Determinants of bicycle commuting and the effect of bicycle infrastructure investment in London: Evidence from UK census microdata.

Worldwide, concern about physical inactivity and excessive car dependence has encouraged ambitious targets and policies to promote cycling. But policy making is hindered by limited knowledge about why cycling prevalence and trends vary greatly between different geographic areas (e.g. in London (UK) <1% cycle to work in Harrow compared to>15 % in Hackney) and individuals (e.g. by age or gender). The role of cycle infrastructure investment in explaining part of these patterns and trends is also unknown. We linked individual-level data on 317,117 London commuters (including 11,199 cyclists) in the 2001 and 2011 UK census to relevant geographic data, including on area-level cycling infrastructure investment during the period. Whilst cycle commuting increased over time on average, concentration curves and indices demonstrated that in contrast with England as a whole, cycling in London shifted from being dominated by commuters with lower socioeconomic status to commuters with higher socioeconomic status. In our first set of regression analyses, we showed that observed differences and time trends in cycling prevalence were partially explained by area-level differences in topography, greenspace, footpaths and crime levels and by differences and changes in population structures. In the second, we conducted a cost-effectiveness analysis which showed that expenditure on cycling infrastructure was associated with increased cycling at a marginal rate of £4915 per additional commuter cyclist, with some variation between groups: ethnic minorities were more responsive, and females, older people and those with lower socioeconomic status appeared less responsive. If planned increases in expenditure in England for the period 2020-25 were as cost-effective, and were sustained for the whole decade, our study suggests that commuter cycling prevalence could increase in England by 0.5 to 1.1 percentage points (this equates to a 16% to 34% increase in commuter cycling prevalence if compared to 2011 levels). More research is necessary to assess the impact on broader measures of cycling, active travel and overall physical activity, and to determine whether such expenditure constitutes good or equitable value for money.