טכניון מכון טכנולוגי לישראל
הטכניון מכון טכנולוגי לישראל - בית הספר ללימודי מוסמכים  
Ph.D Thesis
Ph.D StudentKoren Moran
SubjectSocial Learning in Adaptive Environments
DepartmentDepartment of Industrial Engineering and Management
Supervisors Professor Rann Smorodinsky
Professor Itai Arieli


Abstract

A market is an economic system that is comprised of agents, often with conflicting incentives, and a set of rules for allocating resources (i.e., production, consumption, prices, etc.).   A crucial element in determining the quality of the resulting allocation is the system's ability to aggregate and utilize the agents' private information.  Agents maximize their respective utilities, and may refrain from full disclosure, which may be exploited by others with conflicting interests, resulting in a suboptimal allocation.
We study markets where product values are unknown.   The firms’ information is also available to the consumers, who retain additional private information.  We ask, when do markets result in an efficient allocation? 
  
First, we study the influence of pricing competition on information aggregation.  We introduce a model of a duopolistic competition in a common value market.  Two firms compete over a single consumer.  The consumer receives a noisy signal over the value of both products.     We say that a firm is engaged in predatory pricing if it sets its price low enough as to ensure the consumer will buy from it independently of her signal and its competitor's price.   We find that predatory pricing is never optimal when signals are unbounded, or when the probability that the consumer is nonconformist, is negligible.  A property we call vanishing likelihood.  
  
Next, we assume that there are many consumers who act simultaneously.    Predatory pricing implies that consumer decisions are signal independent, and, it does not guarantee that the prevailing firm is the superior one.   In the complementary case, the consumer could buy from each firm with positive probability. Therefore, her decision adds information about the state of nature. When consumers are abundant, the distribution of purchases between the two firms fully aggregates the consumer information.  
  
A dynamic version of the model, where consumers arrive sequentially, and observe all previous actions, is an extension of the social learning model.  Only with endogenous alternative prices.   We show that social learning occurs if and only if signals are either unbounded or bounded with a vanishing likelihood.
 

An additional market where information plays a crucial role is that of crowdfunding campaigns. Inspired by popular platforms as ``Kickstarter" and ``Indigogo”, we model crowdfunding campaigns as a game in which society decides on the supply of a public good.  Agents choose whether or not to contribute. Contributions are collected, and the good is supplied, whenever total contributions pledged exceeds a predetermined threshold.  We study a case where the public good is excludible, agents have a common value, and each agent receives a private signal about the common value. We ask how well crowdfunding performs from the firm's perspective in terms of market participation, and how it performs from the perspective of society, in terms of efficiency.  We study the implication of crowdfunding on social welfare by introducing a firm to our model. We find that crowdfunding yields the optimal social welfare.  However, in some cases, the increase in welfare comes at the consumers expense.