Berlin Greens Know the Price of Everything and Value of Nothing

While trying to hunt down some numbers on the costs of the three new U5 stations, I found media discourse in Berlin about the U-Bahn expansion plan that was, in effect, greenwashing austerity. This is related to the general hostility of German urbanists and much of the Green Party (including the Berlin branch) to infrastructure at any scale larger than that of a bike lane. But the specific mechanism they use – trying to estimate the carbon budget – is a generally interesting case of knowing the costs more certainly than the benefits, which leads to austerity. The underlying issue is that mode shift is hard to estimate accurately at the level of the single piece of infrastructure, and therefore benefit-cost analyses that downplay ridership as a benefit and only look at mode shift lead to underbuilding of public transport infrastructure.

The current program in Berlin

In the last generation, Berlin has barely expanded its rapid transit network. The priority in the 1990s was to restore sections that had been cut by the Berlin Wall, such as the Ringbahn, which was finally restored with circular service in 2006. U-Bahn expansion, not including restoration of pre-Wall services, included two extensions of U8, one north to Wittenau that had begun in the 1980s and a one-stop southward extension of U8 to Hermannstrasse, which project had begun in the 1920s but been halted during the Depression. Since then, the only fully new extension have been a one-stop extension of U2 to Pankow, and the six-stop extension of U5 west from Alexanderplatz to Hauptbahnhof.

However, plans for much more expansive construction continue. Berlin was one of the world’s largest and richest cities before the war, and had big plans for further growth, which were not realized due to the war and division; in that sense, I believe it is globally only second to New York in the size of its historic unrealized expansion program. The city will never regain its relative wealth or size, not in a world of multiple hypercities, but it is growing, and as a result, it’s dusting off some of these plans.

U8 is the north-south line from Wittenau to the southern leg of the Ring – the intersection station, Hermannstrasse, is unlabeled.

Most of the lines depicted in red on the map are not at all on the city’s list of projects to be built by the 2030s. Unfortunately, the most important line measured by projected cost per rider, the two-stop extension of U8 north (due east) to Märkisches Viertel, is constantly deprioritized. The likeliest lines to be built per current politicking are the extensions of U7 in both directions, southeast ti the airport (beyond the edge of the map) and west from Spandau to Staaken, and the one-stop extension of U3 southwest to Mexikoplatz to connect with the S-Bahn. An extension to the former grounds of Tegel is also considered, most likely a U6 branch depicted as a lower-priority dashed yellow line on the map rather than the U5 extension the map depicts in red.

The carbon critique

Two days after the U5 extension opened two years ago, a report dropped that accused the proposed program of climate catastrophe. The argument: the embedded concrete emissions of subway construction are high, and the payback time on those from mode shift is more than 100 years.

The numbers in the study are, as follows: each kilometer of construction emits 98,800 tons of CO2, which is 0.5% of city emissions (that is, 5.38 t/person, cf. the German average of about 9.15 in 2021). It’s expected that through mode shift, each subway kilometer saves 714 t-CO2 in annual emissions through mode shift, which is assumed to be 20% of ridership, for a payback time of 139 years.

And this argument is, frankly, garbage. The scale of the difference in emissions between cities with and without extensive subway systems is too large for this to be possibly true. The U-Bahn is 155 km long; if the 714 t/km number holds, then in a no U-Bahn counterfactual, Berlin’s annual greenhouse gas emissions grow by 0.56%, which is just ridiculous. We know what cities with no or minimal rapid transit systems look like, and they’re not 0.56% worse than comparanda with extensive rapid transit – compare any American city to New York, for one. Or look again at the comparison of Berlin to the German average: Berlin has 327 cars per 1,000 people, whereas Germany-wide it’s 580 and that’s with extensive rapid transit systems in most major cities bringing down the average from the subway-free counterfactual of the US or even Poland.

The actual long-term effect of additional public transport ridership on mode shift and demotorization has to be much more than 20%, then. It may well be more than 100%: the population density that the transit city supports also increases the walking commute modal split as some people move near work, and even drivers drive shorter distances due to the higher density. This, again, is not hard to see at the level of sanity checks: Europeans drive considerably less than Americans not just per capita but also per car, and in the United States, people in New York State drive somewhat shorter distance per car than Americans elsewhere (I can’t find city data).

The measurement problem

It’s easy to measure the embedded concrete of infrastructure construction: there are standardized itemized numbers for each element and those can be added up. It’s much harder to measure the carbon savings from the existence of a better urban rail system. Ridership can be estimated fairly accurately, but long-term mode shift can’t. This is where rules of thumb like 20% can look truthy, even if they fail any sanity check.

But it’s not correct to take any difficult to estimate number and set it to zero. In fact, there are visible mode shift effects from a large mass transit system. The difficulty is with attributing specific shifts to specific capital investments. Much of the effect of mode shift comes from the ability of an urban rail system to contribute to the rise of a strong city center, which can be high-rise (as in New York), mid-rise (as in Munich or Paris), or a mix (as in Berlin). Once the city center anchored by the system exists, jobs are less likely to suburbanize to auto-oriented office parks, and people are likelier to work in city center and take the train. Social events will likewise tend to pick central locations to be convenient for everyone, and denser neighborhoods make it easier to walk or bike to such events, and this way, car-free travel is possible even for non-work trips.

This, again, can be readily verified by looking at car ownership rates, modal splits (for example, here is Berlin’s), transit-oriented development, and so on, but it’s difficult to causally attribute it to a specific piece of infrastructure. Nonetheless, ignoring this effect is irresponsible: it means the carbon benefit-cost analysis, and perhaps the economic case as well, knows the cost of everything and the value of little, which makes investment look worse than it is.

I suspect that this is what’s behind the low willingness to invest in urban rail here. The benefit-cost analyses can leave too much value on the table, contributing to public transport austerity. When writing the Sweden report, I was stricken by how the benefit-cost analyses for both Citybanan and Nya Tunnelbanan were negative, when the ridership projections were good relative to costs. Actual ridership growth on the Stockholm commuter trains from before the opening of Citybanan to 2019 was enough to bring cot per new daily trip down to about $29,000 in 2021 PPP dollars, and Nya Tunnelbanan’s daily ridership projection of 170,000 means around $23,000/rider. The original construction of the T-bana cost $2,700/rider in 2021 dollars, in a Sweden that was only about 40% as rich as it is today, and has a retrospective benefit-cost ratio of between 6 and 8.5, depending on whether broader agglomeration benefit are included – and these benefits are economic (for example, time savings, or economic productivity from agglomeration) scale linearly with income.

At least Sweden did agree to build both lines, recognizing the benefit-cost analysis missed some benefits. Berlin instead remains in austerity mode. The lines under discussion right now are projected between 13,160€ and 27,200€ per weekday trip (and Märkisches Viertel is, again, the cheapest). The higher end, represented by the U6 branch to Tegel, is close to the frontier of what a country as rich as Germany should build; M18 in Paris is projected to be more than this, but area public transport advocates dislike it and treat it as a giveaway to rich suburbs. And yet, the U6 branch looks unlikely to be built right now. When the cost per rider of what is left is this low, what this means is that the city needs to build more infrastructure, or else it’s leaving value on the table.

Philadelphia and High-Speed Rail Bypasses (Hoisted from Social Media)

I’d like to discuss a bypass of Philadelphia, as a followup from my previous post, about high-speed rail and passenger traffic density. To be clear, this is not a bypass on Northeast Corridor trains: every train between New York and Washington must continue to stop in Philadelphia at 30th Street Station or, if an in my opinion unadvised Center City tunnel is built, within the tunnel in Center City. Rather, this is about trains between New York and points west of Philadelphia, including Harrisburg, Pittsburgh, and the entire Midwest. Whether the bypass makes sense depends on traffic, and so it’s an example of a good investment for later, but only after more of the network is built. This has analogs in Germany as well, with a number of important cities whose train stations are terminals (Frankfurt, Leipzig) or de facto terminals (Cologne, where nearly all traffic goes east, not west).

Philadelphia and Zoo Junction

Philadelphia historically has three mainlines on the Pennsylvania Railroad, going to north to New York, south to Washington, and west to Harrisburg and Pittsburgh. The first two together form the southern half of the Northeast Corridor; the third is locally called the Main Line, as it was the PRR’s first line.

Trains can run through from New York to Washington or from Harrisburg to Washington. The triangle junction northwest of the station, Zoo Junction, permits trains from New York to run through to Harrisburg and points west, but they then have to skip Philadelphia. Historically, the fastest PRR trains did this, serving the city at North Philadelphia with a connection to the subway, but this was in the context of overnight trains of many classes. Today’s Keystone trains between New York and Harrisburg do no such thing: they go from New York to Philadelphia, reverse direction, and then go onward to Harrisburg. This is a good practice in the current situation – the Keystones run less than hourly, and skipping Philadelphia would split frequencies between New York and Philadelphia to the point of making the service much less useful.

When should trains skip Philadelphia?

The advantage of skipping Philadelphia are that trains from New York to Harrisburg (and points west) do not have to reverse direction and are therefore faster. On the margin, it’s also beneficial for passengers to face forward the entire trip (as is typical on American and Japanese intercity trains, but not European ones). The disadvantage is that it means trains from Harrisburg can serve New York or Philadelphia but not both, cutting frequency to each East Coast destination. The effect on reliability and capacity is unclear – at very high throughput, having more complex track sharing arrangements reduces reliability, but then having more express trains that do not make the same stop on the same line past New York and Newark does allow trains to be scheduled closer to each other.

The relative sizes of New York, Philadelphia, and Washington are such that traffic from Harrisburg is split fairly evenly between New York on the other hand and Philadelphia and Washington on the other hand. So this really means halving frequency to each of New York and Philadelphia; Washington gets more service with split service, since if trains keep reversing direction, there shouldn’t be direct Washington-Harrisburg trains and instead passengers should transfer at 30th Street.

The impact of frequency is really about the headway relative to the trip time. Half-hourly frequencies are unconscionable for urban rail and very convenient for long-distance intercity rail. The headway should be much less than the one-way trip time, ideally less than half the time: for reference, the average unlinked New York City Subway trip was 13 minutes in 2019, and those 10- and 12-minute off-peak frequencies were a chore – six-minute frequencies are better for this.

The current trip time is around 1:20 New York-Philadelphia and 1:50 Philadelphia-Harrisburg, and there are 14 roundtrips to Harrisburg a day, for slightly worse than hourly service. It takes 10 minutes to reverse direction at 30th Street, plus around five minutes of low-speed running in the station throat. Cutting frequency in half to a train every two hours would effectively add an hour to what is a less than a two-hour trip to Philadelphia, even net of the shorter trip time, making it less viable. It would eat into ridership to New York as well as the headway rose well above half the end-to-end trip, and much more than that for intermediate trips to points such as Trenton and Newark. Thus, the current practice of reversing direction is good and should continue, as is common at German terminals.

What about high-speed rail?

The presence of a high-speed rail network has two opposed effects on the question of Philadelphia. On the one hand, shorter end-to-end trip times make high frequencies even more important, making the case for skipping Philadelphia even weaker. In practice, high speeds also entail speeding up trains through station throats and improving operations to the point that trains can change directions much faster (in Germany it’s about four minutes), which weakens the case for skipping Philadelphia as well if the impact is reduced from 15 minutes to perhaps seven. On the other hand, heavier traffic means that the base frequency becomes much higher, so that cutting it in half is less onerous and the case for skipping Philadelphia strengthens. Already, a handful of express trains in Germany skip Leipzig on their way between Berlin and Munich, and as intercity traffic grows, it is expected that more trains will so split, with an hourly train skipping Leipzig and another serving it.

With high-speed rail, New York-Philadelphia trip times fall to about 45 minutes in the example route I drew for a post from 2020. I have not done such detailed work outside the Northeast Corridor, and am going to assume a uniform average speed of 240 km/h in the Northeast (which is common in France and Spain) and 270 km/h in the flatter Midwest (which is about the fastest in Europe and is common in China). This means trip times out of New York, including the reversal at 30th Street, are approximately as follows:

Philadelphia: 0:45
Harrisburg: 1:30
Pittsburgh: 2:40
Cleveland: 3:15
Toledo: 3:55
Detroit: 4:20
Chicago: 5:20

Out of both New York and Philadelphia, my gravity model predicts that the strongest connection among these cities is by Pittsburgh, then Cleveland, then Chicago, then Detroit, then Harrisburg. So it’s best to balance the frequency around the trip time to Pittsburgh or perhaps Cleveland. In this case, even hourly trains are not too bad, and half-hourly trains are practically show-up-and-go frequency. The model also predicts that if trains only run on the Northeast Corridor and as far as Pittsburgh then traffic fills about two hourly trains; in that case, without the weight of longer trips, the frequency impact of skipping Philadelphia and having one hourly train run to New York and Boston and another to Philadelphia and Washington is likely higher than the benefit of reducing trip times on New York-Harrisburg by about seven minutes.

In contrast, the more of the network is built out, the higher the base frequency is. With the Northeast Corridor, the spine going beyond Pittsburgh to Detroit and Chicago, a line through Upstate New York (carrying Boston-Cleveland traffic), and perhaps a line through the South from Washington to the Piedmont and beyond, traffic rises to fill about six trains per hour per the model. Skipping Philadelphia on New York-Pittsburgh trains cuts frequency from every 10 minutes to every 20 minutes, which is largely imperceptible, and adds direct service from Pittsburgh and the Midwest to Washington.

Building a longer bypass

So far, we’ve discussed using Zoo Junction. But if there’s sufficient traffic that skipping Philadelphia shouldn’t be an onerous imposition, it’s possible to speed up New York-Harrisburg trains further. There’s a freight bypass from Trenton to Paoli, roughly following I-276; a bypass using partly that right-of-way and, where it curves, that of the freeway, would require about 70 km of high-speed rail construction, for maybe $2 billion. This would cut about 15 km from the trip via 30th Street or 10 km via the Zoo Junction bypass, but the tracks in the city are slow even with extensive work. I believe this should cut another seven or eight minutes from the trip time, for a total of 15 minutes relative to stopping in Philadelphia.

I’m not going to model the benefits of this bypass. The model can spit out an answer, which is around $120 million a year in additional revenue from faster trips relative to not skipping Philadelphia, without netting out the impact of frequency, or around $60 million relative to skipping via Zoo, for a 3% financial ROI; the ROI grows if one includes more lines in the network, but by very little (the Cleveland-Cincinnati corridor adds maybe 0.5% ROI). But this figure has a large error bar and I’m not comfortable using a gravity model for second-order decisions like this.

High-Speed Rail Doesn’t Depend on Megaregions

On my Discord channel, I was reminded of the late-2000s work by some institutional American urbanists about the concept of megaregions. Wikipedia has a good summary of the late-2000s discourse on the subject. In short, there are linear ties across the East Coast from Boston to Washington (“BosWash”), with more or less continuous suburban development in between, and some urbanists tried to generalize this concept to other agglomerations of metropolitan areas, not usually successfully. The American work on this carved most of the country’s population into 10 or 11 megaregions, sometimes annexing portions of Canada, as in the Regional Plan Association’s America 2050 program:

There is a lot to critique about this map. Canada has a strong self-conception as a distinct entity from the United States; while there’s a case for lumping Vancouver with Seattle and Portland as the Pacific Northwest, lumping Toronto with the Midwest is irresponsible. The Hampton Roads region is not meaningfully a periphery of the Northeast, but is rather Southern (for example, it is heavily militarized, and the South has consistently higher enlistment rates than the Northeast). The Rio Grande Valley is not especially connected with New Orleans.

But the core of the program is to propose this as the basis of high-speed rail investment, and that’s where it fails the most visibly. When one of my Discord channel participants posted the map in the channel about high-speed rail, I started talking about my gravity model, and pointed out some patterns that emerge.

For this, consult a table of ridership between any pair of American or Canadian cities in the main connected component of my proposed map:

The table omits Texas, California, and the Pacific Northwest. But it includes lines that I initially considered and rejected, going to Kansas City and Birmingham; the reason is that when I calculated it by hand I omitted very weak long-range connections such as between Boston and the Midwest, whereas the table can automatically calculate them and add them in, producing an estimate of 5 million annual riders between Boston and the entire Midwest region. These extra connections make weak lines like those to Birmingham and Kansas City appear stronger, so those lines are included; it’s plausible they could even justify a connection to Texas via both New Orleans and and Tulsa, but those are not included (and would at any case not impact the analysis below).

The following table includes some connections between two adjacent cities in the table, with their total projected passenger counts. Those are very high numbers, higher than you’d expect; this is because they lump in a great many city pairs – for example, New York-Philadelphia includes all connections from New York, Boston, and Albany to Philadelphia and points south and west, and those sum to a much higher number than just the internal trips on the Northeast Corridor, let alone just trips originating in New York and ending in Philadelphia or the reverse. Also, as a note of caution, there may be small inaccuracies if I mistakenly tabulated very weak markets like Chicago-Charlotte as going via the wrong path; they do not change the main conclusion.

City pairRidership
Boston-New York39,299,133
Boston-Springfield25,482,364
New York-Philadelphia/Harrisburg139,860,707
Philadelphia-Washington110,010,205
Washington-Richmond64,145,050
Richmond-Raleigh50,425,578
Raleigh-Greensboro42,654,519
New York-Albany57,773,629
Philadelphia-Harrisburg65,639,871
Harrisburg-Pittsburgh61,110,782
Pittsburgh-Cleveland62,352,156
Cleveland-Toledo56,482,182
Cleveland-Columbus46,046,790
Buffalo-Cleveland41,584,062

Some observations jump from this (partial) table:

  • New York-Boston is much weaker than a lot of segments that are by themselves far weaker than the Northeast Corridor. The reason for this is that a full 31.1 million annual riders on New York-Boston are internal to the Northeast Corridor, whereas the other city pairs require large swaths of the network to be built to have such high traffic.
  • From Philadelphia to points west, traffic density is fairly consistent. There’s no separation between a Northeastern and Midwestern megaregion evident in the data: Cleveland has about the same traffic density going east and west, as does Pittsburgh. Rather, it’s the connections between the East Coast and the Midwest, chiefly Philadelphia-Pittsburgh-Cleveland but also the Empire corridor between Albany and Cleveland, that create high ridership.
  • Washington-Atlanta is a tail gradually weakening with distance from the Northeast Corridor, rather than an independent corridor.

Outside the US, the same observation about the irrelevance of megaregions to high-speed rail is true. The European attempt to describe a megaregion, the so-called Blue Banana, was constructed explicitly to exclude France – but the highest-traffic density intercity rail link in Europe is between Paris and the bifurcation splitting toward Lyon and Dijon. Frankfurt-Mannheim is a close second, but French intercity trains average around 220 km/h and German ones around 130 km/h depending on the line, and the actually existing high-speed rail network gets higher peak traffic density than the medium-speed one.

Ultimately, high-speed rail as a mode of transportation is a means of connecting metropolitan areas. Whether they fall into megaregions or not is immaterial – some strong links connect distinct regions, like Northeast-Midwest, with higher demand for traffic than some of the internal connections.

Quick Note on My New York Trip

I am back in Europe now (in London until Tuesday), but I was in New York for nearly three weeks, and it was interesting reconciling what I was seeing with what everyone else is saying about the city. It and my March 2022 trip were both enlightening in a way because I’d last been in the US at the end of 2019, so many New Yorkisms that I was used to in the 2000s and 2010s suddenly jarred me as foreign to what I had grown used to in Europe.

As one might expect based on the subject of this blog, I took the subway a lot. I took it so much that I was using weekly passes, and the last week I had a weekly pass for just three days and still I took 13 trips on those days, justifying its cost (which is like that of 12 single trips). I saw things, and notably didn’t see others.

What I did see: abject unreliability. I snapped a photo whenever the train arrival board was showing something weird, like low frequency or bunching; if you’re reading this post as it’s being posted and not going on a deep archive run, then go to my Twitter media and look at the last few weeks of pictures. Out of 19 days, something was going wrong 10 times, usually on the train I used to get between my Queensbridge hotel and Marron, the F train – and that’s without counting a few trips when the train frequency looked good but then I was delayed 10-20 minutes due to incidents. Something would always come up: signal failure, medical emergency, mechanical failure, cascading delays. Uday Schultz, a railfan who scares me with the depth of his knowledge of operations, maintenance, and rail history, points out how one such delay compounded due to bad interlining.

This is not normal. Berlin has delays but nowhere nearly this often – not on the U-Bahn but also not on the S-Bahn, whose interlining complexity is comparable to that of the New York City Subway. Low-frequency sections due to single-tracking for maintenance exist in Berlin, but it’s rare, and trains do not run worse than every 10 minutes except on the suburban periphery of the city. Over a similar period of time in Berlin I might see an incident bad enough to complain to BVG about it on Twitter maybe once or twice, not 10 times.

What I didn’t see: significant crime. I point out that I was staying near Queensbridge because the area is negatively stereotyped by suburbanites and city residents with I-hate-(the-rest-of-)the-city identity politics. Nothing there looked scary, at any time of day. There’s a large housing project there, which I mostly associate with people playing the Halloween theme song on 10/31 for what I imagine was a showing of the film and with some people wearing delightfully scary costumes. The worst I saw was someone selling swipes illegally when there was an unusually long line for the ticketing machines; there were cops on the platform who must have passed this person by and apparently done nothing.

I point this out because the city is convinced that the subway is dangerous. There are annoying announcements all the time: “this is an important message from the New York City Police Department…” It makes for some awful user experience – there’s no possibility of quiet on the train, for which those announcements contribute more than anything, since the panhandlers are much less common and the background noise is easier to tune out. People who speak limited English or can’t make out the phonemes garbled over bad announcer systems learn to tune everything out, including the occasional useful announcement of service changes.

And the police loves how annoying it is, which it justifies by appealing to safety theater. When Sarah Meyer tried reducing the annoyance levels, she ran into some real and some made-up technical problems, and one political problem in that nobody in management cares about UX. The police said they need those announcements, annoying and counterproductive as they are (telling tourists to watch their belongings gets them to grasp their wallets in fear, alerting every thief to the location of the wallet on their person); nobody at the agency thought to push back. In the last few days, a new disturbance has been added: the conductors announce at nearly every stop that cops are on the platform should people need assistance. This is in a safe city. Just stop this.

Quick Note on Transit Expansion and Development

I’ve been thinking a lot about where subway extensions can go in New York. One of the appendices we’re likely to include down the line in the Transit Costs Project is a proposal for what New York could do if its construction costs were more reasonable, and this means having to think about plausible extensions. Leaving aside regional rail and systematic investments for now, this may roughly be it:

The full-size image (warning: 52 MB) can be found here.

The costs depicted are about twice as high as what I wrote in 2019 with Nordic costs as the baseline, because nominal Nordic costs have doubled since then, partly due to updating price levels from the early 2010s to the early 2020s, but mostly because of the real cost explosion in the Nordic countries. These costs are about $200 million/km in outlying areas, $300 million/km in Manhattan or across water, somewhat less than $100 million/km above ground or in an open trench, and higher than $300 million/km when reconstruction of existing tunnel complexes is proposed; everything is rounded to the nearest $100 million, which creates some rounding artifacts for short extensions that cancel one another out.

But the precise map is not what I think is the most interesting. The point is to build to the frontier of the cost per rider that is acceptable in American cities today, so by definition the marginal line for inclusion on the map, such as the D extension to Gun Hill Road to meet with the 2 train, is also socioeconomically marginal. What I think is more interesting is how important transit-oriented development is for the prospects of lines beyond the most obvious ones (Second Avenue Subway Phase 2, 125th Street, Utica, Nostrand, IBX, and maybe also the 7 to College Point).

The current land use in New York is largely frozen from the middle of the 20th century; the 1961 zoning law was the watershed. Since then, change has been slow, in contrast with rapid redevelopment in places that have chosen a pro-growth path. If the pace of change stays slow, then fewer lines are viable; if the city instead chooses not to keep anti-developmental neighborhood interests in the loop, then more are.

This, in turn, feeds into growth plans. Nordic and Italian planning bundles the question of where the regional housing growth goes with where the subway goes. (Our other positive case study, Turkey, works differently; the answer to both questions is “everywhere.”) This means that subway service goes to areas where substantial quantities of transit-oriented development will be permitted and built, often in negotiations with NIMBY municipalities that would rather just get the infrastructure without the housing; in Stockholm the scale involved is tens of thousands of units per tranche of Nya Tunnelbanan.

In the case of New York, this affects the shape of the map above more than anything. The 6 extension to Coop City is likely good either way, but the other radial extensions in the Bronx are more questionable and depend on where new housing in the borough will be built. The same is true in Queens: more housing in Northeast Queens may argue even in favor of further lines not depicted on the map, for example extending the 7 even further.

How Sandbagged Costs Become Real

Sandbagging is the practice of making a proposal one does not wish to see enacted look a lot weaker than it is. In infrastructure, this usually takes the form of making the cost look a lot higher than it needs to be, by including extra scope, assuming constraints that are not in fact binding, or just using high-end estimates for costs and low-end estimates for benefits. Unfortunately, once a sandbagged estimate circulates, it becomes real: doing the project cleanly without the extras and without fake constraints becomes politically difficult, especially if the sandbaggers are still in charge.

Examples of sandbagging

I’ve written before about some ways Massachusetts sandbags commuter rail electrification and the North-South Rail Link. In both cases, Governor Charlie Baker and the state’s Department of Transportation are uninterested in commuter rail modernization and therefore ensured the studies in that direction would put their fingers on the scale to arrive at the desired conclusion. As we will see, the electrification sandbag is one example of how sandbagged estimates can become real.

In New York, the best example is of sandbagging alternatives. Disgraced then-governor Andrew Cuomo wanted to build a people mover to LaGuardia Airport in the wrong direction, and to that effect, Port Authority made a study that found ways to sandbag other alignments; here at least there’s a happy ending, in that as soon as Cuomo left office, the process was restarted and the rapid transit options studied the most seriously are the better ones.

Another example I have just seen is in Philadelphia. There have long been calls for extending the subway to the northeast along Roosevelt Boulevard; Pennsylvania DOT has just released a cost estimate of $1-4 billion/mile ($600 million-$2.5 billion/km). The high end would beat both phases of Second Avenue Subway, in an environment that both is objectively easier to tunnel in and has a recent history of building and operating services much less expensively than New York.

How to sandbag public transportation

An obstructionist manager who does not care much for public transit, or doesn’t care about the specific project being proposed, has a number of tools with which they can make costs appear higher and benefits appear lower. These are not hard to bake into an official proposal. These include the following:

  • Invocation of NIMBYs as a reason not to build. The NIMBYs in question can be a complete phantom – perhaps the region in question is supportive of transit expansion, or perhaps there was NIMBYism in recent memory but the NIMBYs have since died or moved away. Or they can be real but far less powerful than the obstructionist says, with a recent history of the state beating NIMBYs in court when it cares.
  • Scope creep. Complex public transportation projects often require additional scope to be viable – for example, regional rail tunnels often require additional spending on surface improvements for the branches that are to use the tunnels. How much extra scope is required is a subtle technical question and there is usually room for creative innovation for how to schedule around bottlenecks (whence the Swiss slogan electronics before concrete). The obstructionist can take a maximalist approach for the scope and just avoid any attempt to optimize, making the costs appear higher.
  • Scope deflection. This is similar to scope creep in that the project gets laden with additional items, but differs from scope creep in that the items are what the obstructionist really wants to build, rather than lazy irrelevances.
  • Excessive contingency. Cost estimates are uncertain and the earlier the design is, the more uncertain they are. Adding 40% contingency is a surefire way to ensure the money will be spent, as is citing a large range of costs as in the above-mentioned case in Philadelphia.

How sandbags become real

Normally, the purpose of a sandbag is to block or delay the entire project; the scope deflection point is an exception to this. And yet, once a sandbagged estimate is announced, it often turns into the real cost. Philadelphia was recently planning subway expansion for not much more than the international cost, but now that numbers comparable to Second Avenue Subway are out there, area advocates should expect them to turn into the real cost, absent a strong counterforce, involving public dismissal and humiliation of people engaged in such tactics.

The reason for this is that cost control doesn’t always occur naturally, unless one is already used to it. It’s very easy to waste money on irrelevant extras, some with real value to another group (“betterments”), some without. Second Avenue Subway has stations that are two to three times as big as they needed to be, without any sandbagging – different requirements just piled up, including mechanical rooms, crew rooms with each department having its own space, and additional crossovers, and nobody said “Wait a minute, this is too much.” The station designs are also not standardized, again without a sandbag, and it’s very easy to promise neighborhood groups bespoke design just to make them feel important, even if the bespoke design isn’t architecturally notable or useful for passengers.

Likewise, if there’s any conflict between different users, for example different utilities and infrastructure providers in a city, then it takes some effort to rein it in and coordinate. The same situation occurs for conflict between different users of the same tracks on mainline rail: it takes some effort to coordinate timetables between local and long-distance rail services. The planning effort required is ultimately orders of magnitude cheaper than the cost of segregating the uses – hence the electronics before concrete maxim – but people who don’t care for coordination can find ways to define a project in a way that makes additional concrete (on mainline rail) or extra work with utilities (in urban subways) seem unavoidable.

Moreover, a betterment, non-standardized design, concrete-instead-of-electronics, or scope deflection occurs in context of other people’s money (OPM). If a light rail project pays for a municipality’s streetscaping, the municipality will not try to value-engineer any of it, resulting in unusually high costs.

In New York, one of the reasons for high accessibility costs on the subway, beyond the usual problems of procurement and utility conflict, is scope deflection. The agency doesn’t care about disabled people, and treats disability law as a nuisance. Thus it sandbags elevator installations by bundling them with other projects that it does care about, like adding more staircases or renewing the station finishes, and charging those projects to the accessibility bucket and telling judges how much it is spending on mandated accessibility.

Political advocacy is unwittingly one of the mechanisms for this cost blowout. Transit advocates tend to value transit more than the average person, by definition, and therefore are okay with pushing for projects at higher costs than are acceptable to most. Once a sandbagged budget is out there, such groups often say that even at the higher estimate the project is a bargain and should go ahead. And once there’s political support, it’s easy to spend money with nothing to show for it.

I Gave Two Talks About Construction Costs Yesterday

We gave two talks about construction costs yesterday, as I said in my invite earlier this week. There are no slides to upload, so I’ll just give brief overviews.

The 11 am talk had with Aaron Gordon as moderator and comprised me, Eric, Elif, and Marco, in front of an audience of about 40, including a few people in official capacity from the MTA or the more reform-oriented sections of politics. It was recorded, and the video has been uploaded via the Marron channel. The four of us went over our backgrounds and what brought us to this issue, and then we talked about what we’d done – we tallied around 200 personal interviews and correspondences and countless academic and gray studies reviewed – and what the conclusions are (see above link for some of them).

Audience questions were markedly friendly, and so were followup conversations Eric had with people at the MTA about this; Eric and I had spent the previous day catastrophizing about what if we’d encounter a hostile audience with questions insisting that no, New York can’t possibly be an order of magnitude more expensive to build subways in than our comparison cases, but none of that happened there.

The political response is also interesting. I’m not going to name names but I’ve found it striking that there’s interest in this from both politicians who ideologically identify with the radical left and the Democratic Socialists of America and ones who ideologically identify with the neoliberal movement (currently rebranding itself as New Liberals, in parallel with the New Democrat Coalition).

In a way, it’s not too surprising. Both groups are motivated by ideology and not by the petty concerns that lead to NIMBYism and to the politics of delay for its own sake. More subtly, while the term neoliberalism evokes a retreat from state methods and toward privatization, in practice the people who use the label today talk about state capacity all the time, they just have a vision of the state that centers efficiency. The sight of a New York that can, on its present capital budget, build 200 km of rail tunnel in 10 years while also completing investments in accessibility and high-capacity signaling is uplifting to such movements, even if those movements may disagree about driverless trains.

This does not mean everyone is on board, unfortunately. I can’t tell what exactly goes on at the MTA; clearly, there are some people there who are unhappy, although I can’t tell who except in the broadest, least certain outline. In politics, I will say that the people I’ve talked to are not nearly as well-known or powerful as Alexandria Ocasio-Cortez or the staff of Pete Buttigieg.

The 8 pm talk was much less formal and was just me in front of a crowd of about 25 that was more advocacy-oriented. It was from the start the secondary event, designed for people who would like to come but couldn’t make it during business hours. I expected 12 people and got 25, with an awkward signup process at the lobby of the building, for which I am grateful to security for being understanding. I managed to possess the AV system in the room with the help of an audience member and share my screen to showcase some more examples and talk more about our report, but there was no recording.

Audience questions covered a variety of topics: the applicability of our work to California High-Speed Rail (I went on a long rant about the problems of early commitment), how the different factors mentioned in the link at the start of this post interact, what the role of utilities is, etc.

A more interesting question, which I didn’t immediately have an answer to, was what advocates can do about it. People don’t vote based on subway construction costs, or at least not directly. I did bring up the political popularity of mani pulite and the anti-corruption reforms in Italy that have helped bring down costs, and, echoing more experienced activists who I’d asked, recommended that people raise the issue with their state legislator, member of City Council, or mayor if they’re in an inner suburb and not the city. In an American context, there is no criminal corruption that we’ve found, unlike in Italy in the 1970s and 80s, but instead of mani pulite, a popular process for making government more efficient is viable. Even people whose entire political career is built on wrecking the ability of the state to do anything talk about how It’s Time to Build or about Getting to Yes.

I want to say I’m optimistic based on what we saw, but not everything has gone as smoothly, and there are people in powerful positions who should not have them – they just didn’t show up this time. So we’ll see; I’ll know much more at the end of the year.

I Gave a Talk About Through-Running

The ReThink NYC online panel earlier today was strange in a lot of ways: in delivery, in tone, in emphasis. Perhaps the full slide deck will be uploaded and I will be able to more easily point this out. For now, look at my slides; they’re a very condensed version of this post, criticizing the Empire State Development report saying that through-running at Penn Station is impossible.

The technical issue is that as you can see, my slides are a Beamer PDF. The version that I delivered was line-by-line, as is the norm for math presentations; you can click through to see what it means and why every presentation I upload on this blog is modified to be slide-by-slide and therefore has “2” in the file name. Everyone else was on PowerPoint or Google Slides, with centralized control; I took control for my portion, which was not designed around having an assistant who I tell “next slide please” periodically, and the system wasn’t as responsive to my clicks as I’d hoped.

The tone issue is that somehow I was the least offensive person on the panel. Moderator Sam Turvey was complaining that the MTA called the panel a private event as a reason not to send anyone to attend; I just stuck to some technical critiques, even with my background of calling for people to be fired here and on Twitter. I’m not sure how that came to be. But I somehow was the most polite person to the decision makers, I think, and that’s always jarring, when within the Transit Costs Project team I’m the least polite and least charitable (why should I be charitable to $2 billion/km subway builders?).

And then there’s the emphasis issue. I was trying to give a 10-minute technical primer about the value of through-running and suggest one way of doing so (in practice, more like 15 minutes – everyone ran over). There are some differences between my concept and ReThink’s that I think are worth going over:

  • On the level of crayon, I think through-running at Penn Station should connect to Grand Central (similar to the old Alternative G from the early 2000s). ReThink prefers pure East River through-running, I can’t tell whether via the existing tunnels or via a new two-track tunnel (called Alternative S in the 2000s, S standing for Sunnyside), which you can see one version of on Tri-State’s generally excellent report on the subject.
  • My conception of commuter rail is a predominantly urban service, using infrastructure that can then also be used for secondarily important suburban service. I wrote the linked blog post after seeing some discussion on Twitter, without realizing what ReThink was planning; next day, they told me about their conception of commuter rail as a system for decentralizing employment to suburban centers.
  • I think much more about non-crayon issues like junctions, high platforms, electrification of tails than do other advocacy organizations. That’s what I mean by electronics before concrete: fix the surface issues before or during construction of tunneled megaprojects.
  • I’m pretty rigidly against expansion of the footprint of Penn Station. It’s unnecessary (see for example this post), and so expensive it should only be done if absolutely critical; it’s fine to make compromises on platform amenities to avoid such expense. ReThink is against the full demolition of the block south of Penn Station but is open to moderate expansion of the footprint, as is Tri-State.
  • I’m openly YIMBY. I think Penn Station is the best place in the United States to put new commercial skyscrapers – the area is very well-served by mass transit, and the commuter trains are underfull by the crowding standards used to determine subway service. I see fully recovered rail ridership where I live and where I last lived and slower but noticeable corona recovery in New York. ReThink… all I’ll say is that they’re not YIMBY.

And none of this was really discussed. I can’t tell if it’s because everyone ran over, or because audience questions had a different focus, or because some of the other panelists were more critical of the plans to redevelop the area around Penn Station than of the technical merits of different paradigms of rail service. In a way, that kind of advocacy space is the wrong space to decide technical matters like Grand Central vs. no Grand Central through-running, but it might be useful to introduce the options and go over some pros and cons.

The Transit Costs Project Conclusion is Out!

Here it is. This is the result of many months and years of work, and a lot of editing, and it should not be viewed as my work but rather as joint work of mine with Eric Goldwyn, Elif Ensari, and Marco Chitti. People should read the report, which talks about how to build in-house capacity and institutional support that does not involve American-style micromanagement and politiciziation.

We’re going to present on this in person at NYU in a day and a half, on Wednesday 10/26, at 11 am (moderated by Aaron Gordon) and again at 8 pm for people who can’t make it during work hours; this is at Marron’s office at 370 Jay Street on the 12th floor, room 1201. (I’m also separately on this panel about through-running, online, 10/25 at 6 pm New York time.)

We’ve managed to decompose much of the cost premium of New York over low-to-medium-cost comparison cases, in the following manner:

Labor

Labor costs are a total of 20-30% in our comparison cases (Turkey, Italy, Sweden). Sweden is the most relevant, as the highest-wage example; Citybanan’s costs were 23% labor. In the Northeastern United States, labor is 40-60% of the cost. Picking 25% vs. 50% as the respective averages, this means that labor costs in the Northeast are three times what they should be, and the difference contributes to a factor of 1.5 cost difference. This includes both blue- and white-collar labor – this isn’t just overstaffing of unionized workers (although that exists too) but also different agencies such as utilities demanding that their own supervisors be in the tunnel during construction. In Boston, the overhead ratio was 40-65% higher on the Green Line Extension than the norm for Bostonian construction.

Soft costs and design

In New York, and as far as we can tell across the Anglosphere, design and management add a hefty additional share. Of note, what counts as soft costs differs by country. For example, insurance is a soft cost in Italy, but in New York it’s bundled into the regular contracts. British cost breakdowns list contingency separately, but American ones do not. Taking just the project management and design contracts – what counts as soft costs for New York contracting – they add 21% on top of the other contracts. The norm in France and Italy is 5-10%. However, 21% is on top of an inflated base: while extra physical construction means a roughly proportionate increase in oversight costs, the extra labor costs do not, and so, relative to a right-size labor cost (that is, overall project cost falling by a factor of 1.5), this is 31%. This contributes a factor of about 1.2.

Procurement

Procurement problems, including lack of competition, poor management of the contractors (called “the [name of agency] factor” where this can be any American transit agency), change order litigation, risk compensation, and contingency, overall double New York’s overall construction costs. Some of this is recent enough to only have been instituted when Andrew Cuomo was governor, like debarment, a heavyhanded attempt to blacklist contractors who run over the estimated cost that leads to higher initial bids for risk compensation. But the privatization of risk goes back earlier and the closedness to working like in the rest of the world goes back much earlier. Moreover, the tendency to privatize risk and alternate between micromanaging contractors and not knowing how to supervise them at all appears pan-US.

As a note of caution, it’s perhaps best to think of procurement and soft costs together as contributing a factor of about 2.5: under different definitions from New York’s, for example those of Britain, some procurement problems like contingency and excessive contractor profit (due to risk compensation – this isn’t a freeroll for the contractors) are folded into the soft cost account.

Overbuilding

Subway stations should be built cut-and-cover, in a box barely longer than the longest train that is expected to run through them. Italian and French examples are maybe 5-10% longer than the train, and Odenplan on Citybanan is 17% (250 m box, 214 m trains). American stations are often oversize: Second Avenue Subway’s two mined stations are on average about 100% longer, and cut-and-cover 96th Street is almost 200% longer. Moreover, designs must be standardized across each project, whereas in the US they are not, to the point that there were two distinct escalator vendors for the three stations of Second Avenue Subway.

This is not seen as nicer passenger spaces – those stations still look pretty crummy compared with the standardized stations of the Nordic countries or Italy or Turkey. It goes without saying that non-standardized escalator placement does not make stations more pleasant. Moreover, the extra space is just used for back offices with full segregation between different functions and work teams that no legacy station has anywhere, or unnecessary crossovers; Odenplan looks much nicer without much superfluous digging. Political insistence on signature stations in the United States leads to waste without any improvement in user experience resulting from it.

This factor also absorbs conflict with utilities, which is seen in decisions to dig too deep and build mined stations, avoiding cut-and-cover even when the costs are more favorable. (Utility relocation costs should be reduced too, but those are second-order in New York.)

In New York, stations are 77% of the hard costs; systems and tunnels are 23%. Cutting station costs by a factor of 3 (or slightly more, counting utility conflict) means cutting overall costs by a factor of a little more than 2. In fact the overall cut should be bigger because there’s some overdesign in the systems as well. The paused, restarted, and budget-overrunning Paris Metro Line 1 extension budget is split as 30% stations, 55% tunnels and systems, but that’s for trains half the size of New York’s, and Länsimetro Phase 1, with trains three quarters the size of New York’s split about evenly between the two. With systems and tunneling made cheaper as well through scale and standardization, the overall cost difference is a factor of 2.5-3.

What does this mean?

It means you should read the report, linked at the very start of this post. But mostly it means the causes of high American (especially New York) costs are institutional, and fixable, without a revolutionary upending of the legal or social system. We can’t tell you how New York can build for the costs of Nuremberg or Turkey, both around $100 million per km, but $200 million per km, slightly higher than Italy and slightly lower than Sweden, is achievable. Moreover, because institutional problems with procurement and soft costs occur throughout (and also conflict with utilities, a bigger issue for smaller projects than subway expansion), the same reforms that should bring down tunneling costs should also bring down the costs of non-tunneling improvements like elevator accessibility and platform edge doors.

S-Bahn Frequency and Job Centralization

Commuter rail systems with high bidirectional frequency succeed in monocentric cities. This can look weird from the perspective of rail advocacy: American rail advocates who call for better off- and reverse-peak frequency argue that it is necessary for reverse-commuters. The present-day American commuter rail model, which centers suburban commuters who work in city center between 9 am and 5 pm, doesn’t work for other workers and for non-work trips, and so advocates for modernization bring up these other trips. And yet, the best examples of modern commuter rail networks with high frequency are in cities with much job centralization within the inner areas and relatively little suburbanization of jobs. What gives?

The ultimate issue here is that S-Bahn-style operations are not exactly about the suburbs or about reverse-commutes. They’re about the following kinds of trips, in roughly descending order of importance:

  • Urban commuter trips to city center
  • Commuter trips to a near-center destination, which may not be right at the one train station of traditional operations
  • Urban non-work trips, of the same kind as subway ridership
  • Middle-class suburban commutes to city center at traditional midcentury work hours, the only market the American commuter rail model serves today
  • Working-class reverse-commutes, not to any visible office site (which would tilt middle-class) but to diffuse retail, care, and service work
  • Suburban work and non-work trips to city center that are not at traditional midcentury hours
  • Middle-class reverse-commutes and cross-city commutes

The best example of a frequent S-Bahn in a monocentric city is Munich. The suburbs of Munich have a strong anti-city political identity, rooted in the pattern in which the suburbs vote CSU and the city votes SPD and Green and, increasingly, in white flight from the diverse city. But the jobs are in the city, so the suburbanites ride the commuter trains there, just as their counterparts in American cities like New York do. The difference is that the same trains are also useful for urban trips.

I don’t know the ridership by segment in Munich, but I do know it in Berlin, as of 2016 (source, p. 6):

Daily ridership on the Berlin U- and S-Bahn by interstation, in thousands; the Ring encircles city center, meeting the radials at Ostkreuz, Gesundbrunnen (north), Westkreuz, Schöneberg (south), and Südkreuz (also south, one stop east of Schöneberg)

Between Ostkreuz and Hauptbahnhof, just west of the meeting point with the North-South Tunnel, the east-west Stadtbahn has 160,000 daily riders. The proper suburbs are mostly less than 10,000 each, and even the more suburban neighborhoods of the city, like Wannsee, don’t contribute much. Overall, the majority of S-Bahn traffic is urban, consisting of trips taken either within the Ring or in the more urban outside-the-Ring areas, like Pankow, Steglitz, and especially Lichtenberg.

The high-frequency model of the S-Bahn works not because there is a mass of people who work in these outer areas. I don’t know the proportion of jobs in the Berlin region that are within the Ring, but I doubt it’s low. For reference, about 35% of Ile-de-France jobs are in a 100 km^2 blob (about the same area enclosed by the Ring) consisting of Paris, La Défense, and the suburbs in between. New York likewise has about 35% of metro area jobs in a 100 km^2 blob chosen to include Manhattan and the major non-Manhattan job centers like Downtown Brooklyn, Long Island City, and the Jersey City waterfront. I imagine Berlin should be the same or even somewhat higher (this proportion is inversely correlated with city population all else being equal) – Berlin is polycentric but all of its centers are on or within the Ring.

Rather, the reason the high-frequency model works is that there is a lot more ridership in urban areas than in low-density suburbs generating strictly unidirectional trips. The main users of the S-Bahn are city residents, or maybe residents of dense inner suburbs in regions with unusually tightly drawn city limits like Paris. If the highest demand is by people whose trip is 20 minutes and not 90 minutes, then the trains must run very frequently, or else they won’t ride. And if the highest demand is by people who are traveling all over the urban core, even if they travel to the central business district more than to other inner neighborhoods, then the trains must have good connections to the subway and buses and many urban stops.

In this schema, the suburbs still get good service because the S-Bahn model, unlike the traditional metro model (but like the newer but more expensive suburban metro), is designed to be fast enough that suburb-to-city trips are still viable. This way, middle-class suburbanites benefit from service whose core constituency is urban, and can enjoy relatively fast, frequent trips to the city and other suburbs all day.

I emphasize middle-class because lower-income jobs are noticeably less centralized. I don’t have any European data on this, but I do have American data. In New York, as of 2015, 57% of $40,000-a-year-and-up workers worked in Manhattan south of 60th Street, but only 37% of under-$40,000-a-year workers did. Moreover, income is probably a better way of conceptualizing this than the sociological concept of class – the better-off blue-collar workers tend to be centralized at industrial sites or they’re owner-operators with their own vans and tools and in either case they have very low mass transit ridership. The sort of non-middle-class workers who high-frequency suburban transit appeals to are more often pink-collar workers cleaning the houses of the middle class, or sometimes blue-collar workers with unpredictable work assignments, who might need cross-city transit.

In contrast, the sort of middle-class ridership that is sociologically the same as the remnants of the midcentury 9-to-5 suburban commuters but reverse-commutes to the suburbs is small. American commuter rail does take it into account: Metro-North has some reverse-peak trains for city-to-White Plains and city-to-Stamford commuters, and Caltrain runs symmetric peak service for the benefit of city-to-Silicon Valley commuters. And yet, even on Caltrain ridership is much more traditional- than reverse-peak; on Metro-North, the traditional peak remains dominant. There just isn’t enough transit-serviceable ridership in a place like Stamford the way it looks today.

So the upshot of commuter rail modernization is that it completely decenters the suburban middle class with its midcentury aspirations of living apart from the city. It does serve this class, because the S-Bahn model is good at serving many kinds of trips at once. But the primary users are urban and inner-suburban. I would even venture and presume that if, on the LIRR, the only options were business-as-usual and ceasing all service to Long Island while providing modern S-Bahn service within city limits, Long Island should be cut off and ridership would increase while operating expenses would plummet. The S-Bahn model does not force such a choice – it can serve the suburbs too, on local trains making some additional city stops at frequencies and fares that are relevant to city residents – but the primacy of city ridership means that the system must be planned from the inside out and not from the outside in.