The New York report is still a few months to go (the next ones to release on our list will be Istanbul and Italy), but from interviews, we can get a rough estimate of where the problems are. I want to stress rough and there are still extra sanity checks we need to do, both locally to New York and in comparisons. But a picture of the breakdown of what’s going wrong is starting to emerge.
Procurement issues roughly double costs. This includes everything related to contracting, including the change order regime, the lack of competition between contractors for large projects, the qualify-to-bid system, risk compensation by contractors, and so on.
This is not quite the same as the change order problem seen in California, where Tutor Perini infamously bids low and overcharges on change orders; New York has regulations to prevent that, but the cure is sometimes worse than the disease. Overspecified contracts reduce flexibility and increase costs; normal design-bid-build systems let the building contractors make small modifications based on meter-scale geology and short-term changes in material prices, but New York is inflexible. The design-build solution is more flexible but removes a layer of public oversight and increases private-sector risk, so costs do not go down.
There is a blacklisting system, called disbarment, but it’s too onerous and goes further than just finding repeat offenders like Tutor and blacklisting them, so contractors bid higher to avoid the appearance of cost overrun. Some of this is not seen in Second Avenue Subway – disbarment is a Cuomo policy that was not present then, and must therefore be seen as part of the per-km cost increase from phase 1 to phase 2 even though phase 2 is easier.
Roughly everyone involved in infrastructure construction in New York has told us some variation of the story that New York has complex underground utilities and getting around them is a challenge. This is not exactly what is going on – there’s stuff underground in Paris too – but is correct on the level of agency turf battles. All of the following problems are institutional, coming from our sources:
- Utilities such as Con Ed and DEP don’t maintain up-to-date databases of what’s precisely down there, so every time there’s a project, there’s a surprise, like an electric cable that was laid around the footprint of a car that was parked during construction rather than moving as it was supposed to.
- The knowledge that the utilities have is closely-guarded because who is the MTA to tell them to divulge secret information like where the pipes are?
- The MTA doesn’t run interference for its own contractors, so the contractors are on their own when dealing with the utilities, not just for underground work but also routine things like elevator installation and elevated station repairs.
- The utilities demand that their own workers supervise changes to their infrastructure. There is no state- or city-level coordination of infrastructure construction and maintenance.
- The utilities use subway capital money as OPM for their own wishlists for larger pipelines and upgraded infrastructure.
Finally, there is a fear-itself problem: dealing with utility relocation is so painful that the MTA avoids it even when doing so would be cheaper than digging deeper. I cannot give a precise estimate for the impact of utilities writ large, but a factor of 2, counting the fear-itself factor (which is large), is about reasonable.
Labor productivity in New York is low, and wages are elevated due to the use of local workers (in Stockholm the tunnel workers come from all over Sweden, and wages and benefits are high but somewhat less than in New York). This covers both blue- and white-collar labor: many different agencies, departments, and regulators demand to have their own supervisor down there looking at the work being done.
This is the part where I’m least certain of the impact. The reason is that labor productivity numbers beyond patterns that have been covered amply before are hidden behind layers of subcontracting – there is far less transparency on this in the US than in any other country we have looked at. Moreover, management-labor relations are so strained that planner interviews are not a perfectly reliable source; figuring out how much of the overstaffing problem was about too many white-collar supervisors and not just sandhogs took a while.
The impact of labor is best explained as, US infrastructure breaks down as 40-50% labor, vs. 20% in Turkey; I don’t have a breakdown in Sweden yet, where wages are high but labor efficiency is very high to reduce costs. In isolation, some labor costs scale with everything else. For example, contractor risk is added as an overhead on top of everything and is not part of the breakdown between labor, materials, and equipment. Likewise, delays induced by unexpected utilities require paying overtime during the wait, so labor costs scale to some extent. However, using the wrong material because the contract micromanaged what the contractors could use would be seen as material costs.
Design is clearly an important factor but is even harder to quantify than labor. This does not include the preference for mined stations over cut-and-cover stations in SAS phase 1 – that was partly fear of utility work – and station costs were at a far larger premium than tunneling. It’s better to look at the project work for the 96th Street station, which was as expensive as the mined 72nd and 86th Street stations, and was an unusually large dig (7 blocks long) due to unusual but understandable geological issues. But even so, the station was costlier than it should have been based on dig volume.
Some design problems are defensive. Two notable ones:
- TBMs have to work 24/7 to avoid a cave-in. However, due to either real or predicted neighborhood complaints about noise, trucks did not remove the muck overnight. The overnight storage chamber for the muck cost $20 million. I do not know if this made things better; my personal experience in Berlin is that the worst time for truck noise is 6-9 am and this doesn’t help with sleeping.
- The MTA spent a lot of effort on shoring up a building that was damaged due to ground settlement in construction. The building was subsequently demolished anyway to make room for a taller building for transit-oriented development.
In addition, the speed of tunneling in New York is notably low. I do not know why, nor do I know the net cost impact of it, though at this point I suspect that by itself it’s a secondary issue to defensive design; we are investigating further.
The biggest question of the impact of design questions is to what extent the mined stations for SAS phase 1 were defensive against community impact (as mentioned by some of our sources) as opposed to utilities work. But, again, phase 2 stations are cut-and-cover except 125th Street, and costs are higher as procurement keeps getting worse.
Fire safety is another aspect. Contrary to what too many New Yorkers have told us, the American fire code in NFPA 130 is not unique to the United States. China uses the same code, as does Turkey; Spain uses a very similar code. However, the interpretation of the code leads to visible scope that does not exist elsewhere, including full-length mezzanines and obtrusive ventilation structures.
Procurement and utilities together explain a factor of 4 cost difference. The rest is more hazily estimates; labor gets us to a factor of 5-6 (and it will get more precise in the next few months). Design is the big question mark of whether we can promise a factor of 7 or 15 reduction in costs.
Second Avenue Subway is a factor of 7-8 too expensive relative to the global average, but the global average isn’t necessarily the baseline – much of it includes Anglosphere subway with some of the same problems as New York, and non-Anglo examples near the average, such as Paris, have their own visible problems with costs by comparison with lower-cost examples. The baseline is not the lowest-cost places either – they tend to do unique things like building very short platforms (though Stockholm and some of the extensions in Italy clock at 140 meters to New York’s 180, and Istanbul has lines at 180). But a baseline around $150 million/km is reasonable to aim for. So a factor of 15 reduction relative to SAS phase 2 is exactly what we look for, and a factor of 7 still gets New York to be better than Germany and Japan and close to Paris.
So we’re getting close. Not quite there yet – but something that looks like an actionable recipe, provided reasonably competent political leadership looks achievable.