Simulating prize fund rewards

by James Packard Love

At a meeting last year at Columbia, Barry Nalebuff suggested we consider testing the prize fund approach by simulating payoffs, beginning with past drug approvals, to look at how a particular system might have worked, given the drugs that were approved. We intend to do some of this.

Data
Data on drug approvals is easy to get. Excellent data exists on usage -- privately collected and owned by IMS. But IMS data is very expensive, even for older data. We might use data from medicaid outlays or some other public sorce, to get a sense of market shares (as measured by units), as well as relative prices of products introduced in a given year or time period.

The main point of this exercise would be to see how differnent approaches, such as those proposed by Aidan (directly related to QALYs), as well as others (such as this) might work.

Any suggestions regarding data or approach are appreciated.