Testing, disclosure and approval

Abstract

Certifiers often base their decisions on a mixture of information, some of which is voluntarily disclosed by applicants, and some of which they acquire by way of tests or otherwise. We study the interplay between the information acquisition of certifiers and the information disclosure of applicants. We show that the inability of a certifier to commit to the amount of information to be acquired can result in a reduction of information disclosed. Among other consequences, given the choice between two information acquisition technologies, the certifier may prefer to commit to the inferior technology, in the sense of being either more expensive or less accurate.

Test Design Under Falsification

“…By introducing the agency, in the form of costly state falsification, to the standard information design 9 setting of Kamenica and Gentzkow (2011) or Bergemann and Morris (2016), we add to a growing literature on information design when an agent can react to the experiment by undertaking an action that alters its informational content. For example, the agent can choose whether to take the test in Rosar (2017), or to disclose additional information in Bizzotto, Rüdiger, and Vigier (2020) and Terstiege and Wasser (2020). 10…”

“…By introducing the agency, in the form of costly state falsification, to the standard information design 9 setting of Kamenica and Gentzkow (2011) or Bergemann and Morris (2016), we add to a growing literature on information design when an agent can react to the experiment by undertaking an action that alters its informational content. For example, the agent can choose whether to take the test in Rosar (2017), or to disclose additional information in Bizzotto, Rüdiger, and Vigier (2020) and Terstiege and Wasser (2020). 10 Frankel and Kartik (2021) and Ball (2021) study the optimal design of linear scores in a setting in which the agent has a privately known gaming ability (akin to our publicly known cost-scaling parameter γ) and the receiver has a continuum of actions and seeks to most accurately match the agent's fundamental type, which is the analog of our state of the world, and is multidimensional in Ball (2021).…”

Optimal and Robust Disclosure of Public Information

“…Lipnowski et al (2020) and Bloedel and Segal (2020) consider a rationally inattentive receiver who can learn less information than what a sender provides, with substantially different modeling assumptions. On the other hand, Bizzotto et al (2020) and Matysková and Montes (2021) consider a receiver who can acquire additional information after receiving a sender's signal, which is closer in spirit to our model. One of their findings is that additional information acquisition can be detrimental to the sender and the receiver.…”

Impacts of Public Information on Flexible Information Acquisition

“…In Lipnowski et al (2020), a receiver pays costly attention to a state, but the information is less than that provided by the sender. Bizzotto et al (2020) and Matysková and Montes (2021) consider a receiver who can acquire additional information after receiving a sender's signal at no cost, which is closer in spirit to our model. The models in the above papers are based on the Bayesian persuasion framework (Kamenica and Gentzkow, 2011), and the sender's payoff is independent of the receiver's information cost.…”