Market design methods have been increasingly used in important applications like school choice. Most existing methods ask participants to rank their alternatives in order. In contrast, this project studies a class of methods that elicits numerical scores for each alternative. Such preference intensities provide more information than just the ordering over the items, and therefore the methods developed in this project can generate better assignments of the scarce items. By soliciting numerical preferences, the placement of students in schools, for example, can be more efficient, and priorities in school choice can be also allowed. In addition, the assignment methods developed in this project can be used when agents care about how other agents are assigned. Such allocations with peer effects also have important applications in college dormitories or shared offices.
In this project, the submitted intensities of preferences are used to operate a pseudo-market on a computer, without using real money. Methods computing competitive equilibria to pseudo-markets have been used less in practice and many hurdles need to be overcome. This project studies theoretical, computational, lab experimental, field test and structural empirical work on pseudo-market mechanisms, with key applications being school choice and the allocation of shared offices. The investigators develop a new computational algorithm to compute a competitive equilibrium to a pseudo-market. Given the multiplicity of equilibria, equilibrium selection is carefully addressed. Lab experiments investigate the concerns that agents may have non-expected-utility preferences and can manipulate the mechanism by colluding in groups. A variety of research methodologies are combined to implement and evaluate pseudo-market mechanisms in school choice, where students have priorities at each school that must be respected.
