In political contexts, information plays a key role in shaping the decisions that politicians, voters, and other agents (e.g., bureaucrats, regulators) make. For instance, a citizen’s vote in an election may be determined by information she receives about the candidates in the fray, while a politician’s vote on a piece of legislation may be influenced by information about the bill. Oftentimes, this information comes from sources that have a strategic incentive to manipulate information, and agents recognize this. Ads aired on television are paid for by campaigns and Super PAC’s, news agencies have ideological biases, and information about the effects of legislation are often produced by interest groups with the largest stakes in the bill. Clearly, these interests create strategic distortion in the information we all consume. This leads to an important question: how can information producers release information to best suit their interests? And, how should consumers of information—recognizing the objectives of the information producers—interpret what they consume? Moreover, what are the social consequences of this strategic interplay? This course will cover influential theoretical models of information design relevant to a variety of different political settings, that seek to answer precisely these questions.

Unit 1: Bayesian Persuasion We review first the canonical model of persuasion with commitment, or “Bayesian persuasion” (Kamenica and Gentzkow (2011)). We then focus attention on Alonso and Camara (2016), a variant of this canonical model in which an information designer (politician) can design a policy experiment (public signal) that is observed by a multitude of voters, each with different preferences over two alternative policies. The politician’s objective is to design an experiment in order to persuade a critical mass (e.g., majority) of voters to choose her preferred alternative over the other. Taking off from where Kamenica and Gentzkow end their paper, Alonso and Camara thus ask how an information designer would design a single signal to cater to a heterogenous population of receivers. The model has implications for what we observe in advertisement campaigns for ballot proposals or referenda, and for information that an executive may provide to the legislative branch in favor of a piece of legislation;

Unit 2: Adversarial Information Design We next study models at the cutting-edge of a new adversarial information design literature. In many cases, agents’ payoffs depend not only on their own actions and a state of the world (as in Unit 1), but also on the actions of other players. For instance, in collective action problems, individuals take costly action with the intent of realizing a valuable goal, achievable only if a sufficient fraction of the population does the same. A classic example from political economy is the ‘regime change’ game, in which enough people from the population must protest the regime to overthrow it. In such an environment, agents provided with information by the regime can support a large set of beliefs—over not only the state of the world, but the beliefs of other players, and, whether they will act, or not. Whether there are sufficiently many people willing to protest to bring about a change is, of course, crucial in dictating whether an agent would take it upon herself to protest against the regime (a costly action). Adversarial information design asks, “what is the optimal signal structure to the population the regime should design, assuming the worst case scenario for itself (the best case scenario for protesters)?” Papers studied in this unit include Li, et al (2023) and Basak and Zhou (2020);

Unit 3: Institutional Design with Information Design How information transfers are permitted to take place can have a large effect on learning and social outcomes. Consider the process by which pharmaceutical companies persuade doctors or a regulatory authority to approve new drugs. For example, in the U.S., the regulatory regime for new drugs has developed to grant more and more power to the evaluating authority. Earlier, companies could bring drugs to the market and simply let consumers (doctors and patients) decide on adoption. In 1906, the Pure Food and Drugs Act introduced penalties for mis-branding medicines (i.e., mis-reporting the true state, if revealed). The creation of the FDA in 1938 required drug sponsors to submit safety data to the FDA for approval. Further amendments in the 1960’s have led to an FDA that commits to an approval standard before research starts. These standards help to improve the standards of drugs that are brought to the market, but they also introduce additional costs to the firm, since providing information (e.g., designing and running clinical trials) is costly. Thus, the institution in which information design takes place affects the incentives to bring the product to market. Since both information and the product are valuable to the consumer, this trade-off is non-trivial. We explore such problems of institutional design in this unit, through the paper Henry and Ottaviani (2019);

Unit 4: Student Presentations Students will give a ten-minute presentation on a paper in the information design literature with political economy applications. Students can select from a list of papers, or select their own with the approval of the instructor. The exercise is intended to give students the opportunity to present the key ideas from research within a small amount of time. This is a critical skill for conveying research ideas to an economics audience-- whether academic, government, or corporate.