Pay for performance isn’t easy to execute, and is more suited to stable industries and jobs where success can be easily quantified, Nobel laureate Bengt Holmström said in Stockholm Dec. 8 in a lecture as part of Nobel Prize week.
Holmström, who is a professor at both MIT Sloan and the MIT Department of Economics, shared the 2016 Nobel Memorial Prize in Economic Sciences with Harvard University professor Oliver Hart. His lecture was titled “Pay for Performance and Beyond.”
Holmström’s research dates to the 1970s. He has conducted work on contract structure, employee contributions relative to their compensation, outside ownership of firms, and how to structure compensation to incentivize performance across all tasks, not just the most quantifiable.
Unexpected origins for an economist
Holmström explained that his work has focused primarily on motivating workers, focusing on contracts, though he never planned to be an economist at all. He started out as an applied mathematician in Finland, where he was hired by what he called a “progressive firm” to incorporate a corporate planning model.
“It was a wonderful way to learn how a big company works. I worked with a CFO, a tough guy who took a liking to me because I said what I thought,” he said. There, he worked on incenting models and researched how to incentivize people properly.
“I realized incentives were an interest of economists, too,” he said.
Solving the principal-agent problem
Holmström cautioned that “pay for performance” is complicated for a variety of reasons. He distilled this as the “principal-agent problem.” Broadly speaking, principals are employers; agents are employees. They often have different motives. Contracts can be difficult because their preferences aren’t aligned and agent performance is imperfectly measured, he said. Contracted pay is then based on measured outcome. But his work has demonstrated that when an employee’s performance pay focuses solely on short-term cash flow, for instance, it could backfire and risk the long-term health of a company.
Understanding the informativeness principle
Holmström discussed his informativeness principle, which weighs risks versus incentives. He explained that an ideal contract should link payment to all outcomes that can “potentially provide information about actions that have been taken.”
To explain this, the Nobel committee uses the example [PDF] of the manager whose actions influence her company’s share price but not share prices of other firms. Should a manager’s pay depend only on her firm’s share price? No: Because share prices reflect other factors in the economy outside the manager’s control, only linking compensation to the firm’s share price will reward the manager for good luck but punish her for bad luck. It’s preferable to link the manager’s pay to her firm’s share price relative to those of similar firms, the committee explains, and not just on the share price she can work to control.
“The old accounting principle was that pay should only depend on variables an agent can control,” Holmström said. Meanwhile, the harder it is to measure a manager’s effort—and it often is, he said—the less pay should be based on performance.
“The level of effort is not the simplest case to study; it’s about the most complicated. The behavior of the model is so erratic but seems simplistic,” he said.
High-risk industries do better with fixed salaries; steadier jobs with more transparency can embrace additional performance measures, he said.
The trouble with multitasking
He went on to discuss multitasking, which makes it harder for employers to monitor output. To dissuade an employee from concentrating on tasks where performance is easier to measure, weaker incentives are better. For instance, if teachers’ salaries hinge mainly on quantifiable results, like test scores, those teachers might spend too little time teaching more qualitative skills, like creativity. A fixed salary, independent of performance measures, would lead to a more balanced effort.
“Some tasks are easy to measure; others are hard, and you have to consider even though you can easily observe one, quantity is easy to measure, quality is harder. These are the dangers of multitasking,” Holmström said.
He illustrated the pitfalls of over-incentivizing by pointing to “scandals” like Wells Fargo, where employees manufactured fake bank accounts to get bonuses, because they were paid for the number of accounts created. He also mentioned the Gulf oil spill, where excessive output incentives may have led to safety compromises.
“It’s not that [employees] are evil, by the way. It’s just that there’s so much expected of them, and they feel like they have to manufacture somehow in order to hold onto their jobs,” he said.
“Don’t incentivize competing tasks,” Holmström said. To properly provide incentives, pay more for an important task or pay less for competing tasks. When a task is hard to measure, low or no incentives might be best.
Ultimately, he said, “Employees want to be appreciated.” He discussed alternative incentives, like job design to match the needs of an employee, a non-pay-for-performance approach.
“There’s a lot beyond pay for performance,” Holmström said.