Data scientist: Avoid one ‘toxic worker’ don’t hire two ‘superstars’
Michael Housman’s research career has been dedicated to what he calls “decoding the workforce”: using large-scale data to understand why people stay, leave, and perform at work. A recurring theme across his work is that individual talent matters far less than the social environment employees operate within.
This insight is central to a recent Harvard Business School working paper co-authored by Housman and economist Dylan Minor. Drawing on data from approximately 50,000 hourly frontline workers across multiple firms, the study examines the organizational impact of so-called toxic workers.
In the paper, toxic workers are defined as employees whose behavior harms people or property—through actions such as harassment, fraud, bullying, theft, or serious policy violations. These behaviors do more than create isolated problems; they tend to spread, undermining morale and increasing attrition among otherwise solid performers.
One of the study’s most counterintuitive findings is that toxic workers are often highly productive. They complete tasks quickly and may outperform peers on narrow output metrics, which helps explain why organizations tolerate them longer than they should. However, this productivity often comes at the expense of quality, ethics, and team stability.
When the authors compared toxic workers with “superstars,” the financial contrast was stark. A top 1% performer contributes an estimated $5,303 in additional annual value. By comparison, simply avoiding the hire of a toxic employee saves roughly $12,489 per year, primarily by preventing the downstream turnover of coworkers who choose to leave rather than work in a hostile environment.
These estimates are conservative. They exclude harder-to-measure costs such as litigation, regulatory penalties, reputational damage, and long-term erosion of trust. Housman notes that in professional or higher-paid roles, the true cost of toxic behavior can easily rise into the tens of thousands of dollars per employee.
The study also identifies traits associated with a higher likelihood of toxic behavior. Employees who display excessive self-regard, overconfidence, and a rigid insistence that rules should always be followed are statistically more likely to be terminated for misconduct. These signals, the authors argue, can often be detected during hiring through thoughtful assessments and structured reference checks.
Ultimately, the research challenges organizations to rethink how they define talent. Focusing solely on output allows corrosive behavior to survive. A more effective approach evaluates employees across multiple dimensions—productivity, integrity, and corporate citizenship. From both a cultural and financial perspective, avoiding toxic workers is one of the highest-return decisions an organization can make.
