via Swad Komanduri
Policy interventions, particularly when utilizing payment changes to guide behavior, rely on underlying assumptions of behavioral responses to incentives to achieve stated goals. Value-based payment models used for healthcare provider payments are current manifestations of such schemes. While the theoretical assumptions that providers will change their behavior in ways that align with the incentives of the payment model are credible, there remain challenges in being able to statistically formalize and measure the desired behavioral changes. Simulations, particularly agent-based simulations that incorporate learning behaviors, can help shed light on this system.
One method to do this would be simulate how observed outcomes would change over time given a variety of provider responses to incentives. Such techniques have been used in climate change models or in COVID response models to be able to assess if macroscopic behavioral adjustments are ‘on track’ for the desired outcomes, and shed some more light onto the underlying mechanics driving the overall changes. By incorporating these methods with administrative models used for payment, we can better assess not only whether value-based payments are working, but also measure how they compare to expected behavioral changes.