Health is a complex dynamic process that impacts many economic decisions in ways that remain poorly understood. This paper comprehensively reviews how health is modelled in the literature, showing that baseline models typically fail to take into account how persistence and frequency of health shocks vary by past health history and magnitude and direction of past shocks. Methods from the earnings dynamics literature are adapted to produce improved health persistence estimates. This paper also investigates how medical biomarker data can be incorporated in dynamic models of health as a proxy for underlying health. There is significant scope for further work in this area as more medical data becomes available to researchers.
