Mechanism Reform for Task Allocation
Working Paper 32369
DOI 10.3386/w32369
Issue Date
Revision Date
Reforming an existing system for allocating tasks among agents introduces additional political and institutional constraints relative to designing one in isolation. We develop a general mechanism-design framework for using data on agents’ performance to improve outcomes while ensuring that no agents are made worse off relative to the status quo. As an illustration, we apply our results to the assignment of Child Protective Services investigators to maltreatment cases. Simulations show the mechanism reduces false positives (unnecessary foster care placements) by up to 14% while also lowering false negatives (missed maltreatment cases) and overall placements.