Modelling the Mobility Changes Caused by Perceived Risk and Policy Efficiency

Abstract

In many countries, governments have implemented non-pharmaceutical techniques to limit COVID-19 transmission. Restricting human mobility is one of the most common interventions, including lockdown, travel restrictions, working from home, etc. However, due to the strong transmission ability of the virus variants, further rounds of interventions, including a strict lockdown, are not considered as effective as expected. The paper aims to understand how the lockdown policy and pandemics changed human mobility in the real scenario. Here we focus on understanding the mobility changes caused by compliance with restrictions and risk perceptions, using a mobility index from the Google report during three strict lockdown periods in Leeds, the largest city in the county of West Yorkshire, England, from March 2020 to March 2021. The research uses time-varying z-scores and Principal Component Analysis (PCA) to simulate how local people dynamically process and perceive health risks based on multi-dimensional daily COVID-19 reports first. Further modelling highlights exponentially increasing policy non-compliance through the duration of lockdown, probably attributable to factors such as mental anxiety and economic pressures. Finally, the proposed nonlinear regression model examines the mobility changes caused by the population’s dynamic risk perceptions and lockdown duration. The case study model in Leeds shows a good fit to the empirical mobility data and indicates that the third lockdown policy took effect much slower than the first. At the same time, the negative impact of the epidemic on population mobility decayed by 40% in the third lockdown period in contrast with the first lockdown. The risk perception estimation methods could reflect that the local population became increasingly accustomed to the COVID-19 situation, and local people rationally evaluated the risks of COVID in the third lockdown period. The results demonstrate that simulated risk perceptions and policy decay could explain urban mobility behaviour during lockdown periods, which could be a reference for future decision-making processes.

Keywords

• urban mobility
• dynamic risk perception
• data-driven model
• policy analysis