Our Project
Eric Goldwyn and I have just signed a two-year grant contract for our big construction cost project; we’re working via NYU, him on-site in New York and me remotely in Berlin (at least for now).
We’re hiring!
Our budget includes extensive spending on people who are not Eric or me. For now, we are looking for part-time grad student work to help with data collection. We have an ad circulating internally around NYU, but it’s also open to outsiders:
We are looking to hire up to three students from around the university to help us compile data on infrastructure costs from around the world, especially those of urban subway lines. Specifically, we are looking to understand the drivers of costs—why do public transportation infrastructure projects in one city cost more than in another city? We have already begun collecting this data on projects from around the world and would like to extend these efforts.
We are particularly looking to extend our coverage outside countries where information is readily available in English, such as the English-speaking world or Western Europe. The information we need consists first of all of headline costs of public transportation infrastructure costs, but then of more detailed breakdowns, such as construction techniques, ancillary projects, financing mechanism, the environmental review process, the legal situation, labor size, etc.
The ideal candidate will thus have the following skills:
- Reading fluency in a foreign language used in a country or countries with major ongoing urban rail construction, such as Chinese, Korean, Spanish, Russian, or Arabic.
- Either preexisting familiarity with engineering terms (such as “cut-and-cover”) or the ability to learn them quickly.
- Database software, such as Excel or more advanced statistical analysis software.
- Data analysis or data science techniques useful for small N.
- Self-motivation and independence, for example in finding relevant information to add to the database.
- Good oral and written communication skills.
- Punctuality and promptness – deadlines are written in stone.
Drs. Alon Levy and Eric Goldwyn will supervise all students and work side by side with them to extract the most value from the data. There will also be opportunities to collaborate on writing projects related to the data collection efforts.
If you’re interested, please email both me at alon@pedestrianobservations.com and Eric at elg229@nyu.edu. If you only have half the listed skills, you should probably still apply, especially if these skills include a foreign language that other applicants won’t know.
We will also hire more people as time goes by. We have a budget for a postdoc-level research scholar, so if you’re graduating this year please keep in touch – I’ll post updates when we know our exact timeline, since the grant period isn’t neatly in line with academic years, but it’s a minimum one-year full-time position at NYU, one that for visa purposes is treated as an academic job (thus exempt from the work visa cap).
The goal
The ultimate deliverable in the project is a long report – I’m guessing mid-3 figures number of pages – detailing why American costs are high and what can be done about it. The report should include the following:
- The database of construction costs, broken down to include not just headline costs but also details about construction methods, construction costs by component (stations, tunnels, systems, etc.), rock type, procurement methods, and other relevant variables.
- A highlight of what the important variables are for explaining differences in construction costs, including hopefully a few sentences about the situation in each major city in the database (or if not each then many, on the order of 30+).
- Potentially related databases of construction costs if we get them in sufficient detail and judge them to be comparable, such as for road tunnels, high-speed rail, rail electrification, surface tramways, and urban rail accessibility retrofits.
- A brief how-we-got-here historical overview covering institutional and engineering background to how American infrastructure construction differs from that of most other countries.
- Six (at least) detailed city-level case studies. New York may or may not end up as one of them; Boston almost certainly will, for work we have been doing about the Green Line Extension. The case study selection needs to happen early – this calendar year, and not near its end – and this means we need to identify solid sources who will speak to us about the historical, institutional, legal, and social factors at play.
- A conclusion synthesizing everything to give a coherent recipe for how American (and really English-speaking in general) cities can reduce their construction costs to rest-of-world levels, and ideally even further to match the costs in cheaper countries like Spain, Switzerland, Italy, South Korea, Romania, and the Nordic countries.
- A higher level of synthesis suggesting what a rail network for New York could look like at the lower costs we are proposing.
If you know sources who can talk to us – for example, people at agencies that are building urban rail outside the English-speaking world – then please reach out to us.
The test
I feel good about this – about the recognition, and about the ability to study comparative costs without the stress of looking for temporary gigs. I’m reaching out to various contacts and contacts-of-contacts in a number of cities that are building urban subways, and if anyone has suggestions for who I should talk to, please shoot me an email or mention what you know in comments, as this is a field with a huge base of knowledge.
But at the same time I feel terrified, because I can fail. The project is not going to completely flop, because the database already exists and there’s even more data out there that we already got but just haven’t published. But from getting even an exhaustive database to being able to make actionable recommendations the route is long, and involves case studies and qualitative research and emailing people who have no reason to have heard of me or Eric and often just don’t respond. I think it’s very likely we’ll be able to come up with a useful writeup, and decently likely that this writeup will include a recipe for building subways in New York for $200-300 million per kilometer rather than $2 billion ($100 million/km, as in Madrid and such, is aspirational).
But it’s not guaranteed. We can fail at any number of places: managing the students, finding detailed enough cost breakdowns to identify where the US fails, having broad enough coverage to write multiple case studies, getting enough experts who’ve built cheap subways to talk to us, and so on. The report I mentioned above will get written and published, but whether there is an actionable conclusion remains to be seen, and even if the conclusion is actionable, I don’t know how politically realistic it will be.
Doing this research without really knowing what we’ll find is frustrating this way. The conclusion may well be “the US needs to bust the construction unions.” I don’t think such a conclusion is likely from what we’ve seen so far, but I cannot 100% rule out that it is a significant factor. Or it may be “the US needs to get rid of common law,” which is even less likely to happen; I thought this was an important factor until 2018 or 2019. What is likelier is that a lot of local notables and small-time bureaucrats may need to be cut out of the loop entirely through more streamlined project reviews with fewer veto points, which is politically plausible but requires a governor or a federal government with a modicum of political courage to execute.
What it means for this blog
I’m going to keep posting, at the usual rate of twice a week averaged over time. If I find interesting snippets, I may post them before releasing the report; to some extent I’ve already been doing that with smaller projects. I am still going to think a lot about issues of network design and urbanist politics and will keep writing about those topics.
My Patreon is still around if people want to give me money even though I’m not lacking for it at this point. I’ve been slouching on some of the rewards as I spent months not freelancing (thus, not getting ideas to mine for extra backers-only posts and polls) but finalizing this contract, and now that I have the contract at hand and the project is starting I can go back to it as promised.
In parallel with the costs project, I am going to keep thinking about network design and come up with proposals like this one for New York and New England or this one for Germany; I’ve been thinking about an integrated America-takt or a Europe-takt, at vastly larger scale than any national plan so far, even China’s (which only covers high-speed rail and has no regional rail worth mentioning). Subject to upcoming election results, the scope of what budget is realistic may be narrow enough that I can think in terms of what a specific dollar or euro figure could do.
This of course relates to construction costs – the lower the costs, the more stuff can be built for the same amount of money. Moreover, at high enough level, absolute costs do matter: a Green Deal with €150 billion investment Germany-wide or a Green New Deal with $600 billion US-wide is a big enough proportion of GDP so as to hit real limits to tax capacity and deficit spending, so reductions in unit costs are in 1-to-1 correspondence with building more green infrastructure.
This is why costs ultimately matter. A single subway project may look like a drop in the bucket of the national budget, but when it’s bundled with the costs of an entire public transportation network, and those costs in turn are bundled with those of other major government priorities, the drop becomes a bucket and then a river and then an ocean. The biggest successes in public transportation are plans that look at everything simultaneously and integrate every aspect of operations and infrastructure, and the more cost-efficient these plans are, the further they can reach. There is no way around it.
Pedestrian Observations 
Great news and strong wishes for success!
Also – “The project is not going to completely flop, because the database already exists.” I once heard this as Stigler’s Law (the U of Chicago economist): Never get a grant to do something you haven’t already done. (Then you do the work to justify the next grant.)
Mostly, negotiations over this and smaller grants took long enough that I built a basic database and then expanded. But seriously, it’s just headline costs for now.
As a 40-year research scientist, I can say that is absolutely true. Though there are some projects that don’t fit that rule (the more interesting ones, of course) and which create a lot more stress as a result. Of course it is a game known by all, ie. by grant reviewers and funders (I want to pre-empt anyone imagining there is something fishy about this; there’s not.) It is one of those stumbling blocks early in a career, ie. the huge effort to both establish a record and get just that little bit ahead of one’s public self …
I’m sure Alon is both aware and ahead of this thing. And of course rather than over-promise and under-deliver, do the opposite as far as possible, under-promise and over-deliver.
I really hope this is very successful, and believe it’s a really (even unusually) valuable project!!
Even if you come along something like “common law is the problem”, there’s still use in knowing exactly how it’s the problem (which may allow ways of at least partially mitigating the effects within a common law framework).
Excel isn’t a database and you probably don’t want to do this with spreadsheets. People stuff things into one because they don’t know how to use anything else but you probably don’t want to do this with spreadsheets. There’s someone at NYU that can give you half an afternoon of advice. The people in the civil engineering school who specialize in I.T. And the civil engineering school has their own librarians who can tell you what they have subscriptions to and how to dig out information. And would be delighted to help now and then.
There is likely reams and reams of stuff about the Big Dig which may provide some insight to Boston. Railroads are piles of carefully assembled gravel with ties and rails on top and highways are carefully assembled piles of gravel with paving on top. Highway projects might be informative. And the Class 1s, for years, have been doing all sorts of things to move more double stacks. With costs published. Doing stuff so more coal moves faster to the coal docks or containers from the container ports isn’t all that much different that carefully assembling a pile of gravel for passenger trains.
In other news the new platform for Atlantic Terminal in Brooklyn has opened, in Jamaica. Someplace in the bowels of the MTA there’s information on how much that cost. Last I heard NJransit was still fiddling around restoring a single track to Andover. Someplace there is information on how much that is costing. it’s more or less isolated from upgrading old stuff because it’s essentially new on ROW they already own. Sometime far far in the future you might even be able to take a train to Scranton again.
Good advice from adirondacker (do you know the specific person at NYU they can talk to, or was that just a general suggestion?). Might I also suggest Chicago CTA or LACTMA for a case study of less-outrageously-overpriced U.S. subway construction in recent decades (at least compared to New York)? And are you going to compare overseas cost-blowouts (Stuttgart 21, perhaps) to American ones, to see what factors are common to both versus what might be uniquely American?
It’s also not too soon to think about how to effectively disseminate the report in a way that gives it at least a chance of real-world impact, assuming it ends up with solid actionable conclusions.
Good luck! Looking forward to the results, and interesting glimpses along the way.
I’m just a lowly bookkeeper. There’s somebody floating around who knows a lot about civil engineering and bookkeeping. It may just be bookkeeper bias but throwing the information at an accounting system means you can massage stuff all sorts of ways. Assuming you spent the effort to capture data that can be massaged. With enough detail. Thinking about things in terms of “what account does this post to?” enforces some discipline.
There are total project costs
There are costs before an elected official(s) has a photo-op with a shovel and costs after.
The costs-before have things you have to do and things that are nice to do. The nice to do things vary by country but the must do things are more or less the same. . . .
Computers are real good at adding up 147 things to give you total project costs. They aren’t very good at taking total project costs and allocating them back out when when each project has slightly different percentages. And where things get lumped together one place and you can find numbers for the separate items in others.
A quick google finds NYU has a center for data science. If they are not involved now they should be as this type of thing is at least related to what they are experts in, if it isn’t exactly what they are experts in. https://cds.nyu.edu/ Graduate students looking to do work in this product should probably look at it as a “dual major” (that term doesn’t apply to graduate studies that I know of, but the concept should) with data science. Or you might be able to recruit some students from there to help out with the analysis efforts. The first step of course is getting in contact with them to figure out what they can do.
There is a whole computer science department to look at as well. A quick look doesn’t show this type of thing as a major interest of anyone there but it would be worth checking to see if some professor knows something/someone of use. There are also other universities that might have someone who can lend a hand as part of their research (one advantage of publish or perish – someone might help for a bit just to be listed on the paper)
The people in the graduate program in construction management have a better grip on what they should be looking at and what is available to manipulate it. At NYU. Or Columbia. Or Stevens Institute in Hoboken – short PATH ride and long walk. Or New Jersey Institute of Technology in Newark, two PATH trains and bus or an interesting long walk across Downtown Newark. And that is as far as I bothered to look when it comes to New York. There is probably a bunch of them in Berlin too. Who may have some suggestions on what to use instead of Microsoft Word and Excel. And a few horror stories about data backup and recovery.
Point well taken, though it might or might not be correct.
I pointed this department out because he has data – not really big data, but data. Given that everything is in a spreadsheet the first priority is to get the data in a form that it can be analysed. Spreadsheets do some analysis well, but others poorly and so staying in spreadsheet form will limit the results that are even possible to get. This department will have expertise in getting the data into a form useful for any analysis – including types that spreadsheets are not good at. Hopefully they also include backups of the data so it isn’t lost in a disk crash or other accident. Getting the database form right is important, but hopefully takes the less than a week.
Once the data is safe what to do with it is a good question. Construction management will know to ask some good questions and so are worth talking to. However I would expect Dr Levy already knows most the questions. Data analysis experts may not know the right questions but sometimes some an outsider who knows data can find something so obvious the experts ignore it, which is a reason to consider bringing this department in once more data is entered.
Of course everything needs to be within the limits of the grant/budget.
“data science”.
Ridiculous. But a great way to demonstrate how projects and budgets go off the rails.
How about hiring a Machine Learning Expert to go with your Docker-based Data Solutions System? Don’t forget the AWS heavy-GPU instances and agility to cope with explosive growth at scale.
In real life, three or four people can use email, a shared version-controlled repository somewhere (perhaps on somebody’s desktop machine), and a simple backup plan (three consumer USB disks, stored off-site, ie in somebody else’s house)
Sticking it git on bitbucket or github would be free and would give you some backup. I agree an excel spreadsheet is probably enough or some simple file based storage.
Actual USB3 drives not USB2 drives that are cheap because the fine print on the specs says “USB3 compatible”. They are already wondering what that means. Cloud storage is cheap and having something automatically back up stuff from the one In Europe to one cloud and automatically back up stuff from the one in North America to a different cloud hasn’t occurred to them. Or what to backup. Because they haven’t decided on what data they want.
If they have good access to NYU’s I.T. infrastructure, filing stuff on Z:/ that is NYU servers is probably a hurdle.
Yeah, the hardest part isn’t the software, it’s finding enough experts in various cities willing to talk to us about the exact situation behind how things work in Milan/Madrid/Barcelona/Toronto/Paris/London/Stockholm/Zurich/Tokyo/Seoul/Taipei/Singapore/Tel Aviv/Shanghai/Beijing/Delhi/Mumbai/Bangkok/Tehran/Istanbul/Mexico City/Santiago/São Paulo.
Couldn’t agree more; a little IT would go a long long way on a project like this.
Very exciting project! As a practitioner, I am incredibly interested in circulating and communicating your results among policymakers once your report is out! Let me know if you need any introductions in the public sector planning community and/or any nonprofit/think tank community here in metro NYC!
I’m not sure who your contacts are in Berlin, but you might try getting ahold of the people who are behind the U5 extension which is a very rare example of a newly built subway under a river in sandy ground right through the 800 year old heart of a city with archeological finds and the need to freeze the groundwater and still relatively acceptable construction costs.
Other than that, Martin Randelhoff of the Zukunft Mobilität Blog once did a comparative article on how public transit is financed and might thus know some sources for construction costs.
Best of luck in your project. And let’s hope you not only find out useful stuff but it eventually gets picked up by decision makers…
Fantastic news!
Maybe american rail / HSR costs are so high because they’re just really dumb at this?
on Friday 20/02/28, CAHSR will release it’s draft EIR of the Bakersfield to Burbank section of the project, this article has a preview picture of their preferred alternative
https://cal.streetsblog.org/2020/02/26/ca-high-speed-rail-authority-releasing-draft-plan-to-connect-to-l-a-county/#disqus_thread
Setting aside the totally superior alternative of using Tejon, the problem here is not the mountain crossing but the high desert crossing.
Particularly that LONG diagonal section connecting the CA 58 to the CA 14 freeways, which is mostly aerial, with some tunnels and very little at grade.
Diagonal property takings almost exclusively, and billions worth of property to be purchased up. And diagonally shaped property takings have been the HARDEST property parcels to acquire for CAHSR so far.
But even more why is this aerial monstrosity dumb? Because alongside CA14 highway is over 40 km of double track at grade rail, all the way to CA54 highway. And it has only THREE roads crossing it. versus the more than thirty road crossings of the diagonal route. Build along THAT right of way, where it already works, not where you’re going to require 40 km of aerial superstructure to traverse the desert.
Maybe they are just really really dumb?
If you mean Palmdale-Mojave, then one problem with using the legacy ROW is that north of Rosamond it’s not quite straight enough, so deviations are needed. I don’t particularly mind this cutoff (subject to the general issue of the Tehachapi alignment being stupid). I do mind the aerials – property values in the High Desert are low enough they can build the line at-grade until it hits the mountains.
Then again, this line isn’t going to get built, so whatever, if they want to provide public works for consultants and planners, they’re welcome to.
Hmmm. The half million people in the Antelope Valley don’t have the urge to go to Los Angeles where there are 3.7 million in the city itself but the quarter of a million in Binghamton have peculiar urges that must be satisfied with train to the the 700,000 in Syracuse. Intriguing.
Almost as if the construction costs of high-speed and legacy rail are different.
The leagcy rail, during the heyday of speedy rail travel, took two hours from Binghamton to Syracuse. Making that an hour wouldn’t be cheap and unless you make it an hour a bus is faster.
The Binghamton thing, one way or another, seems to have a certain appeal to everyone – maybe because its kind of in the middle and everyone’s crayon routes are tempted to go through there. Its kind of a topological thing too where the headwaters of the Delaware and Susquehanna rivers almost meet; probably why various RRs went through there. It’s like Rome that way (and in fact I think one of those railways actually went to Rome.
When you think about the routes from NYC to Chicago there was the NYC’s Water level route, the Penn RR’s lets go over the top route and whatever the B&O and C&O did going west from DC. The NY Central route was pretty far north of a straight line route and the others somewhat south, the closest being the Penn RR’s, but also the most formidable, mountain-wise.
Of course, you can see where I’m going with this – plow your HSR through the relatively mole-hill terrain of the Southern Tier and you’ve got a nice reasonably straight run.
To get really cute, you can take one of the various inland routes of the Boston-NYC route that goes through Hartford and plow west from there too, somewhat along I-84, sometimes not. Fork in the road at Danbury or Brewster to go to NYC, and otherwise continue sort of straight across over the Hudson where there a several (mediocre, but not terrible) ways to get to Binghamton (i.e. I-84, Rte 17/I-86(?) or one of those old RR ROWs that go more or less west).
Add in a connector from NYC itself, like over the Tappan Zee bridge and up the Thruway to wherever the lines meet up and you’ve got the super NY/NE to Chicago HSR, ….. AND ……. it also goes to Binghamton.
Please accept my congrats too.
It has appeal to the quarter of million people in metro Binghamton who want to go someplace else. Or people in other places who may have heard it exists. It misses the third of million in Utica, the three quarters of million in Syracuse, the million in Rochester and just over a million in Albany. The 100,000 in Ithaca will be thrilled. And misses the half million in Scranton but they are in Pennsylvania.
You want to funnel all of New England through Danbury that’s not enough capacity.
Regarding legacy linesin Central New York and Northern Pennsyvlania…
Adirondacker, you clearly have not spent much time in Central New York. It is 1 hour 20 minutes by bus from Binghamton to Syracuse, if it’s on time (which it isn’t); in practice you have to leave more than 2 hours. By car it’s a fairly reliable 1:30.
Also, it would in actual fact be cheap to make the route about an hour by train; the rail line is essentially where it needs to be, and about the same curviness as the Interstate (in fact they are basically following the same route), so it’s just a matter of getting the tracks up to par. If absolutely necessary a bypass immediately south of Syracuse could avoid the worst chicane. The only reasons it ran as slow as 2 hours in the legacy days were (a) steam engines suck massively, even worse than diesel, (b) it had a ton of stops, and (c) it was never a priority route. A revival would use engines with half-decent acceleration and have only one intermediate city stop (Cortland), though possibly a couple of stops within Syracuse. Heck, get battery-electric trains if you really need the acceleration.
On the topic of priority routes, everyone in the Southern Tier wants the Binghamton-Scranton-NYC route, which is both cheap and goes in the direction most people want to go. The latest study says that reviving Scranton-NYC is about $289 million, including all the bridge reconstruction (they actually sent divers into the river to really properly assess the bridge) and one intermediate station at the Delaware Water Gap.
Extending from there to Binghamton is purely a matter of negotiating with Norfolk Southern, requiring nothing in the way of track upgrades.
Scranton to the Delaware Water Gap is one of those ludicrously curvy legacy routes from back when trains couldn’t handle grades. But the value of avoiding NYC-area traffic, which extends all the way to the NJ border, means people will tolerate the slowness there.
There is no “cheap” for demand that can be met with a 25 passenger bus once an hour. And that would be stretching credulity. No matter how hard you clap, you aren’t going to change that there aren’t many people in Binghamton. Or Ithaca. And NYSDOT has maps with traffic counts. There is no traffic between Cortland and Ithaca and if there was the Census Bureau wouldn’t have made it, it’s own micropolitian area. Get a grip on reality.
And there is nothing to have train station for at the Water Gap. It’s a park. You can almost spit at downtown East Stroudsburg from the river and that is where rational people want to put a station. And a park-n-ride someplace else, for people who clog traffic on the Delaware River bridge going to work in New Jersey.
The question isn’t whether people in Palmdale deserve a relatively fast electric train to LA, it’s if they deserve a train to Bakersfield
None of them deserve nothing because they aren’t one of the chosen, just build hyperloop between LA and SF and those yokels can just suck up bus fumes.
Palmdale is the planned transfer for trains to Las Vegas, plus those heading to the Bay Area.
Congratulations Alon!
I am extremely suspicious that “be able to blacklist corrupt contractors, and have in-house agency people with enough expertise to do so” is going to be on the list of recommendations. This is something which seems to happen routinely in functional places, and the worst cost overruns seem to come in places where this doesn’t happen and known-corrupt contractors are rehired.
I hope you can end with a sensible recommendation for disqualifying bad contractors, since it’s hard to get this right.
Idk if useful for this project but I see the U5 extension in Frankfurt is missing:
The project: total length 2750 meters (source for technical details https://www.sbev-frankfurt.de/de/das-bauprojekt/planfeststellung/)
– two separated tunnels tubes (837 meters) built with TBM (inner diameter 5.90 meters)
– a cut and cover station box built within the TBM section ( about 170 meters – platforms 110 meters)
– cut and cover tunnel (180 meters)
– a ramp (143 meters)
– a combined road/rail cut and cover tunnel 470 meters (600 meters including ramps) under a park (Europagarten) (12.7 meters wide, two ramps)
– the rest (990 meters) including 3 stations is above ground
Initial cost projection from the Bau und Finazierungsvorlage published by the city’s magistrate in 2013- https://www.stvv.frankfurt.de/download/M_75_2013.pdf
217,3 Mio. € (construction related expenses 168,5 Mio €)
the itemized expenses are a little bit sketchy as the city and VGF share the costs butt the city only lists its own expenses which vary for each item (split across all construction related expenses is 77 / 23)
cost for the underground station – €60 million
tunnel (837 meters) built with NATM – €59 million
ramp+ cut and cover tunnel (180+143 meters) -€18.8 million
tunnel Europagarten (470 meters) – €18.4 million
Updated Bau und Finazierungsvorlage 2015 https://www.stvv.frankfurt.de/download/M_87_2015.pdf
322,1 Mio. € (minus €35 million tax write off due to transfer of the project to a city-owned special purpose company)
-tunnel now built with TBM instate because Hessische Landesamt für Umwelt und Geologie didn’t think NATM was approvable in this area
-underground station now being built in “offener Bauweise” instate of “Deckelbauweise”
Update 2018 https://www.fr.de/frankfurt/gallus-ort904318/frankfurter-europaviertel-kostet-millionen-euro-mehr-11489583.html
€371,3 million (€406 million including tax write off)
Tunnel Europagarten has been completed (at €36 million) but other section have been delayed for two years
extra cost bomb disposal – €10 million (+13 months)
extra cost delayed construction contracts -€10 million
inflation – € 17.1 million
extra cost overhead (consulting, planning, construction supervision) – €34,5 million
Update 2019/2020
after 240 meters the TBM is damaged and has to be partially recovered / restart TBM summer 2020 / cost increase not determinated yet.