Mineta Shows How not to Reduce Construction Costs
There’s a short proposal just released by Joshua Schank and Emma Huang at the Mineta Transportation Institute, talking about construction costs. It’s anchored in the experience of Los Angeles more than anything, and is a good example of what not to do. The connections the authors have with LA Metro make me less confident that Los Angeles is serious about reforming in order to be able to build cost-effective infrastructure. There are three points made in the proposal, of which two would make things worse and one would be at best neutral.
What’s in the proposal?
The report links to the various studies done about construction cost comparison, including ours but also Eno’s and Berkeley Law’s. It does so briefly, and then says,
Often overlooked are the inefficiencies and shortcomings inherent in the transportation planning process, which extend far beyond cost, to the quality of the projects, outcomes for the public, and benefits to the region. Rather than propose sweeping, but politically unfeasible, policy changes to address these issues, we focused on more attainable steps that agencies can take right now to improve the process and get to better outcomes.
Mineta’s more attainable steps, in lieu of what we say about project delivery and standardization, are threefold:
- Promise Outcomes, not Projects: instead of promising a concrete piece of infrastructure like a subway, agencies should promise abstract things: “mode-agnostic mobility solutions that ‘carried x riders per day’ or ‘reduced emissions by x%’ or ‘reduced travel time by x minutes,'” which may be “exclusive bus lanes, express bus services, or microtransit.”
- Separate the Planning and Environmental Processes: American agencies today treat the environmental impact statement (EIS) process as the locus of planning, and instead should separate the two out. The planning process should come first; one positive example is the privatized planning of the Sepulveda corridor in Los Angeles.
- Integrate Planning, Construction, and Operations Up Front: different groups handling operations and construction are siloed in the US today, and this should change. There are different ways to do it, but the report spends the longest time on a proposal to offload more responsibilities to public-private partnerships (PPPs or P3s), which should be given long-term contracts for both construction and maintenance.
Point #2 is not really meaningful either way, and the Sepulveda corridor planning is not at all a good example to learn from. The other two points have been to various extents been done before, always with negative consequences.
What’s the problem with the proposal?
Focusing on outcomes rather than projects is called functional procurement in the Nordic countries. The idea is that the state should not be telling contractors what to do, but only set broad goals, like “we need 15,000 passengers per hour capacity.” It’s a recent reform, along many others aiming to increase the role of the private sector in planning.
In truth, public transport is a complex enough system that it’s not enough to say “we need X capacity” or “we need Y speed.” Railways have far too many moving parts, to the point that there’s no alternative to just procuring a system. Too many other factors depend on whether it is a full metro, a tramway, a tram-train (in practice how American light rail systems function), a subway-surface line, or a commuter train. In practice, functional procurement in Sweden hasn’t brought in any change.
In the case of Sepulveda, LA Metro did send some of the high-level P3 proposals to Eric and me. What struck me was that the vendors were proposing completely different technologies. This is irresponsible planning: Sepulveda has a lot of different options for what to do to the north (on the San Fernando Valley side) and south (past LAX), and not all of them work with new technology. For example, one option must be running it through to the Green Line on the 105, but this is only viable if it’s the same light rail technology.
The alternative of microtransit is even worse. It does not work at scale; over a decade of promises by taxi companies that act like tech companies have failed to reduce the cost structure below that of traditional taxis. However, it does open the door for politicians who think they’re being innovative to bring in inefficient non-solutions that are getting a lot of hype. The report brings the example of a New York politician who was taking credit for (small) increases in subway frequency; well, many more politicians spent the 2010s saying that San Francisco’s biggest nonprofit, Uber, was the future of transportation.
The point recommending P3s for their integration of operations and infrastructure is even worse. The privatization of state planning has been an ongoing process in certain parts of the world – it’s universal in the English-speaking world and advancing in the Nordic countries. The outcomes are always the same: infrastructure construction gets worse.
The top-down Swedish state planning of the third quarter of the 20th century built around 104 km of the T-bana, of which 57 are underground, for $3.6 billion in 2022 prices. The present process, negotiated over decades with people who don’t like an obtrusive state and are inspired by British privatization, is building about 19 underground km, for $4.5 billion. This mirrors real increases in absolute costs (not just overruns) throughout Scandinavia. The costs of Sweden in the 1940s-70s were atypically low, but there’s no need for them to have risen so much since then; German costs have been fairly flat over this period, Italian costs rose to the 1970s-80s due to corruption and have since fallen, Spanish costs are still very low.
As we note in the Swedish case report, Nordic planners take it for granted that privatization is good, and ding Germany for not doing as much of it as the UK; of these two countries, one can build and one can’t, and the one that can is unfortunately not the one getting accolades. The United Kingdom, was building subways for the same costs as Germany and Italy in the 1960s and 70s, but its real costs have since grown by a factor of almost four. I can’t say for certain that it’s about Britain’s love affair with P3s, but the fact that the places that use P3s the most are the worst at building infrastructure should make people more critical of the process.
Pedestrian Observations 
There’s something that keeps getting me about these non-comparative studies: they never once ask to check whether what they’re proposing looks like what the success cases do. This way, Schank and Huang are assuming certain things are impossible that happen here routinely, like regular maintenance of publicly-owned, publicly-planned subway systems (one of the arguments they make for DBOM P3s is that public agencies engage in deferred maintenance). Or they propose things that have been tried and are failing before the eyes of anyone who cares to look.
Assuming your boss cares enough, pointing out that someone else is doing your job much better–possibly for less money–is a great way to end up unemployed.
Alon, interesting that today is LA’s transit f**k-ups. Urbanize LA came out with the plan for the Pershing Square redo today. I’ve never seen a deservedly worst reaction to an LA infrastructure project. I think we’ve officially thrown in the towel today and have decided to support 100 new freeways and let the city choke on its own incompetence.
Are you sure you’ve not just misunderstood the first point? Your criticism is about procurement, while the report consistently talks about planning. Authors such a Flyvbjerg (https://doi.org/10.1177/87569728211049046) and Mackie (https://doi.org/10.1016/S0967-070X(98)00004-3) regard premature commitment to some given solution as a major source of cost overruns and project failures.
The Swedish four step-principle is explicitly formulated to ensure that all alternatives have been considered, so when the Swedish bureaucracy puts an itemised T-bana project out for tenders, it did not start as “let’s build a subway!” but “how can we address this issue?” What the Mineta report suggest sure sounds to me like best practice, albeit adapted to the organizational peculiarities of American public sector.
Definitely in the planning process the public sector should be considering more options.
One of the reasons HS2 went bad was that they pre-decided to terminate it at Euston and build it through the Chilterns so the other options were sandbagged.
Uber has been nearing profitability for a while, and posted positive profit in Q2 2023.
When you’re mode-agnostic to the point where microtransit is seriously considered as an alternative to rapid transit, it’s just ignorance. Let’s keep an open mind, sure, but also learn from the real world.
It’s common in the private sector to try to keep up with the “best practices” in the industry, but I wonder how much of this is actual productivity increases vs adopting the latest fad. In general, I guess this process does increase productivity, the proof is in the pudding of persistent economic growth. But transit construction seems particularly bad at this, is there a reason why?
Maybe it’s good practice to put microtransit in your initial report along with a quick calculation showing why it’s the wrong mode. Because microtransit is seriously considered in the popular imagination.
One possible explanation: risk. The prime contractor doesn’t — and shouldn’t — get massive margins on the project. But that means that from their point of view, cost variability is magnified. A pretty small overrun, 10-20%, can eat all profits they expected to make from “the main thing the company was doing for several years”. A bad overrun, 50+%, can bankrupt the company. Thus the company will choose variance minimization or, in other words, programmatically avoiding possible surprises. They will choose e.g. the well-understood mature construction method — and if they have more design input, correspondingly the less controversial route.
Now, there are obvious responses:
– VC-funded startups. First, even startups playing it safe are expected to fail (80-95+%), thus in a sense it doesn’t hurt more to also play some Russian roulette on the side; meanwhile, any single startup failing doesn’t kill the VC. But more importantly, the startups are in an industry where spectacular success is a possibility. In construction, a 95% cost underrun, or equivalently, trying to upgrade, say, Boston-Providence and finding that by doubling the budget you can build HSR all the way to Richmond, is simply not the sort of thing that can happen.
– Itemized and/or cost-plus DBB (or 60%->B). Yes. The whole point is that at level of granularity, the costs are in fact highly predictable, thus the contractor bears very little risk (and at any case, the sums involved are small enough not to bankrupt them). If the grand experiment with a new construction technique or bold route choice goes sideways, the risk is borne by the state; that’s the whole point.
– Unfortunately, when the state tries to privatize itself, risk is among the first things they try to get rid of. You see, they tend to operate under a similar “magnification” problem — “Transit project massively over budget, how can the Mayor sleep at night at this waste of tax dollars?” can be the political career equivalent of company bankruptcy, even if in absolute terms (i.e. international comparison) the project is still doing well, and the risk was absolutely worth taking even in retrospect. Moreover, Alon asserts that politicians completely go overboard in this direction, and treat “Mrs. Grundy says she will be thoroughly displeased if they dig up the street” as a large cost and try to avoid it.
To sum up: risk is like a heap of garbage, nobody wants it, and generally are willing to pay a price to have to put up with less of it, whether by getting someone else to take it or for there to be less of it in the first place. The ideal solution would be an agency with a large budget (for the financial side) and “self-confidence” (for the political side) owning the risk.
You make very good points here, especially about how *everyone* wants to avoid risk, even the ones best positioned to shoulder it. Average costs go down when the state bears the construction risk, but it doesn’t mean politicians like the accountability that comes with it. After all, sometimes they do fuck up and people get justifiably angry.
I think the way out of this is pointing out that privatised planning is a failed model which makes people angry anyway. Brits are angry about the insane price tag of HS2, and the scope reduction that came with the government trying to keep the budget under control.
And if your state capacity atrophies all the way down like in Brazil, people are generally angry all the time at the government for not being able to get anything done. The symptoms are people complaining about corruption and supporting political outsiders.
The millennium dome going over budget didn’t cause Tony Blair to lose the 2001 election.
While you will get bad newspaper headlines I really don’t elections are lost because projects go overbudget.
But then again, there’s Stuttgart 21, which did have political consequences.
Some of the opposition to that is because the project itself is kinda pointless. Stuttgart isn’t really on the way between any major city pairs other than as one of the two equal routes between Frankfurt and Munich.
But then rather than a massive tunnel under Stuttgart why not improve the route via Nuremberg?
So looking again it improves Paris-Munich, Paris-Zargreb, Paris-Vienna etc. But is it the lowest hanging fruit for those routes?
Frankly as Stuttgart-Strasbourg is little faster than Manchester-Cardiff there must be much be lower hanging fruit to speed up.
When I looked at it for Cologne and Brussels, station reconstruction to speed up approaches and remove flat junctions is usually very good value for money. Makes not only long distance service faster and more reliable but also all regional and local service. Very often that’s where you get most time saved per € spent: https://metrovelododo.wordpress.com/2023/08/21/brussels-cologne-in-115/
I guess the issue with S21 is that there were also long approach tunnels to be built as well, and the whole project was mismanaged.
Yeah that all looks sensible stuff.
That said improving the speed of Brussels-Antwerp and Brussels-Cologne would also get more passengers from London and Paris to take those journeys by train.
That would justify more service from Brussels to those places and especially more service from Brussels to London – you’d probably land up with that being hourly. Obviously also being able to run more Brussels S Bahn service would I am sure add value too, so there’s a lot of winning to go round for lots of people.
Well a major part of how you speed up London-Amsterdam is getting rid of the major slow zone in central Brussels. This alone is 5-7 min saved. And on top of that you also increase capacity, so you’re able to run the extra trains as demand rises.
But yeah, there are other slow zones/capacity bottlenecks on the line, like the Mechelen-Antwerp segment. A bypass of Mechelen along the E19 would be nice.
Saying “Stuttgart isn’t really on the way between any major city pairs other than … Frankfurt and Munich” is like saying Wilmington is only between Baltimore and Philadelphia. A couple places you may have heard of like New York and Washington DC lie on the same route beyond Balt and Phila., while past Munich and Frankfurt is Vienna/Central Europe and (checks notes) the *fourth largest* metro area in Europe (Koln/Rhine/Rhur). Then there is a little fact that Frankfurt is the financial capital of Germany and Munich the capital of the wealthiest German state (aside from city states like Hamburg) which means that being in between them alone is very significant.
Finally, there is the *large* fact that Stuttgart’s metro area is (depending on who you ask) either the same size as Frankfurt or the same size/slightly *larger* than Munich. Stuttgart-Frank and Stuttgart-Munich are themselves major city pairs. People outside Germany get blinded by tOktoberfest in Munich and international to Frankfurt and don’t realize that Hamburg has the busiest rail station in Germany or that the Rhur might be larger than Berlin (again, depending on who you ask; ESPON, OECD, etc.) or that Stuttgart exists and Baden-Wittenburg has an economy 60% larger than Hesse (Frankfurt) with the same GDP per capita.
@Onux, Stuttgart is only between Frankfurt and Munich on one valid route between the two. The route via Nuremburg is also equally valid.
The question is whether the 9 billion is worth it to speed up access into Stuttgart for people who live there, plus the trains to Lake Constance, plus any Paris-Munich/Vienna/Zagreb passengers.
I would have thought for €500m-1bn or so you could do a pretty good job of speeding up the existing approaches into Stuttgart.
For the longer distance passengers I would have thought you could get better bang for your buck by spending some money upgrading Stuttgart to Munich and/or Frankfurt-Nuremburg and/or Inglostat-Munich for the Frankfurt-Munich services. The trains only average 80mph/130kmh between Strasbourg and Stuttgart and 67mph/110kmh between Stuttgart and Munich. I would be very surprised if the speed loss was all in the approaches to Stuttgart.
Given how much larger Stuttgart is than Nuremburg, it is the primary route Frank-Munich, especially when you consider that it also serves the Rhine-Neckar, which is by itself a significant fraction of Nuremburg’s size.
Stuttgart 21 isn’t just about the new HBf and its “approaches.” The project includes the Filder Tunnel and new ROW to Wendlingen which together constitute 25km of track with 250kph speed. This is 30% of the total HSR track planned as part of the Stuttgart-Augsburg project.
Actually strictly right now there is only a train every 2 hours from Frankfurt to Munich via Stuttgart, whereas via Nuremburg it is hourly.
“a train every 2 hours … via Stuttgart, whereas via Nuremburg it is hourly.
Given the population along each corridor this is a misapplication of resources. I assume it has to do with the slowness of the line through the Swabian Jura. Let’s see what the schedule looks like after the Stuttgart-Augsburg line improvements (which includes Stuttgart21 works like the Filder Tunnel) are complete.
It reads as an ideological document based on wishful thinking. I mean, they tout the separation of “environmental clearance” and planning as a universal solution while only giving one example of such, and in the same breath saying they don’t yet know whether it’s going to work in that very example. Good luck when somebody sues, I guess.
It’s true that the California planning-by-initiative process is a recipe for dashed expectations, but it’s hard to see how this would be improved by promising aspirational outcomes that may be equally unachievable and (unlike a tangible project) can never be conclusively attributed to the funding in question. The core unexamined premise seems to be that a voting constituency acts like a collective economist whose special skill is holding public agencies accountable for the achievement of metrics: a folk belief beloved of populist politicians and graduates of master’s programs in public policy alike.
To some extent, this is consultants justifying their value to a system that doesn’t really need them. As a consultant, you can go study low-cost projects and learn the things they do differently. But if you do that, you might soon realize that if you came back and actually promoted those things to the folks at home, you’d never get any traction because you’d be telling people to give up other things they valued much more than low costs (like, for example, the ability to Invent It Here). So the only way for you to still get paid is to come up with plausible-sounding tweaks to the process that comport with the ideology of the people who are funding you.
We’ve spent 100 years tilting the playing field in favor of automobile & highway infrastructure. “Mode-agnostic” planning of individual projects is playing on the slanted field instead of fixinig the slant.
Speaking of PPPs, I’m learning that the proposed Kuala Lumpur to Singapore HSR was cancelled very recently because Malaysia was worried about the costs and tried to get Singapore to agree to using mostly Malaysian companies (also there was a change in government). Malaysia is now trying to revive the project under the dreaded PPP format with the stated goal of getting private funding to as close to 100 percent as possible.
https://www.channelnewsasia.com/asia/malaysia-myhsr-kl-singapore-high-speed-rail-project-concept-proposals-3622841
Sounds like such a bad idea.
But it’s startling how much infrastructure is being planned and built in Southeast Asia. I guess the region is just booming. Tons of subways and railways getting put in all over the place. The subways especially are interesting because they’re being built in megacities choked by traffic in hopes of reducing the congestion. May I suggest Alon take a look at these initiatives? I’d be interested in knowing if it actually is possible to turn Manila, for example, into a transit city. This website presents a roundup of projects as of 2022:
https://futuresoutheastasia.com/southeast-asia-railways-report-2022/
*Private funding, construction, and operation for the KL to Singapore HSR, rather.
Any city can be a transit city, if everyone is in walking distance of high quality transit.
In huge dense cities like Manila, this shouldn’t be hard to achieve (just make a subway grid every 1-2km) although Manila has rather bad planning so they are kind of making a mess of their metro expansion.
It probably wouldn’t reduce congestion. It would allow people to get around without being in it.
Yeah, these PPPs fail regularly. Brazil tried to do one to build high-speed rail and couldn’t find any takers, which is why São Paulo-Rio is the second most populous domestic corridor in the world of its length without high-speed rail. Indonesia (which currently has the most populous one) kept trying and eventually had to pay top renminbi to CRH to build its system. California HSR believed it could find private concessionaires to fork over $10 billion in 2008. Texas Central went from “pure private construction” to “can we get federal financing please?”.
Southeast Asia ex-Singapore has some horrifically underbuilt subways. I wrote about this at the beginning of the year.
To build off of a comment in the last thread, I think that #2 could be a very good improvement. American agencies do make EIS the locus of planning, which is bad both conceptually (all projects have some environmental impact, the focus should be on benefit with an aim to balance impact, instead EIS’s of thousands of pages make the focus on the impact) and practically (the explosion of soft costs related to design and consultants).
It would be better if the environmental report comes first. This is similar to how Alon recommends professional planners should design projects completely, then present them to politicians or voters for a decision, with non-engineers not getting a say in the engineering. In this case environmental professionals should get a chance to study the area and specifically define the risks/impacts/mitigation (can’t build here because it is a historic burial ground, cutting trees in this zone requires replanting, this waterway must allow this many m3/sec of flow for floods, etc.). But they should not get to review or set criteria after the project is designed. This way you would not get endless rounds of design, EIS, redesign, EIR, redesign, lawsuit, etc., plus costs from uncertainty. Instead the designers would plan the project with clear environmental criteria, just as the building code is clear at the start, and after design you can fund the project and start building instead of waiting for EIS.
There is the risk that environmental engineers would produce a BANANA (build absolutely nothing anywhere near anyone) report that precludes any practical project. But if they do they become the bad guys making unreasonable requests (“the Crenshaw corridor is protected wetland?!?! But its all paved LA sprawl!) and it is easy for politicians who want to get something done to put them in their place. With the current plan the designers become the bad guys if they try to debate an unreasonable environmental report (“The report said your subway under a paved street requires wetland mitigation, why aren’t you listening to the experts!”) There is a huge bias for specific recommendations to be followed, but not for a priori requests to be honored.
Best of all would be to have environmental criteria set region wide every decade or so, by law, just like zoning regulations and the building code. This way you can be certain the “EIS” isn’t hamstringing any specific project and you can divorce environmental cost and time from all projects; the cost becomes government overhead while project design can start right away with the current “EIS”. Of course California shows that bad zoning codes can cause problems for everyone, but nothing can fix bad government, and at least then you would know where the problem is (just as CA is slowing taking zoning authority away from cities). In contrast EIS production is opaque to the public AND politicians, while even in San Francisco an electrical engineer doesn’t need to commission a report that re-derives Maxwell’s laws before submitting for an electrical permit.