精算师的三个核心问题Three Core Questions from Actuaries
当健康管理方拿着"干预效果报告"去找保险公司谈合作，双方经常鸡同鸭讲。健康管理方说的是"参与者健康改善了X%"，险企精算师问的是"你怎么证明这个改善是你造成的，而不是这批人本来就更健康？"When health management providers take "intervention effect reports" to insurance negotiations, both sides often talk past each other. The provider says "participants improved X%"; the actuary asks "how do you prove this improvement was caused by your intervention, rather than by these people already being healthier?"
- 问题一：参与者是否存在选择偏倚？Question 1: Is there selection bias among participants?传统报告无法回答；PSM报告提供匹配前后SMD对比。Traditional reports cannot answer this; PSM reports provide pre/post-matching SMD comparison.
- 问题二：健康改善是否可归因于干预本身？Question 2: Is health improvement attributable to the intervention itself?传统报告只有相关性；PSM报告提供ATT+95%CI。Traditional reports only show correlation; PSM reports provide ATT+95%CI.
- 问题三：赔付降低的ROI是否合算？Question 3: Is the claims-reduction ROI worthwhile?传统报告是定性描述；PSM报告将ATT换算为预期赔付节省。Traditional reports give qualitative descriptions; PSM reports convert ATT into expected claims savings.
| 评估维度Assessment Dimension | 传统报告Traditional Report | PSM证据报告PSM Evidence Report |
|---|---|---|
| 选择偏倚控制Selection Bias Control | ❌ 未处理Untreated | ✅ SMD <0.1 |
| 因果归因方法Causal Attribution | ❌ 简单对比Simple comparison | ✅ PSM |
| 效应量报告Effect Size Report | ❌ 仅p值p-value only | ✅ ATT+95%CI |
核心结论Key Takeaway
险企评估预防干预效果，问的不是"效果好不好"，而是"效果能不能被因果归因、能不能换算成赔付数字、能不能支撑精算决策"。PSM因果证据报告直接对应精算部门的评估框架。Insurers don't ask "is it effective?" They ask "can the effect be causally attributed, converted into claims numbers, and used to support actuarial decisions?" PSM causal evidence reports directly correspond to this evaluation framework.