Abstract. I study strategic communication between a partially informed receiver and an informed expert who is motivated by influence: she values her advice being followed as an end in itself. The central result is that, despite lacking commitment power, the equilibrium behavior of the expert is identical to the optimal commitment solution of the corresponding Bayesian persuasion problem. Thus, a desire for influence provides an alternative microfoundation for this canonical framework when commitment is not feasible. Beyond this equivalence, I show that the receiver's level of informedness determines the direction of causality between advice and action: as the receiver becomes more informed, the expert ceases to be an "opinion leader" and becomes a "follower". The receiver gains the most from communication when she is neither too ignorant nor too well informed.

Abstract. I examine the factors that determine whether a grassroots social movement reaches the necessary size to achieve its goal. I study a collective action problem where identical individuals who value the common goal sequentially decide whether to join the movement. The model has two key ingredients: (i) The movement is facing a freeriding problem (i.e., while individuals want the movement to succeed, they would rather have others bear the cost of participation) and (ii) The necessary number of members to achieve success is ex-ante unknown but it can be revealed as the movement grows in size. The central insight is that an increase in cost of participation, such as harsher and more likely punishment for members of the movement, can lead to a drastic surge in membership.

Abstract.   We study the interaction of observational learning and free-riding through a dynamic volunteer's dilemma: If one individual exerts costly effort, a public good that benefits everyone equally is produced. The value of the public good is uncertain, and each individual receives a private signal about it. The central mechanism is that individuals make inferences about each other's private signal by observing their action, while aware of their free-riding incentives. The main findings concern the probability of public good provision and information aggregation. First, provision probability is non-monotonic with respect to population size: If the population is small, increasing it beyond a certain threshold leads to an upward jump in provision probability. Above this threshold, provision probability gradually decreases with population. There is a unique and finite population level that maximizes provision probability, and the ``bystander effect" hinders provision only if the population is already large. A similar non-monotonicity arises in provision probability with respect to effort cost. Second, due to free-riding incentives, there are no information cascades in equilibrium even when the population is arbitrarily large: Learning becomes slower as population grows, and social learning never outweighs the private signal.

  Abstract.    This paper analyzes tacit collusion in an oligopoly market where consumers are affected by past prices. In particular, we study an infinite horizon Bertrand competition between two identical firms where today’s demand for the good at a given price is higher if it is a discount relative to past prices and lower if the price has been raised. First, we find that history dependent demand leads to overpricing (relative to the myopic profit maximizing price), as the firms consider it an investment in future demand that can later be exploited through discounts. Second, firms can sustain monopoly pricing only up to an upper bound: prices above it trigger a sharp discount followed by gradual stabilization. This behavior yields a discontinuity in price adjustment following a negative cost or demand shock: If the shock size is below a certain threshold the prices decrease gradually, following the typical ``feather" behavior. If the shock size is above the threshold, the transition is through a sudden and large “fire sale”.

Abstract. This paper looks at the sources of informative communication in elections. I ask why purely office seeking political parties are (partially) honest about their positions on a polarizing issue, even when they face no repercussions for lying. I show that informative equilibra are possible only when (i) there are sufficiently many parties competing for office, and (ii) the share of opposing views is sufficiently balanced among the population of voters. The mere competitiveness of the elections forces a self-reinforcing mechanism of honesty: Some parties find it optimal to claim the minority view, as opposed to joining the many candidates who side with the majority. For the claims to convert into votes, however, they must be informative. The analysis also yields a methodological contribution to the study of competitive cheap talk: I show that the complex multi-sender cheap talk game can be represented and solved as a simple normal form game on beliefs, subject to an additional plausibility constraint.