Health dynamics and reporting bias at retirement: An analysis using high-frequency data

We introduce a novel approach to identify state-dependent reporting bias in subjective health measures. The central idea is that health operates as a stock, making abrupt shifts in self-reported health (SRH) following retirement more likely to reflect reporting bias than actual changes. To capture such shifts, our analysis integrates three key elements: (1) differentiating stock and flow outcomes based on classical health theory; (2) leveraging an identification strategy inspired by regression discontinuity design; and (3) exploiting a unique high-frequency dataset on monthly health and retirement. Traditional estimates find a decline in SRH after retirement over longer periods; however, this decline steadily diminishes as the observation window narrows, showing no evidence of state-dependent reporting bias. Our analysis of short-term health dynamics also emphasizes distinguishing stock and flow health outcomes in policy evaluations.