Rationalizable Screening and Disclosure under Unawareness
Alejandro Francetich, Burkhard C. Schipper
TL;DR
This paper studies a procurement contract problem in which the principal is initially unaware of some agent types and may be exposed to new types through the agent’s disclosure. It develops a dynamic framework of extensive-form games with unawareness and a Δ-O prudent rationalizability solution that relies on log-concave marginal beliefs and reverse Bayesian updates to compare contract menus across different awareness levels. A key finding is that when the principal is unaware only of high-cost types, those types have an incentive to raise awareness, with three types yielding the result that the two low-cost types gain extra information rents by disclosing; with more than three types, the incentive structure can change, potentially excluding some lower-cost types from dedicated contracts. By contrast, when the principal is unaware only of low-cost types, no type raises awareness, and the principal’s standard menu (possibly with redundant contracts) remains optimal; the analysis formalizes how awareness interacts with screening and information rents in a robust, belief-perturbation framework. Overall, the paper contributes a rigorous mechanism-design perspective on disclosure and learning under unawareness, linking discrete concave screening, reverse Bayesianism, and log-concavity to dynamic awareness and offering insights with implications for real-world procurement and disclosure settings.
Abstract
We analyze a principal-agent procurement problem in which the principal (she) is unaware some of the marginal cost types of the agent (he). Communication arises naturally as some types of the agent may have an incentive to raise the principal's awareness (totally or partially) before a contract menu is offered. The resulting menu must not only reflect the principal's change in awareness, but also her learning about types from the agent's decision to raise her awareness in the first place. We capture this reasoning in a discrete concave model via a rationalizability procedure in which marginal beliefs over types are restricted to log-concavity, ``reverse'' Bayesianism, and mild assumptions of caution. We show that if the principal is ex ante only unaware of high-cost types, all of these types have an incentive raise her awareness of them -- otherwise, they would not be served. With three types, the two lower-cost types that the principal is initially aware of also want to raise her awareness of the high-cost type: Their quantities suffer no additional distortions and they both earn an extra information rent. Intuitively, the presence of an even higher cost type makes the original two look better. With more than three types, we show that this intuition may break down for types of whom the principal is initially aware of so that raising the principal's awareness could cease to be profitable for those types. When the principal is ex ante only unaware of more efficient (low-cost) types, then \textit{no type} raises her awareness, leaving her none the wiser.