Yes, cars really are bad for the environment

Transit agencies tout themselves as fundamentally “green,” a real solution to global warming and environmental pollution. Every so often, however, an anti-transit activist will complain that buses pollute more than cars or point to SMART’s decision to run diesel trains and say they are just making the pollution worse.

As it turns out, the transit agencies are often right, and the anti-transit activists are often wrong, though not always.

I measure a transit agency’s efficiency in terms of passenger miles per gallon-equivalent, or pMPGe. This is how much greenhouse gases the transit agency creates, in terms of how much gasoline you’d need to burn to create the same amount of gases, divided by the passenger miles traveled carried by the agency.

Before I begin, however, you should note what this is measuring: how efficient the transit agency is with its current ridership. This isn’t the same as a car’s MPG because, when you choose to use the car, you’re burning more fuel than would otherwise have been burned. That adds pollution to the environment, even if it’s an electric car.

When you get on a bus or train, however, you’re making the whole system more efficient: they’re going to use that fuel or electricity whether you’re on board or not.

With that in mind, let us proceed.


Ferryboat 13
Demand Response 9
Trolleybus 134
Vanpool 91
Commuter Bus 58
BRT 44
Motor Bus 31
Heavy rail257
Aerial tramway 165
Cable car 131
Light rail 126
Streetcar 100
Commuter rail 85
Hybrid rail 50
Inclined plane 48
Monorail & AG 38
Alaska Railroad 28

Glossary: pMPGe: Passenger-miles per gallon-equivalent of gasoline. PMT: Passenger miles travelled.

Modes that run on electricity do outstandingly, in part because gasoline is about as dirty of a fuel as coal when it comes to greenhouse gases. Trolleybuses, aerial tramways, and heavy rail (like subways) are all very good at moving people with minimal environmental impact. Next up are modes that are mostly full for their journeys and run long distances: commuter rail, commuter buses, and vanpools are firmly in this category.

California’s personal car fleet does respectably. If we just look at the vehicles themselves, they get about 24 pMPGe on average, aggregating together diesel, electric, hybrid, and electric cars. Adding the typical passenger load of 1.6 people per vehicle and that jumps up to 38 pMPGe. California predicts that by 2050, these numbers will be 43 and 69, respectively. As for Marin, its self-image as a bastion of “green” driving, the vehicle fleet is just 2 percent more efficient than California’s as a whole.

Buses that don’t run on electricity (motorbuses) are middle-to-low level, dragged down mostly because diesel is such a dirty fuel as far as GHGs are concerned but also because of low loads on many buses that serve areas that are difficult or impossible to serve will with transit.

Ferries are terrifically poor in this metric, getting a measly 13 pMPGe. In the Bay Area, that’s in part because of significant one-way peak traffic, creating a lot of near-empty trips back out to the suburbs to get more people. Even in Seattle, where there is a lot of two-way traffic out to the Puget Sound islands, they barely scratch 30 pMPGe.

Finally, scraping the bottom is demand-response service. These are Dial-A-Ride and paratransit services designed as a last-resort transportation option for the poor, elderly, and disabled. These systems aren’t designed to be efficient, so it’s no surprise they underperform. Those that do well don’t use diesel fuel, relying instead on cleaner biodiesel, gasoline, or ethanol.


BART 441
ACE 109
Caltrain 101
VTA 51
AC Transit 35
WestCAT 31
SolTrans 31
Santa Rosa CityBus 29
Unitrans (Davis) 29
Golden Gate Transit 27
Marin Transit 26
Napa VINE 26
Tri Delta Transit 25
SamTrans 23
County Connection 19
Sonoma County Transit 18
Petaluma Transit 11

If we add up all the modes each agency operates, BART wins at a whopping 441 pMPGe. The next competitor, Phoenix’s Valley Metro Rail, isn’t even close at 247. Rounding out the top five are Pennsylvania’s Department of Transportation commuter rail at 211, New York City’s MTA at 169, and the South Shore Line (a commuter train linking South Bend, Indiana, to Chicago) at 164. These are all, expectedly, dominated by electric heavy-rail modes.

Closer to home, Marin Transit and Golden Gate Transit both get relatively high marks for their bus services (31 and 46 pMPGe) though less for their other services (demand-response is 8 and 6, while the ferry gets just 14). It’s too early for SMART to have the numbers needed to calculate, but it’s likely to be in the 50-70 pMPGe range based on peer systems and the efficiency of the vehicles (1.1 gallons of diesel per mile).

Hiding under raw efficiency

Something we don’t see in these numbers is the car trips avoided because transit enables denser, walkable/bikeable places [1] and the corresponding walkable lifestyles. San Francisco’s Muni buses are terribly inefficient because of the hills and their age, but for many, the trip to the store or a restaurant can be made on foot because the built environment – enabled by Muni – allows it. In other words, the seeming inefficiency of some public transit modes, like the bus, can be offset by all the walking or biking or skateboarding or whatever trips that never even involve a motorized vehicle. In many cases, it could more than offset agency’s fundamental inefficiency.

Another benefit from agencies like Golden Gate Ferry or WETA, which don’t do much for greenhouse gas emissions on its own, is that they relieve San Francisco of the burden of building parking for those commuters. That makes The City more walking-friendly and relieves the highways of some of its traffic.

To account for this, I modeled a target efficiency multiplier or the needed number of additional car-passenger miles a transit agency would need to offset in order to be at least as efficient as driving from a GHG perspective. Golden Gate Ferry, for instance, has a multiplier of 2.65, meaning it would need to offset 2.65 times its current PMT (so, for every 1 passenger mile on the ferry, 1.65 would need to be avoided elsewhere.) Marin Transit’s bus system has a multiplier of 1.24, so it would need to offset 0.24 car-passenger miles for every 1 it carries.

If the multiplier is less than one, then that’s how much ridership the transit agency would have to lose to be as efficient as California’s car fleet. Caltrain, for instance, has a multiplier of just 0.37 – it would need to have a ridership just 37 percent the size of its existing ridership to be as inefficient as cars.

Transit could be greener, but it’s pretty green already

When you hear about a new train that runs on diesel or see a near-empty bus pull away in a cloud of black exhaust, it’s easy to imagine that these modes are pretty terrible for the environment. Though they could use cleaner fuels, they more than make up for whatever inefficiency they have by carrying more people and by allowing people to walk or bike and forego vehicles entirely.

So, salute that half-empty bus next time you see it: it’s probably doing more to keep your community clean than you realize. Especially if it runs on something other than diesel.


First, I found the number of gallons of fuel burned and kilowatt-hours (kWh) of energy consumed by every transit agency in the National Transportation Database (NTD) [2]. I then found how much of 3 different pollutants – CO2, N2O, and CH4 – are produced per gallon of each fuel [3] and how much are produced per kWh of electricity in the appropriate part of the national electrical grid each agency operates within [4]. I divide the number of passenger miles traveled on each agency’s mode, also according to the NTD [5], by the number of gallons or kWh to determine the number of passenger miles per unit of energy.

Each of the three pollutants can be translated into CO2-equivalent (CO2e) units for comparing their global warming potential impact. CO2 is worth 1 CO2e, CH4 is worth 25 CO2e, and N2O is worth 298 CO2e. The emissions from fuel and electricity are converted into CO2e per passenger mile traveled.

Finally, I compare this to how much CO2e is released by burning a single gallon of gasoline to determine the passenger miles traveled per equivalent gallon of gasoline, or pMPGe.

To determine the pMPGe of an agency, I performed the same set of calculations but on the aggregate number of passenger miles, gallons of fuel consumed, and kWh of electricity used for the agency across all modes. To determine the pMPGe of a mode, I performed the same set of calculations but aggregated by mode instead of by agency.

My somewhat-messy full table, including active Excel formulae, is available here.

Works Cited

[1] Kenneth Joh et al., “The Walking Renaissance: A Longitudinal Analysis of Walking Travel in the Greater Los Angeles Area, USA,” Sustainability 7, no. 12 (July 10, 2015): 8985–9011.

[2] National Transportation Database, “2016 Annual Database Energy Consumption” (Federal Transit Administration, September 29, 2017).

[3] Environmental Protection Agency, “Emission Factors for Greenhouse Gas Inventories,” April 4, 2014.

[4] “Emissions & Generation Resource Integrated Database (eGRID2014v2)” (Environmental Protection Agency, February 27, 2017).

[5] National Transportation Database, “2016 Annual Database Service” (Federal Transit Administration, September 29, 2017).