Car-Free Households Are Booming in San Francisco

Image: Michael Rhodes

San Francisco is quickly adding residents, but very few cars.

Between 2000 and 2012, the city has seen a net increase of 11,139 households, and 88 percent of them have been car-free. That’s according to an analysis of U.S. Census data by Michael Rhodes, a transportation planner at Nelson\Nygaard and a former Streetsblog reporter. One net result of this shift is that the proportion of San Francisco households who own zero cars increased from 28.6 percent in 2000 to 31.4 percent in 2012, the fifth-highest rate among large American cities.

The stats show that the city’s average car ownership rate is declining, even as the population is growing. The data don’t distinguish where specific households are foregoing cars, so this doesn’t necessarily mean that the residents of all the new condo buildings going up are car-free. But the broader effect is reverberating throughout the city, whether car-free residents are moving in where car-owning residents previously lived, or residents are selling their cars.

This finding flies in the face of complaints from NIMBYs who protest new housing developments that forego parking, based on a faulty assumption that new residents will own cars anyway and take up precious, free street parking. That’s one of the arguments heard from proponents of the cars-first Proposition L, who complain that “the City has eliminated the time-honored practice of creating one parking space for every new unit.”

“A lot of people who are moving here are choosing it because it’s a place you can get around without a car,” said Livable City Executive Director Tom Radulovich. “People will self-select. If convenience for an automobile is their criterion, there’s a lot of places in the city and elsewhere” to live.

Radulovich noted a number of changes in the 21st century that have made it easier not to own a car. San Francisco has expanded its bike lanes, car-share services now exist, and taxi service has improved (besides the new “ride-share” apps like Uber and Lyft). Muni, BART, and Caltrain ridership have also increased to record levels over the years.

And, Radulovich noted, even the new wave of tech workers tends to get to work on shuttles — unlike those in the dot-com boom of the 90s, who favored living near highways 101 and 280 so they could easily drive to Silicon Valley.

“There’s a lot of people who get to work on the tech buses — those might’ve been car owners a decade ago,” he said.

Across the country, Americans are driving less, and millennials in particular are generally more interested than earlier generations in car-free city living. Radulovich pointed out that it’s often older residents who insist that the city build more free parking (see: Prop L). “There’s definitely a set of cultural expectations, from those raised in the ’50s and ’60s, for car ownership and automobility,” he said.

According to the Census‘ American Community Survey, 31.4 percent of SF households surveyed in 2012 were car-free. That’s up from 29.8 percent in 2009 and 28.6 in 2000. (Note: The 2013 SFMTA Transportation Fact Sheet incorrectly cited the most recent stat as 21 percent.)

Even among the 12 percent of new households since 2000 that do own cars, many are car-lite — not each individual owns a car. SF has actually seen a net decline in two-car households despite growth in total households, while the number of households with three or more cars increased by slightly more.

Here’s a more detailed breakdown of changes in car ownership rates since 2000, provided by Rhodes:

Unit type Change
All units +11,139
Car-free +9,837
1 car +1,242
2 car -1,123
3 or more cars +1,183

The statistics show the efficacy of providing better alternatives to car ownership, but they also underscore the importance of building more housing in walkable, transit-oriented neighborhoods throughout the Bay Area.

“Certainly, we need to find places in those [walkable San Francisco] neighborhoods where we can build. The good news is, if the new households that are coming to those neighborhoods are car-free, then we can spend a lot less space on parking and use that for housing, and the automobile-oriented impacts [like traffic] of increased density may not materialize,” said Radulovich. “But San Francisco’s a small place — and if we’ve got a whole generation that wants walkable, urban places, and this is the only place in the Bay Area that offers that, then there’s gonna be a lot of competition for it.”

  • Dave Moore

    This is an extremely misleading way to characterize these numbers. Just because you mention it below the lead doesn’t make it any less disingenuous. It’s not that 88% of new households are car free. It’s that there has been a minor shift of households to ones without cars. In 2000 there were approximately 350,000 households. If there are 11,000 more of them with no cars then it’s a shift of 3.1% over 12 years, or less than 3/10 of a percent per year. That is still an interesting number and especially if you could show a steady trend over that time it might be compelling.

    It would be more interesting to show the total increase in population and the total increase in the number of cars to see whether the per capita rate has decreased. I expect it has but not in a way that allows you to make it appear so drastic.

    It’s too bad, because you might be able to make your point in a more honest and straightforward way.

  • Dave Moore

    For details on the number of households see

    I accidentally used a slightly inflated number of 350K. That was units but only 329,700 were occupied in 2000. So it’s a 3.34% increase over 12 years, still less than 3/10 of a percent per year.

  • jmeinken

    The article is addressing a common argument about urban policy that every new housing unit should require a new parking space because one household = 1+ new cars. The point of the article is that SF has been adding housing while the number of cars has only increased by 1/10 as much.

    Agreed that the graphic is misleading, but I don’t think the intention of the author was to trick people.

  • While I really wish I could take this story to heart, I don’t know how to reconcile it with the DMV stat covering essentially the same period showing vehicle registrations in San Francisco going up by about 10,000 cars.

  • They say “don’t become a statistic”, and I guess I didn’t, since I didn’t join a new household in that period of time. But I did sell my car and go car-free in 2009. So there’s that.

  • Sanfordia113

    Is this the household survey performed by door-to-door survey workers? Does this survey data include car ownership self-reporting, or is that figure obtained from DMV records? If the latter, it is an inaccurate representation, as many people who bought or inherited a car before living in SF do not bother to re-register the vehicle license, because doing so will not only increase vehicle license fees, but also increase insurance premiums.

  • cwalkster

    SFMTA issues an annual Transportation Fact Sheet. They list the number of households without a vehicle using ACS data
    in 2010 was 30.3%,
    in 2013 was 21.0%

    The author wants thinks he can mislead the readers by obscuring the big picture.

    SFMTA fact sheet shows the number of households that own two cars
    in 2010 was 21.2%
    in 2013 was 28.1%.

    The author writes “SF has actually seen a net decline in two-car households since 2000 despite growth in total households.” This is the complete opposite of what SFMTA says.

    We can’t trust what M Rhodes writes. He hides sources. And is more interested in distorting facts to prove his points.

  • Michael Rhodes

    The SFMTA fact sheet data is incorrect. They are aware of this and are in the process of correcting it. The source for the data they cite (3-year-average ACS data) is here:

    As you can see, the actual share of households with no vehicles is 30.7%.

  • Michael Rhodes

    It’s US Census data, so it’s collected by door-to-door survey workers. Agreed, DMV data for the address of vehicle registration is very inaccurate for a number of reasons, so I don’t generally trust it.

  • Michael Rhodes

    In that case, you are part of the statistic, since (as the figure states) it’s the “net increase”, i.e. including existing households that sold vehicles (which could be offset by new households that have vehicles).

  • Michael Rhodes

    DMV vehicle registration data isn’t very accurate. People often don’t update the address on their vehicle, among other issues. I suspect there’s also some lag in their reporting. The data ultimately isn’t design for analysis at this fine grain, unlike US Census data, which tells you exactly what the margin of error is.

  • Michael Rhodes

    That is the point I intended to make. I certainly didn’t want to trick people — hence including “net increase” in the title, which indicates this is about new households, not about the total population. I also asked Aaron to explain this in the article, which he did quite clearly.

  • Michael Rhodes

    The graphic is all about addressing people’s concerns about growth and increased traffic, demand for parking, etc. As it shows, our shift towards lower vehicle ownership appears to be making it possible to grow without adding many vehicles. It ultimately doesn’t matter whether that’s because new residents don’t have vehicles or because existing residents are selling them. It’s probably more of the former than the latter, but it’s definitely some of both.

  • Guest

    The wording of the ACS data is unclear. The numbers are a percentage of workers 16 and over with “no vehicle available”, “1 vehicle available”, etc. It doesn’t specify ownership. So while the percentage without access to a vehicle decreased in the survey, do we know how much that reflects increased car ownership vs increased car availability via car sharing?

  • binaslice

    Fascinating comments by the ‘You Must Own a Car’ lobby. I trust you’re not all auto dealerships. Regardless, our car-centric world is changing. This is being lead by the Y Generation. So, Mr and Mrs Negatory wake up and smell the coffee.

  • I added ACS stats on total household car ownership in the article so as not to give the impression that we’re “obscuring the big picture.” As Michael noted, the 2013 SFMTA Fact Sheet cited those stats incorrectly.

  • M.

    Reported at the SFMTA Board’s Policy and Governance Committee meeting yesterday: From 2011 to 2014, private car travel went from 55% to 50% in the City of SF. More refined parameters of that stat should appear when the update is posted on the SFMTA’s site soon.

  • Andy B from Jersey

    Despite all the debate about the accuracy of the numbers, I think what you have here are people being priced out of car ownership in SF just as they are in NYC. It’s just too damned expensive to store a car in these cities and considering how much space goes for a premium in these places, that make sense. I don’t think the access to transportation alternatives is the primary causal factor. Having access to varied alternative transportation means is more of a reality of car storage being so impractical and damned expensive!

  • murphstahoe

    Step 1: Move to San Francisco
    Step 2: Leave your car registered in Modesto
    Step 3: Your neighborhood adds RPP
    Step 4: Get Parking permit tickets
    Step 5: Register your car in SF

    More RPP = more people in compliance. RPP areas have been increasing

  • murphstahoe

    People aren’t being priced out of car ownership in SF because the price of owning a car is going up. The price of a car has stayed nominally flat. Parking costs from the City have not risen much in absolute costs, aside from parking tickets which are avoidable (as for the most part are meters, and RPP stickers are close to free). If you get housing without parking, you pay nothing on that end and parking on the street is free.

    People are being priced out of car ownership in SF because *housing* costs have skyrocketed, squeezing budgets to the point that the car becomes an easy way to make a major chop in your budget. At some point, when it comes down to brass tacks, if you have a choice between having a car and having a place to live, almost everyone chooses a place to live.

    And sadly, all that parking is a large part of the reason that housing costs are so high.

  • I’m actually inclined to go with what @murphstahoe:disqus posits, but only because anecdotally I lived in SF for many years but registered my car at my parents’ address in Berkeley because of some perceived (or real, I can’t recall) savings in insurance and fees. Is that the case for everyone else? Who knows. But with so much noise in the data, I don’t think we can rejoice just yet about such a tremendous gain in non-motorist households.

  • If more existing cars are being registered locally, it seems like that would obscure some of the car-free household growth, if anything. But as Michael said, census data doesn’t rely on registration.

  • cwalkster

    Don’t understand how DMV data isn’t very accurate.They track every single vehicle.

    As of Dec.2013 DMV says there are
    397,238 cars; 57,466 trucks; 22,610 motorcycles;
    Total of 477,314 registered cars, trucks, motorcycles in SF.

    If a vehicle owner does not update their new address to DMV they cannot register their vehicle. The US Census doesn’t send vehicle registration forms each year to vehicle owners.

    US Census data are estimates.You admit that when you write “unlike US Census data, which tells you exactly what the margin of error is.”

    DMV registration data is always more accurate than US Census estimates.

  • cwalkster

    Could M Rhodes please tell us where we can read your analysis/report?

  • gneiss

    You are falling victim to the difference between precision and accuracy. While the DMV data is precise, it is not accurate. As Upright Biker points out below, many people have relatives that live outside the city where registration information can still be sent, allowing those residents to have their vehicles registered out of the city. Conversely, some people may have their vehicles registered here, but located in a different community. As the DMV doesn’t track this information, it is difficult to get an accurate margin of error for a count of the number of vehicles in the city and how they are assigned to each household.

    By contrast, the US Census has collect enough data over the decades and they have experience with that data to come up with highly refined margin of error calculations on their door to door surveys that can better reflect the number of vehicles at each household. This gives a more ‘accurate’ representation of the nature of car ownership then does the DMV data.

  • sebra leaves

    I don’t know who owns them but there are a lot more cars on the road than bikes or pedestrians, and is a huge demand to keep SF traffic flowing smoothly regardless of where the cars are registered or how long they stay in town. The proponents of Prop L and their followers are not going to continue to allow a vocal minority to run their lives into the ground. They are fighting back and we shall see how many of the voting public blame the SFMTA.

  • jd_x

    “The proponents of Prop L and their followers are not going to continue
    to allow a vocal minority to run their lives into the ground.”

    Oh, those poor motorists getting their lives run “into the ground”! You know, just like the nearly 1000 pedestrians and cyclists who literally get run into the ground by cars. But kudos to you for the most over-exaggerated, blown-out-of-proportion comment I’ve read on this site in a while. Keep it up: you’re doing a fantastic job making the case against the car-centric, anachronistic, irrational nonsense that is Prop L.

  • I just read a tragic story about a 2-year-old girl who literally had her life run into the ground here in SF by the very mode you so pathetically defend. You disgust me.

  • Cars take up so much space, the whole street could be filled with them, but they still contain fewer people than the small group of pedestrians standing at the corner.

  • Jeffrey Baker

    How many cars do the car sharing services own, and where are they registered?

  • Tim Bracken

    And a red-light-running bicyclist killed a pedestrian in SF a couple years ago. Does that mean all bicyclists are “disgusting” and “pathetic?” Sweeping generalizations like this are absurd and tarnish any argument you’re trying to make.

  • It’s the “having their lives run into the ground” statement that triggered that remark, if you didn’t catch on.

  • Tim Bracken

    Sorry if I missed your meaning. Seemed like you were painting all drivers with a broad brush, when in fact last night’s hit-and-run driver is as similar to other drivers as Chris Bucchere is to all cyclists.

  • JJ94117

    Except that you are calling out 1 incident over the course of how many years? versus the number of cyclists/pedestrians hit by drivers in the same amount of time. Heck, today’s headlines alone lists more than that. Oh, and Mr. Bucchere was prosecuted. Can’t say that for the vehicle drivers, even when they are caught on camera.

  • It’s Sebra Leaves personally to whom I was directing the remark. This is a person whose mindless support of all things auto centric, even in the face of little girls repeatedly dying in crosswalks under the wheels of cars, makes one question the morality of that person.

  • Tim Bracken

    No doubt about it, given the relative sizes of their vehicles, reckless drivers kill far more people than reckless cyclists. Those reckless drivers should be prosecuted aggressively. In my Disqus comment history, you’ll see that I’ve gotten exasperated at the number of times the SFDA seems to be light on reckless drivers (“Well Loved Bartender Erik Dean Killed In Motorcycle Accident,” “Mustang Driver
    Allegedly Responsible For SF Hit-And-Run Rampage Out On Bail,” etc.).

  • cwalkster

    I’d say gneiss is getting confused with precision and accuracy. We are not dealing with experiments where time is measured in seconds, miliseconds, microseconds.

    Calif. DMV is the only state agency that strictly tracks Calif. vehicle registration. They know exactly how many Calif. vehicle registrations were issued each year. There is no margin of error.

    The US Census bureau estimates data and doesn’t have the same information DMV keeps. You say US Census “gives a more accurate rep. of car ownership than does DMV data.” Lets see if you want to put your money where your mouth is.

    I gave a link to DMV vehicle registration for 2013. Please show US Census estimates of vehicle registration for SF in 2013. And we’ll compare them.

  • murphstahoe

    Because *nobody* paints all cyclists with a broad brush….

  • gneiss

    Michael Rhodes did that all ready. Here is the link again:

    Please cite where in the DMV data they indicate the margin of error based on people registering their vehicles in communities other than the ones they live in.

  • Tim Bracken

    Of course they do, and that’s an equally absurd generalization.

  • BP

    Many factors contribute to people deciding whether to adopt a car-free lifestyle or personal ownership of an automobile and access to transit, bicycle lanes or car sharing options may be secondary. Multiple unit buildings that actually having off-street parking often charge a fee for rental of a parking space separate from the rent for the dwelling unit. Having a residential permit parking sticker doesn’t guarantee an on-street parking space available in the neighborhood. Also some people judge the viability of maintaining a personal automobile by the value of the car versus the parking tickets and penalties owed, i.e., once the fines exceed the value of the vehicle the owners often abandon the vehicles to the City to take possession. These personal decisions are likely to have greater weight in determining whether to maintain a private automobile will grow in importance unless mass transit can provide a reliable and comfortable service to the population.

  • oakland_biker

    If you want to live in one of the densest cities in the US where a car is an unnecessary luxury, isn’t that a price you should be willing to pay?

  • Michael Rhodes

    This is just a one-off analysis of basic US Census data, but I will include the underlying figures below for anyone who is interested.

    2000 US Census

    Occupied housing

    No vehicles

    1 vehicle

    2 vehicles

    3 or more vehicles

    2012 American Community Survey, 5-year averages (US Census Bureau)

    Occupied housing

    No vehicles

    1 vehicle

    2 vehicles

    3 or more vehicles

    Change from 2000 to 2012 (5-year average)

    Occupied housing

    No vehicles

    1 vehicle

    2 vehicles

    3 or more vehicles

    2012 (5-year average):

  • cwalkster

    Sorry gneiss, M Rhodes did not post the US Census estimate of vehicle registration for SF in 2013. The link was for 3 year estimates 2010-2012.
    You didn’t answer my question. Please show US Census estimates for vehicle registration for SF in 2013.

    DMV data shows paid vehicle registration for SF.
    Why would someone pay vehicle registration for SF and not live in SF?

    I do not know where DMV posts data that indicates the margin of error of people registering vehicles outside of they live. DMV only works on address supplied by the vehicle owner. If the vehicle owner wants to provide false info, there is nothing DMV can do.

  • SF Guest

    If SF car owners aren’t being priced out of car ownership then why are motorists paying $7/hr meter rates; why is there a proposal for a substantive VLF increase which doesn’t contribute to the ease of motorists finding parking; why is there a proposal to expand meters to residential areas; why is there resistance to overturning Sunday meters; and why is there a $500M transportation bond? (Rhetorical question) You can put whatever spin you want but the idea behind all these fees is to demote car ownership.

    If my examples are not an attempt to price SF residents away from car ownership, what is it? Regardless of the reasons no other Bay Area city’s rates are as high as SF so I would have to agree people are being priced out of car ownership in SF.

  • coolbabybookworm

    Your last paragraph is precisely why the DMV information isn’t necessarily reliable. It works on a state-wide level for the most part but not city wide. As many people have posted here, it’s common to not register in the city you live or update your address when you move. I remember SFMTA data of some neighborhoods considering RPP showed that less than 30% of overnight parked cars were registered to that neighborhood. It’s not common for people to register in SF and live elsewhere, but it is somewhat common for people to register elsewhere and live in SF. My friend had Wisconsin plates registered at her parent’s for years before she bothered changing it.

  • Dave Moore

    I’ll posit that if anything the DMV records undercount registrations in SF as insurance rates are higher here. And that the later numbers are more likely to undercount than earlier numbers because more people have been moving here from elsewhere at a greater rate.



    In net there was a very slight reduction in the rate of vehicle per resident from .573 to .570, mostly accounted for by an absolute reduction in the number of trucks registered in the city of 1175. Auto rates stayed about constant while motorcycle rates rose.

  • cwalkster

    Didn’t receive an answer: where is the analysis available for reading?

    2013 SFMTA Transportation fact sheet most likely citing 2012 DMV paid vehicle registration in SF shows:
    385,442 cars; 56,694 trucks; 21,693 motorcycles. Total of 463,833.

    US Census estimate of vehicles available in 2012 in SF is 346,842.

    Don’t know what the US Census counts as vehicles, only cars; cars and trucks; cars, trucks and motorcycles, …

    If the US Census only counts cars as vehicles, the difference between DMV and US Census car data is 38,600. An 11% error.

    If the US Census counts cars, trucks, motorcycles then the difference between DMV and US Census vehicle data is 116,991. A 33.7% error.

    Your US Census only data produces in the best case an 11% error. Your analysis is not good.

  • Michael Rhodes

    There is no 2013 Census data for vehicle registration yet. Comparing DMV’s registration data for 2000 with 2013, the rate of vehicles registered per capita in 2000* was 0.584, which dipped to .580 per person in 2013. The rate of autos per person (excludes trucks, motorcycles, etc.) has inched up 3% from 0.460 autos per person to 0.474 autos per person. One possible explanation is that while car-free households are more common, households with 3 or more cars may actually own more cars now than they did in the past.

    In terms of data accuracy for the DMV, the issue is that people who move to SF are often slow to change registration. In periods of fast growth, that means our vehicle registration may be undercounted. This was certainly the case in 2000, and in 2013 as well. It’s difficult to say how the rate of delayed registration has changed over time (the VLF is now lower, which may incentivize registering in CA faster), so it makes these statistics hard to compare at a fine grain.

    Another important piece of evidence is that congestion hasn’t gotten worse in the last few years as the economy has grown. Ultimately, that’s the bottom line that matters.

    *-Vehicle registration for 2000 is actually from 12/31/1999, the closest data available.


    2000 vehicle reg data:
    2013 vehicle reg data:

    2000 population:

    2013 population:

  • Maron

    Michael cited exactly where the data comes from. It’s from the US Census, the gold standard on car ownership rates. His analysis is impeccable.

    By the way, the decline in car ownership in SF is great fucking news if you absolutely, positively must drive in the city.


Census: 95% of New SF Commuters Since 2006 Don’t Drive Solo

As San Francisco’s economy booms, a lot more people are commuting, and very few are doing it in a car. Between 2006 and 2014, the city saw a net growth of about 86,400 commuters, and 95 percent of them don’t drive, according to data from the US Census American Community Survey. The ACS numbers provide the best available […]

SF Chronicle Regurgitates Misinformation From the Free Parking Crowd

The SF Chronicle printed an op-ed this weekend, written by the Republican-backed group that aims to “restore balance” on San Francisco’s streets. And by “balance,” they mean enshrining a status quo where cars, not people, get the lion’s share of the public streets, in the form of more pavement and more traffic. Unfortunately, the Chronicle […]

Studies Show Car Traffic in San Francisco is Dropping

Car traffic has dropped in San Francisco in recent years, despite an economic boom and a growing population, according to studies by the SF County Transportation Authority. A newly updated study (reported by SF Weekly) by the SFCTA counted fewer cars at 11 of 15 intersections during evening peak hours this year, compared to earlier […]