Rank Uncertainty in Organizations

Using Divide and Conquer to Improve Tax Collection: Theory and Laboratory Evidence

“…Incomplete information over rank. Halac et al (2020) show that providing players incomplete information over their rank can improve the effectiveness of divide-and-conquer strategies. Our analysis suggests that information can be quite helpful if it does not require high degrees of sophistication to be interpreted.…”

“…In particular, the work of Abreu and Matsushima (1992) on virtual implementation highlights the value of more sophisticated divide-and-conquer schemes to achieve full implementation. Divide-and-conquer schemes also play an important role in the literature on contracting with externalities, including Segal (2003), Winter (2004) and more recently Halac et al (2019Halac et al ( , 2020. Halac et al (2020) also studies the impact of information design, but allows for schemes in which the rank of agents is not common knowledge.…”

“…Divide-and-conquer schemes also play an important role in the literature on contracting with externalities, including Segal (2003), Winter (2004) and more recently Halac et al (2019Halac et al ( , 2020. Halac et al (2020) also studies the impact of information design, but allows for schemes in which the rank of agents is not common knowledge. In contrast to our results (which pertain to revealing information about the compliance of others), they find that appropriately designed information about others' rank can increase the principal's surplus.…”

“…Halac et al (2020) show in a related context that incomplete information over rank can increase the principal's surplus. Since we have experimental implementation in mind, we prefer to focus on a relatively simpler class of mechanism.…”

Organized Information Transmission

“…In the expectations in (30), taken over (s ≻ s i , ω), player i can deviate to any a ′ i and also misreport to all players ≻ s -below him by switching to any positive probability message profile s ′ ≺ s ′ i = (s ′ j : j ≺ s ′ i) such that the set of i's predecessors at s ′ is the same as at s. 17 One subtlety in (30) is that upon observing s i , i learns his rank in the total order ≺ s , because he can infer it from { j : j ≺ s i}, whose messages he is asked to forward and can manipulate. Hence, unlike in Halac, Lipnowski, and Rappoport (2020) and Morris,Oyama,17 We abstract from deviations to message profiles where { j : j ≺ s i} ⊆ { j : j ≺ s ′ i} as those may be detected as misreports at some point in the hierarchy.…”