective contributions warranted. The overconfidence wasn't general. It clustered around social skills specifically. Among self-promoted managers, there was a strong negative correlation between how highly they rated their people skills and how they scored on an objective test of emotional perceptiveness.
In plain terms: the people who were most sure they were great with people were measurably worse at reading other people's emotions.
Your self-nomination pipeline doesn't just fail to filter for competence. It actively selects for a specific type of incompetence: people who overestimate their interpersonal skills and undervalue the contributions of their team members. The researchers found that self-promoters listed significantly more "management" and "leadership" skills on their LinkedIn profiles than their randomly assigned counterparts. They don't just believe they're ready. They broadcast it in ways that reinforce the organization's perception.
The confidence you're interpreting as leadership presence might be the exact trait that predicts failure. And the quiet contributor who never asks for the promotion? The data suggests they might outperform the volunteer.
How to rethink your leadership identification process:
- Decouple expressed interest from candidacy by proactively assessing a wider pool of employees for management potential, including those who haven't self-nominated. The study found no performance advantage from wanting the role.
- Add objective skill assessments to your talent review process rather than relying on manager judgment, self-advocacy, or peer perception. Self-reported "people skills" showed a negative correlation with actual emotional perceptiveness among self-promoters.
- Audit your current pipeline for confidence bias by comparing the traits of people who enter your leadership track against the traits of people who actually succeed once they're in management roles. If the entry criteria and the success criteria don't match, your filter is broken.
- Normalize opt-in assessments for all employees rather than only those who express interest, so that high-potential people who would never raise their hand still have a path into the pipeline.
2. The single best predictor of management success isn't what most organizations measure
If you were designing a manager assessment from scratch, you'd probably include personality inventories, emotional intelligence tests, maybe a situational judgment exercise. Most L&D programs do exactly this. The Big Five personality traits, leadership style assessments, and 360-degree feedback instruments dominate the manager development landscape.
This study tested all of those against actual causal measures of management performance.
Personality traits did not predict managerial success in any meaningful way. Extraversion, conscientiousness, agreeableness, openness, emotional stability. None of them. Political skill (a validated measure of workplace influence ability) didn't predict it either. Emotional perceptiveness, measured by a well-established test of reading emotions from facial expressions, showed no significant relationship with management performance among randomly assigned managers.
Two things did predict success: fluid intelligence (the ability to solve novel problems) and economic decision-making skill. Of these two, economic decision-making was the stronger and more consistent predictor across both the lab experiment and the field validation.
Economic decision-making, as the researchers defined it, is the ability to strategically allocate scarce resources, specifically attention and effort, across competing demands. It requires understanding comparative advantage intuitively, resisting cognitive biases like anchoring, and making good decisions in complex environments where you can't attend to everything at once.
In a multi-billion-dollar retail chain with 500 stores, economic decision-making skill was the strongest predictor of store manager performance. A one standard deviation increase in this skill was associated with a 0.19 standard deviation improvement in manager performance, translating to roughly $794,000 in additional annual sales per store.
Most manager development curricula invest heavily in personality awareness, communication styles, and emotional intelligence. Those aren't bad things to develop. But the evidence here says they aren't the things that predict whether someone will make their team more productive. The skill that matters most is the ability to look at a complex situation with competing demands and limited resources and make smart tradeoff decisions quickly.
That's a trainable skill. But it's not one most organizations are training.
How to realign your assessment and development approach:
