Working Paper: NBER ID: w20376
Authors: Anirban Basu
Abstract: This paper uses Roy's model of sorting behavior to study welfare implication of current health care data production infrastructure that relies on solicitation of research subjects. We show that due to severe adverse-selection issues, directionality of bias cannot be established and welfare may decrease due to new data. Direct diversification of treatment receipt may solve these issues but is infeasible. Unifying Manski's work diversified treatment choice under ambiguity and Heckman's work on estimating heterogeneous treatment effects, the paper proposes a new infrastructure based on temporary diversification of access that resolves the prior issues and can identify nuanced effect heterogeneity.
Keywords: health care; welfare implications; data production; treatment effects; diversification
JEL Codes: C01; C9; D61; I18
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| existing health care data production mechanisms (I10) | interpretable treatment effect parameters (C22) |
| voluntary participation in research studies (C90) | interpretable treatment effect parameters (C22) |
| adverse selection issues (D82) | interpretable treatment effect parameters (C22) |
| randomized fractional coverage (C46) | treatment effect estimates (C22) |
| Learning Through Diversification (LTD) framework (J24) | welfare outcomes (I38) |
| LTD framework (L24) | treatment effects (C22) |
| LTD framework (L24) | marginal benefits (D61) |
| treatment choices (D87) | welfare outcomes (I38) |
| treatment effectiveness (C90) | welfare implications (I30) |
| data production design (D20) | welfare implications (I30) |