The paper by Green and Stokey (1983) compares tournaments, reward structures based on rank order, with individual contracts in a model where a risk-neutral principal employs many risk-averse agents. Each agent's output is a stochastic function of their effort level plus a common additive shock. The principal observes only the agents' output levels. The key findings are: 1. **Absence of Common Shock**: If there is no common shock, optimal independent contracts dominate optimal tournaments. This is because the common shock introduces additional noise into the payoff function, which is costly for the principal. 2. **Diffuse Common Shock**: If the distribution of the common shock is sufficiently diffuse, optimal tournaments dominate optimal independent contracts. This is because the common shock reduces the randomness in any agent's compensation by filtering out the common shock term. 3. **Large Number of Agents**: For a sufficiently large number of agents, the optimal tournament can perform as well as optimal independent contracts. In this case, the rank order of an agent's observed output is a very accurate estimator of their output net of the common shock, making tournaments efficient. The paper also discusses the robustness of tournaments against lack of information or agreement about the distribution of the common shock. It concludes that while tournaments may not be optimal contracts, they can be effective in large groups where the inefficiency due to information loss is negligible.The paper by Green and Stokey (1983) compares tournaments, reward structures based on rank order, with individual contracts in a model where a risk-neutral principal employs many risk-averse agents. Each agent's output is a stochastic function of their effort level plus a common additive shock. The principal observes only the agents' output levels. The key findings are: 1. **Absence of Common Shock**: If there is no common shock, optimal independent contracts dominate optimal tournaments. This is because the common shock introduces additional noise into the payoff function, which is costly for the principal. 2. **Diffuse Common Shock**: If the distribution of the common shock is sufficiently diffuse, optimal tournaments dominate optimal independent contracts. This is because the common shock reduces the randomness in any agent's compensation by filtering out the common shock term. 3. **Large Number of Agents**: For a sufficiently large number of agents, the optimal tournament can perform as well as optimal independent contracts. In this case, the rank order of an agent's observed output is a very accurate estimator of their output net of the common shock, making tournaments efficient. The paper also discusses the robustness of tournaments against lack of information or agreement about the distribution of the common shock. It concludes that while tournaments may not be optimal contracts, they can be effective in large groups where the inefficiency due to information loss is negligible.