Could Anticipating Gaming Incentivize Improvement in (Fair) Strategic Classification?

As machine learning algorithms increasingly influence crucial decisions in areas like loan approvals and hiring, understanding human strategic behavior in response to these systems becomes vital. We explore strategic manipulation and improvement actions by individuals facing algorithmic decisions, the algorithm designer’s role in shaping these strategic responses, and the fairness implications. We formulate these interactions as a Stackelberg game, where a firm deploys a (fair) classifier, and individuals strategically respond. Unlike previous research, our model incorporates both different costs and stochastic efficacy for manipulation and improvement. The analysis reveals different potential classes of agent responses, and characterizes optimal classifiers. Based on these, we highlight the impact of the firm’s anticipation of strategic behavior, identifying cases when a (fair) strategic policy can motivate improvement while reducing manipulation.