Can curbside bicycle lanes buffer COVID-19 ridership losses? A case study from Melbourne, Australia

Abstract

Many cities implemented temporary or permanent improvements in safe cycling infrastructure during COVID-19 in a bid to encourage cycling. However, many studies found that cycling during COVID-19 was primarily for recreation and exercise. In contrast, cycling for commuting declined, largely because commuting overall (especially into cities) was partially replaced by working from home. In 2020 and 2021, the City of Melbourne fast-tracked a range of cycling infrastructure upgrades in an effort to attract more workers back into the city using this sustainable transport mode. The upgrades were relatively modest – ranging from extending curb-protected lanes through a single intersection, to replacing a few kilometres of painted lane with a protected bike lane. This study examines how patterns of bicycle riding have changed in response to the COVID-19 pandemic, using bicycle count data from 15 automatic counters within 5 km of Melbourne’s city center. In addition, this paper isolates the impact of COVID-era infrastructure upgrades against the backdrop of huge fluctuations in bike riding from 2019 to 2022. Multi-level generalized linear modelling found that weekday bicycle counts were significantly dampened during lockdown and remain below pre-COVID levels on weekdays and at sites on off-road bicycle paths. However, when controlling for lockdown stage and seasonality, the average daily count at sites near upgraded infrastructure were 7.4 % higher (855 per day) than counts from sites that were never, or not yet, near upgrades (796 per day). This increase is notable given that almost all counter locations saw declines in bicycle counts even through 2022. These findings are particularly relevant for cities that are grappling with whether to continue with temporary or long-term cycling infrastructure upgrades. This study suggests that even relatively modest upgrades to cycling networks can have a measurable impact on cycling rates.