Feedback Design and Preference Elicitation: Field Experiments in Digital Economics

Abstract

Chapter 1 experimental method that elicits individual risk attitudes repeatedly. This section proceeds with a discussion of the research background and motivations against which the field experiments in this thesis are conducted. The discussion does not attempt to be exhaustive, but rather aims to provide a general overview of the points of departure for each experiment. Section 1.2 contains a detailed description of the chapters in which the experiments are discussed. Section 1.3 offers concluding thoughts. Digital economics as a research field is concerned with whether and how digital technology changes economic activity (Goldfarb and Tucker 2019). Digital technology enables information about one's behavior and choices to be stored in bits, which substantially reduces the cost of storage, computation, and transmission of data. Goldfarb and Tucker (2019) identify five types of costs that have fallen significantly in this regard: search costs, replication costs, transportation costs, tracking costs, and verification costs. This thesis zooms in on two types in particular: tracking and replication. Tracking is the ability to follow anyone's individual behavior. 1 Replication means that data, once digital, can be replicated at near zero marginal cost. The ability to track individual behavior likely reshapes principal-agent relationships in many domains. In traditional offline settings, an agent's effort is oftentimes unobservable, and in such cases the principal may be forced to design a contract based on aggregate information about the agent's activities (Hölmstrom and Milgrom 1991). In digitalized environments, an agent's effort may be completely observable to the principal. While this increases the principal's scope for using pay-for-performance schemes to achieve the desired effort level of the agent, such schemes are often not feasible in unionized establishments or in settings where the culture is such that financial incentives are seen as undesirable. Alternatively, the principal can use the data on observable effort to design and evaluate non-financial incentives that can motivate the agent, such as tailored performance feedback. The data can also be used to acquire a better understanding of agents' preferences, choices, and actions. This thesis considers three principal-agent settings in more detail and explores the potential of Student behavior In recent years, there has been much interest in the use of digital technologies in education (Escueta, Nickow, Oreopoulos, and Quan forthcoming). Existing research in this area has mainly focused on comparing courses with 6 Chapter 1 it rules out behavior in which lack of effort today can be compensated for with more effort in the future, though crucially at the expense of flexibility. In the field experiment, a novel digital commitment mechanism is developed and tested that is fully consistent with students' initial study plans. The mechanism, inspired by pacesetters in sports, incorporates the motivating effect of goal bracketing in a dynamic fashion and preserve...