Shifting Towards Cycling in the Greater Stockholm Area: A Comparison Between 2005–2006 and 2011–2013

Shifting Towards Cycling in the Greater Stockholm Area: A Comparison Between 2005–2006 and 2011–2013


Chengxi Liu, Swedish Road and Transport Research Institute


By using Swedish nation travel survey 2005-2006 and 2011-2013, this paper develops a series of mode-destination choice models with more detailed model specification on cycling as a mode alternative.


Cycling is becoming an important mode of transportation in many European metropolitan areas. In Sweden, although bicycle use has declined since the mid-1990s, some recovery has been seen within the last ten years, particularly in urban areas and their surrounding suburban regions, like the Greater Stockholm area (Petersen et al., 2015). However, the reasons underlying such a shift are still unclear as the bicycle infrastructure in Stockholm has not changed significantly over the past ten years. Therefore, this study utilises Swedish national travel survey data from 2005–2006 and 2011–2013 and estimates a series of nested-Logit models (mode-destination choice models) to assess the potential factors underlying such a shift in cycling. The mode choice is placed on the upper level in the nested-Logit model structure while the destination choice is placed on the lower level. Some of the level-of-service variables (travel time by car and public transport, and travel distance by car between a given origin and destination pair) are taken from the Swedish national transport model, Sampers. An open-sourced trip planner ( that does routing on openstreetmap ( is used to query travel time by walking and cycling. Relationships between walking/cycling travel time and car distance are then regressed for different origin-destination pairs. These established relationships are then used to create a travel time matrix for walking and cycling from the existing car distance matrix that is available in Sampers.
The estimation results reveal that cyclists have a lower cycling travel time sensitivity and are willing to cycle to further destinations in 2011-2013 than 2005-2006. It was also more attractive to cycle to zones with a high population density in 2011–2013 than in 2005–2006. Those changes result in a more even distribution of logsum measures (accessibility measures) for cycling over space. Regarding mode choice, different cycling trends are found for different socio-demographic groups. Middle-aged adults (age 41-60) and elders (age 60) cycle more often for their non-work trips than adults (age 31–40), and more so in 2005–2006 than in 2011–2013. More children in a household corresponds with a greater likelihood of cycling for both work and non-work trips. This tendency is more substantial in 2011–2013 than in 2005–2006, thereby indicating that cycling has become more popular, though at the expense of private cars among those who have children in the household, particularly for their work trips. Given these findings, future research should focus on capturing attributes other than travel time that influence bicycle destination choice and those which include cycling as a travel mode for the access/egress part of public transport trips.


Association for European Transport