📈 economics

Incentive Contracts and Peer Effects in the Workplace

This paper analyzes how firms should design optimal incentive contracts in collaborative teams by examining how performance-based pay cascades through peer networks, revealing that the allocation of incentives depends on the interplay between worker centrality, output risk, and production technology complementarity, with implications for organizational structure and workforce investment.

Original authors: Marc Claveria-Mayol, Pau Milán, Nicolás Oviedo-Dávila

Published 2026-04-17

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Marc Claveria-Mayol, Pau Milán, Nicolás Oviedo-Dávila

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine a company as a giant, living ecosystem where everyone's work affects everyone else's. If the person next to you is working hard, you might feel motivated to work harder too (a "peer effect"). But if they slack off, you might feel tempted to do the same.

This paper asks a big question: How should a boss decide who gets the biggest bonuses?

Should the boss just pay the most important person (the CEO) the most? Or should they look at the invisible web of connections between employees and pay the people who can "spark" the most energy in others?

Here is the breakdown of their findings, using simple analogies.

1. The "Spark Plug" vs. The "Fireworks"

The authors found that the best way to pay people depends on how risky the business is.

When things are stable (Low Risk): The boss should look for "Spark Plugs." These are people who might not be the most famous or highest-ranking, but they are in the middle of the network. If you give them a bonus, they light a fire under their neighbors, who light fires under their neighbors, creating a chain reaction of productivity.
- Analogy: Think of a campfire. You don't just throw a match on the biggest log; you throw it on the kindling in the middle. That small spark spreads the fire to the whole pile. Sometimes, paying the "middle" worker more is smarter than paying the "top" worker more.
When things are chaotic (High Risk): The boss should pay the "Fireworks." These are the people at the top of the pyramid or the most central figures. When the future is uncertain, the cost of making mistakes is high. The boss needs to focus incentives on the people who have the biggest direct impact on the final result, rather than relying on complex chain reactions that might fizzle out.
- Analogy: In a storm, you don't rely on a chain of dominoes; you just pay the person holding the umbrella to make sure the whole team stays dry.

2. The "Domino" vs. The "Swiss Watch" (Production Styles)

The paper also looks at how the company makes its product.

The Domino Effect (Substitutable Work): Imagine a team of writers. If one writer is slow, another can pick up the slack. Here, the boss pays based on who can influence the most people.
The Swiss Watch (Modular Work): Imagine building a rocket. If one tiny bolt fails, the whole rocket explodes. Every single part is essential. In this case, the boss has to pay attention to the "bottlenecks."
- Analogy: If you have a team of 100 engineers and 1 manager who has to approve every single line of code, that manager is the "bottleneck." Even if they aren't the most "central" in terms of who they talk to, they are the most critical. The paper shows that in these "Swiss Watch" companies, the boss might have to pay the manager massive bonuses compared to the engineers, creating a huge pay gap, simply because the manager is the single point of failure.

3. The "One-Size-Fits-All" Trap

Many companies have a rule: "All Junior Associates get the same bonus." The paper calls this Wage Benchmarking.

The Problem: Imagine you have two Junior Associates. One is a "Super Connector" who talks to everyone and helps the whole team. The other is a "Lone Wolf" who works in a silo. If you pay them the exact same bonus, you are wasting money.
- Analogy: It's like giving the same amount of fertilizer to a giant oak tree and a small sapling. The oak tree (the Super Connector) could produce twice as much fruit if you gave it more, but the sapling doesn't need it. By forcing them to have the same pay, the company loses potential profit.
The Finding: The more different the workers are in their "connectivity," the more money the company loses by treating them all the same. If everyone in a job title is equally connected, equal pay is fine. But if some are super-connected and others aren't, equal pay is expensive.

4. Building the Best Team (Organizational Design)

Finally, the paper asks: Should a boss invest in training individuals or in team-building exercises?

The Rule of Thumb: If the team is already somewhat connected (everyone knows at least one other person), investing in team-building is usually better than training individuals.
- Analogy: If you have a group of people who already know how to pass a ball, teaching them to pass better (team building) is more valuable than teaching one person to run faster (individual training). The "network effect" multiplies the value of the team.
The Tipping Point: Once the team is connected enough that a "giant component" forms (where everyone is linked, directly or indirectly), the boss should stop worrying about individual skills and start focusing entirely on how well the team works together.

Summary

This paper tells us that pay isn't just about a job title; it's about a job's position in the social web.

In a flat, risky organization, pay the top leaders.
In a stable, connected organization, pay the "connectors" in the middle.
In a high-stakes, modular organization (like rocket science), pay the bottlenecks (the managers) huge amounts.
And finally, don't treat everyone the same if their ability to influence others is different, or you will leave money on the table.

The authors essentially provide a "map" for bosses to navigate the complex terrain of who to pay, how much, and why, based on the invisible threads that tie their workforce together.

1. Problem Statement

The paper addresses the design of optimal wage contracts in firms where workers collaborate in teams and their productivity is influenced by peer effects (network externalities). Specifically, it investigates how firms should structure performance-based pay when:

Effort Costs are Interdependent: A worker's cost of exerting effort depends on the effort levels of their colleagues (modeled via a network $G$ ).
Information Asymmetry: The firm cannot observe individual effort (moral hazard) but can observe total output.
Risk: Workers are risk-averse, creating a trade-off between providing incentives and insuring workers against output risk.
Constraints: Firms may face institutional constraints (e.g., wage benchmarking) preventing personalized contracts, or face modular production technologies where output depends on the weakest link between teams.

The core question is: How does the structure of the peer network and the production technology dictate the optimal allocation of performance pay (incentives) across workers?

2. Methodology

The authors extend the standard moral hazard model with multiple agents (Holmstrom, 1982; Holmstrom & Milgrom, 1987) by integrating network games (Ballester, Calvó-Armengol, & Zenou, 2006).

Model Setup:
- Firm: Risk-neutral, maximizes expected profit.
- Workers: Risk-averse (CARA utility), choose effort $e_i$ .
- Production: Output $X = \sum e_i + \varepsilon$ (linear) or $X = \min(\sum_{k \in \text{module}} e_k) + \varepsilon$ (modular), where $\varepsilon \sim N(0, \sigma^2)$ .
- Peer Effects: Modeled through the cost function: $\psi_i(e) = \frac{1}{2}e_i^2 - \lambda e_i \sum_{j} g_{ij} e_j$ $ψ_{i} (e) = \frac{1}{2} e_{i}^{2} - λ e_{i} \sum_{j} g_{ij} e_{j}$ .
  - $\lambda > 0$ : Strategic complements (effort reduces marginal cost for others).
  - $\lambda < 0$ : Strategic substitutes (free-riding).
- Contracts: Linear wage schemes $\omega_i = \beta_i + \alpha_i X$ .
Solution Approach:
- The firm solves a constrained optimization problem maximizing expected profit subject to Individual Rationality (IR) and Incentive Compatibility (IC) constraints.
- Using the first-order approach, the IC constraints are replaced by workers' best-response functions, leading to a Nash equilibrium effort profile $e^* = (I - \lambda G)^{-1} \alpha$ .
- The authors utilize matrix algebra and spectral graph theory (eigenvalues, eigenvectors, Bonacich centrality) to derive closed-form solutions for optimal incentives $\alpha^*$ .

3. Key Contributions

The paper makes three primary theoretical contributions:

Integration of Network Games and Contract Theory: It derives closed-form optimal incentive rules that explicitly link wage structures to network topology (specifically Bonacich centrality) under moral hazard.
Dual Role of Incentives: It identifies that incentives serve a dual purpose: (1) Amplification (triggering network spillovers to increase output) and (2) Cost Reduction (making colleagues more tolerant of risk, thereby lowering the marginal cost of incentivizing them).
Extension to Modular Production and Coarse Contracts: It generalizes the baseline model to "O-ring" style modular production (where output is limited by the weakest team) and analyzes the profit loss when firms are forced to offer identical contracts to groups of workers (wage benchmarking).

4. Key Results

A. Linear Technology with Personalized Contracts (Baseline)

Optimal Incentive Rule: The optimal performance pay $\alpha^*$ for worker $i$ is a linear transformation of the vector of Bonacich centralities ( $C' \mathbf{1}$ ):
$\alpha^* = W C' \mathbf{1}$
where $C = (I - \lambda G)^{-1}$ and $W$ is a weighting matrix capturing cost interactions.
Mechanism:
- Amplification: Workers with high centrality (who influence many others directly and indirectly) generate larger marginal benefits.
- Cost Reduction: Incentivizing a worker reduces the marginal cost of incentivizing others if they share "common influence" (i.e., they both influence the same downstream targets).
Risk Sensitivity:
- Low Risk: Firms may incentivize workers who are excellent "cost reducers" (those who align well with others) even if they are not the most central.
- High Risk: The cost of risk dominates. Incentives become a monotonic transformation of centrality; the most central workers receive the steepest incentives.
Organizational Structure: In "flat" firms (shorter paths, wider spans of control), the wage profile is steeper, and the gap between top and bottom earners is larger compared to "tall" firms. This explains empirical trends in earnings inequality without assuming differences in managerial talent.

B. Modular Production (O-Ring Technology)

Setup: Output depends on the minimum performance across $K$ distinct modules.
Result: Incentives depend on modular shares ( $\mu_k$ $μ_{k}$ ), which act as shadow prices for each module's contribution.
- Small Modules: Workers in small modules (e.g., a single bottleneck manager) receive disproportionately high incentives because their failure halts the entire process.
- Every Worker Essential: If every worker is a separate module, incentives are allocated inversely to in-degree (number of incoming links). Workers with few incoming links face higher effort costs and are more prone to shirking, requiring higher incentives.
Wage Profiles: Modular production creates sharp discontinuities (jumps) in wage profiles between modules, unlike the smoother profiles in linear production.

C. Wage Benchmarking (Coarse Contracts)

Setup: Firms must offer the same contract to all workers in a specific occupational category (group $k$ ), despite heterogeneity in their network positions.
Result:
- Incentives are allocated based on the aggregated centrality of the group ( $TC' \mathbf{1}$ ).
- Profit Loss: The loss in profit due to the inability to discriminate is proportional to the within-group variance of Bonacich centrality.
- If all workers in a job category have identical centrality, wage benchmarking causes no profit loss. If centrality varies significantly within a group, the firm loses efficiency because it cannot tailor incentives to the specific network positions of individuals.

D. Organizational Design Implications

Profit Decomposition: Firm profits are linked to the spectral properties of the network. Profits are maximized when:
- The leading eigenvalue ( $\mu_1$ ) is high (dense, expansive networks).
- The variance of the leading eigenvector ( $Var(u_1)$ ) is low (centrality is evenly distributed).
Investment Thresholds: The paper establishes a threshold for investment strategies. If the expected number of connections per worker ($np$) in an Erdős-Rényi graph is $\geq 1$ , investing in team-building (strengthening peer effects) yields higher returns than investing in individual human capital.

5. Significance and Implications

Explaining Earnings Inequality: The model provides a theoretical rationale for the steepening of earnings profiles and increased inequality in flat organizations. It suggests that inequality arises from the structure of peer interactions and the optimal need to concentrate incentives on highly central nodes, rather than solely from differences in innate talent.
Policy and Management:
- Wage Transparency: Policies forcing wage compression (benchmarking) may reduce profits if they ignore the underlying network heterogeneity of workers within job titles.
- Organizational Design: Firms should consider the "effective size" of their workforce (based on network centrality distribution) rather than just headcount. Flattening hierarchies can be profitable if it optimizes the centrality distribution.
- AI and Automation: The framework offers a lens to study how AI might alter peer networks (e.g., shifting from modular to more integrated or vice versa) and necessitate new contract designs.
Empirical Predictions: The paper predicts that performance pay should track Bonacich centrality, and that profit losses from standardized pay scales should correlate with the variance of centrality within job categories.

In summary, the paper demonstrates that network structure is a fundamental determinant of optimal contract design, fundamentally altering how firms should allocate risk and incentives to maximize productivity.