Historical patterns and sustainability implications of worldwide bicycle ownership and use

The transport sector contributes about one-quarter of global fuel-related greenhouse gas emissions, with passenger cars responsible for half of these emissions. Demand for passenger road transport is expected to triple by 2050, and technological measures alone (electrification, lightweighting, fuel efficiency) will be insufficient to meet climate targets. Behavioral change and modal shift from cars to bicycles for short trips are also necessary. Despite bicycles’ long history and multiple uses, they currently play a marginal role in transport in most countries, with a few exceptions where pro-bicycle policies exist (e.g., the Netherlands, Denmark). There is substantial untapped potential to increase cycling to reduce emissions and improve health by counteracting sedentary lifestyles. However, global-scale understanding is limited by data gaps on bicycle production, stocks, and use, the weak correspondence between ownership and use, and literature that is largely city- or region-specific. This study aims to compile a global dataset of bicycle production, trade, stock, ownership, and use (1962–2015), compare historical bicycle versus car ownership trajectories, classify countries into development types, and assess potential climate and health benefits of increased cycling.

Prior research documents the limited role of bicycles in most countries and highlights the health and environmental benefits of active transport. Studies typically focus on a handful of cities or regions and rely on travel surveys (e.g., Stockholm, Flanders, London, and various U.S. cities), with recent work using counters and lane data for European cities. Evidence shows built environment, individual behavior, and trip characteristics affect cycling, and infrastructure expansion can increase cycling. Nonetheless, there is no comprehensive, dynamic, global analysis of bicycle production, stock, and use; and data on ownership and use are scarce, inconsistent, and often city-based. This study addresses these gaps by compiling and modeling global country-level data over five decades.

Scope: Conventional human-powered bicycles only; 60 countries/regions covering ~95% of global GDP and bicycle production, imports, and exports in 2015.

Stock and ownership estimation: A dynamic material flow analysis (MFA) was used to estimate historical in-use bicycle stocks (1962–2015). Apparent consumption (inflows) was calculated as production plus imports minus exports. A top-down lifetime-delayed stock model applied survival rates to annual inflows to derive in-use stock. Initial stock before 1962 was assumed zero. Per-capita ownership was computed as stock divided by population. Equations: I_ct = P_ct + IM_ct − EX_ct; ST_ct = ST_c,t0 + Σ(I_ci × SR_ci); OW_ct = ST_ct / PO_ct.

Data collection and cleaning: Bicycle production data from UN Industrial Commodity Statistics and national sources; bicycle trade data from UN Comtrade, cleaned for code consistency, missing units inferred from monetary values via average prices, and gaps interpolated. For countries without production, production set to zero. Bicycle and car modal shares were compiled from literature (travel surveys, reports); city-level values averaged to approximate national shares when necessary, with exceptions (e.g., China adjusted). For countries lacking modal share data, imputation used income-group quartile/median values. Car stock data compiled for comparison. Sensitivity analysis varied key inputs (production, trade, lifetimes) by ±10%.

Use patterns and correlates: Correlation analyses tested relationships between bicycle modal share and candidate drivers: bicycle ownership, car ownership, income level, population density, and traffic safety (traffic death rates). Linear regressions quantified associations.

Scenario analysis of climate and health benefits: Two thought-experiment scenarios assumed global adoption of current cycling levels of Denmark (1.6 km per capita per day) and the Netherlands (2.6 km per capita per day). Increased cycling distance was computed from national populations, current and target cycling distances, and assumed 365 cycling days. Carbon savings equaled increased cycling distance multiplied by the difference in per-km emission intensities of cars versus bicycles, adjusted by the share of short car trips and the shift efficiency from car to bicycle. Health benefits were assessed with WHO’s HEAT model, estimating prevented adult (20–64) all-cause mortality from increased physical activity. Traffic crash mortality increases were also estimated proportionally to increased cycling distance to derive net prevented deaths.

• Global production: Bicycles rose from 20.7 million units (1962) to 123.3 million (2015), CAGR 3.4% (higher than cars’ 3.0%). Cumulative bicycle production (1962–2015) was 4.65 billion units, 2.4× cumulative car production.
• Production geography: In 2015, China produced 65.7% of global bicycles; it became the largest producer after 2002. Other top producers: Brazil (~5%), India (~4%), Italy (~2%), Germany (~2%). The top-5 sales share fell from 67% (1962) to 44% (2015) as China’s share declined.
• Stocks: Global bicycle stock reached ~1.9 billion units in 2015. Top stock countries: China, U.S., India, Japan, Germany (>54% combined). Global car stock grew to ~1.1 billion by 2015; top: U.S., China, Japan, Germany, Russia (~50% combined).
• Ownership patterns: Per-capita bicycle ownership globally increased until ~1995 and then stabilized around 0.29 bicycles per person. Many European countries (e.g., Denmark, Netherlands, Norway) exceed one bicycle per capita; Japan leads Asia (~0.95 per capita in 2015). Five ownership-development types were identified: Types 1–2 (low/middle-income, higher bike than car ownership but below global medians; Type 2 shows rapid motorization and often declining bike ownership); Type 3 (high-income, rapid car growth with modest bike growth); Type 4 (highest car ownership, high bike ownership; e.g., U.S., Canada, Australia, New Zealand); Type 5 (highest bike ownership, high but saturated car ownership; many industrialized European countries).
• Use vs ownership: Bicycle ownership explains only ~36% of variation in bicycle modal share (2015), indicating ownership alone does not drive use. Average bicycle modal share across 60 countries is <5%. The Netherlands and Denmark exhibit both high ownership (>1 per capita) and high modal share (>20%).
• Car use: Car ownership correlates well with car modal share (R2 = 0.57, p = 0.000). Type 4 countries reach up to ~74% car modal share; countries with high ownership but lower car share (e.g., Switzerland, Japan, Czech Republic) have strong public transport.
• Determinants of cycling use: Higher car ownership tends to reduce bicycle modal share; higher income with supportive culture/infrastructure (e.g., NL, DK), higher population density, and better traffic safety are associated with higher bicycle use. High traffic death rates (e.g., Thailand, Brazil, Russia) correlate with lower cycling.
• Climate benefits: If everyone cycled like Danes (1.6 km/day), global carbon emissions could drop by ~414 Mt CO2/year (~98% of UK’s 2015 emissions). Dutch levels (2.6 km/day) yield ~686 Mt CO2/year savings (~86% of Germany’s 2015 emissions; ~20% of 2015 global passenger car CO2).
• Health impacts: Current cycling prevents ~0.17 million deaths/year (2015). Under Danish and Dutch scenarios, net prevented deaths could reach ~0.34 million and ~0.62 million per year, respectively, after accounting for increased cycling crash fatalities (~0.09 and ~0.16 million deaths, respectively).

The study demonstrates that while bicycles are abundant globally, their use for daily transport remains limited in most countries. Ownership alone is a weak predictor of cycling activity; instead, infrastructure, safety, urban density, culture, and car dependence critically shape use. The five historical ownership-development types highlight diverse transport transitions: many countries undergo rapid motorization with stagnant or declining bicycle ownership, whereas a subset of high-income European countries combine high bicycle ownership with moderated car ownership, reflecting mature, multimodal transport systems. The strong link between car ownership and car modal share underscores ongoing car dependence, yet cases with high car ownership and lower car use show the mitigating role of robust public transport. The substantial potential climate and health co-benefits from adopting Dutch or Danish cycling levels suggest significant global gains are possible through policy and infrastructure interventions—e.g., protected lanes, traffic calming, education, and measures that internalize car externalities. These findings address the research question by quantifying global historical stocks and use, clarifying mismatches between ownership and use, and estimating the sustainability implications of shifting short car trips to bicycles.

This work compiles the first global, country-level dataset (1962–2015) of bicycle production, trade, stocks, ownership, and use, and classifies countries into five development types relative to car ownership. It shows that high bicycle ownership does not necessarily translate into high use, and that strategic policies and infrastructure are needed to realize bicycles’ potential. Scenario analyses indicate large climate mitigation (up to ~686 Mt CO2/year) and public health benefits (up to ~0.62 million net prevented deaths/year) if global cycling matched Dutch levels. Future research should: expand temporal (post-2015) and spatial coverage beyond 60 countries; integrate more city- and subnational-level, sensor- and GPS-based data; investigate causal mechanisms linking infrastructure, safety, culture, and built environment to cycling; and assess the impacts of emerging systems (e.g., bike sharing, e-bikes) on stocks and use.

• Data uncertainties: Production, trade, and use data are incomplete or inconsistent; ownership and use estimates rely on modeling and imputation.
• Causality: Analyses are primarily correlational due to limited bottom-up, city-level datasets; causal inference is not established.
• Coverage: Temporal coverage ends in 2015; spatial coverage includes 60 major countries. Post-2015 developments (e.g., bike sharing, e-bikes, pop-up lanes) are not captured.
• Assumptions: Initial stock set to zero in 1961; survival/lifetime parameters and interpolations introduce uncertainty. Modal share approximations use city averages for national estimates in some cases.
• Health assessment: HEAT-based estimates exclude air pollution disbenefits (assumed negligible at typical cycling durations) and initially ignore crash risks in the positive-benefit calculation; net effects are subsequently estimated but rely on proportional scaling from current fatalities.