Highlights • Use of pseudo-panel methods to explore the demand for cycling in Denmark. • Division into age and city size cohorts in the period 2006–2019. • Decomposition of behavioral changes into a selection and mileage effect. • Evidence that the demand for cycling is declining, mainly due to a selection effect. • Evidence that the decline is strongest for rural areas and younger generations.

Abstract In this paper, we investigate the dynamics of cycling demand for the population of Denmark. Using pseudo-panels based on large-scale cross-sectional data, we analyse the temporal stability of cycling demand preferences for different age cohorts in combination with residential city sizes. Cycling demand is decomposed into two effects. Firstly, a population ’selection’ effect that explains the probability of being a cyclist, i.e. engaged in cycling activities. Secondly, the conditional demand for cycle mileage provided that the respondent is a cyclist. The joint probability model is estimated as a Gamma Hurdle model. The study reveals several empirical findings, of which three stand out. Firstly, overall cycling demand in Denmark over the period is in decline. Secondly, it is shown that this is mainly a selection effect. Hence, the main driver of the observed decline is essentially a shrinking cycling population rather than a decrease in trip distances for those who travel by bicycle. Thirdly, the decline is strongest for younger generations, particularly those residing outside the larger cities. With Denmark being an international forerunner for bicycling and with a cycling culture developed over many decades, we believe these findings can be relevant to mitigate similar long-term changes in other countries.