The Value of Bosses

GAICIL Insights

How much is a good boss worth?¹ Everyone has an opinion; The Value of Bosses, by Edward P. Lazear, Kathryn L. Shaw and Christopher T. Stanton, has a number. Using company data reporting daily output for technology-based services workers and recording which supervisor each worker is assigned each day, the authors estimate how much supervisors enhance productivity. Boss effects are large and significant, and bosses vary substantially in quality.

The headline comparison deserves to be framed and hung in every HR department. Replacing a boss in the lower 10% of quality with one in the upper 10% raises a team's total output by more than adding one worker to a nine-member team would. Under a separate normalization, the average boss is about 1.75 times as productive as the average worker. Workers assigned to better bosses are less likely to leave the firm, and poorer quality bosses are themselves less likely to remain. Comparative advantage plays a limited role: sorting good bosses to good workers only slightly increases total output.

The authors note that although economists assume managers matter, neither the mechanisms nor the size of the effects had been documented before. The paper fills that gap with granular measurement. Some boss effects persist; much of what bosses do is idiosyncratic and fleeting.

Enter artificial intelligence, which is quietly applying for the supervisor's job. AI tools increasingly monitor output, give feedback, schedule shifts and coach. The evidence shows good supervision raises output substantially. If AI can replicate or augment the practices of the best bosses, it could lift teams stuck with weaker managers and narrow the large quality gap the paper documents. The bottom-decile boss with an excellent AI assistant might start performing like a top-decile one, a productivity gain larger than a new hire.

Caution is warranted, though. Boss effects are large partly because bosses differ in ways that are hard to codify. If much of their contribution is fleeting and idiosyncratic, software may struggle to capture it. AI is likely better at tracking performance than at motivating people or defusing conflict. The gains from AI-assisted supervision will depend on which parts of good management the tools can actually support. The parts that are easiest to automate may be the parts that matter least.

Retention adds a twist. Workers with better bosses stay longer. If AI-driven management feels impersonal or intrusive, turnover could rise and swallow the productivity gains. Conversely, AI that frees supervisors from routine monitoring may let them spend more time on the human side of management that keeps good people. The same technology can drive workers out or keep them in, depending on design.

Measurement is the deeper opportunity. Linking daily output to supervisors yields precise estimates of supervisor value. As work is digitized, such data become common. Analytics can help firms identify which supervisors add value and why, supporting better selection, training and promotion. Because boss quality varies so much, simply putting the right people in supervisory roles could produce large productivity effects. Promotion decisions may be among the highest-return choices a firm makes.

Policy implications follow. Productivity debates obsess over technology and capital. This paper shows supervisor quality is a comparably important factor within firms. Supporting management training and the diffusion of management practices, including through AI tools that encode good practices, could raise productivity broadly.

The broader message: people who manage people create substantial value. A very good boss lifts team output by about as much as an additional team member. As AI transforms work, the real question is how technology and managers will combine. Tools that help weaker bosses perform like stronger ones could yield large gains; tools that erode the human elements of supervision could cost firms their best workers.

One question to put to executives: before buying AI to replace managers, test whether it can make the worst managers better. That may be where the money is.