We all have our boss horror stories. The underminer. The bad communicator. The credit hog. The snake. Then again, if we’re lucky, we’ve all had those amazing bosses as well—the supervisor who encourages all employees to take their work up to the next level; the manager who makes everyone around them look better.
But how much of an effect does a good or bad boss have on workers, really? Harvard Business School Assistant Professor Christopher Stanton sets out to ask that question in The Value of Bosses, a paper recently published in the Journal of Labor Economics—and finds out the answer is, quite a lot.
“If you have a better boss on a team, you get more out of each individual worker”
Academics and practitioners alike are interested in how to construct the best teams to get the most productivity out of people working together. But comparatively little attention has been placed on those people supervising teams. In part, that’s because it’s difficult to separate the performance of the boss from the performance of the individual workers he or she oversees.
“Bosses may get lucky and have subordinates who can do their job well—or, in other settings, they can get really unlucky and have one person who poisons the whole bunch,” says Stanton, who co-wrote the study with Edward Lazear and Kathryn Shaw of the Stanford Graduate School of Business, where he began the research as part of his dissertation in 2011.
In order to isolate the effects of bosses on workers, Stanton and coauthors worked with a technology-based services company that tracked all of its workers’ transaction times. Importantly, supervisors were rotated on an ongoing basis, so workers would have different bosses every few months. Looking at the company’s data, Stanton and colleagues discovered a wide range of worker performance.
“There was tremendous variety in the productivity of workers doing the same task compared to other workers who looked similar at the start,” Stanton says. But for particular workers, their individual performance fluctuated in a predictable pattern according to which boss they worked with at a given time—with some bosses just clearly better than others. “In our setting, idiosyncratic effects of bosses on certain workers were quite small—on average, Boss A was uniformly better than Boss B for everyone.”
Measuring boss quality
To measure boss quality, the researchers looked at transaction times for the workers under them on any given day (due to confidentiality agreements, he can’t say what the transaction is—but a good analogy is the amount of time it takes for a grocery clerk to ring up a customer; or a call center employee to troubleshoot a customer’s problem). To a lesser extent, they also looked at the amount of time employees actually spent with customers compared to other duties; and customer surveys of worker quality.
When they examined all of this data, they concluded that replacing a boss who was in the bottom 10 percent of the distribution with a boss who was in the top 10 percent had the same effect as adding another whole worker to a nine-person team—a huge effect for such a small variation in quality.
“That’s because the effect of a boss is multiplicative,” says Stanton. “If you have a better boss on a team, you get more out of each individual worker.”
“These results suggest the most important peer is your supervisor”
As a consequence, it may be tempting to assign more subordinates to better bosses. But that might not always be the right recommendation. In situations where it’s important for a boss to have dedicated one-on-one interactions with employees, there may be diminishing returns when more workers are added to a team under his supervision.
“You might be tempted to fire the worst supervisor, and have the best supervisor take over that person’s role for more people. That might create congestion effects,” he says. “You don’t want the boss overseeing 100 people if they can only spend 5 minutes a day with each person.”
Overall, however, the effects of their study point to the outsized influence that supervisors up and down the chain of command—not just upper managers or C-level executives—can have on worker performance. “If you recognize that,” says Stanton, “it should inform how you recruit people, promote people, and structure trial periods.”
Especially in work situations where it may be difficult to fire managers after they are hired, Stanton recommends companies think carefully about the screening procedures they put in place, including longer trial periods to measure performance before officially hiring a new manager.
“You need to have provisional steps in place that allow you to determine whether the person will be a good boss or not.”
What makes a boss good or bad?
One thing that Stanton’s study can’t say is what exactly makes a boss good or bad—whether its education, experience, temperament or other systemic factors. Given the enormous impact a boss can have, however, it’s an area ripe for future exploration.
“There is a growing literature in the social sciences about the importance of peer effects—many people think about how to form teams and what the right team should look like,” says Stanton. But according to this data, the effects of substituting one worker for another are actually relatively tiny compared to the effect of substituting a boss.
“These results suggest the most important peer is your supervisor.”