European Commission logo
English English
CORDIS - EU research results

Information Aggregation in Elections

Periodic Reporting for period 4 - InfoAggregation (Information Aggregation in Elections)

Reporting period: 2020-01-01 to 2020-06-30

Elections are the foundation for democratic decision making. This research program examines the effects of biased and privately informed entities—election organizers—on the ability of elections to aggregate information: Existing theory demonstrates that large electorates can reach correct decisions by aggregating information dispersed among many voters. However, existing theory does not account for the ubiquitous presence of biased organizers who intend to affect the election outcome. Examples of biased organizers may include a CEO holding a shareholder vote, a regional government holding a referendum, and political parties in general elections. This project develops and analyzes new models of voting that account for the effects of biased organizers on information aggregation. I am interested in whether elections are still effective instruments of collective choice in the presence of “biased organizers.” If they are not, then how can we design elections in a robust way? I am particularly interested in the possibilities of manipulating elections by manipulating the number of voters and the information they have (“persuasion”). A particularly salient application is shareholder voting as an element of corporate control.
The research conducted in the project has lead to a better understanding of these questions and their answers, as described below.
The focus at the beginning of this project was my work on information aggregation in elections in which the number of voters is state-dependent. We use this model to study the possibility of a biased organizer to manipulate elections by deliberately choosing the number of voters depending on the underlying state. In collaboration with my co-author, Mehmet Ekmekci, I have extended and revised preliminary work on the topic. We have studied, in particular, elections in which the number of voters is Poisson distributed, building on previous work by Roger Myerson. We have published an article in the Review of Economics Studies, "Manipulated Electorates and Information Aggregation" (2020) and a working paper, "Information Aggregation in Poisson-Elections". This part is related to my work on information aggregation in auctions in which the number of bidders is uncertain and state-dependent. We have also worked with Kailin Chen on "Fishing for a Veto." There, a biased organizer samples voters sequentially in order to find a veto against a project that the organizer wants to stop. Somewhat surprisingly, the actions of the organizer enhance, rather than dampen, information aggregation.

Over time, I also started working on information aggregation in the presence of an informed adversarial sender. In collaboration with my co-author and student, Carl Heese, I have developed a theory of persuasion in elections in which voters are heterogeneous and have access to other information. We have published a working paper, "Persuasion and Information Aggregation in Elections." In this working paper we show that many of the assumptions in Bayesian Persuasion can actually be substantially weakened, including the knowledge of the sender and its commitment power. Moreover, Carl Heese has studied further the process of information acquisition and its effect of information aggregation in his working paper "Voter Attention and Distributive Politics."

Finally, I have started to study non-binding elections more recently in collaboration with Mehmet Ekmekci. Such elections are particularly relevant in the context of shareholder voting because shareholder votes are typically not binding for the management. Thus, we study whether elections still aggregate information even if the decision maker does not commit to a particular outcome as a function of the election outcome. We are especially interested in the role of participation costs as a screening instrument. Other important applications are polls, petitions, and protests. We have published a working paper on "Informal Elections with Dispersed Information." Our main finding is that with costly participation such informal political processes enable significantly more information transmission than what was previously thought possible based on the analysis of models with purely "cheap talk." Moreover, we show that the economic forces determining information transmission are quite different, rather than the bias of the receiver and the inference from being pivotal, standard economic insights regarding the provision of public goods take center stage, leading to substitution and complementarity effects.

Research from this project has been widely disseminated. The research papers are all publicly posted and we distributed the papers to many researchers in the area. Moreover, research from this project has been presented in over 100 seminars around the world, in addition to being presented at many workshops and conferences.
All three subprojects extend existing theory beyond the current state of the art. The first considers state-dependent electorates and its endogenous emergence. The second considers persuasion when receivers have private information about their own preferences and the relevant state of the world. The third considers informal elections and the role of participation costs. In each case, my co-authors and I have constructed and solved by a new class of models of elections that provide us with a better understanding of the effectiveness of elections in aggregating dispersed information in a variety of contexts.