It can be surprisingly hard to gauge how much effort an employee is putting in at work.
After all, even a lazy employee can manage to look busy whenever the boss walks by. Team projects allow slackers to ride on the coattails of their harder-working colleagues, and tasks that are more subjective in nature, such as providing high-quality customer service, further muddy the waters. Throw in a bit of noise—everyone can have good days and bad days—and some companies face a perfect storm in making fair assessments of their workers.
Yet a company’s success often hinges on ensuring that employees are, in fact, hard at work.
“The firm cannot see my exact actions, and therefore they have to motivate me to take the right actions,” says George Georgiadis, an associate professor of strategy at Kellogg.
One popular way to motivate employees, of course, is to tie compensation to performance. And there are a number of studies that explore the optimal way for firms to structure these performance pay contracts.
But critically, most simply take for granted that there is some observable behavior that is a rough proxy for effort and tie that behavior to compensation. They do not explore the actual process by which this observation occurs—thus glossing over just how difficult, and costly, it can be.
“In practice, firms have a huge array of pieces of information that they can collect,” says Georgiadis. “And it’s unclear which pieces of information are the most useful for assessing an employee’s actions. What this paper tries to do is look at the problem of choosing what pieces of information you want to collect and how to use that information to design a performance pay contract.”
The researchers find that, for all of the “information acquisition” scenarios they studied, the optimal compensation scheme for firms to offer their employees is also a very simple and popular one: the single-bonus contract. That’s when employees receive a base wage, but have the potential to earn a fixed bonus if their performance passes a certain threshold.
Collecting Evidence about Employees
To underscore the challenges facing organizations, Georgiadis lays out the following not-so-hypothetical scenario—his own role as a Kellogg professor.
“I choose how hard I want to work, how much diligence I put in all my work, how much effort I put into my teaching, how much effort I put in service to the university. And that generates a bunch of observables: it generates my teaching evaluations, how many papers I publish, how many papers I write, how many invitations I get to speak at assemblies and conferences, how much my peers respect me, how much time I spend at the office, how much coffee I drink, and so on,” he says.
Previous researchers have simply taken the “performance measure” on which a compensation contract is based as a given. But this is unrealistic—firms expend a lot of resources to identify more effective performance measures. So Georgiadis and colleague Balazs Szentes of the London School of Economics take a different approach.
The contract’s very simplicity is notable, because contracting models generally point to very complex optimal contracts.
They assume that companies can choose how many such observables to collect. The idea being that, as a company collects more and more information, it becomes increasingly confident that it knows whether an employee is working hard.
At least theoretically—if somewhat disturbingly—a company could collect an infinite amount of information about employees in order to be as certain as possible about their performance: installing cameras throughout the office to observe personal interactions, monitoring their emails, interviewing each of their collaborators, and noting who attends their presentations and what their body language says about how effective they are. But this would cost a company a great deal of time and money.
As such, the researchers assume that companies must balance the trade-off between accuracy and cost.
“At some point I can stop collecting information because I’m sure enough that you either worked or shirked,” says Georgiadis. A company that wants to be 55% sure that an employee was working hard will collect a certain amount of information, he explains; a company that wants to be 90% confident must collect substantially more information.
“More precise information is generally more expensive to the firm,” says Georgiadis.
The researchers model the process of collecting information about an employee’s effort using a mathematical model originally associated with the dispersion of physical particles over time. They assume that when an employee works, he or she generates information that is correlated with the effort that they put forth—information about, say, their sales generated or widgets manufactured or papers published.
The model assumes that, after observing the first piece of information, the company will continue to collect another piece of information with some probability, and so on.
“I observe one piece of information, and I make a first assessment about whether the employee worked hard. And then I can collect one more piece of information, and I update that assessment. And then I can choose one more piece of information, which allows me to update my assessment further, and so on. At some point, I’m going to say, ‘enough,’” says Georgiadis.
The more positive the information collected about an employee is, the closer they are to their bonus threshold and, thus, the less additional information is needed in order to push them over a threshold.
Georgiadis explains how this mathematical model plays out in the real world of employee evaluations.
“We walk into a room to deliberate on an employee and we’re not sure if that person deserves merit or not?” he explains. “So we start to acquire information. So we’ll say, ‘Let me read this paper. Let’s look into the evaluations. Let’s see what the students wrote in their comments.’”
But the longer the deliberations take, the costlier it is. Because the people deliberating could be using their time to do other work. So, at some point, the cost of deliberating further outweighs the value for more information.
Think about a jury, he says. “At some point, there’s enough consensus to go one way or another. And for different crimes, there’s a different standard of proof. One way to think about this is the firm can choose the standard of proof, in a sense. So you’ll acquire information, and at some point you stop.”
Consider a Fixed Bonus
By modeling this deliberation process, the researchers were able to shed light on the optimal way to structure contracts.
“The main result of the paper is that you end up with a contract that is remarkably simple,” says Georgiadis. Namely, the researchers find that a single bonus contract is the best way to provide employees with sufficiently high-powered incentives in the most cost-effective way possible.
“The contract pays a base wage, and if the information that you collect is sufficiently positive, then you pay a fixed bonus,” says Georgiadis.
The contract’s very simplicity is notable, he explains, because contracting models generally point to very complex optimal contracts.
“One of the big puzzles in contract theory is that typically, the optimal contract is very, very complex,” says Georgiadis. These complex contracts “depend on the assumptions of the model in very subtle ways. This is not a desirable trait, because as economists, we make assumptions of reality, because otherwise we cannot make any predictions. When you get a result which depends in such a sensitive way on the assumption that you’re making, then you’re not too comfortable.”
What makes a single-bonus contract effective? It allows employers to provide their workers with strong incentives while varying their pay as little as possible. And having relatively consistent pay is an important factor for keeping employees happy. When pay varies widely, workers cannot budget for their own expenses ahead of time, and economists have found that they then expect to be paid more generously to offset that uncertainty.
Popular and Useful, at Least for Some Roles
The research explains why this type of contract is so popular among employers. It also explains when it is and isn’t particularly useful.
For example, the benefits of this contract are only realized when employees don’t have a lot of opportunities to game the system, says Georgiadis, since a single-bonus contract is prone to gaming. Say an employee knows she will be evaluated based on her sales figures and she knows she has already sold enough to safely earn the bonus. She may slack off for the rest of the sales period. Or perhaps she knows that her early sales figures put her hopelessly off track for a bonus. This may also lead her to slack off.
Additionally, it may not be the best option for a very large workforce, when it is unlikely that the evidence about every employee’s performance will be deliberated individually, one at a time.
“But, if you’re evaluating a fairly small number of analysts at a hedge fund, then this is a much more reasonable thing. Same thing at a university: tenure decisions are done one by one,” Georgiadis says.