@article{169b226847074f8b80649fc9ec3ec79c,
title = "Deliberating collective decisions",
abstract = "We present a dynamic model of sequential information acquisition by a heterogeneous committee. At each date, agents decide whether to vote to adopt one of two alternatives or continue to collect more information. The process stops when a qualified majority vote for an alternative. Three main insights emerge from our analysis and are consistent with an array of stylized facts regarding committee decision making. First, majority rule is more vulnerable than super-majority rules to the disproportionate influence of impatient committee members. Second, more diverse preferences, more patient members, or more unanimous decision voting rules lead to lengthier deliberation and more accurate decisions. Finally, balanced committees unanimously prefer to delegate deliberation power to a moderate chairman rather than be governed by a rule such as unanimity.",
keywords = "Collective learning, Optimal stopping, Sequential likelihood ratio test, Swing voters",
author = "Jimmy Chan and Alessandro Lizzeri and Wing Suen and Leeat Yariv",
note = "Funding Information: Next, we show that such (gˆ,Gˆ ) is an equilibrium outcome of the two-stage decision process with any decision rule kd.Wefirstnotethatitisanequilibriumforeachagentitovoteforαifθ≥viandtovoteforβotherwise,becauseα is preferred to flipping a coin (which is in turn preferred to β) whenever θ ≥vi, while β is preferred to flipping a coin (which is in turn preferred to α) whenever θ <vi. Given such equilibrium second-stage strategy, and given the fact that Gˆ >v2m−1, α will be adopted with unanimous support for any kd if the belief reaches the upper cutoff. Likewise, β will be adopted with unanimous support for any kd if the belief reaches the lower cutoff. In the first stage, suppose each agent i adopts the strategy (gi,Gi)=(φi(Gˆ ),Φi(gˆ)).As ri goes to 0, we have Gi>v2m−1 for every i. Given this first-stage strategy, the span of control IG(σ−i) for any agent i over the upper cutoff is above v2m−1. Similarly, the span of control for any agent over the lower cutoff is below v1. For any decision rule kd which is nonunanimous, it is not feasible for any agent to unilaterally deviate to obtain an indecisive outcome, given the strategy profile of other agents. Furthermore, it is not profitable to unilaterally deviate to delay or hasten the adoption of either alternative, because (gˆ,Gˆ ) solves the constrained maximization problem 1 for the one-stage problem. If the decision rule kd is unanimous, it is feasible for an agent to deviate by changing his thresholds in the first stage and withholding his support for an alternative in the second stage to obtain an indecisive outcome. But doing so is worse than just changing the thresholds in the first stage and voting sincerely in the second stage, because an indecisive outcome is worse than α foranyθ∈IG(σ−i),andisworsethanβforanyθ∈Ig(σ−i).Sincethelatterdeviationisunprofitable,deviationtoforce flipping a coin cannot be profitable for any belief within an agent{\textquoteright}s span of control. ‖ Acknowledgments. This article is the result of a consolidation of {\textquoteleft}Does Majority Rule Produce Hasty Decisions?{\textquoteright} by Chan and Suen, and {\textquoteleft}Sequential Deliberation{\textquoteright} by Lizzeri and Yariv. Ben Chiao, Jacques Cremer, John Morgan, Nicola Persico, Debraj Ray, Nikita Roketskiy, and Erik Snowberg provided very helpful conversations and feedback. We thank the editor, Marco Ottaviani, and six anonymous referees for many useful comments and suggestions. We gratefully acknowledge financial support from the Gordon and Betty Moore Foundation (grant 1158), the National Natural Science Foundation of China (project no. 71171125), the National Science Foundation (SES 0551014), the Shanghai Dongfang Xuezhe Program, and the Research Grants Council of Hong Kong (project no. HKU753911). Publisher Copyright: {\textcopyright} The Author 2017. Published by Oxford University Press on behalf of The Review of Economic Studies Limited.",
year = "2018",
month = apr,
day = "1",
doi = "10.1093/restud/rdx028",
language = "English (US)",
volume = "85",
pages = "929--963",
journal = "Review of Economic Studies",
issn = "0034-6527",
publisher = "Oxford University Press",
number = "2",
}