Spring 2011 Seminar Series

DATE: April 7th
LOCATION: E62-550
TIME: 4:15pm
Reception immediately following in the same room

SPEAKER:
Stefanos Zenios

TITLE
Evidence-Based Incentive Systems with an Application in Health Care Delivery

ABSTRACT
Recent legislations directed Medicare to revamp its decades-old system for reimbursing dialysis treatments. Two particular foci are on the risk-adjustment of payments and on the transition towards a pay-forcompliance system. To design an optimal payment system that incorporates these features, we develop an empirical method to estimate the parameters of the multi-task principal-agent model underlying Medicare’s dialysis payment system. Medicare (the principal) has a stake in the production of patient health outcomes, but delegates the treatment decisions to a dialysis provider (the agent). The provider chooses how much effort to put into each task of the treatment process. Medicare observes each patient’s aggregate health outcome (downstream outcome) which is a noisy signal of the provider’s effort in each task. Also observed are several other health outcome measures (typically process-compliance measures to be referred to as intermediate outcomes) which are noisy signals of the provider’s efforts in a subset of tasks. Medicare can either reward the provider for aggregate downstream outcomes (current practice) or for performance on the intermediate measures (pay-for-compliance). We use the model and parameter estimates to answer the following questions: How should patient outcomes be risk-adjusted, and what welfare gains can be achieved by doing so? Can a pay-for-compliance system based only on the intermediate performance measures currently identified by Medicare achieve first-best? Our main findings are: 1) In addition to preventing selective admittance of patients, risk adjustment can also lengthen the hospital-free life of admitted patients by two weeks per patient per year without increasing Medicare expenditures; 2) The current set of spotlighted intermediate measures are not comprehensive enough for use alone in a pay-for-compliance system.

 

(Joint work with Donald Lee, Yale School of Management)