We estimate that walking and cycling an additional kilometre may result in GHG emissions up to 0.26 (95% UI 0.12 to 0.53) and 0.14 kgCO2e/km (95% UI 0.06 to 0.28)), respectively, when additional energy expenditure is fully compensated with increased food intake in the most economically developed countries. Our current best estimate for the most economically developed countries suggests that the actual emissions associated with walking and cycling will reflect partial compensation of energy intake - from 19% to 96% of the required energy intake. There is wide variability in emissions required to compensate for walking and cycling between countries, representing nearly a 5-fold difference between the most and least economically developed countries.
IPCC estimates that direct emissions from cars range from 0.08–0.21 kgCO2/km1, with a similar range of estimates in a more recent ‘well-to-wheel’ life cycle assessment (in the range of 0.15 kg/km to 0.26 kg/km19,20,21). Factors such as the composition of the vehicle fleet (age, manufacturer, size, etc), fuel type and where it is obtained from, driving behaviour, and road patterns are relevant. When comparing vehicle emissions with our estimates for walking and cycling, it is important to remember differences in the sources of emissions. Vehicular emissions are primarily drawn from non-renewable stocks, whereas estimates of dietary emissions include both renewable and non-renewable sources. Nevertheless, our estimates suggest that the net emissions associated with the ‘fuel’ required for driving, walking, and cycling may be comparable in some settings. Our intention was to provide an estimate of emissions required for walking and cycling, and explore the likely extent of compensatory intake. Our study demonstrates that overall assessments of the emissions impact of transport interventions should consider emissions impacts associated with diets. Taking account of walking and cycling emissions may suggest that car share schemes could have a bigger positive emissions impact than increasing walking, and (all else being equal) interventions that decrease vehicle use through increased cycling will have greater emissions benefits than those that increase walking.
Active transport has many advantages: more pleasant urban living, reduced air pollution, reduced non-communicable disease. But, to maximise the effect on GHG emissions achieved by increasing active transport, we need to address dietary patterns. Emissions associated with active transport will be lower if walking and cycling are powered by with low carbon dietary options, and/or associated with less than full compensation thereby resulting in lower obesity rates.
The GHG emissions associated with many modern diets can be reduced without health penalties. First, the way we as a society use energy to produce food can be altered. Like transport, agriculture is responsible for a substantial proportion of emissions, up to one-third of all anthropogenic GHG emissions by some estimates22. Second, given emissions associated with different food groups (given current food systems) range widely - from 0.02 (sugar, legumes) to 5.6 gCO2e/kcal (ruminant meats) in one global study18, consumer switching to less energy intense foods could reduce dietary emissions by up to 70–80%22. Third, food waste matters: the emissions footprint of the edible food that is lost or wasted each year is 3.5 GtCO2e/year23, equivalent to 50% of all transport emissions1. In high-income countries reductions in emissions are largely proportional to the magnitude of meat and dairy reduction22. That is, in order to reduce GHG emissions we need to encourage changes in what we eat as well as reducing motorized vehicle use.
At present, the magnitude of the impact of increased active transport on obesity is unclear and is likely to vary by setting and intervention. Quantifying the extent and nature of compensatory energy intake is needed to establish the health impact and GHG emissions impacts associated with active travel. Our results demonstrate that even in the countries with high dietary emissions, the emissions associated with walking and cycling may be considerably lower than the emissions associated with driving if individuals only partially compensate for the increased energy expenditure.
We estimate the compensation of additional energy expenditure with increased food intake ranged from 19% to 96% in one UK study that reported a BMI reduction with increased active transport10. For walking, this would translate to a range in diet-related emissions between 0.03 to 0.13 kgCO2e/km, equivalent to 12% to 63% of the emissions associated with the fuel for an average size car in the UK (0.21 kgCO2e/km24).Future research is needed to establish how compensatory behaviours vary between settings and across different interventions. For example, we would hypothesise that a media campaign that encourages active transport as a way to lose weight may be associated with lower levels of compensatory food intake and thereby may achieve greater population level reductions in obesity than an equivalent campaign encouraging active transport to reduce congestion.
To our knowledge, this is the first international estimate of GHG emissions associated with food intake required per kilometre travelled by active transport. Our results demonstrate the importance of including emissions associated with food intake when estimating the net GHG emissions impacts of interventions to increase active transport.
Our estimates are indicative of general patterns (i.e. that emissions required to fuel walking may be similar to emissions associated with fuelling a car when diets are highly carbon-intensive) but merit further consideration in more specific contexts. For example, how the emissions of walking or cycling compare to those of driving for specific trips in specific cities.
We limited our scope to emissions associated with the fuel required to walk and cycle a kilometre and do not include embodied emissions (e.g. emissions associated with the manufacture of cars or the construction and maintenance of transport systems). For interventions that switch selected (car) trips to walking or cycling, fuel-vs-fuel comparison may be appropriate, noting that the energy required to obtain and transport the fuel should be included (i.e. a well-to-wheels approach). Ideally, assessments of specific interventions should incorporate a wider view that considers the fuel-vs-fuel trade-offs, embodied emissions, and broader impacts (e.g. health and social costs of emissions), but adopting a wider view is a demanding and complex undertaking.
We use a single, global study as the source of both dietary emissions and energy availability18. As the data reflect energy availability, it is not subject to individual-level under-reporting present in survey data that would (falsely) deflate estimates of excess energy required for walking and cycling. Whilst estimates of energy availability will result in an overestimate of the energy intake required, they more accurately reflect the emissions associated with additional energy intake as they account for the amount of food lost and wasted prior to consumption. This means that our estimates implicitly account for food waste, assuming that food waste patterns do not change under increased active travel.
The estimates of dietary GHG emissions used are based on grouping countries at similar levels of economic development, using a consistent methodology. However, these results may mask variation in the dietary greenhouse gas emissions between countries with similar levels of economic development. This means that the true variation in emissions associated with walking and cycling is likely to be wider than represented here.
We explore the sensitivity of our results to assumptions around compensation of energy intake by modelling BMI changes and estimated emissions per km for a population based on a UK cohort study. The UK study was selected as it was the only longitudinal study we identified that presented sufficient data to estimate the extent of compensatory energy behaviour. However, our estimates of compensatory behaviour are based on aggregate results and not an analysis conducted at the individual level. As neither diet nor energy intake were assessed in the study, the observed change in weight may easily be confounded by other lifestyle changes accompanying transport mode shift (e.g. change in diet, income, or family circumstance). The reported change in BMI corresponds to a wide range in compensatory energy intake that is likely to be highly heterogeneous across individuals. The sensitivity of emissions estimates to compensatory behaviours in our example clearly demonstrates the need for better quantification of BMI changes and compensatory behaviour in the context of active transport.
If increased energy expenditure from active transport was not fully compensated, then both the food-related emissions associated with active transport will be lower and a (modest) contribution will be made to lowering obesity rates. Given the high impact of obesity on healthcare systems, and the high emissions associated with healthcare (e.g. 10% of total emissions in the USA25), the indirect benefits of active transport (such as through reduced health system burdens) onto emissions may be substantial, and warrant further research.