Fair Prediction with Endogenous Behavior
This addresses fairness in consequential domains like criminal justice, providing a theoretical justification for specific fairness metrics, though it is incremental as it builds on existing fairness debates.
The paper tackles the problem of incompatible fairness notions in machine learning for criminal justice by modeling societal incarceration rules and agent behavior, showing that equalizing type I and type II errors across demographic groups minimizes the overall crime rate, while other fairness notions do not.
There is increasing regulatory interest in whether machine learning algorithms deployed in consequential domains (e.g. in criminal justice) treat different demographic groups "fairly." However, there are several proposed notions of fairness, typically mutually incompatible. Using criminal justice as an example, we study a model in which society chooses an incarceration rule. Agents of different demographic groups differ in their outside options (e.g. opportunity for legal employment) and decide whether to commit crimes. We show that equalizing type I and type II errors across groups is consistent with the goal of minimizing the overall crime rate; other popular notions of fairness are not.