| Popis: |
In the first essay, we model information transmission in a rumor-spread approach, wherein the goal of the Rumor Maker (RM) is to start a rumor that can spread to the largest possible number of Decision Makers (DM). In the model, an RM designs and commits to a rumor-generating mechanism. After the state is realized, a rumor of the state is generated and sent to the first DM. Each DM needs to choose a binary action by playing a guessing-the-state game, and also decides whether to spread the rumor or block it. The DMs move sequentially in a prefixed order. A DM transmits a rumor to the next DM only when she expects the following DM’s gain from the rumor to be larger than the cost of reading the rumor; otherwise this rumor will be blocked. In addition to the rumor, DMs have a common signal and private signals about the state. We show that when DMs’ private signals dominate common signal in terms of accuracy, always telling the truth is one of the best-spread rumors. However, when DMs’ common signal dominates, the unique best-spread rumor tells the truth with a certain probability. We also show that as the cost of reading a rumor and the accuracy of DM’s signals increase, the best spread rumor converges to the truth.In the second essay, we study an endogenous timing learning model over a star network, in which there is 1 central player connected with n periphery players. Players in each period face two options: make an irreversible investment or wait for another period. Players receive a binary private signal on the profitability of investment at the beginning of the game, and also observe neighbors’ actions in past periods. We show that there exists a threshold of network size (N ̅) : when the size of the network is small (n≤N ̅), in equilibrium, periphery players use a pure strategy which fully reveals their private signals to the central player; when the network is large (n>N ̅), periphery players use mix strategy and only partially reveal their signals to central player. The central player waits in the first period and then makes the final decision on whether to invest at all in the second period (based on the number of first-period investment of periphery players). Among the equilibriums, the central player works as a crowdsourcing platform, collecting information from some peripheral players and deliver it to the rest players. We also show that Asymptotic Learning does not occur in star network: the probability of central player making the right action does not converge to 1 as the size of the network increases to infinity, which indicates a failure of information collecting in star structure.In the third essay, we develop a search and matching model in housing market with restricted purchase policy, where an agent cannot own more than one house. I show that in equilibrium, the housing price is lower than its value in free market. By comparing restricted purchase policy with restricted price policy in a search and matching model, I show that to reach a certain amount of price decrease, restricted purchase policy generates a larger welfare cost and a lower temporary excess return for prospective buyers. I also show that in stationary equilibrium, the excess returns in both restricted purchase and restricted price market equal to zero, which indicates the source of excess return is not stable restricted policies, but the policy shocks. |
