Essays on Information in Politics and Social Decisions

Abstract

This thesis consists of two parts. The first part consists of two chapters. The second part has one chapter.<br />

How does the information of citizens shape the democratic process? This is the question asked in the first part. Each chapter of the first part proposes a specific economic model and analyzes a particular dimension of this question.<br /> Chapter 1: "Persuasion and Information Aggregation in Elections", which is joint work with Stephan Lauermann, analyzes the scope of persuasion of voters by interested third parties. How manipulable are elections by third parties who hold private information and can strategically release relevant information to affect voters' behavior. Examples are numerous: in a shareholder vote, the management may strategically provide information regarding a potential merger through presentations and conversations; similarly, lobbyists provide selected information to legislators to influence their vote. We show that a manipulator can ensure that a majority of a large electorate supports his favorite policy simply by releasing some additional information to the voters. Moreover, persuasion does not require detailed knowledge about the citizens, the precise distribution of their preferences or their previous information. With very little knowledge about these, a third party manipulates by sending out private signals randomly to the citizens. A numerical example shows that persuasion is effective in elections with as few as $ voters.<br />

Chapter 2: "Voter Attention and Distributive Politics" studies how citizens paying attention to politics (or not) affects election outcomes, social welfare and its distribution. Demographic groups care differently much about different issues: e.g. older people care more about healthcare issues, while changes in education policy are more relevant to citizens with children. People that care more, pay more attention. We show that this attention effect shifts election outcomes into a direction that improves the overall welfare of a society. Elections often lead to outcomes that maximize a weighted welfare rule: the implicit decision weight of each voter is higher when he cares more about the issue voted on; however, less so when information is more cheap. In general, the decision weight is proportional to how informed the voter is. These results are important as they stress that information is a critical determinant of democratic participation. They imply that uninformed voters have effectively almost no voting power, and that elections are susceptible to third-party manipulation of voter information.<br />

Taken together, the first two chapters shed light upon two general topics. First, political actors seek to influence the citizens’ opinions and behavior through propaganda, by the diverting of attention of the citizens, or by spreading false information, even more so in the digital age. How manipulable elections are through such informational tools? The first two chapters point out that the scope of manipulation is rather large. These insights may serve as a starting point for studying related questions, that, I believe, are highly relevant and deserve further analysis. The second broader topic this thesis touches upon is how the incentives of individuals shape their political beliefs. The second chapter points out that the size of the incentives matters since it affects the precision of people’s beliefs, and thereby their implicit decision weight in elections. Studying the interaction of incentives and political beliefs has a positive motivation: we have a very limited understanding on how people form political beliefs, let alone why beliefs differ so much. But it also has a normative motivation since it informs about the consequences of economic interventions that shape the incentives.<br />

The second part of the thesis is devoted to the social dimension of incentives and their role for belief formation, taking a step back from the political environment, however. Much empirical evidence has shown that many people depart from maximizing their self-interest, if doing so benefits others. For example, people donate to charity (e.g. DellaVigna, List, and Malmendier, 2012), pay postage to return misdirected letters (e.g. Franzen and Pointner, 2013), and share wealth with strangers in laboratory dictator games (e.g. Forsythe, Horowitz, Savin, and Sefton, 1994). This means that these individuals' decisions are not sorely governed by their material desires, but also by "social motives". The recent research on motivated reasoning shows that many people deviate from complete egoism in order to `feel moral'(for a review, see Gino, Norton, and Weber, 2016). It argues that, in social decisions, individuals can behave selfishly without a guilty conscience if they can make themselves believe that the selfish decision harms no others (for a review, see Gino, Norton, and Weber, 2016).<br /> In Chapter 3: "Motivated Information Acquisition in Social Decisions", which is joint work with Si Chen, we ask: when do people stop acquiring information before a decision where pursuing one`s own material benefits might harm others. Examples include medical examinations that help a doctor to decide between treatments with different profits, media consumption of voters before casting a ballot on ethically controversial policies, or consumers choosing to get informed about potential ethical issues of the products they would like to buy.<br /> Using a laboratory experiment, we provide causal evidence that having a selfishly preferred option makes individuals more likely to continue their inquiry for information when the information received up to that point suggests that the selfish behavior harms others. In contrast, when the information received up to that point suggests that being selfish harms nobody, individuals are more likely to stop acquiring information. In some sense, individuals are fishing for excuses to behave selfishly until they find them.<br /> We also provide a theoretical model, drawing on the Bayesian Persuasion literature Kamenica and Gentzkow, 2011). The model shows that the information acquisition strategy documented in our experiment can be optimal for a Bayesian agent who values the belief of herself not harming others but attempts to persuade herself to behave self-interestedly.<br /> Further, we empirically and theoretically provide results regarding the externalities that might not be obvious at first sight. Although one might think that strategic information acquisition must lead to more negative externalities when motivated by selfish interests, our model shows that also the reverse can happen: for some agent types, motivated information acquisition improves the welfare of the others affected by the decision. This counter-intuitive result rests on the observation that an "unmotivated" agent faces a moral hazard problem: when unmotivated, some agent types acquire only a small amount of information due to, for example, the satisficing behavior (Simon, 1955). The agent's selfish preference for one option over the other can mitigate this moral hazard problem by causing her to acquire more information in order to make sure that she chooses her least-preferred option only when certain that it is harmless to others. This result implies that delegating information acquisition to a neutral investigator might lower the welfare of the others affected by the decision.