The World Bank prices PM2.5 at $8.1 trillion for 2019 (World Bank, 2022). Studies that price a worker's lost day use area-level readings, not what a commuter inhales. On a motorcycle route, the area reading was too low for the rider (Tsai et al., 2008). So the rider is the line item the bill cannot see.
A monitoring station is a box somewhere in the city. It is not a rider stopped at a red light behind a bus.
In Taipei, motorcycle commuters had the shortest trip of four modes, 28.4 minutes, and the highest PM2.5, 67.5 µg/m³ on average (Tsai et al., 2008).
Now price that. The World Bank puts the 2019 health cost of PM2.5 at $8.1 trillion (World Bank, 2022), and the gap between the station and the visor is not in it.
Key takeaways
- The World Bank put the 2019 health cost of PM2.5 at $8.1 trillion, 6.1% of global GDP on a purchasing power parity basis (World Bank, 2022).
- On the same Taipei routes, motorcycle commuters breathed 67.5 µg/m³ of PM2.5, the highest of four modes, and fixed-site data under-estimated their exposure (Tsai et al., 2008).
- In Stockholm, each 10 µg/m³ of PM2.5 raised the odds of a sick-leave episode by 8.5%, at a station average of only 4.2 µg/m³ (Kriit et al., 2024).
- Studies disagree on riders' lungs: Bangkok motorcycle taxi drivers had lower lung function at 50 µg/m³ PM10 or more (Arphorn et al., 2018), while personal sampling in six Thai provinces found no individual link (Samana et al., 2025).
The research behind this
Peer reviewed measurements this post is built on. Each card says what was measured, what was found, and what it cannot tell you.
Commuter PM exposure by transport mode
- What they measured
- Personal particulate exposure of commuters using motorcycles, cars, buses and the rapid transit system on the same routes in Taipei. Sample size is not in the abstract.
- What they found
- Motorcyclists had the shortest trip (28.4 minutes) and the highest PM2.5, 67.5 µg/m³. Fixed-site data under-estimated their exposure.
- What it can't tell you
- One city in 2008. The abstract gives the direction of the gap, not its size.
PM2.5 and sick leave in Stockholm
- What they measured
- 1.5 million individual sick-leave episodes from 2011 to 2019, matched to PM2.5 at the main urban background monitor, in a case-crossover design.
- What they found
- Odds of a sick-leave episode rose 8.5% per 10 µg/m³ of PM2.5, and 4% of episodes were attributable, about €17 million a year in lost productivity.
- What it can't tell you
- A very clean city (4.2 µg/m³ average) and one station. Applying the slope to a rider in Hanoi is an extrapolation.
PM2.5 and productive life-years lost in Brazil
- What they measured
- Daily national all-cause deaths in Brazil from 2000 to 2019 against PM2.5, converted into productivity-adjusted life years (PALYs) lost.
- What they found
- PALYs lost rose 5.11% per 10 µg/m³ of PM2.5, and about 268 billion US dollars of cost was attributed to PM2.5 exposure.
- What it can't tell you
- Deaths only, at population level. It says nothing about any one commute mode.
Lung function of Bangkok motorcycle taxi drivers
- What they measured
- Lung function of 1,283 motorcycle taxi drivers and 600 enclosed-vehicle taxi drivers in Bangkok, against annual ambient PM10 estimates.
- What they found
- Motorcycle drivers had lower FEV1/FVC, and those at 50 µg/m³ PM10 or more were 2.82% lower again.
- What it can't tell you
- Cross-sectional, so association only. Exposure was estimated from regulator data, not worn sensors.
Personal PM and PAH on Thai motorcycle taxi drivers
- What they measured
- Worn air samplers on motorcycle taxi drivers in six central provinces of Thailand, January to March 2023.
- What they found
- Personal PM was independent of station readings. Pathum Thani drivers averaged 410.9 µg/m³ PM2.5, and individual lung function did not correlate with personal PM.
- What it can't tell you
- Winter only and small samples (22 PM10 and 25 PM2.5 samples at four workstations). The authors flag PM2.5 reading above PM10 as unexpected.
Commuter VOC exposure on bus, motorcycle and bicycle
- What they measured
- Volatile organic compounds (VOCs) breathed by bus, motorcycle and bicycle commuters in Taiwan, and the inhaled dose for each.
- What they found
- Benzene, toluene, ethylbenzene and xylene made up 58 to 68% of VOCs. Motorcyclists had higher VOC concentrations than cyclists, but cyclists had the highest dose.
- What it can't tell you
- Gases, not particles. A particle filter does not remove them.
What the studies actually measured
Six papers sit behind this post. Only one of them compares a rider with a monitor, so it carries most of the weight.
Tsai et al. (2008) put personal samplers on commuters using four modes on the same Taipei routes. Motorcycle riders had the shortest trip and the highest PM2.5. Fixed-site data explained the variation in their exposure well (r² of 0.64), but the regression slope was above 1 for motorcycles and below 1 for cars and rapid transit.
In plain words: the station follows the rider's good days and bad days, but reads low across the board.
Kriit et al. (2024) and Wen et al. (2024) do the pricing. Kriit counted 1.5 million sick-leave episodes in Stockholm and matched them to one urban background monitor. Wen converted Brazilian deaths into productive life years lost. Both turn micrograms into money.
Arphorn et al. (2018), Samana et al. (2025) and Yang et al. (2023) measure riders directly, and they do not agree with each other. That is useful. It shows what a clean answer would need.
Ride with filtered air
Filters haze PM2.5 on every ride
Easi Breezi clips onto any helmet and feeds you H11 HEPA-filtered air. Bike-powered, no batteries, no charging.
The bill is real, and it counts air at the wrong address
The World Bank estimates that PM2.5 cost the world $8.1 trillion in 2019, equal to 6.1% of global GDP on a purchasing power parity basis (World Bank, 2022). Ambient air accounts for $6.43 trillion of it and household air for $1.67 trillion (World Bank, 2022).
The World Bank's price on PM2.5 health damage, 6.1% of global GDP at purchasing power parity.
Source: World Bank, The Global Health Cost of PM2.5 Air Pollution, 2022This is a careful number. The report says the 40% real-terms rise over its 2013 estimate (World Bank, 2022) came mainly from changed exposure-response functions, higher estimates of exposure and the new inclusion of illness, not from a sudden jump in smoke.
It is also a population number. It prices the air over a region, which is the right tool for a ministry and the wrong tool for a person at a junction.
The papers that price a single worker work the same way. Kriit used one urban background monitor for a whole city. The assumption is quiet and almost always true for an office worker: your lungs are near the monitor's air.
Why a fixed monitor reads low for a rider
Riders spend the trip inside the plume. In Taipei, motorcycles idling at traffic lights increased commuters' exposure, and the authors concluded that proximity to traffic emissions contributes to a person's PM exposure on the way to work (Tsai et al., 2008).
Car and rapid transit commuters showed the opposite pattern. A station over-estimated what they breathed.
A passive helmet vent does not fix this. It lets in the same kerbside air at the same particle load, and at idle there is almost no ram air at all.
Treat your city's AQI as the floor for your commute, not the ceiling. The Taipei data say the station is low for riders, never high.
Two studies disagree about riders' lungs
The Bangkok study and the Thai personal sampling study look at the same kind of worker and reach different answers. Both are worth reading together.
| Arphorn et al. (2018) | Samana et al. (2025) | |
|---|---|---|
| Who | 1,283 motorcycle taxi drivers, 600 enclosed-vehicle taxi drivers | Motorcycle taxi drivers, six central provinces |
| Exposure method | Annual ambient PM10 estimated from regulator data | Worn personal air samplers |
| Lung function result | Lower FEV1/FVC, and 2.82% lower again at 50 µg/m³ PM10 or more | No significant link between individual PM and individual lung function |
| Design limit | Cross-sectional; exposure is an area estimate | Winter only; small samples |
The honest reading is that neither design settles it. Area estimates can blur real differences between drivers, and short personal samples are noisy. What both agree on is that the area reading and the personal reading are not the same thing. Samana found personal PM independent of station readings (Samana et al., 2025).
A particle filter does not touch the gas share
Part of a rider's exposure is gas, and gas is a different problem. Benzene, toluene, ethylbenzene and xylene made up 58 to 68% of the VOCs that Taiwanese commuters breathed (Yang et al., 2023).
Motorcyclists' VOC dose was 5.08 µg per minute. Cyclists' was higher at 7.09, because they breathe harder for longer (Yang et al., 2023).
So motorcycles are not the worst mode on every measure. They are the worst on fine particles in Taipei, and a HEPA stage only addresses particles. No single filter fixes every part of the exposure.
Priced to 2031: the bill rises, the rider stays off it
Here is the part that is my own arithmetic. Label it as such.
The OECD projects that welfare costs from premature deaths in its member countries will go from about $1.4 trillion in 2015 to $3.4 to 3.5 trillion in 2060 (OECD, 2016). Interpolating between those two points, with either steady growth or a straight line, puts 2031 at roughly $1.9 to 2.1 trillion.
That is 37 to 53% above 2015.
| Year | Cost (US$ trillion) | Basis |
|---|---|---|
| 2015 | 1.4 | OECD estimate |
| 2031 | about 1.9 to 2.1 | Synthesised interpolation |
| 2060 | 3.4 to 3.5 | OECD projection |
Three limits matter. The OECD figure covers member countries, which excludes Indonesia, Thailand and Vietnam. The two end points come from one model, so the middle is a guess. And the 37 to 53% rise is a change in a modelled price, not a forecast anyone has published.
What the chart cannot show is the rider's share. That is the point. Every extra trillion is priced from area-level air.
Here is how to read the two coefficients without overreaching. Kriit found sick-leave odds up 8.5% per 10 µg/m³ in a low-exposure city (Kriit et al., 2024). Wen found productive life years lost up 5.11% per 10 µg/m³ in Brazil (Wen et al., 2024).
If a rider's air runs 10 µg/m³ above the station's, a model fed station data misses one step of that size. It does not mean riders lose 5% of anything. The slopes are borrowed from other populations, and linear extrapolation is a stretch.
The prediction that is mine, and only mine: by 2031 someone will price a commute with a worn sensor instead of a station. Samana et al. already show that the method exists. Until then, the rider is off the ledger.
For the dose version of this same ledger, see the Air Ledger to 2031. For why an index can hide a peak, see what the AQI number does not measure.
Closing the gap at the face
The gap in this post sits between the air at the monitor and the air at the rider.
Easi Breezi's unit draws kerbside air through a replaceable H11 HEPA filter, which stops 95% of PM2.5, and delivers it to the face with its own fan instead of relying on a vent. It runs off the vehicle's power, with no battery.
It filters particles only. It does not remove gases like the benzene in Yang et al.'s data, and it makes no health promise. It acts on the air the rider breathes, which is the number the bill misses.
Electric bikes do not close this gap either, as the 2031 electrification post shows.
- Treat the station reading as a floor. In Taipei it was low for motorcyclists and high for car and rapid transit commuters.
- Measure your own commute for a week. A small personal PM2.5 sensor on your bag strap shows how far your trip runs from the station.
- Check which monitor a cost headline used. If it is a city station, it prices the average resident, not the person on the bike.
- Filter the air at your face. See how the active HEPA unit works, and keep the limits above in mind.
Frequently asked questions
How much does air pollution cost the global economy?
The World Bank puts the 2019 health cost of PM2.5 at $8.1 trillion, equal to 6.1% of global GDP on a purchasing power parity basis (World Bank, 2022).
Do air monitors under-read motorcycle commuters?
In Taipei they did. The slope between fixed-site data and personal PM2.5 was above 1 for motorcyclists, which means the station under-estimated them (Tsai et al., 2008).
Does air pollution cause sick leave?
It is associated with it. In Stockholm the odds of a sick-leave episode rose 8.5% per 10 µg/m³ of PM2.5, a link seen even at a low station average of 4.2 µg/m³ (Kriit et al., 2024).
Do motorcycle helmets have air filters?
Standard helmets rely on passive vents, which let in the same kerbside air. Active filtration, with a fan and a HEPA filter, is a separate add-on.
Is motorcycle commuting the worst mode for pollution?
On fine particles in Taipei, yes. On VOC dose in Taiwan, cyclists were higher (Yang et al., 2023). Mode rankings change with the pollutant.
Sources
- World Bank (2022). The Global Health Cost of PM2.5 Air Pollution: A Case for Action Beyond 2021. worldbank.org
- Tsai DH, Wu YH, Chan CC (2008). Comparisons of commuter's exposure to particulate matters while using different transportation modes. Science of the Total Environment. doi.org/10.1016/j.scitotenv.2008.06.016
- Kriit HK, Forsberg B, Nilsson Sommar J (2024). Increase in sick leave episodes from short-term fine particulate matter exposure: a case-crossover study in Stockholm, Sweden. Environmental Research. doi.org/10.1016/j.envres.2023.117950
- Wen B, Ademi Z, Wu Y, et al. (2024). Ambient PM2.5 and productivity-adjusted life years lost in Brazil: a national population-based study. Journal of Hazardous Materials. doi.org/10.1016/j.jhazmat.2024.133676
- Arphorn S, Ishimaru T, Hara K, Mahasandana S (2018). Considering the effects of ambient particulate matter on the lung function of motorcycle taxi drivers in Bangkok, Thailand. Journal of the Air & Waste Management Association. doi.org/10.1080/10962247.2017.1359217
- Samana K, Ito K, Suthienkul O, Ketsakorn A (2025). Analysis of PM-bound polycyclic aromatic hydrocarbons exposure among motorcycle taxi drivers in six central provinces in Thailand in winter. PLOS One. doi.org/10.1371/journal.pone.0336587
- Yang HH, Dhital NB, Lai YH, Chan TY (2023). Intermodal comparison of commuters' exposure to VOCs between public, private, and active transportation. Environmental Monitoring and Assessment. doi.org/10.1007/s10661-023-12125-5
- OECD (2016). The Economic Consequences of Outdoor Air Pollution. oecd.org
Written by Ash, mechanical engineer and founder of Easi Breezi, building an active HEPA filtration system for motorcycle helmets (patent pending). Based between Hong Kong and Bali, riding daily in the traffic this blog writes about.