Image from Eric Fischer: General Electric, BART Crosses the Golden Gate Bridge.
Something I have read on Twitter lately is the assertion that transit-oriented development (TOD) doesn’t work in the suburbs. This mostly comes from a place of arguing that new homes will cause irreparable harm to suburban areas resting on the assumption that suburbanites are naturally inclined to drive. While studies of real-world TOD show that it is less effective in the suburbs, it still does its job of reducing traffic from new homes quite well.
What’s a suburb?
The squishy idea of what constitutes a suburb is the first piece of the question. Is it the mileage from the regional central business district (CBD)? The time from the CBD? Is it the built form? Certainly, the form of the suburbs of Paris is different from the suburbs of Amsterdam, which are different from the suburbs of New York City, which are different from those of Dallas, which are different from those of San Francisco. “Suburb,” in this broadest sense, means those places that on the whole send workers into the CBD but for which the CBD is too far (temporally, not spatially) to be part of most non-commute trips.
Marin, whose urban corridor is 25-45 minutes away from its regional CBD, the Financial District in San Francisco, certainly qualifies.
What is TOD?
TOD, short for Transit Oriented Development, is basically development with higher density and a mix of uses focused around a transit station, typically some kind of rail. It’s supposed to encourage people to use transit, to walk more, and to sell their car or at least leave it at home.
Although TOD is often paired with driving-demand management – giving out transit passes, free ZipCar memberships, separating parking rent from home rent – the concern among those concerned about TOD in the suburbs is related exclusively to TOD’s density and urban form. A later post will examine the non-urban aspects of TOD, but today TOD simply means the housing density and mix of uses.
What the literature says
Distance from a CBD is a factor, but it is only one of many. Other interventions like the street grid and the mix of uses also play their own roles. Regionally, the cost of parking at the CBD, the cost of driving to the CBD, and the speed of transit relative to driving are also factors. So, even if the distance is quite far – Novato rather than Sausalito – those other things can outweigh the distance. All these factors, including distance, are called D-variables.
To see how effective any one D-variable is, a researcher will often look at the elasticity of driving: if the D-variable changes by 1 percent, driving changes by some other percent. In one meta-analysis, distance from the CBD was found to have an elasticity of 0.29 [1], so if distance goes down by 1 percent, driving (as measured in miles driven per day, or VMT) drops by 0.29 percent. Sausalito, being 66 percent closer to the Financial District than Novato, then, would see about 19 percent less driving on average than Novato, all else being equal. Another study put the elasticity at 0.22 percent, so Sausalito would see about 14 percent less driving [2].
This makes intuitive sense, too: commuting is often the longest trip of any household and, with the CBD being the center for most jobs in the suburbs, you’d expect less driving if the jobs weren’t so far away.
What is more interesting is how other D-variables influence travel. Though none seem to be as powerful as distance to the CBD, they are powerful together and are easier to control. Land-use mix, intersection and street density, the percentage of intersections that are 4-way, residential density, the jobs/housing balance, and distance to a transit stop are all important D-variables, though just how important depends on the study.
Because all these D-variables are measured in percentages they can be particularly powerful in suburban settings where driving is high, and the environment is built for driving. It’s easier to build 1 percent more four-way intersections and make the land-use 1 percent more mixed than in an urban setting where all such improvements have already been done. Plus, because the typical suburbanite already drives a lot, cutting it by, say, 10 percent (from 30 miles per day to 27) yields more reward than cutting a city-dweller’s from 5 miles to 4.5 miles.
That isn’t to say that cities shouldn’t deploy these same TOD tactics – far from it! Cranking up the D-variables should be done everywhere. Nor should planners ignore the dramatic impact of just living closer to the CBD: San Francisco, Oakland, San Jose, and Palo Alto should all be the first choice for new homes and businesses. But if new development does happen outside of these areas, like in Marin, then it should be done to minimize its impact on existing residents by using these strategies.
Academics and the real world
These analyses are based on observations of the real world, not abstract modelling and guesstimations, with examples taken from around the US and the world. An earlier study by Robert Cervero and G. B. Arrington [3] looked at 17 different TODs, including ones in Fremont, Walnut Creek, Union City, Hayward, as well as the DC, Philadelphia, and Portland metro regions. Each one had fewer driving trips than expected, posting as few as 8 percent the number of trips the Institute for Transportation Engineers (ITE) suspects – 0.55 trips per day per household rather than 6.72 – though the average was a more modest 46 percent cut.
Another study examined the travel habits of 62,000 households in 15 American metropolitan areas from Atlanta to Houston to Seattle [4]. This study’s resulting models are more complex than the elasticities from Cervero and Arrington but are similarly grounded in real-world observation.
There are examples in each sample set that fit train-as-transit-hater and Cato Institute senior fellow Randal O’Toole’s theorizing that transit-oriented development doesn’t work (including two examples in O’Toole’s native Portland) but the studies go further than O’Toole – who tends to stop when the data fits his thesis – and attempt to explain why that occurs.
What about Marin?
Marin is already driving a bit less than ITE would suspect at 6.12 trips per day per household [5]; this is likely due to its extremely low rate of driving to school (just 50 percent [6]) and its relatively low rate of driving to work (67 percent compared to 74 percent for California as a whole [7]).
But let’s say we applied these studies to two different versions of development: one which spread 1,500 homes out over San Rafael broadly (1 building of 20-40 units every few blocks) and one which concentrates those 1,500 homes within a quarter-mile of the San Rafael Transit Center. Let’s also say the new residents are identical to the average San Rafael household. Using the information we have available, what would the overall effect be on San Rafael?
Dispersing the homes across the city would increase the density of the city by 6.5 percent, slightly change the city’s land-use mix, and alter the city’s jobs-housing balance. Overall, depending on the model, the projects would lead an increase in driving between 5.2 and 6.1 percent. That’s less than the 6.5 percent increase in total density and total unit count, but it would still be significant.
Concentrating the homes near downtown San Rafael wouldn’t have much effect on the city at large but would dramatically alter the neighborhood. That would increase neighborhood density by 50 percent and make the land-use of the area 61 percent more diverse. It would also focus these homes 83 percent closer to a major transit station than the average San Rafael home. Overall, such a plan would increase driving in the city by between 4.0 and 5.3 percent – significant, but noticeably less than the dispersed plan.
Because Marin is laid out linearly, any new home would have commute-time impacts both where it’s located and everywhere to their south on the freeway. The benefit of the dispersed plan is that neighborhood streets wouldn’t see much impact, even as traffic would increase more for everyone passing through or living in San Rafael. The benefit of the concentrated plan is that the impacts on the county as a whole would be less even though impacts on downtown San Rafael would be more.
Both the dispersed and focused plans could see even fewer trips if those 1,500 homes were predominantly senior housing (few-to-no commute trips, which is when there isn’t road capacity for them), predominantly affordable housing (higher-income people drive more), or predominantly smaller housing (meaning smaller households, which don’t generate as many trips). Other policies, like separating the rent for parking from the rent for the apartment or giving residents Clipper cards, could also cut driving.
What it all means
For there to be no new traffic whatsoever, the county would need to fight traffic not just from new homes but existing ones as well. Faster transit trip times to and from downtown San Francisco would be exceptionally helpful for this goal, with new bus shoulder lanes on Highway 101 and extending the planned Van Ness BRT lanes to the Golden Gate Bridge via Lombard. Electrifying SMART and extending it to the Transbay Terminal would be an expensive $5.5 billion proposition, but it would put San Rafael and southeast Marin within half an hour of San Francisco. Higher parking costs within The City and a higher toll would also suppress driving.
For driving within Marin, higher parking costs in central San Rafael, a better urban protected bike lane system, and allowing homes to be converted to office space (thereby increasing land-use diversity) would incentivize people to leave their cars at home.
Still, the closer people are to jobs, the less driving they do, so the bulk of new housing in the region should go to San Francisco, Oakland, the Caltrain Corridor, and San Jose. But to the extent that new housing is built in Marin, it should go as much as possible into dense homes close to San Francisco-bound transit: the three ferry terminals, the San Rafael and Novato transit centers, and the San Anselmo Hub.
The idea that dispersing development around the county will be less impactful than focused TOD is only correct if we look at particular neighborhoods. If we zoom out to the county and region, however, it’s obvious that TOD will mean less impact.
Works Cited
Image: Selection from General Electric. BART Crosses the Golden Gate Bridge. April 1, 1961. Catalogued by Eric Fischer on Flickr.
[1] Mark R. Stevens, “Does Compact Development Make People Drive Less?,” Journal of the American Planning Association 83, no. 1 (January 2, 2017): 7–18.
[2] Reid Ewing and Robert Cervero, “‘Does Compact Development Make People Drive Less?’ The Answer Is Yes,” Journal of the American Planning Association 83, no. 1 (January 2, 2017): 19–25.
[3] Robert Cervero and G. B. Arrington, “Vehicle Trip Reduction Impacts of Transit-Oriented Housing,” Journal of Public Transportation 11 (September 1, 2008).
[4] Reid Ewing et al., “Varying Influences of the Built Environment on Household Travel in 15 Diverse Regions of the United States,” Urban Studies 52, no. 13 (October 1, 2015): 2330–48.
[5] California Air Resources Board, Mobile Source Emission Inventory - EMFAC2017 Web Database, version 1.0.2 (Sacramento, CA, 2018).
[6] “Marin Safe Routes to Schools Program Evaluation” (San Rafael, CA: Safe Routes to Schools, September 2016).
[7] U.S. Census Bureau, “Table B08006: Sex of Workers by Means of Transportation to Work,” in American Community Survey 2016 1-Year (Washington, DC: U.S. Census Bureau, 2017).