Friday, August 20, 2010

Induced demand and transit

Jarrett, as he usually does, has another interesting post up on Human Transit on the subject of whether or not rail attracts more riders than bus.  He points to some research done by a Munich transit planner (who subsequently joins the conversation in the comments) suggesting that this is the case, comparing various mid-sized (under 500,000) European cities with trams and without, noting that the cities considered in the study who have trams generally have higher ridership than those without.

Jarrett asked the fundamental chicken-and-egg question:  Was the construction of streetcars (in most cases, lines running in an exclusive right-of-way, functionally more similar to the MAX Yellow Line than Portland Streetcar) a driver of demand, or a response to it?  In other words, "did the trams cause the ridership, or did the ridership cause the tram?"

The chicken...

This is a good question.  A big technical advantage of rail over bus is vastly greater passenger capacity--and many successful rail lines have been built in bus corridors that got too crowded.  A common argument against rail projects (or against capital-intensive busways) in corridors that aren't already crushloaded to the gills at three minute headways, is "the existing demand doesn't justify the investment".   One commentor, "Danny", asserted that:

All successful high speed rail systems were capacity expansions for crowded low speed rail systems. All failing systems were constructed under the impression that speed causes ridership. 

While I respectfully disagree with his use of the universal quantifier "all"; and I  disagree with the first half of his statement completely (I can think of many counter-examples), the second half of the sentence has a lot of truth:  Many "failing" systems (by which I assume he means systems with usage levels far less than predicted) were constructed based on ridership increases which didn't come too pass.  If you build a rapid transit line in place of a local bus line and cancel (or reduce in frequency) the latter, you already have established a reasonable floor for the utilization of the rapid transit line--the bulk of the bus passengers will switch.  OTOH, if you are anticipating additional ridership that isn't there... there's a far greater chance you will be wrong.

...or the egg?

But, there's also a good chance you'll be right.  The Munich transit planner who produced the above-referenced paper surfaced in the comments, and had this to say. 

When you have cities of a similar size and density usually you should expect a similar ridership of public transit. But obviously, in average, it is not like that. So I sent Jarrett the data as one indicator for this "tram bonus", as we call it, that is able to get people out of their cars, like a bus system would never be able to do it.
Even the successful BRT systems in different cities do usually just channel a demand that already exist, a high demand of captive riders. Who ever has a car in those cities still uses it. 
We believe in the ability of a tram system to attract more people, people who would never set a foot into a bus, and we invest a lot of money in that. And: no matter if we are right or wrong with this "belief", it does work indeed. We see the numbers, we see the ticket sales and it makes sense, even from the economic perspective. It does not make sense for any bus line, but you can expect to increase ridership in a dimension of 50% to 100% on a line, compared to the bus service. If that makes sense for you, then do it.
Doug Allen, a local transit advocate (who was involved in the planning for the initial MAX line, and still works for TriMet--his words, obviously, aren't necessarily the opinions of his employer) added his two cents:
I would agree with "TransitPlannerMunich" that experience with the conversion of a bus route to tram, in a city with a mature system of bus and rail, can provide insight into whether passengers prefer rail over bus.

The implementation of the "Yellow" light rail line in Portland Oregon also showed that a rail line that is shorter and less frequent than the bus service that it replaced can attract higher ridership, no matter how irrational this may seem.
Doug refers to the now-gone 5/Interstate, which ran from Hayden Island along Interstate Avenue to downtown Portland.  The Yellow Line doesn't presently serve the island, and runs at 15 minute headways, so in at least two respects, it's inferior to the bus service it replaced.

Neither Doug or "TransitPlannerMunich" offered any explanation why they observe a presence for rail; formal research on the "tram bonus" is inconclusive.  And other commenters wondered whether or not reconfigurations of bus service (converting parallel downtown routes into feeder routes) might boost rail ridership without boosting overall commute share.  Many properties which riders attribute to bus and rail, are often properties of the local implementation thereof--if, for example a transit agency operates 40' diesel busses through rough neighborhoods with no capital improvements on the bus routes, alongside modern light rail that sticks to the nicer parts of town--don't be surprised if local residents characterize the bus as dirty, noisy, slow, unreliable, and full of unsavory characters.  Bus can be clean and fast, and rail can be slow and uncomfortable; there is much overlap in the service parameters for both.   On the other hand, if there is a demonstrated community preference for one mode over the other--even if entirely irrational--that's something that ought to be factored into planning.

Induced demand

Which leads us to the subject of induced demand-- the "Field of Dreams" problem. ("If you build it, they will come").  If the supply of a given resource is increased, then more of it is often consumed.  This is especially true when the existing supply is insufficient (or barely sufficient) to meet the existing demand; there's often lots of pent-up demand that is more than happy to consume the new supply.  A related phenomenon is value-induced demand; if you increase the value (quality) of a resource, without increasing the cost, demand will also rise.  Both phenomena are in the realm of Economics 101.

Induced demand is frequently invoked in arguments concerning construction or expansion of freeways, where congestion relief is cited as a justification for the bulldozer and the mixer.  Highway opponents frequently point out that when freeway capacity is added, it frequently fills up with additional traffic, causing a failure to deliver on the expected congestion relief.  (Freeways suffer from a fundamental scalibility problem as well--additional traffic, even if below capacity, increases the chance of a wreck, stall, or other incident which severely impacts service.) 

But induced demand works for transit as well.  Transit opponents frequently point out the same thing about transit projects sold on the basis of congestion relief--even if the transit service attracts motorists out of their cars, other motorists often take their place.  But the principle applies:  When transit service improves, either in capacity or in quality, more people will elect to use it.  The improvement can be manifested in values which are easily measured--coverage, capacity, frequency, speed, reliability--or in those values which are less tangible and may reflect social facts or personal biases (comfort, prestige, sex appeal, environmental benefits, etc).  Either way, if you make a service more attractive to the population, more people will use it.  And the advantage that transit has over the automobile (considering only the geometric aspects of it, and ignoring the numerous negative externalities of cars which provide ample additional reasons for transit construction) is one of scalability at high density.
The reverse phenomenon is also readily observed:  when you cut service, ridership levels go down--decreases in line frequency or other service parameters will often drive riders to other modes, or to not make trips at all.

How badly do you want it?

These phenomena can be explained by the property of elasticity--a concept which is the way economists answer the question, "how badly do you want it?".  (Actually, it measures the inverse of that question--a lower elasticity means that the product or service is "wanted more badly").  Many transit advocates who know little about economics understand the fundamental concept of elasticity--the concepts of "choice riders" and "transit-dependent" refer to two different populations who express different elasticities with regard to their transit-using habits.  Choice riders express a high elasticity--meaning that their consumption (use) of transit is likely to change as the value proposition changes; if the fares go up or the frequency goes down, they stop riding, and if service improves, they may switch back to transit.  Transit-dependent riders express a low elasticity--they'll put up with a lot because they have no other choice, and conversely, improving service isn't likely to attract more transit-dependent riders, as those who need the service are already using it.  A third community to consider is the transit-averse:  those who won't use transit under any circumstances.  They also express a low elasticity, as improvements to the system are unlikely to entice them to increase their consumption (usage), which will remain at zero.

One interesting phenomenon concerning elasticity is that it goes up in a recession.  Even if the capacity remains adequate and the coverage remains the same, when a recession occurs, cuts to service are more likely to negatively affect ridership during hard times.  When a recession occurs and people lose their jobs, two things happen:  unemployed riders for whom transit demand was previously inelastic suddenly find it elastic, as they no longer need to be at work by a certain time.  And congestion on roadways decreases, increasing the relative value of driving for those who can drive (which is the majority of the population in most parts of the US)  .   This is why "death spirals" are a pressing problem for transit agencies during a recession--loss of revenue leads to service cuts which leads to loss of riders which leads to loss of revenue which leads to... you get the picture.  (When a transit agency finds itself in this situation, as TriMet appears to be, you had better hope that the pattern of panic/cut/cringe converges at a level well about zero).

Its all about the values

When publicly-operated rapid transit projects are proposed, there are often several reasons offered to justify the expenditure of public funds:
  • Increasing coverage
  • Increasing capacity
  • Increasing service quality
  • Decreasing operating costs
  • Environmental outcomes
  • Land-use outcomes
  • Economic development
The first three are important to transit users (and potential users).  The others are important to other constituencies (the transit agency, the public at large, etc) and won't be considered further.  Whether any of these are important to a given rider or group depends on what the rider or group's goals are.  And if a person takes a more minimalist view of transit--s/he may not be interested in attracting riders to the system--and consider expenditures for that purpose a waste of money. While I can't point to research to back it up, it seems apparent that many of those who adapt the minimalist point of view are those for whom transit demand is most inelastic.  the transit dependent, who have every reason to be skeptical of proposed new projects--many of which will benefit some other community and adversely affect the particular service they depend on; and the transit-averse, who won't ride the system at all, and may view the entire enterprise as a big waste of money.

This dichotomy may affect one's answer to the question posed by Jarrett at the top of this article.   It's a common psychological phenomenon to project ones thoughts and beliefs onto others.  Thus, if an individual's personal demand for transit is inelastic (or one's own personal value system WRT transit focuses on tangible service parameters such as coverage, capacity, and performance, and disregards things like amenities or social acceptance), s/he may assume that this is true for others--and question the claim that improving transit quality may increase ridership.  Likewise, if a person's personal transit demand is highly elastic (or is motivated by factors specific to a particular mode choice), s/he may assume that increasing service levels (including by conversion to rail) will axiomatically drive up demand--and further assume that what is important to him/her, is important to everybody, thereby causing an overestimation of the induced demand.

The truth, of course, lies somewhere in between, and ought to be determined empirically rather than by ideological catfights between different communities and constituencies.


  1. so would it be better to add transit service during recessions than to cut it? might you actually be cutting costs by adding more buses and trains? its certainly seems counter-intuitive, has there been any study of this?

  2. You wouldn't cut costs; which only go up when service is added. It's possible you might be able to cut losses, but that's a risky thing to plan for--even in good times, most transit systems (including TriMet) have a farebox recovery ratio less than one.

    But recessions are damaging to enterprises of all types--transit is no exception.

  3. A further comment from "TransportPlannerMunich": He points to a German Wikipedia article on the so-called "Scheinenbonus" (track bonus)--a phenomenon which appears to be more greatly documented in the German transport literature. Google Translate does a fair job on the article (its conversion to English is readable), and there are numerous references. Unfortunately, much of the discussion focuses on vehicle noise, rather than ridership. Still interesting though.

  4. i wasnt trying to sound uninformed about transit costs but was sort of wondering if when you've dug yourself into a hole, if perhaps the solution may be to actually keep digging. the worst is that death spiral that you talk about, then the whole system unravels. i mean look where the portland transit system had fallen to when trimet began in 1969, rose city transit was carrying 60,000 riders a day (the blue lines couldnt have been carrying that many additional, say at most a combined 80,000), now its 325,000/day (not including c-tran).

    i do notice that the 72 is not seeing any cutbacks, i take it due to its profitability. i wonder if there is anyway to set up other bus lines to function like it, whatever it is doing, it seems to be working.

    what is the farebox recovery for the hillsboro-gresham line? that line must carry at least 80,000 riders/day when you include the other lines running duplicate service on the line (ie red line gateway-beaverton, green line gateway-rose quarter, and of course blue line hillsboro-gresham). even if the numbers are just for the blue line, i have to believe it is very high.

    i love the day passes, but they seem awfully underpriced at $4.75. they are only 15 cents more than a roundtrip all-zone (2 x $2.30 = $4.60). as much as i dislike fare increases, this would have been the place to make a rather painless raise. i'd rather reasonable fare increases to cutbacks. and a regular fare of $2.05 is a just hassle, at least make it something logical like $2.25 (and then stash the extra away for a rainy day or make less severe cuts).

  5. Really good post, one thing though:
    "The first three are important to transit users (and potential users). The others are important to other constituencies (the transit agency, the public at large, etc) and won't be considered further. Whether any of these are important to a given rider or group depends on what the rider or group's goals are."

    Having more passengers should be in the interest of any other passengers because more passengers means additional investments to the service can be made, headways reduced, etc. With highways it is the opposite, more cars means more congestion.

    This is the killer feature of rail economics, it gets more economical the more people use it.

    By the way, I am not quite sure this is understood that side of the pond, and propably explains a lot differences of opinion regarding trams: German style tramways cost about 5 to 10 million euros per kilometer to build, about the same as a dedicated busway. This is achieved by relying on signal priorities instead of grade separation and in general limiting any kind of underground construction to where it is absolutely necessary, new Scandinavian systems for example follow this model and have similar costs. French style trams cost something like 20 million per kilometer, but that style involves rebuilding entire streets, squares, etc. and really ought not to be considered purely a transit investment. Most European specialists I know of simply can not understand why US light rail projects have so high costs.

  6. Re: the day passes (poncho's comment), I was kinda wondering about that, too.

    A slight increase to an even $5 would put it on par with a day pass in San Diego. A day pass on SEPTA (Philly) is $11, MTA (NYC) is $8.25, and Chicago is $5.75.

    How many daily passes does TriMet sell, and who typically buys them? Any numbers on that somewhere?

  7. TriMet sold 76,594 daily passes in June, according to its monthly report. This brought in $363,821.50, or a little less than 4% of TriMet's total passenger revenue of $10.08 million.

    TriMet doesn't supply data on purchasers in that report.

  8. Thanks, R A Fontes. So a 25 cent increase on daily passes would bring in roughly another $19,000 a month, assuming purchases remains constant. Or let's round down a bit to $17,000 to be safe, since daily passes probably sell at their best in the summer season.

    Would be interesting to see who buys these. My hunch is that it's largely tourists and others who are least likely to mind, or even notice, a slight increase. Forgive me if I'm stating the obvious here, I really don't know either way.

    Was an increase on daily passes ever considered?

  9. The big question is: Who buys daily passes? Tourists? Occasional local transit users who want to make a round trip and would rather not have to purchase a second ticket for their return journey, so pay the extra $.15 as a convenience? People who need to make >2 trips per day, separated by sufficient time that three+ tickets would be otherwise needed?

    Were I to advance an uninformed proposal, I'd have the following:

    * 2-hour ticket, like today, at the current all-zone fare ($2.30)
    * 4-hour ticket, costing, say, $4.
    * 6-hour ticket, costing the same as 2 2-hour tickets. (Or perhaps slightly less--say, $4.50). The six-hour window is designed to permit many round-trips to be made, leaving sufficient time for activities in between, but exclude commuters.
    * All-day ticket, costing more than the current all-day ticket ($5-$6 sounds like a good adult price; the $11 charged by SEPTA is ridiculous).

    I'm also tempted to get rid of the zone system altogether, and instead a propose 30-minute ticket costing, say, $1.25. Unlike the current 1-2 zone ticket, which is good for 2-hours, and thus can be used for both legs of many short round trips, these tickets would only be useful for one leg (and explicitly prohibit use for a return trip) --and the 30 minute limit is intended to exclude crosstown trips. Encouraging short trips is a good thing (many of these are for non-commute trips, and would be made during off hours); the current fare structure penalizes them too much.

    One issue concerning time-limited tickets, is what to do when someone's ticket expires en route (which can happen to anybody should the system be delayed). How do TriMet fare inspectors deal with someone whose 2-hour ticket has expired? I would assume that obvious violators (someone travelling three hours beyond the time limit) would be booked for evasion, but someone who can plausibly claim to have boarded during the window and is tem minutes over woudln't be fined or kicked off...

  10. i'm a regular local rider and i'm starting to use them a lot more. being that they are unlimited use passes, not surprisingly you use more service than you otherwise would if you paid per ride... you might just hop on a bus headed in your direction and ride 5 blocks or so. i got really turned to day passes using the NYC transit system as a visitor.

    a big thing about them is no need to fumble for change, rush to buy a ticket from a TVM as the train pulls into the station, or save a couple bills in your wallet for a return trip. if you see a bus or train coming all you need to do is hop on.

  11. @poncho:

    I'm assuming that you're not a sufficiently regular rider for a monthly/yearly pass? How often do you find yourself purchasing day passes? It's interesting that TriMet gives a substantial price break on a 7-day pass (which costs the price of about 4 1/2 day passes); a very miniscule break on top of that for week, 2-week, and monthly passes, and a medium break on a year pass (12 months for the price of 11).

  12. re: what to do with time-limited tickets

    TriMet's current rule is that the time stamped on a validated ticket or at the cut line of a transfer is the latest time one can board a vehicle. So a rider could board the blue line in Hillsboro with a transfer from a local bus with just one minute remaining and ride legally all the to Gresham.

  13. R A - I think Scotty was asking how a fare inspector, during an onboard fare check at, let's say, 8:50 at Orenco / NW 231st, would know whether the person whose ticket that expired at, say, 8 PM, had boarded at Rockwood (legally) or at Beaverton TC (probably illegally).

    I kinda wonder about that, too. Is it an honor system kind of thing, and how do they judge one's honesty? Do they even ask or just grant everybody a 60 or 90 minute window? I only rarely ride MAX, although I ride buses multiple times daily. And I always have a monthly pass, so on the (extremely rare, like once a year if that) occasions I run into a fare inspector (usually descending the stairs onto the platform at Hollywood TC), I just flash my pass and that's that.

  14. That policy (as long as you board before the ticket expires, you can ride on the same vehicle) would probably prevent short-duration tickets from replacing the zone system.

    Of course, I've seen more inspectors in my occasional travels to Hong Kong, then I have in Portland. :)

  15. true, i'm not frequent enough to get a monthly pass. i use it mostly for errands or a trip out of central city, so when i use it, i use it a lot. dont have a car, dont want a car. much i can do by foot. taking some classes at pcc so i use the 44 often... get a day pass for that and make a lot of other trips that day.

  16. I generally ride on TriMet 3-4 days per week and I typically buy/use day passes. It's precisely for the reasons you mentioned:

    - I'm going round-trip to some place where I need an all-zone; and find it easier to just punch one "D" pass rather than deal with payment on both legs.

    - I'm traveling 2-z and think that there's a likely chance that I might be in a situation where I'd need 3 or more tickets.

    The thing about day tickets, however, is that the QFC near my house generally doesn't have more than 6 or 7 of them on-hand at any given time. TriMet doesn't make a book of them, like they do the 2Z or AZ tickets -- even if you go to Pioneer Square, they have to count them out one by one.


Keep it clean, please