Executive summary
- The English-speaking world has a poor track record when it comes to complex infrastructure like metro rail.
- Ireland has a poor track record of achieving value for money on its infrastructure spending.
- Many European countries can deliver metro rail at approximately a quarter the cost of English-speaking ones.
- Experts agree: the common factor behind the most efficient European builders is their use of in-house expertise.
- MetroLink could realistically cost anywhere between €9.5 billion and €24.4 billion. Hiring the expertise needed to see the project through will cost in the order of €60 million.
- Ireland should learn from the most efficient builders. It should empower Transport Infrastructure Ireland (TII) to hire in-house technical staff, and give them the autonomy they need to deliver.
The English-speaking world has an infrastructure problem
Any traveller could tell you that (Continental) Europeans do infrastructure well. The trains are fast. The airports are big. The train stations are nice.
In the English-speaking world, infrastructure is a mixed bag. Infrastructure feels older in the UK and the US. And there’s no shortage of anecdotal evidence that something is wrong with infrastructure in English-speaking countries – from children’s hospitals to high-speed rail to power plants and metro lines.
When it comes to the efficiency of its infrastructure spending, Ireland is closer to the inefficient UK than the efficient Europeans. The following chart, from a 2017 study by the IMF, showed the perceived quality of advanced countries’ infrastructure (Y-axis) against the amount of money invested in infrastructure (X-axis). The countries on the green line were the ones that built infrastructure most efficiently. Ireland was one of the countries furthest from the green line, ie one of the least efficient builders of infrastructure.
Why is European infrastructure perceived as the highest quality? Why is Ireland less efficient than the best builders in Europe? These are difficult questions because infrastructure is complex, projects are unique, and it’s hard to make apples-to-apples comparisons.
One clue can be found in a new field of research that compares the construction costs per kilometre of metro systems in different places. One study compared 946 tunnelled metro projects in 59 countries on a per kilometre basis. Another study made the same comparison across 284 projects. These studies are helpful because they allow for apples-to-apples comparisons among a large number of big infrastructure projects. They allow us to see what the most efficient, and least efficient, builders have in common.
The research was carried out by NYU’s Marron Institute of Urban Management and, separately, at the Eno Center for Transportation Research. The Transit Costs Project was the result of five years of study by seven full-time researchers. The Eno Centre’s study looked at 284 separate metro projects.
We can draw three simplified lessons from these studies.
The first lesson is that there’s an enormous difference between the most efficient metro-building countries and the least efficient ones. The most efficient countries can build one kilometre of tunnelled metro for between $100m and $200m; in the least efficient countries, a kilometre costs $600-$1,000 million. At the scale of individual projects, it’s even worse. New York’s new 2nd Avenue subway extension cost $4.2 billion per kilometre; Paris’s new Line 11 extension cost $223 million per kilometre.
The second lesson is that there’s no obvious relationship between a country’s wealth and its metro construction cost. One might think that metros would be cheapest in low-income countries, where labour and other input costs are lower. But there’s approximately zero correlation between national income and metro construction cost.
The third lesson is that the most efficient builders are mainly European (they include Italy, Spain, Portugal, Sweden, Turkey and South Korea); and the least efficient builders tend to be in English-speaking common-law countries (the UK, the US, New Zealand, Hong Kong, and Singapore).
The following chart is ominous for Ireland and MetroLink, the new underground metro line linking Swords in North Dublin with the city centre. The chart shows countries’ average construction costs per kilometre. Ireland’s common-law kin are found at the right hand side of the chart.
MetroLink construction is coming soon, and more complex infrastructure projects are coming down the tracks. Before Ireland sets off on this path, it should pause and consider whether it wants to stick with the English-speaking world’s delivery model.
English-speaking countries’ metros cost much more because they don’t employ and empower experts
What accounts for the difference in construction cost per kilometre? The most important difference between models is in their use of in-house expertise. Metro projects in English-speaking countries don’t tend to employ experts directly. Instead of employing experts directly, they hire consultants to manage the project on their behalf.
In European (and East Asian) countries, by contrast, the state employs its own corps of experienced technical staff. This subtle difference in the employment terms of a few dozen project leaders has big consequences for the efficient delivery of the project.
Eric Goldwyn is an NYU Assistant Professor and Program Director of Transportation and Land Use at the Marron Institute, and lead author of the Transit Costs Project. He described the common factor among the most efficient metro builders: “Among the cheapest builders there’s a wide variety of delivery models. Some use on [public-private partnerships], others do not. They rely on consultants to varying degrees. But what they all have in common is a sophisticated state client who can manage the project.”
What’s the big deal? If the project is being run by experienced people, what does it matter about the fine print of their employment contract? Who cares if the engineers take their paychecks from the state or from a private company, hired by the state?
The precise employment status of managers matters because of the question of control. The builder of a metro project can outsource a million tasks to private companies: the digging of tunnels, the manufacture of rolling stock, the financing, the signalling system, the planning application. But one thing it can never outsource is the control of the project. Control of the project must always sit with the owner of the project, the client. That’s a responsibility it can’t shirk.
Why does it matter whether the client is an expert? It matters because the client is the only stakeholder who’s motivated to care about value for money. It’s not possible for the client to pay consultants to deliver value for money on their behalf, because a consultant’s first loyalty is to its own shareholders. A consultant’s job is to make money for their employer, not to save money for their clients.
To be clear, an in-house expert need not be employed directly by the central government. They might work for a department of the national government, a city government, a semi-state company, or a special-purpose vehicle. The point is that their first loyalty should be to the agency whose job it is to deliver the project.
Paul Lewis is the former Policy Director at Eno Center for Transportation and is the lead author of Rail Transit Project Delivery Around the World. Lewis explained why projects in the US tend to be so costly: “The staffing piece is really critical. In the US what happens all too often is they say ‘we don’t have experience’ and they bring in consultants to do it for them.”
Specifically, how do things go wrong when non-experts are in charge? There are three categories of problem: gold-plating, cost creep, and delays.
The first problem is gold-plating. Gold plating is when consultants design nice-to-have but unnecessary features because they’re not incentivised to prioritise value for money. Take, for example, the station designs for New York’s Fifth Avenue Subway extension. On the fifth avenue line, stations are almost twice the length of those used in Europe. Much of the extra space is used to create additional underground storage. If an expert design consultant says a bigger station is safer, better or more efficient, it’s hard for a non-expert to push back.
Dr Marco Chitti, one of the authors of the Transit Costs Project, said of the countries where metro construction is expensive: “They have too many managers. Instead of engineers, they have many people who are good at processes. It’s like someone going to buy a car and having no idea if the car should cost €1,000 or €10,000. It’s fundamental that the government keeps this minimal knowledge.”
The second problem is scope creep. The Project Management Institute defines scope creep as “adding features and functionality without addressing the effects on time, costs, and resources, or without customer approval.”
You get scope creep where there are grey areas – elements of a project that aren’t specifically agreed upon in advance. Complex projects like metros have many grey areas. Because of grey areas, decisions need to be made on the hoof. Without expertise of its own, the state will need to defer to expert consultants on these decisions. Scope creep will be the result unless the client is empowered to hold it back.
“If you depend on consultants to know what you are doing, then you are in real trouble,” said Bent Flyvberg of Oxford University, an expert on megaprojects. “A good balance is where the owner is not outsourcing all the knowledge. A bad balance guarantees a bad outcome.” To be sure, this situation isn’t a moral failing of consultants and contractors. Consultants play an important part in all projects, from the most successful to the least. Problems, where they occur, are less a failing of consultants than of clients who are unable to manage their projects properly.
The third problem is delays. More time on site mechanically increases costs. An efficient project is by definition a quick one.
For a project to be built quickly, there needs to be mechanisms for making decisions quickly. For a complex project like MetroLink, it’s impossible to make every decision in advance. That’s why it’s called a complex project. Decisions will need to be made on the hoof. Which type of transformer should be used? How thick should this wall be?
When a decision is to be made, a non-expert project manager will consult widely and defer to procedures. Decisions-by-procedure take longer than decisions-by-expert. For example, in the proposed MetroLink setup, important decisions must make their way from the Program Director to the Program Board, to the TII Board, to the NTA Board, to the Department of Transport, to the Department of Public Expenditure, NDP Delivery and Reform (DPENDR), and potentially to the Cabinet. In the context of a complex project in which decisions need to be made continually, this process is certain to result in delays.
Countries that build efficiently have a streamlined process that’s built for speed. This doesn’t involve launching straight into a project – on the contrary, the most efficient projects invest lots of time in getting the design right. In their book How Big Things Get Done, Flyvbjerg and Gardner referred to this as “think slow, act fast.”
The time-on-site problem gets compounded when you factor in the incentives it creates for contractors and other partners. It incentivises partners to slow the process down. Imagine midway through the project that a contractor comes up with a proposal for how to save money on a signalling system. The proposal must then be moved through the hierarchy of decision-makers. Reaching a final decision might take many months. The contractor is then entitled to seek extra payment for the extra time spent on site.
The UK’s National Infrastructure Commission was established to study recent infrastructure delivery and recommend areas for improvement. One of its five recommendations was for “Pace, not perfection“. It said: “Government has spent too much time on small-scale changes and repeated consultations. Ambitious goals must be backed up by bold policies and effective implementation. To make the rapid progress required, options must be closed down where the risk of delay is greater than the risk of making a suboptimal decision.”
An ominous early example of MetroLink’s decision-making is the process for the appointment of Dr Sean Sweeney as Project Director. The decision over whether to hire an experienced outsider for the role took three years. Having agreed on the role and picked the candidate, it took six months for final approval to come through. It’s these sorts of processes that explain the 10x cost difference between the most efficient and inefficient metro builders.
A subtle difference in the employment terms of a few dozen project leaders has big consequences for the efficient delivery of the project.
A state can hire as many consultants and contractors as it wants. But ultimately, someone needs to manage the consultants. The question is whether this group should be made up of generalist civil servants or construction experts. Prof Bengt Flvberg, author of How Hard Things Get Done, Emeritus Professor Oxford Said Business School said: “The data do not support the often-seen claim that public ownership is problematic per se and private ownership a main source of efficiency in curbing cost escalation.”
Putting the right people in charge is necessary for efficient megaprojects. But it’s not enough on its own. These people need to be empowered to make decisions. It’s no good having a 30-year metro project veteran in place if they’re expected to go through the same slow processes as any civil servant. The name of the game is making decisions and moving quickly.
The state is exposed to a €13.1bn risk it can mitigate for €52m
MetroLink will be 18.8 kilometre long, of which 62 per cent will be tunnelled. Its central forecast cost is €9.5 billion, or €505 million per kilometre. Is that a realistic estimate of MetroLink’s final cost?
Progress Ireland created a peer set of 36 metro projects against which to judge MetroLink. The projects are metros; more than 50 per cent tunnelled; in the common law jurisdictions of the UK, the US, Hong Kong, Canada, Australia, New Zealand and Singapore; with construction starting in the last 20 years.
The average cost of these 36 projects is €959 million per kilometre in 2023 prices. An average of 91 per cent of the length of each of these projects is tunnelled.
On the one hand, Ireland has never built a metro before, so its first attempt might be expected to be more expensive than its peers. That suggests Metrolink’s final cost per kilometre could be higher than €959 million per kilometre. On the other hand, less of MetroLink (62 per cent) is tunnelled than these peer projects (91 per cent), which suggests its final cost could be lower than peers.
Based on peer countries’ experience, an upper bound range of €700-1300 million per kilometre – €13.1 billion to €24.4 billion in total – seems right. TII’s own upper bound estimate for cost is €24.7 billion. That is, TII says it is 95 per cent certain that costs won’t exceed €24.7 billion.
With MetroLink, the current plan is to entrust delivery to Transport Infrastructure Ireland (TII). TII earned its stripes building the motorway network and the Luas tram lines. But MetroLink will be an order of magnitude more complex than a Luas line. TII will need to learn on the job.
What would it take to turn TII into a competent client for a project of this scale? According to the construction cost researchers, it would need a core team of 10-20 project managers and procurement specialists who have experience delivering a project of this scale and complexity. The Transit Costs Project report recommended: “The procurement process must be based on the principle of public-sector expertise, with an in-house engineering team that is competent enough to do planning and design.”
The market rate for people with experience on projects of this scale is €300-400,000 per year. Annually, including employer contributions, this would sum to €5.8 million. There’s never been a project of this scale in Ireland so they would need to be hired from overseas. Over a nine-year build, the cost would sum to €52 million. This is a significant spend. However, the amount of money at risk in the delivery phase of the project is in the region of €13 billion.
Asking inexperienced project managers to build a megaproject, in the face of expert advice, and international evidence, would seem an unacceptable risk. Paul Lewis, lead researcher of Rail Transit Project Delivery Around the World at the Eno Center, said: “It seems jarring to say ‘we’re going to spend all this public money on these highly-paid public servants’. But the alternative is to have somebody inexperienced managing a megaproject worth ten or twenty billion euros.”
MetroLink will cost somewhere between €9.5 and €22.5 billion. More projects will follow it. A small up-front investment in qualified professionals will pay for itself many times over.
How the best metro builders do it
How do the most efficient countries deliver metro systems for a fraction of the cost of the UK or the US?
Italy builds metro systems for an average of $193 million per kilometre. And it builds lots of metros. It is one of the world’s most efficient metro builders.
There is a degree of variety in how Italy does it because the country has powerful regional governments. Some cities, like Milan and Turin, have built very extensive metro systems quickly and efficiently. Others, like Naples and Rome, were less successful.
The national government’s role is to issue grants to regional transport projects and to issue environmental permits. Regional and city governments have most of the power and responsibility for the delivery of metro projects.
Successful Italian cities got certain things right. Firstly, they hired and empowered experienced people. In Milan’s case, the expertise was built up gradually at the Metropolitano Milanese (MM). Set up in 1955, MM is a state-owned company that sits outside Milan’s civil service and is responsible for metro, sewer and social housing construction. MM has overseen the construction of 104 kilometres of tunnelled metro for an average of just $149 million per kilometre (in 2023 dollars).
Turin’s delivery agency, InfraTo, is a smaller and simpler facsimile of the MM. InfraTo relies to a greater extent than MM on public-private partnerships, consultancy with other cities like Rennes in France, and consultancy with specialised architecture and engineering firms. To oversee construction, in the mid-1990s the city of Turin hired fifteen to twenty professionals to staff InfraTo’s project delivery office. This team consisted mainly of engineers and was recruited from a mixture of the private sector and other public agencies. Despite its small size, InfraTo has overseen the construction of 28.8 kilometre of metro for an average cost of $195m per kilometre (in 2023 dollars).
In Italy, national law specifies three key functions for the delivery of major infrastructure projects. The first is the chief project manager. This official is solely responsible for the project, legally and bureaucratically. The second role is high supervisor. This person is responsible for supervising execution of the contract. They have final say about all major changes in the project that involve cost or scope variations and have the ultimate power to accept or refuse the payment to the contractors based on progress, quality of work and contract adherence. The third function is work supervision. This relates to the “muddy boots” supervision of day to day work. It is labour intensive. This task can either be handled in-house by the delivery agency, or contracted out to a private firm.
The less efficiently built Italian metro programs show how things can go wrong, even when the delivery agency is staffed with experienced professionals. In both cases, the political system shifted their project’s goalposts, which is poison for efficient delivery. Naples over-designed their (admittedly gorgeous) stations. Both cities used a form of contract that invited corruption.
South Korea boosted its internal workforce abilities after the first phase of subway construction in Seoul during the 1970s. Back then, the city got technical help from Japan. Afterwards, in-house staff who learned from the Japanese experts handled the construction. Now, most preliminary project planning is done in-house, and external contractors are hired for the building. Regions looking to build new in-house skills should reach out to countries with successful project delivery to learn from their expertise.
The city of Istanbul is one of the most active and efficient metro builders in the world. Since 1990 Istanbul has built 15 separate metro lines, totalling 408 kilometres at an average cost of just $156 million (in 2023 dollars). Istanbul’s metro lines have been built by the Rail Systems Department of Istanbul’s city government, called the Istanbul Metropolitan Municipality (IMM). The IMM employs a total of 10-15 full-time staff to oversee each project.
Stockholm’s metro system (26 kilometres long, $265 per kilometre in 2023 dollars) was delivered by Trafikverket, the Swedish transport administration. The quality of in-house designs under the civil service system is high. Trafikverket has a generations-long tradition of apolitical engineering.
Make TII a commercial state body
There’s a lot of variety among the successful metro projects. But they have a few things in common. There are high-level principles.
The first is that they take time to get the design stage right.
The second is that they assign day-to-day control over the project to experienced people.
The third is that the delivery organisation has an engineering culture.
The fourth is that the experts are empowered to make decisions.
The fifth is that their political system doesn’t shift the goalposts mid-project.
In Ireland, we’re not a million miles off target. We have taken time over the design stage of MetroLink. The project is being managed by TII, which has an engineering culture and a track record of delivering transport infrastructure. The political system has backed the project and shown few signs so far of meddling.
Ireland falls down in two ways. First, TII doesn’t have enough people with experience delivering projects of this scale. The hiring of Dr Sean Sweeney from New Zealand is a good start. But he needs to be joined by perhaps fifteen other experienced project managers, procurement experts and engineers.
The second way Ireland falls down is the processes used to make decisions. As a state agency under the Department of Transport, TII is limited in many ways. It is limited in what it can pay specialist staff. It’s limited in its total headcount. And it’s limited in how it can make decisions. As we’ve seen, important decisions must make their way from the Program Director to the Program Board, to the TII Board, to the NTA Board, to the Department of Transport, to DPENDR, and potentially to the Cabinet.
Ireland already has a regime that’s designed to solve these kinds of problems. Ireland’s commercial state bodies are government-controlled entities that sit outside the normal channels of the civil service. Ireland has commercial state bodies because it recognises that a civil service-controlled forestry service, broadcaster or water utility simply wouldn’t work. The level of control exercised by the civil service would not work for these specialised entities.
The best way to empower TII so that it can deliver MetroLink is to designate TII as a commercial state body. At the stroke of a pen, this designation would give TII much of what it needs to deliver MetroLink and any further complex infrastructure projects.
As a commercial state body, TII would be free to hire and act more flexibly. It could bring in project managers, designers and procurement experts and empower them to make decisions. TII would be entitled to pay the median of the market rate for specialised roles. It would have the autonomy to make decisions and organise itself in the best way to deliver a complex project (as opposed to the best way to comply with oversight procedures).
Commercial state bodies have greater freedom of action than state bodies. The latitude afforded to commercial state bodies comes in part from their governance arrangements and in part from the fact that they have independent, commercial funding.
TII would not, at first, have a source of commercial funding like Bord na Móna or Coillte. It would be dependent on the exchequer. This would be unusual but not totally unprecedented; other commercial state bodies such as EirGrid and RTÉ depend in large part on direct funding. Furthermore, there’s a path to TII developing its own funding stream in the form of property development. Rail companies are natural property developers because the property value their infrastructure creates often greatly exceeds the construction cost of the infrastructure itself. Many templates exist, from Copenhagen to Hong Kong, for how rail companies can fund themself this way. But regardless of whether or not TII ever develops its own revenue streams, it’s worth designating it as a commercial state body so that it has the autonomy to deliver MetroLink and other complex projects.
An oversight regime that’s fit for complex projects
Empowering TII as a commercial state body would be good for TII. But how would the state ensure TII was making good use of the extra freedom? What is the right relationship between the delivery agency and DPENDR?
With too little autonomy, a delivery agency like TII won’t have the skills, autonomy or risk appetite to deliver complex projects. With too much autonomy, a delivery agency could ‘go rogue’, demanding more and more resources from the state without oversight.
Complex infrastructure projects are different to ordinary projects, so they call for a different approach to oversight. In order to achieve value for money, the delivery agency needs autonomy to hire, make decisions and bear risk. As we’ve seen, autonomy is incompatible with traditional bureaucratic oversight mechanisms. So the treasury needs other ways to oversee the agency and ensure long-run value for money.
The following mechanisms are proposed for keeping delivery agencies accountable while giving them enough freedom to deliver value for money.
The next step is that TII be given a simple mandate to deliver value for money. Value for money should not be defined in terms of overspending vs budget but in final cost relative to objective benchmarks.
To ensure accountability, an agency’s board should be appointed by the Minister. The board’s job is to hold the agency to its mission.
The agency’s governance matters a great deal. The board should be recruited only from industry experts with relevant experience.
To recruit a high-quality board, board fees should be increased. Board members of commercial state boards currently receive €12,600 per year. To give a sense of what the market pays – a survey of 1,500 US companies found that boards at companies of comparable size (with revenue of more than €500 million) paid directors an average of €59,000 per year. Shrinking the gap, at least somewhat, would cost the state a couple of hundred thousand euros per year. They would be overseeing an organisation delivering a €9.5-22.5 billion project of national importance.
An agency’s value for money key performance indicators should be clear and measurable – e.g. cost per tunnelled kilometre. KPIs should be set by a panel of international experts, by reference to international experience.
The agency should be subject to ex-post audits of its performance, at the conclusion of all projects. Italy’s Corte de Conti is the government’s spending watchdog. It runs audits on infrastructure projects as follows: “The ex-post audit is essentially a performance or value-for-money audit, which is based on the principle of ‘collaboration’ (not punishment) with the government department being audited. This kind of audit is designed to underpin the political scrutiny of the audited government department and to propose corrective measures with the aim of ensuring a regular and cost-effective management.”
An agency should be dependent on DPENDR’s approval of its budget on a five-year envelope. A multi-year envelope is essential for achieving value for money on complex projects. In its absence, the pressure will be on TII to drain its budget every year. Giving TII a multi-year funding envelope gives it the ability to defer spending decisions when the conditions are unfavourable. A recent study by the UK’s National Audit Office found that the imposition of annualised spending controls in the High Speed 2 rail project increased costs by £2-3 billion.
Conclusion
In 1990, Ireland’s great challenge was solvency. It was struggling to manage the national debt. The government resolved to create a new debt management agency, and staff it with specialists.
In the Dáil at the time, Ivor Callely said: “Specialised financial personnel have proven very cost-effective and create massive savings for the private sector where they are working, and we should avail of that in the public sector. We should not be the losers. We must attract the right personnel by offering the right package. [The NTMA is] being set up to ensure that we can, without public service restrictions, recruit and attract the expertise to work under the Minister for Finance but with increased flexibility.”
It turned out that the Irish state really did need an in-house financial services arm. The NTMA did a great job of managing the national debt, and it was rewarded with further responsibilities. Now the NTMA manages Ireland’s national debt, its sovereign wealth fund, its development finance, and provides commercial advice to state bodies.
In 2024, Ireland’s great challenge is to build infrastructure fit for its flourishing society. It should do what has worked before, in this country and across Europe. It should give TII the autonomy it needs to deliver.
