Beyond Statistics: The Economic Content of Risk Scores
指出保险和信贷市场中广泛使用的统计风险评分掩盖了更丰富的异质性，并可能因应用环境而内生变化；利用Medicare Part D数据实证显示风险评分混淆了真实健康与保险引发的支出反应，理论分析表明即使有完美风险评分，合同下的策略性撇脂动机仍可能存在。
"Big data" and statistical techniques to score potential transactions have transformed insurance and credit markets. In this paper, we observe that these widely-used statistical scores summarize a much richer heterogeneity, and may be endogenous to the context in which they get applied. We demonstrate this point empirically using data from Medicare Part D, showing that risk scores confound underlying health and endogenous spending response to insurance. We then illustrate theoretically that when individuals have heterogeneous behavioral responses to contracts, strategic incentives for cream skimming can still exist, even in the presence of "perfect" risk scoring under a given contract.