부산대학교 도서관

국외단행본

Essays on Prediction [electronic resource].

University of Pennsylvania. Economics.

[S.l.] : University of Pennsylvania. 2024 Ann Arbor : ProQuest Dissertations & Theses , 2024

1 online resource(106 p.)

Source: Dissertations Abstracts International, Volume: 85-12, Section: A.
Advisor: Ordonez, Guillermo L.

Thesis (Ph.D.)--University of Pennsylvania, 2024.

Information has emerged as a primary determinant of outcomes in economic and, broader, social settings. In this dissertation, we focus on its implications for trade and policymaking.The first chapter, The Visible Hand: Price Discrimination Under Heterogeneous Precision, is motivated by firms' increased use of data to inform their decisions. The diffusion of this technological trend has been intensive, comprehensive, and unequal, with the emergence of leading sectors, firms, and business functions. Against this backdrop, I study the efficiency and distributional effects of these changes in a leading business function and sector, retailers' use of data to tailor offers, where firms are heterogeneous in their predictive ability.To study this technological change, I leverage a standard search-theoretic model of imperfect competition, featuring buyers with heterogeneous private valuations for quality, and introduce sellers who observe valuation signals of heterogeneous precision. Signals induce third-degree price discrimination, and their precision largely dictates whether they are used to increase trade or increase markups - impacting aggregate surplus and its distribution. When buyers' valuations are more heterogeneous, imprecisely informed sellers prioritize high markups despite limiting trade, and precision relaxes this tension, not only allowing them to pursue high markups when it is least obstructive but also primarily incentivizing low markup offers that increase trade upon signals indicative of low valuation - increasing aggregate surplus and benefiting (hurting) buyers with a low (high) valuation. However, when valuations are more homogeneous, imprecisely informed sellers prioritize trade, and precision can primarily incentivize high markup offers that limit trade upon signals indicative of high valuation, hurting all buyers and even decreasing aggregate surplus. In either case, precision makes sellers more profitable, but its effect on competitors can be positive or negative. Generally, competitors suffer (benefit) when laggards (leaders) gain precision.The second chapter, The Cost of Discipline, is motivated by a central channel in modern politics, which peaked during the COVID-19 pandemic, whereby the public's assessment of politicians and institutions is informed, in part, by comparing the actions of their political actors against those taken in other political regions for similar problems. This kind of oversight is traditionally considered to be beneficial for the public, decreasing their information gap with governing agents; however, as the pandemic illustrated, it can also homogenize policymaking - a potentially negative practice.Jointly with Guillermo Ordonnez and Helios Herrera, we study the effect of comparing agents' actions on ex-ante principal welfare when agents are (1) better informed about the returns of each action, (2) sometimes have ex-ante preferences in favor of certain actions (bias), (3) are concerned about the public's assessment (popularity), and (4) sequentially solve an identical problem with identical information about returns. A policy of comparisons increases principals' welfare over that achieved by a policy of monitoring only their own agents' actions whenever principals match with agents who have an identical distribution of characteristics (bias and popularity concern). However, when the distribution of agent characteristics is unequal between principals, the latter policy with less oversight can be better for principals. In particular, we show that asymmetry in the average agent's bias or popularity concern between principals can make comparisons detrimental for principals matched with better-aligned agents (greater popularity concern, lower bias), as these primarily incentivize rallying behind actions that are expected to be popular, even when they are worse for principals.An overarching theme between these chapters is the strategic role of inference: sellers leverage it to extract the greatest surplus from buyers, while agents leverage it to optimize the efficiency and popularity of their actions. Its pivotal nature in applications as central as retail and policymaking highlights the importance of understanding the rapidly evolving information space. In this dissertation, I build two theoretical frameworks that challenge popular narratives on these issues and provide policy guidance.

9798382830919