The impact of migration background and ethnicity on car, bus and bicycle use in England

International migration has increased over the last decades, mainly driven by work and study or family reasons, increasing the diversity of the population both globally and in Europe (IOM 2021). Thus, a growing body of research has tried to understand whether and to what extent the travel behaviour of immigrants is similar or different from the native-born population (e.g., Smart 2015; Hu 2017; Welsch et al. 2018; Delbosc and Shafi 2023). The question of whether and how migration and ethnic background impact travel behaviour has gained increasing attention over the last two decades. Most of the research stems from the United States of America (US) and to a lesser extend from other countries with substantial immigrant populations like Canada or Australia. European studies on this topic are rather scarce, despite the growing share of the migrant population in many countries (Lanzieri 2011, eurostat 2022). This is also the case for the United Kingdom (UK), where immigrants and ethnic minority groups account for a substantial and growing share of the population (as we discuss in Sect. "Case study context"). In the UK, this aspect of population diversity is often discussed in relation to income and inequalities (Shen and Kogan 2020; Saunders et al. 2021; Sweida-Metwally 2022) but less so with regards to travel behaviour. In many European countries, there is a general paucity of data sources that allow the investigation of immigration or ethnicity effects on travel behaviour while controlling for socio-demographic, economic and spatial intervening factors, as well as for differences in transport systems or access to transport modes. In order to assess and cater to different travel needs, differences in travel behaviour need to be understood. Existing studies point out that the travel behaviour of immigrants differs from that of the native-born population as they tend to use the car less and public transport more often, particularly in the initial period after migration (Delbosc and Shafi 2023), as we discuss in greater detail in Sect. “Literature on travel behaviour by immigration background or ethnicity”.

A range of terms with overlapping definitions are used in research and public debates on migration and ethnicity. These also tend to vary across countries, reflecting differing cultures, policy traditions and sensitivities around this issue. The terms “immigrant” (in the US context) and “migrant” (in the European context) are often used to identify people who are not native born to the country. Both the US and the UK also distinguish people’s ethnicity (for US: race/ethnicity) e.g. by including items on ethnic group identity in national surveys, which overlaps with the regions of origin for many immigrants. Burton et al. (2008) discuss ethnic group classifications for the UK and point out, that pre-defined categories can only weakly capture a person’s subjective identity (Supplementary Material: Terminology). Nevertheless, these concepts can support the inclusion of certain aspects of the population’s diversity in research or policies.

In this article, we analyse survey data on migration, ethnicity, and everyday travel mode use (frequency of car, bus, and bicycle use) from the UK Household Longitudinal Survey (UKHLS), which offers uniquely rich data and a large, representative sample (University of Essex 2021). This enables us to disentangle immigration- and ethnicity-related effects. The survey data uses the concepts of migration generation (also called migration background) and ethnic group (Sect. "Data"). Based on the literature review (Sect. "Literature on travel behaviour by immigration background or ethnicity"), we include further factors in our analysis such as detailed spatial information on accessibility to services by different modes, and socio-demographic factors. We account for the effect of these factors in a step-by-step approach. This way, we can shed light on what underlies the different travel behaviour of migration and ethnic minority groups.

In this paper we investigate following research questions: (1) Are there differences in travel mode use between groups defined by their migration and ethnicity status? (2) To what extent are these differences accounted for by intervening factors related to socio-economic characteristics and spatial characteristics of the residential area? (3) Is there evidence of ethnic neighbourhood effects on travel mode use? These questions emerge from the literature review in Sect. “Literature on travel behaviour by immigration background or ethnicity”. Notably, our study advances research in this area in three respects. First, by providing representative evidence on migrant travel behaviour from a context (the UK and Europe) for which little evidence on this topic exists. Second, by simultaneously considering and disentangling the effects of migration generation and ethnicity on travel behaviour. Finally, by considering a wide range of factors that might explain or confound this association.

The paper is structured as follows. Following Sect. “Literature on travel behaviour by immigration background or ethnicity”, where we review the relevant literature, Sect. “Data and methods”describes data and methods and provides information about the case study context. Sect. “Results” presents the results of bivariate and multi-variate analyses. In Sect. “Discussion”, we discuss our findings as well as limitations of our approach and conclude our paper by elaborating on directions for further research and policy implications (Sect. “Conclusion”). Additional information is provided in the Supplementary Material.

Against the background of a growing immigrant population, the first publications on this topic analysed similarities and differences in the everyday travel behaviour of immigrants in comparison to native-born individuals, with a main focus on commuting and on public transport demand (e.g., Germany: Hautzinger et al. 1996, US: Myers 1997, Casas et al. 2004, Canada: Heisz and Schellenberg 2004, UK: Gardiner and Hill 1997, Owen and Green 2000, DfT 2003).

Most subsequent research stems from the US, where surveys often include information on race/ethnicity and immigration status (e.g., Blumenberg and Evans 2007; Beckman and Goulias 2008; Liu and Painter 2012; Smart 2015). When compared to the native-born population lower levels of car access and use, and higher levels of public transport use and walking were found. Some of those differences can be attributed to differences between the two groups in terms of socio-economic characteristics and residential location, as many of the more recent immigrants tend to have lower financial resources and lower car ownership. Those differences generally diminish with a longer duration of stay, something which is referred to as “transport assimilation” or “adaption” effect (e.g., Xu 2018; Asgari et al. 2017; Chatman and Klein 2009).

This assimilation effect may play out across generations, with some evidence that second migration generations have travel behaviour patterns more similar to people without a migration background (Harms 2007; Klocker et al. 2015; Welsch et al. 2018; Hu et al. 2021; Mattioli and Scheiner 2022). Yet other studies found a persisting difference, i.e., the “immigration effect”, which still remains after controlling for socio-demographic, economic or spatial factors (e.g., Harun et al. 2021; Chatman 2014; Chatman and Klein 2013). Here, one can hypothesise that unaccounted cultural effects, preferences and attitudes contribute to explain differences in travel behaviour and that those “soft” factors vary across different groups defined by ethnicity or countries of origin (e.g., Welsch 2015; Yum 2019; Shafi et al. 2023).

US studies also found that car-pooling (sharing a private car with family or acquaintances) is more common among immigrant groups (Lovejoy and Handy 2011; Blumenberg and Smart 2014) and is often correlated with living in a neighbourhood with a high share of immigrants (Smart 2015). Similarly, Shin (2017) finds that living in an ethnic neighbourhood with a high ratio of same ethnic/racial group generally affects inter-household car-pooling positively for most ethnic groups.

There are few studies focusing specifically on bicycle use in the US. They show that cycling is more common among recent immigrants and/or certain immigrant or race/ethnicity groups (e.g., with European origin). Also, cycling varies depending on overall spatial factors, (limited) mobility options and support of local social networks (Smart 2010; Howerter et al. 2019; Barajas 2020). In the US, Smart (2015) found that residing in a neighbourhood with high shares of immigrants further increases the probability that immigrants use modes alternative to the car (notably walking and cycling), even after controlling for intervening factors such as built environment characteristics. This might be due to social factors in the neighbourhood such as a higher propensity to conduct social activities in close proximity within the neighbourhood, and learning from the example of fellow immigrants with regard to travel mode choice.

Research on the travel behaviour of immigrants and ethnic groups has also gained attention in other countries in which immigration has played an important role in history, namely Canada (e.g., Lo et al. 2011; Mercado et al. 2012; Newbold et al. 2017; Harun et al. 2021) and Australia (e.g., Syam et al. 2012; Klocker et al. 2015; Churchill 2020, Shafi et al. 2022). Research evidence from the UK and other European countries is still rather scarce. In many European countries, there is no established tradition of collecting information on race, ethnicity, or immigration status within (national) transport-related household surveys, in contrast to the US or the UK. If such data is included, detailed analyses are often limited due to small sample size and/or underrepresentation for the groups of interest, with low response rates possibly due to lower socio-economic status or barriers related to language or literacy. As a result, European research often relies on the limited information available within national household surveys or on ad-hoc, smaller scale surveys (e.g., Netherlands: Harms 2007, Durand et al. 2023, Austria: Fassmann and Reeger 2014, Germany: Welsch et al. 2018, Geis 2019, Welsch 2022, Denmark/Netherlands: Haustein et al. 2020). Similar to studies from North America, European studies also found lower levels of car ownership and car use, and higher levels of public transport use, although when controlling for socio-economic factors, these differences tend to decrease or remain significant only for certain groups of immigrants and/or for the first generation. In countries with higher cycling levels such as Denmark, Germany, Austria and the Netherlands, immigrants tend to cycle less than the native-born (van der Kloof 2015; Fassmann and Reeger 2014; Harms 2007; Harms et al. 2014; Durand et al. 2023). Welsch et al. (2018) find a correlation between cycling socialisation and bicycle use for different immigrant groups, which could partially explain such an immigrant effect. For Denmark and the Netherlands, Haustein et al. (2020) analysed the effect of ethnic neighbourhoods as a proxy for cultural influences on the propensity to use the bicycle and found a small negative influence.

Despite a flourishing tradition of transport research, there is a paucity of quantitative research on the travel behaviour of immigrant and ethnic minority groups in the UK. Participants in the English National Travel Survey (NTS) are asked to self-report their ethnic group and country of birth (DfT 2018), but to the best of our knowledge no dedicated study has used this information. Some reports have looked into the impact of migration on transport, with e.g., Tsang and Rohr (2011), analysing mode use for the journey to work based on the Annual Population Survey. They found that a significant transport assimilation took place within the first few years upon arrival (except for bus use), but information on migration status was limited and important transport related variables such as car access/ownership were missing from the dataset. Based partly on other, smaller-scale data sources and on descriptive analysis, other reports have focused on issues such as the (public) transport needs and related security perceptions of different ethnic groups (DfT 20032003, Syam et al. 2011), cycling (Bowles Green Limited 2008, Goodman and Aldred 2018; Bednarowska-Michaiel 2023; Osei and Aldred 2023) or issues such as road safety (Steinbach et al. 2007) and the transport-related problems of socially disadvantaged groups, including ethnic minorities among others (Wixey et al. 2003; Smith et al. 2006; Lucas et al. 2016, 2019; Vidal Tortosa et al. 2021). A recent study by Mattioli and Scheiner (2022) based on UKHLS data for 2012/13 focused primarily on air travel but found lower car mileage as driver among foreign-born and non-white UK residents, which is partly but not entirely accounted for by differences in socio-economic composition and residential location, and by the fact that their close family members live abroad. The study however does not include detailed information on transport mode availability and accessibility at the neighbourhood level, and as such cannot rule out that this just reflects a pattern whereby migrants and ethnic minorities are more likely to live in areas that are less car dependent.

In summary, and in line with what recent reviews found (Delbosc and Shafi (2023) most quantitative studies consider multiple indicators to explain the different travel behaviour of immigrants and/or ethnic groups. Most commonly, differences in socio-economic factors such as lower income and associated factors (employment status, education level, lower car ownership rates) are found to account for part of those differences (e.g., Newbold et al. 2017; Xu 2018; Welsch et al. 2018). Age and household composition are often included to account for demographic differences between native-born and (newly) immigrated population groups. Gender-related differences e.g., in transport mode use are often found to be more pronounced among immigrant or ethnic population groups, which is sometimes explained with reference to differing social norms or cultural preferences (e.g., Syam et al. 2012, Aldred et al. 2016, Klocker et al. 2015, Shin 2017, Haustein et al. 2020). Spatial and transport-related indicators are not always taken into account in the literature to date. When included, residential location or neighbourhood characteristics are considered (e.g., population density, proximity to the city centre or access to public transport services, sometimes in combination with ethnic composition (e.g., Beckman and Goulias 2008; Chatman 2014; Liu and Painter 2012; Smart 2015)). In many studies that take the factors above into account there remains some unexplained differences in travel behaviour, the so-called immigration effect. While it is often argued that this reflects cultural differences these have mostly been directly investigated in qualitative studies (e.g., Bowles Green Limited 2008, Lovejoy and Handy 2011; Chatman and Klein 2013; Barajas 2020), rather than in quantitative ones.

In this paper, we investigated car, bus and bicycle use frequency in England, focusing on differences associated with migration background and ethnic group identification, and investigating to what extent these associations are accounted for by other factors. We summarise the main associations identified in the bi- and multivariate analyses in Fig. 1. Note that in the figure, the findings for the multivariate analysis refer to Model D in Tables 2–4, not Model E. This is explained as follows: we consider that car and driver licence ownership potentially act as mediating factors between our main independent variables of interest (ethnicity and migration background) and the outcomes of interest (mode use frequency). This implies that, in the fully adjusted models that controls for car and driver licence ownership (Model E), the coefficients for ethnicity and migration background are an underestimate of the true effects of these variables, part of which is direct, and part of which is indirect (i.e., mediated by car and driver licence ownership, which in turn influences mode choice). As such, we believe that the coefficients from Model D provide a better reference for estimating the impact of ethnicity and migration background on mode use frequency.

With regard to car use the bivariate analysis shows that 1st and 2nd generation migrants drive less than the reference group, with particularly low levels of car use among recent migrants. This is consistent with previous research (Delbosc and Shafi 2023). The multivariate analysis shows more nuanced patterns of association. For second generation migrants, there is no clear difference in car use once other factors are controlled for, similar to findings on car milage in the UK (Mattioli and Scheiner 2022) and car use among people with a migration background in Germany (Welsch 2018). For more established 1st generation migrants, who had been in the country for more than 5 years, lower levels of car use are accounted for mainly by differences in socio-economic factors. For more recent immigrants though, lower car use should rather be linked to lower car access and (valid) driving licence ownership in the first few years after arrival, which is consistent with the transport assimilation hypothesis.

In the bivariate analysis, all ethnic minority groups drive less on average than the White British majority. For most groups, this is explained by differences in socio-economic characteristics and residential location. As such, we find very little indication of other social or cultural effects, with the possible exception of the “Other Non-White and Mixed” group, for whom lower levels of car use persist even when controlling for all other factors. Due to the heterogeneous nature of this group, it is challenging to interpret this finding. Overall, car use is the travel mode for which we find the least indication of differences between ethnic groups. Note however that our analysis does not distinguish between drivers and passengers, so it might overlook higher rates of carpooling among certain ethnic groups which in the US context was found to be especially prevalent in ethnic neighbourhoods. Also, our analysis may overlook finer differences by gender, as some studies find that immigrant women tend to have a lower driving licence ownership and car use as drivers, but more car use as passengers, (Blumenberg and Smart 2014, Chatman and Klein 2013, Asgari et al. 2017; Shin 2017), perhaps because of the cost of driver’s licenses (Priya and Uteng 2009).

With regard to bus use, the bivariate analysis shows a positive association for 1st and 2nd migration generation and all ethnic groups. They use the bus more often than the reference groups. In the regression models, this association is explained by ethnicity and other control factors for 2nd generation and more established 1st generation migrants. For recent migrants, the association persists until the mediating factors of car access and driving licence ownership are controlled for. This is consistent with a transport assimilation effect whereby migrants use the bus in the first few years after arrival when they do not have access to cars, but then eventually get access to it and switch modes. For the ethnic groups “Other White” and “Asian or Asian British” higher levels of bus use are overwhelmingly due to differences in socio-economic characteristics and residential location. For the “Other Non-White and Mixed” and “Black or Black British” groups these factors account for only part of the association, which suggests that there might be other uncontrolled factors at play such as attitudes and socialisation, or concentration in areas poorly served by rail-based public transport and more reliant on buses such as South London (Sims et al. 2016, TfL 2019).

With regard to cycling, we find no significant differences between migration generations (in both analyses) other than for 1st generation immigrants who had been in the country for 10 years or less, who cycle significantly less often than others. The reasons for this are unclear but might be due to the perceived unsafe cycling conditions in the UK and/or to the particular cultural connotation of cycling in British culture, which is seen to be dominated by white male (Aldred et al. 2016; ETSC 2020, statista 2023), both of which might discourage migrants or members of ethnic minority groups to use this mode, as we discuss below. However, the absence of an effect for immigrants who had been in the UK for more than 10 years suggests that immigrants increase their cycling frequency over time and can also be interpreted in terms of transport assimilation.

We found significant differences in cycling frequency by ethnic group but in contrast to car and bus use these are not all in one direction. In the bivariate analysis, “Other Whites” cycle more, while “Asian or Asian British” and “Black or Black British” cycle less than the “White British” majority. These differences remain significant when controlling for socio-economic characteristics and residential location, suggesting that other uncontrolled factors play a role. This dovetails with findings from the US context, where a tendency for “white immigrants” to cycle more than the native born (white) population has been observed as well (Smart 2010; Hu et al. 2021).

Previous research suggests that soft factors such as attitudes and socialisation might play a larger role for cycling than for car and bus use. These factors might be more relevant in ethnic neighbourhoods (Welsch et al. 2018; Haustein et al. 2020) and for women, and include high risk perceptions (among low skilled/non-users) (Garrard et al. 2012; van der Kloof 2015; Mohammadi 2018; Prati et al. 2019). Other factors such as a perceived low status of bicycles as transport modes (Welsch 2022) might be a greater barrier to (young) men (as reported for “Asian or Asian British” by Batool and Pangbourne (2022) as well as discrimination or racialised policing for the “Black or Black British” group (Osei and Aldred 2023). Other factors not included in our analysis such as cycling skills, availability of a (suitable) bicycle or basic knowledge about traffic rules might also play a role. There is also evidence to suggest that safe routes, separated bicycle infrastructure and safe bicycle parking facilities tend be more lacking in areas with higher shares of ethnic minorities (Lusk et al. 2017; Vietinghoff 2021). For example, there is evidence to suggest that the “Black and Black British” group is spatially concentrated in areas with worse bicycle infrastructure provision such as South London (Bednarowska-Michaiel 2023; Osei and Aldred 2023). More broadly, the UK has very low cycling levels, which tend to correlate with a less diverse cyclist population with regards to type of cyclists gender, age or ethnicity (i.e., mainly white, male commuters or sportsmen) (Aldred et al. 2016; Goel et al. 2022). This might discourage from cycling people who do not identify as White because they perceive it as a social practice that is “not for someone like me” (Aldred and Jungnickel 2014; Barajas 2020; Batool and Pangbourne 2022; Osei and Aldred 2023). Conversely, higher levels of cycling among “Other Whites” might be explained by greater propensity to cycling in at least some of the countries these people originate from, such as e.g., other European countries.

Further studies could investigate both the vertical cultural transmission as well as possible horizontal cultural transmission similar to Marcén and Morales (2021), who studied the gender commuting gap by culture and cross-country differences. Horizontal transmission takes place through neighbours, friends or the ethnic communities and vertical one through parents, grandparents or other ancestors who probably instil values in their children. Mchitarjan (2015) points out that vertical cultural transmission also depends on someone’s own cultural identity. The children of immigrants, who do not cycle themselves, might thus face additional difficulties in having to learn how to cycle, despite their own families’ non-cycling mobility culture. For the children of the non-migrant majority, these difficulties could also occur when the family has a very car-oriented mobility culture.

With regard to ethnic density, the bivariate analysis shows higher levels of bicycle and bus use, and lower levels of car use, in neighbourhoods with a high share of ethnic minorities. Our analysis of ethnic neighbourhood effects, however, provides only little evidence of their existence (beyond the effect of individual ethnic group membership). We find suggestive evidence of “cross-group” effects, whereby a higher share of members of an ethnic group in the neighbourhood is associated with a change in the propensity to use a travel mode among members of another ethnic group. While the reasons for these associations are unclear, and we cannot rule out that they are spurious, they might be interesting to explore in future research.

In general, we find evidence that the travel behaviour of immigrants tends to converge towards the average of the population after the first few years. Such a transport assimilation effect is one of the common findings among studies on immigrant travel behaviour in various countries (Delbosc and Shafi 2023). Nevertheless, as the more advanced discussion in the US shows, apprehending the extent and the determinants of transport assimilation is complex, as the effect might vary depending on location, ethnic group and cohort. Thus, the process is discussed as a segmented assimilation, which varies in relation to different ethnic groups (e.g., Xu 2018; Yum 2019; Hu et al. 2021). Also, while our findings are consistent with the transport assimilation hypothesis, we cannot rule out that differences between more and less recent first-generation migrants are due to differences in the country of origin, which are obscured by our relatively aggregate ethnic group categories.

Our study provides (to the best of our knowledge) the first comprehensive multivariate analysis of the impact of ethnic minority and migration background on daily travel behaviour in a UK context. Compared to the international literature reviewed in Sect. "Literature on travel behaviour by immigration background or ethnicity", our study has several strengths as it: (i) investigates the impact of both ethnic minority and migration background, both of which are classified in detail; (ii) based on a large, nationally representative dataset in which ethnic minority groups and immigrants were oversampled to ensure a healthy sample size for these groups; (iii) controlling for a wide range of intervening factors, including detailed spatial information on accessibility by travel modes in the neighbourhood. This allowed us to explore the independent variable side in great detail. One limitation of the UKHLS dataset, however, is on the dependent variable side, as information on travel behaviour is collected in a rather simplified way, i.e., with questions on habitual mode use frequency. While the National Travel Survey (NTS) provides better information on travel behaviour in England, it is more limited in terms of availability of relevant independent variables and sample size, which is why we opted for analysing the UKHLS here.³ Future research could investigate the same questions using the NTS, providing triangulation for our findings.

Our analysis approach has two further limitations, which must be kept in mind when interpreting the results. First, we chose to analyse data from the most recent pre-COVID survey wave, in order to provide a baseline picture of the travel behaviour of these groups, which did not exist to date. However, travel behaviour has changed during and after the pandemic, in various countries including the UK. Notably, a reduction in work-related travel has been observed, along with a faster rebound effect for car use than public transport (the latter of which declined drastically during the pandemic), and active travel generally increased during the pandemic but the longer-term impacts on changes in travel mode use are not yet clear (Christidis et al. 2023; ITF 2023, Metz 2024, Lu 2024). For example, cycling decreased in England in 2023 to a pre-pandemic level but was at a peak in 2020 (DfT 2024b). In general, active travel also depends on local circumstances, e.g., whether walking and cycling conditions and infrastructure have been improved (Buehler and Pucher 2022; Buck 2023; Ortar and Rérat 2024). Whether and to what extent these changes in particular have affected individuals with ethnic minority and migration background deserves further investigation. We encourage future researchers to explore these changes, possibly using more recent waves of UKHLS, and the findings of our study as a baseline reference.

Second, our study provides initial evidence of a transport assimilation effect over the life course of migrants. However, as discussed in Sect. "Data", there are some issues concerning the sub-sample of recent migrants in the UKHLS. As such, there is some uncertainty surrounding those findings, which must be kept in mind. Overall, we believe this leads us to underestimate the extent of transport assimilation effects. While we hope that future research will be able to provide more solid representative findings on this point, this will require a deliberate oversampling of recent migrants in surveys, including notably panel surveys such as UKHLS. While the oversampling of ethnic minorities in UKHLS allows us to robustly analyse the main ethnic groups and their differences with regards to transport mode use, variations within each of these ethnic groups (due, e.g., to differences in national cycling cultures within Europe or Asia) are obscured in our analysis. Further research could also explore these differences in more detail and how that compares to other groups.

Overall, our analysis suggests that in England ethnic identification and migration background are two independent, and only loosely associated, determinants of travel behaviour. As such, there is value in exploring the effect of both simultaneously. At the same time, our findings suggest that ethnic identification is a more important factor than migration background. Besides the above discussed transport assimilation, our findings show that certain differences in transport mode use persist among ethnic minorities, a phenomenon which is also referred to as “immigration effect”. A large share of these differences has to do with the socio-economic composition and spatial distribution of these groups, and this is notably the case for car use. Yet for some ethnic groups and some modes, differences in travel behaviour remain significant after taking into account those factors and might be due to soft factors. As discussed above, this is particularly the case for lower levels of cycling among the main non-white ethnic minorities in England (i.e., “Asian and Asian British” and “Black and Black British”).

Our study sheds some light on the complex relationships between migration background, ethnicity and transport mode use, while also pointing at directions for future research in a British and European context. These could include: (i) similarities and difference within and among ethnic groups, including people who do not belong to the main ethnic minority groups; (ii) an investigation of intra- and inter-household carpooling; (iii) a better understanding of ethnic neighbourhood effects and how they play out, including possible “cross-group” effects; (iv) further investigation of the dynamics of migrant travel behaviour, based on longitudinal data; (v) qualitative research insights into socialisation, habits and experiences as well as underlying motives, values and attitudes and specific barriers to travel behaviour, such as the factors that currently prevent or discourage Non-White ethnic groups from cycling in England.

From a sustainable transport policy perspective, our results show that in a car-dependent society like the UK, migrants do not necessarily have more sustainable transport behaviour, as sometimes implied by earlier research (Klocker et al. 2015). While in an initial phase they drive less and use public transport more, their behaviour converges with that of the majority over time. Nevertheless, this particular life phase (the first years after arrival) could offer an interesting entry point for measures aimed at getting immigrants to maintain sustainable travel behaviour. These could build on the insights developed in mobility biographies research (Müggenburg et al. 2015). Our findings also suggest that measures to encourage active travel need to consider ethnic diversity and the cultural and socialisation factors that currently hinder cycling take-up among minority groups.

From a social policy perspective, lower car use among ethnic minorities can entail a risk of social exclusion–especially if there are few viable alternatives or these come with risks (such as exposure to infection during COVID-19; e.g., Long et al. 2023). If ethnic minorities are more reliant on buses, they are also more vulnerable to service cuts and price increases, both of which are ongoing trends in the UK. As such, we argue that transport policy needs to take into account distributional effects not just across income groups, but also across ethnic groups. At the same time, these groups seem to face greater barriers (both “hard” and “soft”) to cycling. Given the well-documented positive health effects of cycling, this could also be seen as an injustice that needs to be addressed.