Category: Shoddy Studies

Rumors of the Death of HSR Greatly Exaggerated

Aaron Renn has a post on New Geography pronouncing American high-speed rail dead. His reasoning: the stimulus spread the money around too much, Republican Governors rejected the HSR stimulus money, rail advocates have called 110 mph legacy lines high-speed rail, the FRA hobbles good passenger rail. All of those factors are true – though some cancel out, e.g. the 110 mph pretend-HSR lines in Wisconsin and Ohio were the first on the chopping block – but California HSR marches on.

Reading California HSR Blog gives an impression that the project is controversial, but in no real risk of disappearing. While some of the money from the canceled lines went to chaff, a lot went to California, which already has enough money to build a demonstration line in the Central Valley and is already looking at leveraging other money it will get to reach either Los Angeles or the Bay Area. Moreover, although the authority still carries over a lot of past incompetence, the current administration of Roelof van Ark is looking at alternatives to reduce costs, such as reducing the number and length of viaducts and even revisiting past alignment decisions. The adults are more firmly in charge today than a year and a half ago.

There’s still NIMBYism, particularly from Central Valley farmers and from suburbs on the San Francisco Peninsula, but the former is no big deal by the standards of what TGV construction has to go through, and the latter has simply led the authority to focus on connecting HSR track to Los Angeles first and use legacy track at slightly lower speed with much less local impact to get to San Francisco. Whether the project will ultimately have a useful starter line or remain a Bakersfield-Fresno-Merced shuttle depends on how much private funding it can attract, but Japan promised to fund 50% of the line, and the authority has had meetings with Spain and China. It’d be enough to do at least LA-Fresno, which is quite useful, if not as good as LA-Fresno-San Francisco.

Moreover, calling HSR dead on New Geography and saying it’s because Republican Governors rejected the money is ironic, in light of who owns the site. Aaron is interested in reform and efficiency; the same can’t be said of New Geography executive editor Joel Kotkin, an anti-urbanist so uninformed and desperate he blamed megacities for AIDS.

Kotkin may be just uninformed, but contributing editor Wendell Cox goes further: he and fellow Reason transportation hack Robert Poole wrote a report claiming, on flimsy evidence, that Florida’s high-speed rail line would have huge cost overruns and ridership shortfalls (a later report released by professional consultants said in fact the line would have been more profitable than expected). The report is a lie, and Rick Scott’s cancellation of the Florida HSR line, based on the report, involved additional lying to the court.

My explanation, hoisted from a comment I wrote on the subject on the Infrastructurist, responding to commenter Colin Prime:

1. The executive summary – i.e. what most people would read – says, “This report estimates that the cost to Florida taxpayers could be $3 billion more than currently projected.” As it turns out, in the body of the report in the section on Flyvbjerg the report says $0.54-2.7 billion, with $1.2 billion as the likeliest. None of these lower figures appears in the executive summary. That alone suggests massive deception.

2. In fact, Flyvbjerg either talks about megaprojects in general or focuses on urban rail. HSR projects don’t run over budget frequently, and when they do, it’s not by 100%. In Norway, a 50% cost overrun on the HSR line to the airport (coming from geological problems) was considered so unusual it triggered a government investigation.

3. Here’s the report on California [the projected per-km cost of the Central Valley segment is much higher than that of the Florida line]: “The California segment is not being built to full high-speed rail standards, because of a legal requirement that the line be usable by conventional Amtrak services if the Los Angeles to San Francisco project is not completed. The line would be upgraded to full high-speed rail standards when and if the much longer route is completed.”

This is technically known as “a lie.” Making the line Amtrak-usable is actually a cost raiser rather than a cost saver, because Amtrak trains are heavier and therefore elevated structures would have to be beefed up. Otherwise the line is built to HSR standards in terms of the expensive bits, i.e. track geometry and physical infrastructure; the only component that may not be included in California in this round is electrification, which is a fraction of the total cost of HSR ($3 million per mile kilometer at Acela costs).

4. In general, of the 11 factors cited for California-Florida differences, the ones on which Florida would be more expensive than California are all small things like stations and electricity; the big items involve physical infrastructure, and there Florida would be cheaper.

5. To support the assertion that HSR can suffer from a ridership shortfall, the report mentions Eurostar and THSR. Unmentioned are the many TGV lines that exceeded projections. The report also makes a spurious comparison to the Acela; it even doubles-down on the Acela comparison, and uses a false comparison to make the Florida line look slow. Florida’s travel time is compared not with end-to-end travel time on the average fast train (an average of 80 mph on the Acela NY-DC, and 140 mph on the Sanyo Shinkansen) but with the fastest intermediate segment on the fastest train of the day, connecting two small cities (100 and 170, respectively). On top of it, the Acela is priced for premium travel, with coach travel provided by the 66 mph Regional.

6. To add insult to injury, Cox and Poole dismiss Florida’s tourism as such: “The metropolitan areas in both markets [NEC and Florida] have substantial tourist volumes.” In reality, the tourist volumes in Florida relative to the metro area size are much larger than in the Northeast, and the Florida line directly serves tourist attractions (airport to Disneyland) whereas the Acela does not (minimal airport service, premium brand).

Given the above issues the study, I’d say calling it a lie is fair.

High-speed rail has challenges, many correctly identified by Aaron. The FRA is an obstacle (though the people most interested in changing it tend to be good transit activists); spreading the money around was a problem. But right-wing populists who can’t govern soon become unpopular, and are thus an ephemeral phenomenon. Rick Scott’s approval rate is 27%, John Kasich‘s is 35%, Scott Walker‘s is 37%. And it’s deeply troubling to go on a website and say that high-speed rail is dead when one of the reasons it’s dead is shoddy or dishonest work done by another contributor to the same website.

Fortunately, in California, the real obstacle is so far not a huge deal (California is planning to run on dedicated tracks, or at least on tracks shared only with commuter trains), and the ephemeral obstacle lost the gubernatorial election. Money is a problem and so is incompetence, but the incompetence seems to be waning, albeit slowly, and the money is likely to materialize. Don’t count HSR out yet.

Quick Note on Air Pollution

Yesterday’s USA Today carried a story about a study from the Harvard Center for Risk Analysis coming up with a huge figure for excess mortality, 2,200 nationwide just from the extra gas consumption caused by traffic congestion. Such a figure is almost certainly too high.

On page 4, the study compares the costs of congestion in terms of wasted time, wasted gas, and excess mortality due to pollution. In 2020, the cost of excess mortality is given as just under $20 billion in the largest 83 urban areas. Since the total amount of fuel wasted due to congestion according to the TTI, on whose data the study is based, is about 3.5 billion gallons, this corresponds to more than $5 per gallon.

With this figure in hand, we can compare the study to studies of car pollution and not just congestion pollution. American studies tend to find much lower costs of pollution and lower percentages of pollution coming from cars than foreign studies, and foreign studies find costs in the $2-3/gallon range. For examples, see here for Toronto and compare with fuel consumption figures coming from carbon emissions figures in the same study; here for Sydney and compare with fuel consumption figures from here; and here for Auckland sourced to this New Zealand study and compare with these fuel consumption numbers. Note that in the US, such figures are considered high-end estimates – see anything on social costs by Mark Delucchi.

The most likely reason for the factor-of-2 discrepancy we obtain is funding sources. The study under discussion was sponsored by the transportation construction industry, and was conducted by a research institute that had ties to the tobacco industry in the 1990s.

The study’s content indeed suggests such interest conflicts. The methodology estimates pollution per unit of VMT; it could just as well have posted figures for total car pollution. And the conclusion, far from suggesting regulations or pollution pricing, is:

long-term policy alternatives for addressing congestion such as traffic management through congestion pricing, traffic light synchronization and more efficient response to traffic incidents, and adding new highway and public transit capacity.

Adding more transit capacity is reasonable, since it displaces car trips. But adding highway capacity means people drive more, which increases rather than reduces pollution. And nowhere does the study recommend a tax on gas, which is what causes this pollution in the first place. This is not serious public health research; it’s lobbying for construction.

Whither BRT?

The Institute for Transport and Development Policy has joined Brookings in publishing a completely pointless transit system ranking, this time focusing on the quality of BRT, the mode of transit ITDP advocates.

I want to like ITDP for its BRT planning guide tome, but this BRT ranking uses random criteria, with bad weightings. Every system is ranked out of 100 points, with points divided into small criteria and subcriteria. On page 17, we see the following:

Off-vehicle fare collection 7
Multiple routes use same BRT infrastructure 4
Peak period frequency 4
Routes in top 10 demand corridors 4
Integrated fare collection with other public transport 3
Limited and local stop services 3
Off-peak frequency 3
Part of ( planned ) multi-corridor BRT network 3
Performance-based contracting for operators 3
Enforcement of right-of-way 2
Operates late nights and weekends 2
Operational control system to reduce bus bunching 2
Peak-period pricing 2

Bus lanes in central verge of the road 7
Physically-separated right-of-way 7
Intersection treatments (elimination of turns across the busway and signal priority) 4
Physically-separated passing lanes at station stops 4
Stations occupy former road/median space (not sidewalk space) 3
Stations set back from intersections (100 feet min) 3
Stations are in center and shared by both directions of service 2

Platform-level boarding 5
Buses have 3+ doors on articulated buses or 2+ very wide doors on standard buses 4
Multiple docking bays and sub-stops ( separated by at least half a bus length ) 3

Branding of vehicles and system 3
Safe, wide, weather-protected stations with artwork (>/=8 feet wide) 3
Passenger information at stops and on vehicles 2

Bicycle lanes in corridor 2
Bicycle sharing systems at BRT stations 2
Improved safe and attractive pedestrian access system and corridor environment 2
Secure bicycle parking at station stops 2

Those criteria are for the most part not bad, but they’re weighted wrong. Observe that off-peak frequency counts for only 3 points, the same as contracting out the operations. It’s actually worse: a system gets 1 point for having any off-peak frequency, even if it’s worse than 15 minutes; 15 minutes is enough for 2 points. Peak-period pricing, which is absent or all but absent from many well-run rail and bus operations around the world, gets 2 points. The core elements of BRT – level boarding, physically separated median lanes, off-board fare collection, signal preemption – have 36 points between them.

In first-world cities, BRT has two uses. One, lower-capacity, slightly lower-quality transit on corridors with less demand. Two, dedicated guideways that can branch out and make local stops in shared lanes in lower-traffic areas, on the model of Brisbane. The full-fat BRT in Guangzhou and Bogota cited by BRT proponents requires a lot of concrete and many operators, and is best-suited to a city with low labor costs.

Many of the features touted for BRT can and should be used for all buses. Off-board fare collection with proof of payment is practiced systemwide in such cities as Singapore, Paris, Berlin, Zurich, and Florence; in conjunction with multi-door boarding, this reduces bus dwell times and increases speed with zero investment in concrete. Signal priority can be practiced independently of all else. Physical separation of lanes requires barriers only a few centimeters wide, and can be done selectively on the most congested and highest-demand segments.

Buses can be great buses; they make bad trains. By all means first-world cities should increase frequency, procure better buses with low floors and more doors, make sure riders know which routes are frequent and which are not, and give buses dedicated lanes when necessary. But the focus on specially branded rail-like BRT only detracts from this goal.

In American cities, BRT is more often than not an excuse to not implement those features on local buses. In New York, not only does the MTA rule out proof-of-payment on non-SBS buses, but also backroom state legislative dealings banned bus camera enforcement of painted lanes except on a closed list of six SBS routes. All this while SBS service levels are comparable to those of local buses in Singapore and many European cities – in fact lower if those local buses have signal priority. This and not low scoring on an arbitrary rubric is what ITDP should have complained about.

Brookings Folly

People who have read Brookings’ awful report saying San Jose is the second most transit-accessible city in the US and New York the thirteenth already know not to trust what Brookings says. Even at the level of collecting facts, it seems to get service frequency wrong, making sprawling suburbs with hourly bus service look like they have service every few minutes.

So it’s not surprising that senior Brookings fellow Robert Puentes’ article about infrastructure in the Wall Street Journal is full of misunderstandings and frankly amateurish claims about US infrastructure problems. Puentes opens with a standard claim that “we do a great job of building new roads” (no mention of the Big Dig, Bay Bridge Eastern Span replacement, or proposed Tappan Zee replacement, each substantially costlier than undersea tunnels in Europe) but smarter investments are needed. He proposes the following:

1. Boosting exports. Puentes complains that US border crossings are congested, and hints that more are needed, for example the proposal for more bridges between Michigan and Ontario. He mentions some interstate cooperation as a solution, but never says anything about international cooperation, which is the real problem.

The Ambassador Bridge carries 10,000 trucks and 4,000 cars per day; the Holland Tunnel, which has the same number of lanes, carries 90,000 vehicles per day. The problem with the bridge is the border crossing, not the infrastructure. Nowhere does Puentes say the US and Canada should build more border checkpoints or process people faster. If Michigan doubles the number of border control booths or halves the time it takes to process a vehicle, it’s equivalent to building another bridge, but at a vastly lower cost.

2. Getting greener. Puentes praises Obama for proposing to put one million electric cars on the road in 2015. Then he talks about charging stations and the need for national standards encouraging them. In reality, the US has 240 million cars on the road, so Obama’s proposal would, even assuming zero-emission electricity, cut car emissions by 0.4%.

On the subject of cars and transit, Puentes wisely mentions that the government funds roads more liberally, but instead of railing against highways to nowhere and high construction costs says “We need equal treatment of all possible transportation projects, so cities don’t have to give up on, say, transit systems that fit their needs and help us go green, just because they cost more than highways.” It’s not that there aren’t examples of severe waste; it’s that Puentes doesn’t seem to care.

3. Adding Innovations. This is a boilerplate blurb for electronic toll collection, bus tracking, and, of course, public-private partnerships. Individually the things proposed are not bad – they’re just the most important things. For mass transit, fare integration and tighter schedule adherence are more important, but were not invented here and involve messy fights with the bureaucrats Brookings represent.

The PPP part indicates what this is really about: kickbacks to technology companies, often defense contractors looking to diversify. Many US transit systems have a smartcard using vendor-locked proprietary technology; defense contractor Cubic is the top vendor. New York’s smartcard proposal is instead a kickback to credit card providers, which is slightly less bad because the standard is open but is still far behind best practices. The best practices do not involve PPPs – instead, agencies develop technologies in-house, or instead rely on open standards. Minimal collusion offers minimal opportunity for corruption.

4. Connecting Workers With Work. Here Puentes repeats his institution’s flawed study’s findings as if they’re universally recognized facts. He does not even say “According to a recent Brookings study” – people are supposed to know it like they know Pearl Harbor happened in 1941. Then, based on said study’s conclusions, he declares the problem is that the poor are disconnected from their workplaces and makes relevant suggestions.

Since the Brookings study got things wrong in the direction of too much transit accessibility, the suggestions are for the most part not bad. The problem is that he says nothing about the problems of connecting people to where they work. The biggest problem for metro area transit is that while downtowns are reasonably connected (e.g. downtown LA workers have a 50% transit mode share), secondary downtowns and suburban job centers are not.

The common theme of all the proposals is that they’re makework for the bureaucrats and consultants who are Brookings’ base. Adopting best industry practices is useless to Brookings fellows, because pointing out that Europe does it better also means that the consultants who should implement reform are European managers. In contrast, PPP means coming up with new standards and new ways of doing things; it’s attractive to government administrators as much as it is to the companies that get the contracts.

The interests of the riders are not the same as those of the service providers. That labor does not have the same interests as riders is clear, but management benefits from bloat just as much: if things run smoothly, managers can’t look like they’re continually saving the day. Thinktanks like Brookings represent certain interest groups, and Brookings’ interest group excludes transit users.

Fatality Numbers vs. Safety

On Streetsblog, they’re waving New York’s relatively high pedestrian fatality rate as evidence the streets are unsafe and much more can be done. The region’s pedestrian death rate is the 13th worst in the nation, about the same as Houston, which is supposed to be evidence of unsafe streets.

John Adams points out that in Britain, the pedestrian fatality rate today is one third what it was in 1922. The roads are much less safe than they were then, when they were narrow and traffic was slow, but there are so few pedestrians today that cars rarely hit them. As a result, looking at absolute death rates means nothing.

Even the Transportation for America study that Streetsblog links to doesn’t fully correct for it. It scales fatality rates based on the pedestrian commute share, which is better than nothing, but still fails to account the huge volumes of people in New York and other walkable cities who take mass transit to work but still walk a lot for their other trips. The proof is in the pudding: the study says Cleveland is the second safest metro area in the US for pedestrians, behind Boston and ahead of New York.

New York has a lot of street safety issues, but it’s still light years ahead of the rest of the US, except for small pockets in Boston, San Francisco, and other compact, walkable cities. The same is true for Manhattan within New York. Ignore complaints that the community board comprising the Upper East Side has the third highest pedestrian fatality count; it also has the third highest population, trailing two outer-urban CBs with fewer pedestrians. At this stage input-based measures such as traffic speed, sidewalk width, stoplight phasing, and the presence of a good street wall and trees are much better than any skewed output-based statistic.

As a corollary, bike lane opponents who complain about the large number of cyclist injuries on protected bike lanes are just as wrong (see here and scroll for comments). There are more cyclists on 9th Avenue than on pre-bike lane Prospect Park West; of course more will be injured. Counterintuitive claims about how bike lanes are less safe than mixed traffic are fun, but they aren’t true.

More on Driving vs. Transit Costs

Thanks to Elizabeth Alexis of CARRD for finding and giving me a link to the AAA’s methodology for computing driving costs, used in APTA’s flawed study about the high household savings coming from switching from driving to transit. The AAA methodology indeed assumes perfect rather than realistic maintenance and tire changing, and has elevated depreciation and warranty charges.

The full list of problems with the AAA methodology, according to Elizabeth:

You are spot on about the misuse of data.  The AAA study is really misleading It represents the costs for someone who buys a new car from the dealer  with the extended warranty, overinsures it, drives it for 5 years, buys a new set of tires and then trades it in to the dealer, getting totally ripped in the process.  If everyone did this, the average car fleet would be 2 1/2 years old (instead of 9).     The only thing this study tells you is that you should never buy a new car and that you are an idiot to do anything but buy used cars off craigslist.

They are also assuming:

1) You buy a new car every five years.
2) Even though you know you will sell the car, you buy the extended warranty.
3) You accept the dealer’s trade-in price (which is very low generally).
4) Even though you know you are going to sell it to the dealer for no money, you go ahead and put on a new set of tires right before doing so.
5) You buy insurance with really low deductibles.
6) Because on average you have a 2.5 year old car, your annual car tax and your insurance are very high (in most states, the taxes are based on the value of the car).
7) And you finance the car @ non-deductible 6% interest. It should be noted that most car loans are 3-5 years.  So if you kept a car after it was paid off… this cost would go away.

A better study for the costs of driving was done by Steven Polzin, of the National Center for Transit Research, who also serves on “several APTA committees.” Using various government survey data, he finds an average saving of $3,600 from giving up a car; this is less than the cost of an average car, since households might give up the lesser used car or take more transit or drive the remaining car more. I encourage everyone to bookmark the study and refer to page 18 for comparative spending on transportation in the US versus the EU-15; it’s a difference of 19.5% of household budget versus about 14%. Any figures for world public transit leaders Japan and Switzerland will be appreciated.

How much does driving cost?

APTA has just come out with a press release touting large savings for households that do not drive but instead take mass transit. The average it claims is $10,000 per year per person nationwide; in some cities it’s higher, and Second Avenue Sagas seized upon the study’s claim of a $14,000 saving in New York.

The only problem: drivers in the US do not spend $10,000 a year, let alone $10,000 more than transit riders. The federal government has a detailed breakdown of household budgets, so that it can compute inflation rates accurately and set cost-of-living adjustments and monetary policy. The New York Times has a nifty graphic breaking down household spending as of 2008, and transportation was 18% of American households’ budgets, of which about 17% is on cars and 1% is on all other forms of transportation. Mean household income in the US is $68,000 per year as of 2009, so we’re talking about $11,560 spent on cars per household. There are on average a bit more than 2 cars per household in the US (246 million cars, about 113 million households), so we’re talking about $5,400 per car. Not $10,000.

At least it’s better than its shilling against a climate bill on the grounds that some of the carbon taxes it would raise from roads would not go to the Highway Trust Fund, to the point that it proposed an alternative that would raise less money for transit. But in either case, it cares less about a mode shift benefiting transit users and the environment than about lobbying for transit operators’ interests.

New York Provincialism, and the MTA’s Flawed Smartcard Report

New York’s MTA recently published a report proposing a next-generation payment system replacing the MetroCard. You can find it here: to read it, download it and add .pdf to the file extension. Sections 4-5 are the most relevant here.

The report is full of little facts about New York such as the number of transactions on each component of the MTA, but never once mentions case studies abroad – Hong Kong’s Octopus, Tokyo’s Suica and PASMO, or the many European cities that are happy with paper tickets. It uses factoids to intimidate more than to explain. For example, it repeats the fact that New York City Transit spends 15% of its revenue on fare collection, but never breaks it down to parts, does cross-city comparisons, or even estimates how much a smartcard system will save; the only purpose of the number is therefore to scare people into doing something.

Despite advertising its intention to save money, the MTA makes no mention of bundling smartcards with proof-of-payment, widely used way to speed up bus boarding and reduce train staffing levels, even with plain paper tickets. On the contrary, the report specifically mentions equipping commuter rail conductors (and not fare inspectors) with card readers, and only mentions inspectors in relation to Select Bus Service and the Staten Island Railway.

Even on the level of checking existing technology use, the report falls short. The MTA rates smartcard options as “medium-low” on “inter-modal interoperability,” on the grounds that they require card validators and card readers. In reality, card validators are cheap: in Singapore, they cost about S$950 per unit (about US$770); placing one at every bus stop and commuter train station and on board every bus door and commuter train door pair would cost $20 million, less than a tenth the cost of a smartcard implementation.

Similarly, the report rates internal transit smartcards’ lifecycle risk as “medium: mature technology, though standards are not.” The closest thing to truth in there is that there are two open standards, Sony’s FeliCa and a separate standard whose top vendor is NXP’s MIFARE, and the ISO chose the standard used by MIFARE over FeliCa (FeliCa was already in place in Japan and Hong Kong, so it still has the most users). In reality, both FeliCa and MIFARE date to the mid-1990s, making them older than the smartphone and broadband Internet.

The report mentions a foreign city exactly once: it says that “The technology risk is mitigated by Transport for London’s adoption of open payments, planned for 2012, and its role in advising the MTA and potentially sharing technology.” Optimistically, it means the MTA listens to other cities when they say what it wants to hear. Pessimistically, both cities are using each other to justify a prior decision. MTA Chairman Jay Walder, the primary proponent within the MTA of the credit card-based smartcard, worked in London until 2007 and was responsible for introducing the Oyster card.

And speaking of London, Oyster is bumpy at best. It is superficially similar to Hong Kong’s Octopus, down to the similar name, but in practice it is much more primitive. Octopus is licensed as anonymous electronic money (in a culture that according to Western stereotype is authoritarian and indifferent to privacy), generating additional profits to the MTR; Oyster is not, and the MTA report makes no mention of this possibility. Octopus comes in more forms than just a card – for example, there is an Octopus watch and an Octopus keychain, making tapping easier since the rider does not need to take out their wallet; Oyster does not, and when riders took out the chip to create a makeshift Oyster watch, TfL fined them even though they were not dodging the fare.

The MTA keeps underperforming because it doesn’t listen to other cities’ experience, unless it’s what it wants to hear. And this is perhaps the worst abuse, because here the person who’s leading the charge for reform in New York has a track record of screwing up abroad. New York has spent decades convincing itself that it is the best city in the world and needs to learn from no other, taking pride in its subway. The result has been a metro area transit mode share lower than that of European cities one tenth New York’s size. Walder speaks like a reformer who tries to change this, but the one time he’s proposing something concrete, it’s the usual New York provincialism.