Set them (almost) free: Discretion in electoral campaigns under incomplete information

Abstract

The paper analyzes a model of electoral campaigning as a problem of competitive delegation. We consider an environment characterized by two sources of uncertainty: uncertainty on the nature of the optimal policy and uncertainty on the candidates' biases. Although voters know whether the candidate is left-or right-wing, they do not know the extent of the bias. In this environment, discretion may benefit voters as it allows the elected politician to adjust his policies to the state of the world. The paper shows that the optimal set of promises must be a closed interval, whose size is decreasing in the expected bias of the candidate. An example where the set of types is finite shows that an increase in the variability of candidates' types may either increase or decrease the voters' willingness to grant discretion to politicians.

The Complexity of Campaigning

“…For instance, when Vermin Supreme says, "When I'm president everyone gets a free pony", we discount the possibility that he plans to give everyone two free ponies; if he did, that would not contradict his promise, but his omission of information would be infelicitous. 5 A potential source of discrepancies between the framework's predictions and the reality of campaigns is that, given an elaborate body of information about a candidate's policies, voters have trouble evaluating the candidate due to the intractability of their utility functions. We will consider the computational complexity of the function problems of determining exact voter utility, but also of decision problems of determining whether the utility meets a given threshold, which are particularly relevant for the "stay-at-home voter" scenario we will discuss in Section 5.…”

“…Political scientists have also taken interest in candidates' decisions about what to say when campaigning; Petrocik [15] found empirical evidence that a candidate will try to focus on issues where the candidate has a good record and their opponents have bad records. Game theorists have a shared interest with Dean and Parikh in what might motivate a candidate to be ambiguous [1,2,5]. The game-theoretic models take into account the interaction between multiple candidates (and in the case of Baghdasaryan and Manzoni's model [5], voters' uncertainty about their most-preferred policies), but often with simplified representations of a platform (e.g., points on a one-dimensional spectrum or probability distributions over a small set of alternatives).…”

“…Game theorists have a shared interest with Dean and Parikh in what might motivate a candidate to be ambiguous [1,2,5]. The game-theoretic models take into account the interaction between multiple candidates (and in the case of Baghdasaryan and Manzoni's model [5], voters' uncertainty about their most-preferred policies), but often with simplified representations of a platform (e.g., points on a one-dimensional spectrum or probability distributions over a small set of alternatives). In contrast, Dean and Parikh's framework abstracts away details of the electoral system like multi-agent interactions and voter strategy to focus on the implications for an individual candidate of communicating using more expressive logic-based statements.…”