Statistical Laws of Human Mobility under Large-Scale Behavioural Shock

Abstract

Human mobility follows robust statistical laws, but whether these laws persist under large, sudden behavioural shocks remains unclear. Here we analyse privacy-enhanced GPS trajectories of over one million users in California and 180,000 in Massachusetts from January to September 2020, spanning the onset and easing of COVID-19 restrictions. During restrictions, individual ranges contracted below a characteristic scale of about 5 km, while short- and medium-range trip lengths and rare long-distance trips persisted. Exploration slowed from super-diffusive to near-diffusive, and visits concentrated on a few habitual locations. Trajectory shape, once rescaled, was preserved, and the entropy reduction is largely explained by fewer visited locations rather than more regular routines. Recovery after reopening was heterogeneous rather than a uniform return to baseline, and the decline began before mandatory orders. These results suggest that restrictions reorganised, rather than simply rescaled, individual mobility, with implications for mobility-based epidemic models.