This paper examines delegation as an alternative to communication in organizations. It shows how a principal may give up control rights to foster efficient information use. The benefits of delegation depend on the congruence between agent and principal, the principal's risk-aversion, and environmental uncertainty. The paper identifies the role of intermediaries in limiting control loss while preserving communication. The paper argues that delegation helps avoid noisy communication and information loss due to differing objectives. While the principal could always rubber-stamp agent recommendations, delegation commits the principal to never reversing decisions. Aghion and Tirole (1997) show that delegation can improve incentives for information acquisition. However, the paper focuses on the information structure, assuming the agent is better informed, and examines how authority allocation affects information use and communication. Key to the analysis is that differences in objectives are systematic, not random. Managers may be short-term biased, status-quo biased, risk-averse, or empire builders. When the principal and agent systematically disagree, the principal will not rubber-stamp naive recommendations but correct for objective biases. Communication is thus strategic and noisy in equilibrium. The central trade-off is between information loss from keeping authority and control loss from delegation. A stylized model is developed where the principal screens projects differing on one dimension. The agent has superior information but different objectives. The agent can only make loose, unverifiable recommendations. Projects cannot be contracted upon, and resources controlled by the principal are needed for implementation. The principal faces a choice between delegating authority or consulting the agent. If she keeps authority, a strategic communication game occurs, with equilibria characterized by partitions of states of nature. The agent introduces noise by specifying which partition element the state belongs to. Crawford and Sobel (1982) show that communication is better when preferences are more congruent. The paper finds that delegation occurs when preferences are sufficiently congruent. The principal keeps control if an uninformed decision yields higher expected utility than a biased one. Delegation is more likely for larger biases when the principal faces more uncertainty. Delegation is also more likely when the principal is risk-averse. The paper also shows that intermediaries with objectives between the principal and agent can improve communication. For small biases, full delegation is optimal. For intermediate biases, delegation to an intermediary dominates full or no delegation. Direct communication between principal and agent never occurs with a uniform distribution. The paper concludes that changing the nature of information can generate more communication. While the agent can produce unverifiable reports, he may sometimes prove his information is true. In extreme cases, this leads to fully revealing equilibria where the principal never delegates. The paper also discusses the implications of incomplete contracting and strategic communication. It distinguishes itself from most incomplete contracting literature by analyzing an adverse selection problem without moral hazard. It also departs from traditional team theory by considering information distortion incentives. The paper shows that with a uniform distribution, the principal optimally delegates authority whenThis paper examines delegation as an alternative to communication in organizations. It shows how a principal may give up control rights to foster efficient information use. The benefits of delegation depend on the congruence between agent and principal, the principal's risk-aversion, and environmental uncertainty. The paper identifies the role of intermediaries in limiting control loss while preserving communication. The paper argues that delegation helps avoid noisy communication and information loss due to differing objectives. While the principal could always rubber-stamp agent recommendations, delegation commits the principal to never reversing decisions. Aghion and Tirole (1997) show that delegation can improve incentives for information acquisition. However, the paper focuses on the information structure, assuming the agent is better informed, and examines how authority allocation affects information use and communication. Key to the analysis is that differences in objectives are systematic, not random. Managers may be short-term biased, status-quo biased, risk-averse, or empire builders. When the principal and agent systematically disagree, the principal will not rubber-stamp naive recommendations but correct for objective biases. Communication is thus strategic and noisy in equilibrium. The central trade-off is between information loss from keeping authority and control loss from delegation. A stylized model is developed where the principal screens projects differing on one dimension. The agent has superior information but different objectives. The agent can only make loose, unverifiable recommendations. Projects cannot be contracted upon, and resources controlled by the principal are needed for implementation. The principal faces a choice between delegating authority or consulting the agent. If she keeps authority, a strategic communication game occurs, with equilibria characterized by partitions of states of nature. The agent introduces noise by specifying which partition element the state belongs to. Crawford and Sobel (1982) show that communication is better when preferences are more congruent. The paper finds that delegation occurs when preferences are sufficiently congruent. The principal keeps control if an uninformed decision yields higher expected utility than a biased one. Delegation is more likely for larger biases when the principal faces more uncertainty. Delegation is also more likely when the principal is risk-averse. The paper also shows that intermediaries with objectives between the principal and agent can improve communication. For small biases, full delegation is optimal. For intermediate biases, delegation to an intermediary dominates full or no delegation. Direct communication between principal and agent never occurs with a uniform distribution. The paper concludes that changing the nature of information can generate more communication. While the agent can produce unverifiable reports, he may sometimes prove his information is true. In extreme cases, this leads to fully revealing equilibria where the principal never delegates. The paper also discusses the implications of incomplete contracting and strategic communication. It distinguishes itself from most incomplete contracting literature by analyzing an adverse selection problem without moral hazard. It also departs from traditional team theory by considering information distortion incentives. The paper shows that with a uniform distribution, the principal optimally delegates authority when