Fraud-proof Mechanism Design

University Of Texas At Austin, Austin TX

Investigators

Abstract

This research focuses on designing better contracts and auction mechanisms in contexts where agents can engage in costly (or risky) fraud. The project extends the theory of mechanism design (which is often considered as the engineering side of economics.). Specifically, this work improves our understanding of the allocation of public housing units, seats in selective schools, human organs, and other goods whose assignments are done without monetary transfers and may be affected by falsified applications. Eligibility is usually based on a score (metric) that proxies the social value of assigning a unit to an individual. However, reliance on the metric creates strong incentives to game it. As a result, practices such as forgery, greenwashing, teaching to the test, or manipulating statistics are commonplace. Infamous examples include doctors escalating their patients' treatments to increase their priority on organ waiting lists, families using a fake address to gain access to a desirable public school, and the various elaborate ways well-off parents facilitated admissions of their children to highly selective universities in the US during the college admissions scandal. Gaming the system leads to misallocation, unfair outcomes, and erodes public trust. Socio-economic groups better at gaming achieve faster access to organ transplants, better school assignments for their children, and so forth. The misallocations are not only unfair but they can also cost lives in some cases. The researcher develops procedures that maximize allocative efficiency while being fraud-proof, meaning individuals cannot benefit from gaming their scores. In contrast to the majority of the mechanism design literature, which assumes that misreporting is costless, the research funded by this award characterizes agents by their natural score, which they can falsify at a cost, and by other privately-known dimensions (tastes, gaming abilities) that they can misrepresent at no cost. The analysis shows how one can leverage falsification costs to design optimal allocation procedures. The developed solution methodology complements techniques stemming from Myerson (1981), which have been widely applied to settings with transfers, as well as the Lagrangian techniques employed by Amador et al. (2006) and others for settings without transfers. The research program suggests a pioneering way of tackling problems of mechanism design without transfers and costly fraud, highlighting novel connections to the literature of optimal transportation theory. Within economics, it extends the main mechanism design paradigm by allowing agents' private information to have 'hard' (costly to misreport) and 'soft' (free to misreport) dimensions. The setting and insights we provide can be applied to the design of rating systems based on which federal funds are allocated (for example, rating systems for Medicaid providers and nursing homes). They can also be applied to the design of accounting and taxation standards. All these systems are often plagued by gaming and manipulations, making our modeling based on costly type falsification relevant. The project also has interdisciplinary impacts on computer science. Computer science theory is concerned with algorithmic manipulations. Given that such manipulations can be costly and that the abstract modeling of algorithms resembles that of mechanisms without transfers, the modeling and solution approach can be applied to the design of manipulation-proof algorithms. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.