General

What should public transit agencies do when a city is convulsed by massive demonstrations carrying a high risk of violence? Over the weekend, many transit agencies shut down rather than risk possible harm to their staff, passengers, and equipment. This left many good people stranded as they tried to leave the demonstrations.

Christof Spieler, a Board member at Houston Metro, has some ideas in a Twitter thread.  He starts with:

Two lessons for transit agencies in the past several days:
(1) Make a plan to never strand riders
(2) Think about an agency’s relationship with the police

— Christof Spieler (@christofspieler) May 31, 2020

As someone who shares the goals of these protests, let me gently lay out why this is so difficult for transit managers.  This is not to defend or uphold any particular choices any agency has made.  The debate about this is urgent and important.  My only point is that it isn’t easy or obvious what transit agencies should have done.  It was absolutely not OK that people were stranded, and there needed to be a solution for that, but the actual solution isn’t so obvious.

I’m in the Canadian Urban Transit Association newsletter this week, with a piece on goals.  Key quotation:

The crisis has revealed a strong new argument for transit, one we should deploy at anyone who wants to judge us on ridership.  Transit has kept running through the crisis as an essential service, supporting people who work in hospitals, grocery stores, utilities. These mostly low-income people would typically have been called “transit dependent.”  But it is they who are holding civilization together right now, so we are all “transit dependent” in this sense.

What’s more, this has always been true.  Transit riders have always been part of the basic functioning of our cities.  Measuring that role solely with ridership would be like measuring the success of the police by how many arrests they make.  The purpose of the police is to provide a base level of security that people can count on.  The purpose of transit is to do the same for urban mobility.  Transit means that people can go places, and thus do things, in a way that isn’t as harmful or expensive as driving.

Read it here.

Covid-19 has understandably caused steep declines in public transit demand, but the pattern of that fall is important.  Peak (rush hour) demand has fallen much more than all-day demand, mirroring a change in travel demand overall.

This chart is from the daily updates we’ve been getting at the Transit app website.  This chart covers all Transit clients, who are heavily US and Canadian agencies but include a few in other countries.

[Caution:  These are not ridership statistics.  Instead, Transit counted queries of its app, which provides realtime information about when the next bus is coming and, in some cities, can be used to pay your fare.  That means this metric probably undercounts relatively unwired people, including low income people without smartphones and those less comfortable with apps.  Still, it’s the only data that can be collated and updated so rapidly.]

Of course people are working at home during the emergency, but some leading companies are planning to continue the practice, and nobody really knows how much office work will return.  You can see conflicting reports on whether working at home is wonderful or terrible, so we can expect continued experimentation as people and companies figure out what they like.  Still, with the virus lingering, it will be a long time before everybody is back in the office, and there’s room to wonder if they ever will be.

Why is the fall of the peak important?  Running service only at rush hour is expensive, for three reasons.

• A vehicle must be owned and stored that isn’t used very much.
• A driver must report to work for just 2-4 hours, which is less efficient, hard on the driver and will cost the agency more per hour of service.
• Most peak demand is massively one-way in the morning and the other way in the evening.  Drivers’ shifts must end where they began, so every bus or train that runs full in one direction has to return empty in the other, often over long distances.

So the fall of the peak, if it were sustained into the future, could be great news.  While the peak is an easy place to rack up lots of ridership, its high costs mean it’s not always the best place to seek productivity (ridership divided by operating cost).  Ultimately that means that there could be all-day markets that would be more productive once the high cost of peaking is taken into account.

There is also the large social justice dimension to the peak.  Peak commuters are far more affluent on average than all-day travelers, because higher wage jobs are more likely to be “nine to five” while lower wage workers, predominantly in retail and services, are more likely to be needed around the clock.  So a decline in peaking could help sustain services that support lower wage people — and remember, these are people whose work everyone depends on.

Of course peak service is justified by the need to mitigate traffic congestion that occurs at that time, but it remains to be seen what levels of congestion will return.  It may go up if we try to run a full economy with social distancing, but it could also go back down after a vaccine.  We’ll have to see.

Peaking has such a huge effect on the life of a city, and the costs and efficiency of transit, that it’s worth taking a quick tour of how different it is in different cities, and how that reflects choices the city made — consciously or unconsciously.  Here’s my home city, Portland.

This is typical of a lot of US cities where the social distancing expectations (both legal and cultural) have been firm over the last two months. The peak is mostly gone. In percentage terms, the difference isn’t huge. Ridership is down about 80% at both midday and PM peak times, but it’s down about 87% on the AM peak, where few people travel other than for peak work shifts and schools. Still, the absolute numbers matter too, because they measure the degree to which an agency will be forced to run expensive peak-only services rather than an all-day pattern.

Note that this typical North American peak is about two hours long in the morning and three hours long in the afternoon.  That’s the result of peak spreading, the widespread tendency for start and end times to vary slightly by employer (and by school).  Compare Sydney, Australia [really this is all of the state of New South Wales, but Sydney is the overwhelmingly dominant market there]

Australian peaks are much sharper than North American peaks — less than two hours long — and as you can see from yesterday (May 25) they are coming back as sharp as ever.  Australian public transit also does most of the work of school transportation, which explains why the afternoon school peak at 3-4 PM is bigger than the evening commute peak at 5-6 PM.  Note that it is the school peak that is returning now, though, while the work commute peak — marked by activity in the 5-6 PM hour, is not so prominent.

When I worked as a transit planning consultant in Australia I was always struck by the huge cost of serving a rush hour that’s so brief.  Yet Australia is far behind North America is adjusting work and school scheduling to spread the peaks out, so that demand (both transit and road) can be served more efficiently.  (Spreading the peak can also create huge savings on infrastructure projects, since if you scale infrastructure to serve the peak one hour, you’ll need a lot more infrastructure than if the same demand is spread over two hours.)

Now look at Paris

The abundance of transit in Paris isn’t just a matter of spending lots of money on it, but on focusing on an all-day demand rather than peak demand. Paris has plenty of peak commuters, but so many people rely on transit for all kinds of purposes that those peaks don’t stand out the way they do in less transit-oriented cities.

Of course, there are some car-oriented cities without much peaking either, generally those built around entertainment and tourism. Las Vegas, say:

This is what happens when you have relatively few rush-hour commuters to office jobs, but massive employment in tourism: shifts starting and ending all the time. But it’s not all tourism. All those jobs that keep society running — working at WalMart or a hospital or a (take-out) restaurant — are visible here because the commute peak doesn’t swamp them in the chart. Note how much smaller the drop is from pre-pandemic days: only about 50% compared to 70-90% in most US cities, because there were fewer peak commuters before. Even with the massive loss of casino and hotel jobs, plenty of low-income people need to get to work.

Finally, a small US state capital with a huge university. Universities tend to be huge consumers of transit, and you can see the effect of both university and state offices shutdowns in Madison, Wisconsin.

The drop was over 90%, and only the PM peak shows the earliest signs of coming back, now to 80%.  [Again, these are Transit app queries, not ridership numbers that can only be provided by the agency.]

All these graphs are scaled to the pre-pandemic demand, so only in Las Vegas do you see the importance the low-wage worker in essential services including essential retail.  But those people are in all of these numbers and are why there was any ridership at all.

Keep an eye on these charts, and on peaking patterns as they emerge in ridership data from the agencies.

This letter went out to our colleagues today:

Friends and colleagues.

First of all, we are deeply grateful to everyone working in public transit agencies right now.  Frontline staff are taking personal risks, while everyone is having to think fast and abandon usual procedures.  If you are in the midst of this, thank you for everything you’re doing.

The goals that justify our work have turned upside down.  I never thought I’d hear transit managers tell people not to use their service, or that the mandate of social distance would have us running big buses and long trains so that they would be as empty as possible.  All our old instincts for how to run good transit systems suddenly feel wrong.

Like many of you, my team and I have been thinking about how Covid-19 changes what we do, how we do it, what stories we tell, and what questions we ask the public.  I wanted to take a moment to share our thoughts as they stand right now.

The most important fact about Covid-19 is that our recovery will be gradual, and we will emerge from it into a different world.  Experts tell us to expect a tentative and slow recovery, in which people may fear the virus for several years.  Here are some of the changes we think are coming.

Most of the goals that we expect public transit to pursue have just turned upside down.  Ridership was an important goal two months ago, but now public transit agencies are telling customers to stay away if they can, restricting service to “essential trips.”  These trips are critical, though, to the functioning of society.  You could even argue, as I did in Citylab recently, that transit is helping prevent the collapse of civilization.

But now transit authorities are having to cut service.  This post examines why, and how they should do it.  It’s long, so there’s a table of contents:

Contents

1. Why Cut Service?
2. Why Not Shut Down?
3. What Service Should be Cut?
4. What Agencies are Doing, and May Do
5. Three Stages of Adaptation and Recovery

1.  Why Cut Service?

Why cut service at all in this situation?  There are two compelling reasons:

• Staff availability.  Some critical staff are sick, and it’s likely that every transit employee is at least worried about being sick.  So right now, there just aren’t enough people showing up for work.
• Saving money.  In the US, the Federal CARES Act will keep transit agencies whole for a couple of months but requires them to continue paying all their staff.  But the financial prospects are dire in the longer term, so some agencies are trying to save money so that they can keep going a little longer if the federal funding dries up.  In countries where no emergency funding is being provided, many agencies will run out of money very soon if they don’t cut their service dramatically.  (Vancouver’s TransLink, for example, has been told not to expect support from the Canadian federal government, and will have to “deconstruct” much of the network as a result.)

2.  Why Not Shut Down?

This is a brutal time to be a bus driver.  Already some services have been shut down by union action and some labor leaders are demanding more shutdowns.  Some transit systems — such as the one in Windsor, Ontario — have shut down completely with no particular attempt to provide alternate solutions for the “essential services” workers.

Apart from the workforce safety issue, would it be cheaper to shut down transit and just subsidize Uber-Lyft rides for all essential services workers? That math only works if you truly lay off the workforce.  This is not what the unions have in mind, and in the US it’s expressly prohibited for agencies taking the Federal CARES act rescue funds.  It also means that when economic activity resumes it will take more time to get the agency restarted, potentially creating a long period when lots of people need transit and it’s just not there.  Meanwhile, while keeping people 6 feet (2m) apart is certainly hard on public transit, it’s impossible on the Uber or Lyft.  You’d need a van to carry just one passenger in addition to the driver.

3.  What Service Should Be Cut?

Easy Options

If service has to be cut, how should it be cut, and with what priorities?  Two cuts are really obvious and most agencies have done them:

• Tourism and recreation services.  These markets have vanished.  San Francisco, for example, has replaced its cable cars with buses.  Many downtown circulators and trolleys, especially those justified by tourism, have shut down.
• Service usually added for rush hour, unless it’s the only service to an area.   The “essential service” trips that are still on transit happen all over the clock with very little peaking.  The huge nine-to-five peak created by white collar workers is nearly gone, and of course the school peak — a big one in many agencies — has disappeared entirely.  Peak-only service is expensive, so it’s a logical first thing to go as long as people along those routes have a local service option to complete essential trips.

Difficult Options

• Cut frequencies.  Most agencies have done it. but it’s dangerous.  A common misconception is that if you cut frequency, people still have “access.”  But lower frequency means longer waits for connections, so this can cause massive increases in travel times.  A trip that takes longer than people have in their day might as well not be served at all.  A good guideline, which several agencies are following, is to cut very frequent services to run only every 15 minutes, the worst frequency at which connections are still viable, but resist cutting further.
• Abandon service to some areas entirely, while focusing on “essential” destinations.  Many agencies have plotted the locations of essential services, especially medical centers, and retained only routes serving those.  The problem here is that the workers who need to get to those places live everywhere.  To make all essential trips possible, you need a degree of citywide access, at least to all areas where incomes match the typical pay of essential workers.
• Cut duration of service.  This totally abandons some essential shift workers, especially in the 24/7 medical world.  Most transit agencies were already offering little or new overnight service, and closure of restaurants and entertainment has eliminated most of the ridership of evening service.  But further reductions of span can still be a problem for essential services workers, who may have to travel all over the clock.
• Increase walking distances consistently.  In a grid network where there are routes about every half-mile or 1 km, just turn off every other route, so that walking distance to transit increases but everyone still can get to some service.  More on this interesting option below.

4.  What Agencies are Doing, and May Do

The most common quick response to the crisis has been to go to Saturday or Sunday schedules.  This effectively does a bit of everything on the “difficult options” list.  It’s totally reasonable as a quick response.  Saturday and Sunday schedules already exist, have already been approved by both union and management, and are relatively easy to turn on.  The normal process of writing, approving, and implementing a schedule takes months, with procedures that tend to be strictly defined in labor contracts.  Many agencies have found their unions to be motivated to help make things work in the crisis, but that approval process still takes time.  So it’s understandable that using weekend schedules was the thing that could be done fast.

Still, several agencies have already managed to implement more carefully designed service cuts.  Cleveland’s RTA, for example, has completed a service cut with revised schedules reflecting detailed thought about every route and how to make a minimally functional and connective system.  In their case, the suburbanization of low income means they can’t cut any coverage, but they’ve turned off their downtown shuttles and carefully chosen frequencies for each route to optimize travel times.

It’s critical that transit agencies do these designed reduced networks they have time to think about it.  Many transit agency Sunday schedules are just the daytime network run less frequently, and often don’t work very well because it can take an hour to change buses.  But transit agencies will be running on reduced demand and resources for some time, so more careful design is needed.

San Francisco and Philadelphia (within the city) are among the cities where the cuts have tended to protect frequency bhy increasing walking distance — though of course this works best in old cities that are relatively walkable.  San Francisco has turned off many routes and thus increased walking distances to about a mile but in return its frequencies are 20 minutes or better everywhere, so once you’ve walked to a functioning route you can be on your way.  Their new “Core” network doesn’t immediately show a consistent route spacing, because their initial design was limited to turning existing routes off or on, but if this goes on I wouldn’t be surprised if they start doing modest restructuring to improve the amount of useful service they can provide with such reduced resources.

5.  Three Stages of Adaptation and Recovery

From what agencies are doing, and the facts behind them, I think we can identify three stages that agencies are going through, or likely to go through, as they deal with the combination of (a) suddenly needing to cut service and (b) retaining the ability to slowly grow service back as demand returns, knowing that the pattern of new demand may be different from the old.

Stage 1.  Emergency cuts.  These cuts, mostly done by now, are necessarily rushed and crude, such as going to weekend schedules.

Stage 2.  Designed cuts.  When time permits, refine the schedules to better reflect actual patterns of “essential services” needs during the emergency.

Stage 3.  Strategic design.  While Stage 2 generally works with the existing network structure, this stage may revise routes or even redesign some services.  You would do this, though, only if you think that redesign will probably continue making sense as demand returns.

Everyone knows that when this crisis passes, they will emerge into a different world with different patterns of demand.  The usual controversy around network changes, where we disrupt patterns that people are used to, will no longer apply, because nobody’s used to anything at the moment, and we all know the new world will be different from the old.  We can anticipate some reasons that the return of service will need to follow new patterns:  People who can drive will be slow to return to transit, but people without a good option of driving will show up in much greater numbers.  So service priorities optimized for the latter may be the best way to bring service back, and that’s not always what existing route structures are designed to do.

In short, like all crises, this is an opportunity.  And because this is a big and enduring crisis, the opportunity is big and enduring too.

When people want to imply that public transit is irrelevant or failing or wasteful, they always point to relatively empty buses or trains.  To the layman it certainly seems as though buses with few passengers aren’t achieving much.

This common line is based on two false assumptions.  It assumes that empty seats indicates waste, which they usually don’t, and it falsely assumes that ridership is transit’s only measure of success, which it isn’t.

Ridership Is Not Transit’s Only Measure of Success

If public transit were the only goal of transit, it would look very different.  For example, it wouldn’t serve many parts of the city, because high ridership is just geometrically impossible in those places, for reasons we explain here.  For example, it would delete a service that almost nobody is using, and it wouldn’t waste five seconds on people’s complaints that they really need or deserve that service.

The boards of directors who govern transit agencies — who are either elected officials or their appointees — never tell their staffs to run the agency this way.  They want ridership, but they also want a lot of other things that often justify running low-ridership service.

The 2020 Covid-19 epidemic proved this point.  Instead of shutting down and laying off most staff, as a profit-making business would logically do, most big city agencies did their best to run service, not just to keep their staff employed but also to enable those “essential service” workers to get around, so that the city could keep functioning.  To do this, and to encourage social distance, they ran intentionally “empty” buses and discouraged people from riding them.  In this, transit revealed itself to be the opposite of a business.  It’s a utility, an essential element of the functioning of a city.

The Math of Efficiency: Run the Biggest Bus You Will Ever Need

In fact, when a transit agency runs a relatively empty bus, they’re doing what they’re told to do, and they’re doing it efficiently.

Transit operating cost is mostly labor.  The cost of a bus is in the driver, not the size of the vehicle.  So it doesn’t cost much for the vehicle to be bigger than needed.

But it costs a lot for the bus to be too small.  Then you have crowding, or you leave passengers behind.

Demand for transit goes up and down during the day, and on different parts of the route.  It also goes up and down for unpredictable reasons.  A school decides to have a field trip, and a normally empty bus is suddenly packed.  Nice or bad weather can change ridership patterns suddenly.

Buses and trains cannot dynamically shrink or expand to match these changing demands, and since the operating cost lies mostly in the driver, there wouldn’t be much to gain by doing that.  For example, you might propose that after the morning rush hour the big buses should be replaced by smaller ones.  But the cost of paying a driver to take a bus back to the base and take out a different one far exceeds any cost in running a larger bus than you need for a few hours.

So transit agencies are smart to run the biggest bus they will ever need, even though that means the bus will be empty at some hours, or on some parts of the route, or even on some days when demand is lower.  They may even run a bigger bus than a route ever needs, because there are also massive inefficiencies in having too many different kinds of buses.  Smart agencies have thought this through and what they are doing, in my experience, mostly makes sense.

The sudden decline in travel due to fear of Covid-19 is obviously affecting public transit.  Preliminary and unpublished numbers shared with me by two US West Coast agencies showed ridership losses of 30-50% from pre-crisis levels.  We’ll see plenty of published numbers soon. The sudden fall in gas prices at the same time could make the ridership impacts even worse.

This is a good time to remind ourselves, and our favorite journalists, that ridership is always volatile and heavily driven by factors outside an agency’s control.  There are many things transit agencies can work on to improve ridership, but (a) those things together amount to a minority of the total forces governing ridership and (b) ridership isn’t the sole metric of success for transit agencies, and sometimes not even a predominant one.

However, transit agencies can do things that will cause ridership to fall further and stay down longer:  They can cut service, as they will be tempted to do now.

It’s not hard to see the dangers:

• Declining fare revenue, due to lower ridership.
• Declining tax revenue, due to economic slowdown.[1]

So what should transit agencies do if they start to run out of money?  Cut service?  If so, how?

First, let’s distinguish between service and capacity.  If revenue falls, many urban transit agencies can trim rush hour capacity without affecting customer mobility very much.  If you’re running a commuter bus every 7 minutes at rush hour, cut that to 10 or 12 or whatever the loads support.  Because peak commuters mostly plan around the schedule, the impact on travel time is trivial, but you’ll save something.  Peak-only service is very expensive, so you can save a lot by trimming that.  What’s more, preliminary numbers I’ve seen show commute ridership falling much more steeply than all-day local ridership, which suggests that the peak should bear the brunt of any temporary service cuts.

By contrast, when you start cutting all-day and all-week service, by reducing frequencies, you start to dramatically reduce the usefulness of network, and this is the most efficient way to drive riders away.  You also trigger social justice impacts, because lower income riders tend to be all-day, evening, and weekend riders, not just peak riders.

Remember, the riders you drive away due to service cuts will stay gone until the service improves again, while those who are just working from home will come back post-crisis if the service is still there.  

Second, if money runs low, you have to question the schedules of your infrastructure projects.

Infrastructure projects are so complicated and involve so many possible points of failure that they require a concerted effort to maintain momentum. All the messaging around how inevitable the project is, and how certain the opening day is, can make it sound like nothing can be slowed down in response to a crisis.  Of course, some contracts do impose costs for slowing and stopping, and those have to be considered.

But if infrastructure work continues while service is being cut, you’re driving away current riders for the sake of future riders, and if the goal is ridership, that makes no sense.  It can also be a social justice problem.

Why do I have to point these things out?  Influential people often tend to be peak commuters, so they defend peak service but care less about all-day service.  They also tend to be invested in infrastructure, and don’t want to hear that those projects might have to slow down.  To many of them, all-day service can seems easy to cut by comparison, since that affects people who don’t comment as much.  But apart from trimming peak capacity, that’s the sure way to turn a temporary crisis into a more lasting one.

[1] These impacts will be cushioned if agencies have (a) a lot of pre-sold fares, such as employer programs and (b) tax sources that rely on less-volatile sources like property or income tax, as opposed to payroll or sales taxes.