Endogenous Surplus Retention in Peer-Based Team Incentives

In the world of work, getting a group of people to collaborate effectively is a persistent puzzle. When a team produces something valuable, such as a software project or a creative campaign, the total output is often greater than the sum of individual efforts. However, dividing that reward creates a dilemma. If the team splits the money equally, individuals may slack off, hoping to enjoy the group's success without doing their fair share. This is known as free-riding. To fix this, many organizations ask team members to evaluate one another, hoping that peer pressure will encourage hard work. But this solution introduces a new problem: people can manipulate the system. A worker might give low scores to rivals to make themselves look better, or friends might trade high scores to boost each other's pay. The result is often a noisy, unfair signal that fails to reward genuine effort.

Researchers have long sought a way to break this cycle. A classic economic insight suggests that to eliminate free-riding, someone outside the group must keep the leftover money, rather than giving it all to the workers. This "budget breaker" ensures that the total payout is less than the total value created, which theoretically aligns individual incentives with the group's success. Yet, in practice, this leftover amount is usually a fixed number decided by a manager, disconnected from the team's actual behavior. The question remains: can the team itself determine how much money is kept, in a way that is fair, resistant to manipulation, and linked to how hard they actually work?

A new study proposes a specific mechanism to solve this, tested across real-world data from universities in China, New Zealand, and Italy. The researchers designed a system where a team's pay is determined by a fixed denominator, a concept that changes how the reward pool is calculated. In a standard peer-review system, if you want to increase your share of the money, you might try to lower your teammates' scores, because the total pot is divided by the sum of all scores. In this new system, however, the denominator is a theoretical maximum number of votes the team could possibly cast, not the actual number of votes received. This means your share of the money depends entirely on the votes you receive, not on how you vote for others. You cannot manipulate your own paycheck by giving bad reviews to your colleagues.

Because the denominator is fixed and often higher than the actual votes cast, the total shares given to the team members do not add up to the full amount of money available. A portion of the revenue is automatically retained by the organization. Crucially, the size of this retained portion is not a fixed rule set by a manager. Instead, it is an outcome of the team's own voting intensity. If the team is generous and votes for almost everyone, very little money is kept back. If they are stingy or selective, a larger share of the money remains with the firm. The researchers call this "endogenous surplus retention," meaning the leftover money is generated by the team's own behavior.

The study analyzed three distinct datasets involving hundreds of students and various grading schemes, including simple yes-or-no votes, coin budgets where students had a limited number of points to spend, and detailed rating scales. The findings confirmed that the mechanism works exactly as the math predicts. In every group studied, the amount of money retained by the organization was an interior value, meaning it was neither zero nor the entire pot, but a specific percentage that varied based on the grading system used. For instance, when students used a generous rating scale, the organization kept about 13 percent of the revenue. When they used a restrictive coin budget, the retention rose to about 50 percent. With simple yes-or-no votes, it climbed to roughly 70 percent. The data showed a precise, predictable relationship: the design of the grading scheme directly determined the size of the retained fund.

Furthermore, the researchers found that the votes students received were a reliable signal of their actual performance. By comparing peer votes to independent grades given by instructors, the study showed that while the peer evaluations were not perfect, they carried a clear, positive signal. Students who received more votes from their peers tended to have higher instructor grades. This suggests that the mechanism successfully separates the ranking of individuals from the size of the total payout. The relative standing of a worker is decided by the votes they receive, while the total amount of money available to be shared is decided by how intensely the team evaluates one another.

The study also explored the trade-offs involved in this design. Using data to simulate different scenarios, the researchers identified a "frontier" where a manager must choose between keeping more money back and maintaining high accuracy in rankings. If a system is too stingy, keeping a large portion of the revenue, the grading scale becomes so harsh that it loses its ability to distinguish between good and bad work. Conversely, if the system is too generous, the retained fund becomes too small to provide strong incentives. The data suggests an optimal middle ground exists. For the teams studied, a retention level of around 13 to 14 percent appeared to offer a strong balance, preserving enough money to motivate the group while keeping the peer evaluations accurate enough to identify top performers.

Importantly, the researchers were careful to note what their work did not prove. While the structural design of the system is sound and the mathematical relationships hold true, the study does not definitively prove that real people will always act honestly without additional rules. The mechanism removes the ability to manipulate the system by lowering others' scores, but it does not automatically create an incentive to vote truthfully. To address this, the author proposes a governance layer that includes rewards for honest voting and costs for low-quality participation, though the effectiveness of these behavioral rules requires further testing. The core finding remains that the system successfully turns the retained surplus from a fixed, arbitrary number into a measurable, schedulable outcome of the team's own engagement.

This approach offers a new way to think about team compensation. It resolves the tension between free-riding and manipulation by making the budget breaker a natural consequence of the evaluation process itself. The organization does not need to guess how much to keep; the team's own voting patterns reveal the answer. By choosing a specific grading scheme, a manager can effectively select the level of retention they desire, knowing exactly how that choice will impact the accuracy of the peer reviews. The study establishes that this is not just a theoretical possibility but a design tool that works in practice, verified across different cultures and educational settings. It provides a clear path for organizations to align individual incentives with group success, using a mechanism that is both mathematically robust and empirically grounded.