Working Papers

This paper investigates strategic information transmission within a hierarchical framework, where information flows through a chain of agents to a decision-maker whose actions affect the payoffs of all agents. Each agent may conceal part or all of the information they receive. It is shown that the analysis can be restricted to simple equilibria where only the initial agent conceals information. This simplification permits hierarchical communication to be modeled as a direct interaction between the initial sender and the decision-maker, subject to recursively defined incentive compatibility constraints imposed by intermediaries. In the binary-action case, irrespective of the number of intermediaries, no more than five agents influence the information conveyed to the decision-maker. The findings highlight the importance of appointing a pivotal vice president to mitigate inefficiencies and enhance the decision-maker's payoff, particularly by selecting a like-minded but stubborn vice president.


We study the classic principal-agent model when the signal observed by the principal is chosen by the agent. We fully characterize the optimal information structure from an agent’s perspective in a general moral hazard setting. Unlike standard information design, only the support of the distribution of beliefs is relevant to the agent. We show that the problem can be mapped into a geometrical game between the principal and the agent in the space of likelihood ratios. We use this representation result to show that coarse contracts are sufficient: The agent can achieve her best with binary signals. Additionally, we characterize conditions under which the agent is able to extract all the surplus and implement the efficient allocation under full information.


Natural gas is a large and growing share of U.S. energy consumption; based on data from U.S. Energy Information Administration (EIA), in 2021, it accounted for 32% of the primary energy consumption, 38% of the electricity generation, and about half of home heating in the U.S. The transportation of natural gas from producer to consumer, however, is heavily regulated and relatively unexamined by economists. Using an extensive panel of daily data on natural gas flows through interstate pipelines from 2005 to 2016, we take the first steps of analyzing this networked market, We first perform demand estimation to derive price-elasticity of demand in different sectors: residential, commercial, industrial, and electric utility. Our results suggest that while demand in all these sectors is relatively inelastic with respect to the average price, electric utility is the most elastic sector and industrial sector is the most inelastic one. We then investigate one of the largest mergers among natural gas interstate pipelines. Our results suggest that this merger had a significant effect on natural gas transportation prices even though the two pipelines were not in the same physical market. Furthermore, we study the role of storage in the natural gas market and quantify the effect of temperature, storage, and pipeline congestion on natural gas prices. Our results suggest that storage and temperature are the main factors explaining the price shocks in the natural gas market where abundant storage dampens the effect of cold winter weather on natural gas prices. Finally, we explore the network effect in the natural gas market, evaluating the effect of changes in the temperature of one geographical region on all other regions. Our results suggest that a change or shock in one geographical region can have a significant effect on prices even in the farthest regions. This observation can be rationalized by the change of natural gas route in the pipeline network from low-demand regions to high-demand regions.


We consider a market-based spectrum sharing system where a band manager allocates K spectrum channels to competing users in each round and settles payments by a usage-based fee. This setting creates two challenges: uncertainty because usages by users are unknown and time varying, and selfishness because users can under-report consumption, which corrupts the learning signal. We formulate this setting as strategic combinatorial bandits problem and propose an algorithm that integrates learning and incentive design. The algorithm combines elimination-based exploration, K-price-referenced exploitation, and a user-report-dependent terminal bonus. We prove that truthful reporting is a dynamic dominant strategy for every user and on the truthful path the system reduces to a standard top-K combinatorial bandit that observes true feedback. The mechanism achieves sublinear regret with respect to an incentive-compatible benchmark. Extensive simulations show robust performance under strategic untruthful reporting, fast identification of top users, and operator revenue close to the benchmark, while avoiding the allocation instability without an incentive design.


In hierarchical Bayesian persuasion, a simple sufficient condition for the equilibrium level of information revealed to the decision maker is that no agent prefers a less informative signal. Although this condition is not necessary, it provides a useful benchmark. Under general conditions, we show that equilibrium outcomes can be well approximated using this principle.


We study indicator design in a principal–agent model with limited liability and continuous effort and performance measures. An agent chooses how performance is aggregated into a contractible signal before the principal offers an incentive scheme. We show that, under general regularity conditions, optimal information structures take the form of binary, threshold-based signals with one or two thresholds. Using a first-order approach, we characterize when such coarse, continuous indicators are sufficient to implement target effort levels and, in some cases, to achieve full surplus extraction. Our results provide a foundation for understanding why simple, rule-based performance measures arise endogenously in continuous environments and clarify the informational limits of efficient incentive provision.


Work in Progress