Risk Adjustment Reform: Navigating Ideas And Tradeoffs (Part 1)

In an individual insurance market, plan payments are risk-adjusted to limit distortions in plan offerings that arise from incentives to select favorable risks. With this objective in mind, risk adjustment is often reduced to a statistical goal—to maximize the fit of a predictive model to spending data—in an attempt to vary per-person payments across plans according to the expected costs of the populations they attract. With the problem framed purely as a prediction problem, the promise of “big data” and advances in predictive methods such as machine learning are often invoked in risk adjustment “solutions.”

However, improving fit (often called predictive accuracy) to limit selection incentives tends to exacerbate other distortionary incentives; it comes at a cost. These...

Today's Sponsors

Today's Sponsor