Tag Archives: San Francisco

The Limits of O’Toole-onomics

Update: Randal O’Toole was kind enough to respond via email. I’ve updated the post to reflect those corrections, and added his full comment and my reply at the bottom of the post.

This thought has been kicking around in my head for a while, but this Next City – The Works post by Stephen J. Smith on commute times in cities finally motivated to me to hash it out.

It’s long been noted that, super commuters aside, human beings tend to have a fairly constant travel time budget. This means that increases in the average speeds of transportation facilities often result in people traveling further distances in the same amount of time, rather than the same distances in less time. It also means that, given an average speed for a mode of transportation, there’s a practical limit to the size of city you can serve primarily with that mode.

For example, in a rural town that predates cars, you can access everything in the town by walking. No matter where you are, nothing would be more than a mile or two away. People might bike or drive to save time out of convenience or to avoid unpleasant weather, but functionally, the town can work without cars. For example, if you’re in Lone Pine, you can get to anything else in Lone Pine just by walking.

Biking expands your reach, and in a small city – say the size of Merced or Santa Maria, maybe even Santa Barbara or Ventura – could provide you access to everything the city has to offer. Now, maybe bicycle facilities in some of those places are sadly lacking, but that doesn’t mean the concept is technically unsound. We could make it work if we wanted to.

If your city gets much bigger than that, though, you need some type of higher speed transportation. There are many possible combinations that work. For example, New York and Boston provide rapid transit to move you quickly across parts of the city, depending on you to walk the last bits of your trip. Places with huge bike usage, like Amsterdam and Copenhagen, provide transit and plenty of bike parking. Phoenix and Houston give you freeways and craploads of car parking. Ignoring environmental, aesthetic, and efficiency concerns, the only requirement is that you increase the amount of distance people can cover in the same amount of time.

In very large metro areas, it’s hard for even freeways and rapid transit to overcome the distances, and as a result, new nodes of development start to spring up – places like Irvine and Tysons Corner – to keep commuting times down to what people will tolerate. And in fact, despite the perception of Orange County as a suburb of LA, 85% of people who live there work there as well. Cross county flows are about the same in each direction – 180,000 live in Orange County and work in LA County, with a similar number doing the opposite.

Okay, we have the technology to build lots of freeways, transit, whatever – so why don’t metro areas just sprawl out into infinity to keep land costs down? Well, working in opposition to things that tend to decentralize cities, like quality transportation and communications, we have agglomeration economies. Basically, people and businesses want to be located as close as possible to the people and businesses that they interact with. If you want to start a movie studio, it makes sense to do it in Los Angeles, where there are lots of people you need like actors, grips, gaffers, show runners, and so on. If you know how to write smartphone apps, it makes sense for you to move a place like San Francisco where there are lots of jobs for people with that skill.

And that brings us to today’s question: what is Randal O’Toole’s answer for a place like Los Angeles, that has grown out to the practical limits of presently available transportation technologies?

First, let me define what I see as the essential points of the Randal O’Toole plan:

  • Public transit can’t compete with the car in modern cities. It’s cheaper to build more roads and use things like congestion pricing. Bus transit is cheaper than rail transit.
  • Centralized land use planning is inherently less efficient than the free market.
  • Things like urban growth boundaries drive up the cost of housing by limiting the amount of developable land and forcing multi-family construction that is more expensive per square foot than single-family residential (SFR).

For the sake of argument, let’s accept these points. In this framework, places like the Bay Area and Portland are unquestionably making bad decisions that will cost a lot of money, hurt their economies, and make the regions less affordable.

And hey, he’s got a point. Throwing open West Marin and all of Clackamas County to master planned suburban development like Clark County would enable you to build a lot of housing relatively close to the centers of San Francisco and Portland. You might not like the idea of the Golden Gate National Recreation Area turning into Daly City, but technically, it would work. In his critique of Plan Bay Area (PBA), O’Toole calculates that currently, 21% of the land area is developed, and by increasing it to 44%, growth could be accommodated by SFR development. Again, that might seem like an unacceptable change to a lot of people in the Bay Area – including, ironically, a lot of the NIMBYs who cited O’Toole’s analysis when fighting PBA – but it would work.

But what about LA?

Other than Ventura County, LA doesn’t have any urban growth boundaries. The developable areas that are protected – the Santa Monica Mountains, the Chino Hills, the San Joaquin Hills – are small in the scheme of the region, and would end up being luxury housing anyway. The boundaries we’re pushing up against, like the San Gabriel Mountains, have topography that is simply too insane for development on a meaningful scale, along with having challenges like insufficient water supply.

Meanwhile, on the fringes of the LA region, the suburban development machine is coming back to life in places like Temecula, Beaumont, and Rialto, and the folks up in the Antelope Valley and the Victor Valley are waiting for their turn. They don’t have any urban growth boundaries, and they’re eager to see your subdivision or industrial park get up off the mat and start growing again. Their problem isn’t controls on land use, it’s slow growth in manufacturing, construction, trade, and logistics.

You know what could help those industries? More construction in the Los Angeles Basin. The parts of the LA economy that are doing well are centered in places like the Westside, and due to agglomeration effects, they want to expand on the Westside, not in Palmdale. But the places where suburban development is happening – Porter Ranch, Santa Clarita – are really far from the Westside. Housing isn’t expensive on the Westside because land use controls are preventing construction of SFRs; the problem is that the undeveloped land where you can build SFRs for under $200k is 90 miles away in Beaumont. What we need is construction of more apartment buildings on the Westside, construction that would almost certainly happen if it wasn’t prohibited by zoning laws and discouraged by onerous permitting requirements.*

To his credit, O’Toole is generally against zoning restrictions as a form of central planning. But his substitute, deed covenants, is even worse. Zoning, at least, can be changed by democratically elected officials, for better or worse. A homeowners association with deed covenants seems to me like a horizontal condo – a neighborhood that has no hope of being redeveloped, no matter how high property values go, because it’s just about impossible to get 100% of that many people to agree on anything. If you believe in letting the market guide development of cities, things like deed covenants are right out. Update: Mr. O’Toole corrects me on the issue of deed covenants. In many areas, deed covenants automatically renew unless 51% of owners vote to get rid of them, which is obviously an easier threshold to reach than 100%. If that’s the case, developers could conceivably buy 51% of the lots and vote to eliminate the restrictions. That still seems like a hard way to do things, and it will prevent the market from responding to demand.

So, what would Randal O’Toole suggest that we do?

*Note that if you follow this logic through, I’m saying that allowing more urban development in LA will encourage more suburban residential, commercial, and industrial development on the edges of the region. I think this is true: construction in the LA Basin will cause growth of construction-related industries, which are the kinds of the uses that need a bunch of cheap land. Contrary to the way many people on both sides of land use debates see it, regional growth is not zero-sum.

Update: here’s his full comment.

You raise a lot of issues. First, LA may not have formal urban-growth boundaries. But LAFCos effectively prevent extension of urban development. Under California law, developers cannot create the special districts needed to support development of unincorporated land without approval from the LAFCos. Under CEQA, such approval would almost certainly require an EIR, whose cost of $15 million or more must be paid by the developer. As a result, development is pretty much restricted to existing incorporated areas. Cities can’t annex without LAFCo approval either. This explains why the L.A. urban area has become the densest urbanized area in the U.S.

Congestion can be fixed through congestion pricing. If the toll revenues generated from congestion pricing are more than needed to operate the roads, then that is a signal that more roads should be built. If not, no need to build more roads.

You misunderstand how covenants work, at least in Texas, Kansas, and many other areas. These covenants typically renew periodically unless 51 percent of lot owners in the neighborhood decide not to renew them. It doesn’t take 100 percent. Developers have been known to persuade homeowners in some Houston neighborhoods to change their covenants to allow different kinds of development.

My thoughts: first, I appreciate the correction on deed covenants.

On the issue of LAFCos (Local Agency Formation Commissions): In California, counties have LAFCos, which can approve or deny applications to incorporate new cities or annex territory to existing cities. For example, not long ago, the LA County LAFCo turned down an application to incorporate East Los Angeles, on the grounds that the city would not be able to raise sufficient revenue to fund its operations. LAFCos also approve or deny applications to add territory to service districts like water and sanitation.

While you theoretically could use a LAFCo to stymie suburban growth by denying all incorporations, annexations, service districts, and so on, that doesn’t seem to happen in practice. LALAFCo recently approved annexations to Santa Clarita and Glendora. Riverside LAFCo has approved four incorporations in the last five years (Wildomar, Eastvale, Menifee, & Jurupa Valley). LAFCOs will naturally reflect the development climate of the county; I’d guess that no one at San Bernardino LAFCo or Riverside LAFCo is that worried about confining development to existing urbanized areas. On top of that, the cities in the Antelope Valley and Victor Valley have already annexed huge swaths of undeveloped desert.

Antelope Valley:


Victor Valley:


Also, let’s not forget that sometimes cities incorporate to prevent more development, like say Malibu or Rolling Hills.

You could write a book about California municipal finances, and I’m no fan of CEQA requiring people to analyze things that can’t be predicted anyway, but that’s a topic for another time!

Bikes and Transit: Frenemies

Programming note: apologies for the sparse posting over the last few weeks. Things should be back to normal soon.

I’m always a little hesitant to wade into debates on cities that I don’t know very well. That’s why this blog borrows heavily from my time in Boston – it’s the only city that I know well enough to analyze without being there. I recently commented on this Systemic Failure piece about Caltrain bumps on Twitter, and unfortunately the conversation ended up basically being “Quiet, you LA blogger, you don’t know anything about Caltrain”.

I won’t lie to you: that’s true. I don’t know anything about Caltrain in particular. I was trying to make two general points, one about bikes and transit in general, and one about the SF Bike Coalition (SFBC) analysis that was linked in the Systemic Failure piece. Neither of these points require knowing anything about Caltrain: (1) paying customers without bikes should not get bumped while bikes are being hauled for free, and (2) the assumption that Caltrain bike boardings should increase at the same rate as bike riding in San Francisco is questionable.

Bikes Fill Seats, Bikes Take Up Space

When ridership is low, providing accommodation for bikes helps fill seats. Outside of San Francisco itself, the patterns of urbanization in California are very different from east coast cities, in a way that often exacerbates the “last mile problem”. By allowing riders to carry bikes, agencies can attract riders who might need to travel a mile or two at each end of their transit trip. This helps fill seats and raises farebox revenue.

However, as ridership rises, you start to get a couple problems with accommodating bikes. One is that dwell times go up due to the time needed to get the bike on or off the transit vehicle. This isn’t a problem if you’re running hour-long headways and you have plenty of recovery time. But it is if you’re running short headways, because dwell time variability will wreck your headway spacing and service reliability. The other is that the vehicle gets too crowded and passengers get left behind, which is especially unfair if paying customers are left behind while bikes are being carried for free.

These problems are more obvious with smaller transit vehicles. A transit bus bike rack can only hold two bikes, and there’s not enough room to bring a bike on the bus because of the seats. Should some seats be eliminated so that an extra couple bikes can go on the bus during off-peak periods? Bikes can also become a space problem on light-rail vehicles; take a ride on the Blue Line during rush hour for example. (When Regional Connector and Gold Line Foothill Phase 2A are complete, the Blue Line from Long Beach to Azusa will be about the same length as Caltrain.)

In the case of Caltrain, the issue was pitched as providing seats vs. providing space for bikes. But really, if trains are getting crowded, you could remove seats to provide standing room for paying customers too. The MBTA did this with several Red Line trains dubbed “Big Red” and still doesn’t allow bikes on the train during peak periods. The question of providing seats vs. standing room is something every agency must address. Longer trips lend themselves towards more seating, but I would note that even in transit systems with high crowding and relatively short trips, agencies still provide some seats.

I’m not singling out bicycles here. For example, transit vehicles can only accommodate so many wheelchairs, and loading/unloading wheelchairs is detrimental to service reliability. Agencies do not eliminate seats in order to accommodate an unlimited number of wheelchairs. If a passenger in a chair is waiting for the bus and the wheelchair spot on the bus is already occupied, that passenger will be bumped (a “pass-up”). The ability to accommodate things like bikes and wheelchairs is partly a value judgment, but it is also partly a matter of geometry. You don’t have to take my word for it – ask Jarrett Walker.

Fundamentally, allowing bikes on transit is easy if your system is running below capacity. The busy transit systems of the east coast do not allow bikes on at all during peak periods and allow a restricted number at other times. Caltrain has already done much more than most agencies; given current ridership patterns, Caltrain’s current bike accommodations seem to make sense. Metrolink would probably do well to modify some of their vehicles to increase bike capacity (there are currently only some trains with bike cars, and the bike cars have a capacity of 18 bikes). Bumps are a major detriment to ridership, and passengers must be able to expect that they will be able to take any train.

For what it’s worth, if you have high transit ridership and high bike usage, you simply need a different solution: bike storage at stations so that riders can have a bike stored at each end. That’s how these guys do it, and I hear they’re pretty good with bikes.

Bike Boarding Potential

My other complaint about the Systemic Failure piece is that the SFBC analysis of bumps assumes that bike boardings could increase at the same rate as biking in SF, bike to work counts in SF, and SFBC  membership, which range from about 55%-85% increases over the analysis period. It’s not clear to me why Caltrain bike boardings should increase at the same rate. Caltrain trips are different kinds of trips – longer, and with less varied purpose. Now, not knowing Caltrain that well, I’m not sure what that means. Maybe it means Caltrain bike boardings should have increased even faster!

There is one reason to think that the potential to increase bike boardings might eventually be lower: system capacity. SF is a pretty committed to bike infrastructure, but the percentage of street capacity that has been dedicated to moving bikes is still very low. There are many streets with no bike lanes, limited bike parking, etc. On the other hand, Caltrain has already committed a significant portion of its capacity to moving bikes:


So Caltrain has dedicated about 10% of its theoretical seating capacity to bikes. There is only so much room on the coaches, and it has to be split among seated passengers, standing passengers, and bikes. Again, maybe Caltrain does not have the right configuration, and maybe they should be providing more space for bikes. But objectively, Caltrain faces greater constraints in accommodating bikes than those faced by city streets.

Brief Notes on Bumps

I would be remiss if I did not point out that as of 2013, Caltrain bike boardings exceed the level predicted by the SFBC analysis, which was completed in 2011. That indicates that there was indeed significant latent demand at the time of the analysis. Contrary to the gist of the Systemic Failure piece, Caltrain has paid attention to bike boardings and increased bike capacity in the interim.

Caltrain’s 2013 counts, conducted over five days in February, found that there were 59 bumps, 25 of which were due to a single incident. Interestingly, all of the bumps occurred during peak periods on reverse direction trains (though directional bias in Caltrain demand is low). Peak loading on those trains ranged from 69%-98% of seated capacity.


The SFBC analysis showed a photo comparison of bike storage areas and seating on southbound trains at 22nd St. Since San Francisco station itself accounts for almost 25% of all bike boardings (and everyone there is obviously going south), 22nd St southbound is by far the most likely place to be bumped. The author was bumped from Train 324, and the photos seem to suggest that there are far too many seats. However, Train 324 has peak load of 91% leaving Hillsdale.

Considering that overall Caltrain ridership is up about 10% per year for the last three years, these points suggest that removing more seats to accommodate additional bikes will result in forcing some passengers to stand. And remember, this is partly a value judgment! Maybe that’s the right thing to do. Or maybe Caltrain should be given the funding needed to run additional trains. I can’t tell Bay Area stakeholders what the right answer is for them. But it seems to me that the discussion needs to start from an understanding that, like a city street, this a case where not everyone is going to get everything they want. There doesn’t seem to be unused capacity available for the taking.