Published on: November 16, 2026 at 11:07 am
The US Office of Personnel Management (OPM) has told federal agencies that no more than 40% of General Schedule (GS) employees can receive the two highest performance ratings this cycle, and that “calibration” programs will keep ratings within that limit. OPM’s diagnosis is correct: When nearly everyone is rated as outstanding, ratings lose their informational value. The cure mandated by OPM, however, rests on an assumption that the evidence does not support.
A fixed cap assumes we know in advance what proportion of employees should receive top performance ratings. For decades, performance management systems have been built on the assumption that performance follows a bell-shaped curve, with most people near the middle and few at the extremes. In a study of more than 600,000 researchers, entertainers, politicians, and athletes, we found that individual performance usually does not follow that pattern. A small group of star performers produces a disproportionate share of the output. More recently, we examined 274 distributions covering more than 200,000 teams and found that only 11% were normally distributed (i.e., bell-shaped). We also found that the degree of performance variation depends on features of the work, such as how authority is distributed within the team.
The implication for the federal workforce is that the true share of top performers will vary across agencies, offices, departments, teams, and jobs. A single 40% ceiling will be too loose in some units and too tight in others. Where it binds, managers will be forced to rate some genuinely excellent employees as merely good. Our research warned about this consequence years ago: When a rating system forces reviews into a shape that performance does not have, stars get rated as average, and, as a result, the people most likely to leave are the ones the organization can least afford to lose.
The early reaction at the US Department of Health and Human Services points to a second cost. Employees told the health news outlet STAT that they view the new self-assessments and calibration process with suspicion. Performance management works only when employees believe their ratings reflect the quality and quantity of the work they actually did.
There is a better path: Measure observable, verifiable results; train supervisors to rate employees against clear standards rather than against one another; and let the data reveal how many top performers each unit has. Then, audit units whose ratings appear implausible. Inflation in employee performance ratings is a measurement problem that requires a measurement-based solution.