Last year, in a provision tucked away in an appropriations bill, Congress instructed the Government Accountability Office to produce a report comparing the construction costs for rail-transit infrastructure projects in the U.S. to those in other advanced countries.
Here are the instructions Congress gave last March:
Not later than 9 months after the enactment of this act, the GAO shall report to the House and Senate Committees on Appropriations regarding the construction costs of transit capital projects in the United States in comparison to other developed G-20 nations, such as South Korea, Japan, Spain, France, Italy and Germany. The GAO shall examine potential cost drivers, including: contracting and procurement, project and station design, routing, regulatory barriers, interagency cooperation and legal systems, but not those which are required by Federal law.
Sixteen months later, the GAO has produced its report, and it doesn’t actually perform the cost comparisons as instructed.
“The complexity of rail-transit construction projects and data limitations, among other things, limits the ability to compare the costs of these projects, according to the stakeholders we interviewed,” the GAO explained. They noted a number of very real problems that make useful international comparisons difficult. For example: Projects are built in different contexts, requiring different construction methods; sometimes they require the purchase of expensive land and sometimes they do not; different authorities report costs by different methods, and there are some cost categories (such as financing costs) that are counted in some reports but not in others.
So, instead of providing cost comparisons, the GAO produced a report on how the Federal Transit Administration is and is not following best practices on cost estimation, and what the FTA could do to help the agencies that build transit-infrastructure projects produce more accurate cost estimates in the future. A worthy subject, but not really what Congress was getting at.
I am sympathetic to the GAO here. The puzzle of why these costs in the U.S. are often triple what they are abroad has been a professional hobbyhorse of mine for a decade, and I recently wrote a feature on New York’s world-beatingly astronomical infrastructure costs for New York. Like the GAO, I have been unable to produce a satisfactory answer to the question of exactly how high our costs are and why. And at the various think tanks and journalistic organizations where I have worked, I have run into the same problems the GAO describes.
The conclusion I have drawn — not just from my inability to adequately answer this question, but the from similar inability of people and institutions with more expertise and more resources at their disposal — is that the transit-infrastructure cost-disease question is unsuited to typical methods of both journalism and policy research. A different (and unfortunately much more expensive) approach will be required to answer the question of exactly how much more expensive our projects are, and why.
The usual way you produce journalism or policy research on a technical question like this is that you hire someone with expertise in journalism or public policy to produce a report. That person calls up relevant experts and asks them to talk, for free, about their areas of expertise. This is one of the great perks of my job: I call up lawyers and other professionals who ordinarily charge hundreds of dollars an hour for their services, and they talk to me for free, whether because they find it interesting, because they have a message they want to get out in the world, or simply because they like seeing their names quoted in print.
This is also approximately what the GAO did to produce their recent report, interviewing nine academic experts, representatives of six construction companies, officials from seven nongovernmental organizations, and one foreign-government official.
But to do truly useful cost comparisons of transit-infrastructure projects, the kinds of questions you need answered — Did this tunnel really need to be built with a form-in-place approach? How much money would precast actually have saved? Do these two financial statements I’m looking at truly follow the same approach for what costs to include? Could this tunnel boring machine have been operated with just nine employees, like they do in France? And on and on and on and on — are so complex and so numerous that you will never obtain the expertise you need for free. Even if the experts are deeply interested in your research question and share your goal of lowering costs, they can’t provide the help you need in their spare time.
Another problem with conducting international comparisons is that many of the people whose expertise you need are not English speakers. When Congress ordered this report, it produced a list of countries it thought might be especially good subjects for a cost comparison. The list includes Spain and Italy, which have reputations for especially cost-effective subway construction. None of the countries that Congress encouraged the GAO to look at are English-speaking.
But when the GAO did some qualitative international comparisons in its report, it drew its examples from Canada, the United Kingdom, and Australia. There is an obvious advantage to such comparisons: The project documents are in English and the participants in the projects speak English. Relevant press coverage is in English. But a problem is that other English-speaking countries — especially Canada — also tend to have very high transit-infrastructure costs compared to world averages. Some researchers have theorized that the common-law system we share with these countries is a driver of those high costs.
Even the best report on cost comparisons I have seen, the Regional Plan Association’s 2018 report on which I relied heavily for my recent article, runs into many of the problems described by the GAO.
The GAO noted that “differences in how sponsors and contractors categorize and report costs, and differences in what costs are included in estimates” is a significant barrier to producing useful cost comparisons. And the RPA, using Federal Transit Administration guidelines for allocating costs, found the MTA had much higher tunneling costs for the East Side Access commuter-rail project than for two recent subway projects in Manhattan. But the MTA objected to the federal methodology. They said tunnel costs for all three projects were about the same when measured using their own standards, which folded certain costs into the tunnels that the FTA would exclude and excluded certain costs the FTA would include.
Which methodology you choose has immense effects on the answer you get: The MTA says the Second Avenue Subway tunnels were nearly four times as expensive as the FTA approach says they were. But if you accept the MTA’s model, then you need to figure out how to reconstruct cost reports from Los Angeles and everywhere else to conform to it if you hope to do comparisons. The RPA wound up deciding it had to stick with the federal approach in order to produce apples-to-apples comparisons, despite the MTA’s objections and the RPA’s somewhat odd finding that the East Side Access tunnels were five times as expensive, per mile, as the Second Avenue Subway tunnels.
So, how can these questions actually be answered? I believe the institution or agency seeking to understand why American costs are so high will need to start by hiring not a policy researcher or a reporter, but an experienced megaproject manager, who would in turn be given a budget to hire construction managers, engineers, accountants, and even translators to essentially conduct forensic audits of past megaprojects. Getting relevant experts to talk for free hasn’t worked; the best solution I see is to bring them on the payroll.
Such an approach would be wildly expensive by policy-research standards; I can easily imagine the budget for a project like this reaching into the eight figures. But a lavishly funded research project into the drivers of high infrastructure costs could easily pay for itself many times over if it actually produced useful answers that made it possible to shave billions of dollars off the costs of megaprojects.
A funny thing about “policy research” is that so much of it is conducted by think tanks seeking to promote an ideological view that predates the research. This does not mean everything that comes out of think tanks is useless; there are a lot of smart people doing good work at these institutions. But typically, the research process is a little backward: There is an answer that people have a strong prior inclination to believe is correct, and then they go out and produce research that points toward that answer.
The infrastructure-cost problem is different. There is a question with a genuinely unknown answer (why are our projects so expensive and what could we do to make them cheaper?) that policymakers across the political spectrum would be very interested in hearing answers to. There could likely be a cross-ideological consensus in favor of many of the effective solutions, if only we could identify them. So what’s needed here is policy research in the truest sense of the word: We need the unknown answers to a question about which we do not have a strong prior.
It would be worth paying to get the answers even if that would be very expensive.