Assessing the Cost of High Speed Rail
As with any major infrastructure project, the question of cost is an important one. And so the San Jose Mercury News on Sunday examined the issue of HSR costs in a front-page article. The article, by Mike Rosenberg and Gary Richards, has some good points about HSR, but overall seems to repeat the common flaw of not assessing HSR costs in context. To properly understand HSR costs, we have to not only look at the project itself, but the cost of doing nothing – which is not zero – and the cost of expanding freeways and airports to meet the travel demand HSR is to meet.
Let’s take a look at the article.
For California, the lure of its new ride — a bullet train system capable of whisking passengers between the Bay Area and Los Angeles — has proved so enticing that the state jumped at the deal, even though it has only a quarter of the money needed.
That’s leading some critics to ask whether the state’s largest project ever could also prove to be its most financially disastrous.
California voters jumped not at a “deal” but at an opportunity to provide for fast, sustainable passenger rail to connect their major cities, in a year when the cost of oil dependence was brought home to voters in a powerful way. HSR shouldn’t be framed as a business proposition, but as a lasting piece of infrastructure that will provide economic growth and meet the state’s travel needs for decades to come.
I don’t fault Rosenberg and Richards alone for approaching HSR as if it were a business proposition. They’re merely reflecting the reality that in the early 21st century, American policymakers no longer think of infrastructure as a public good that should be built because it serves a need, but as an enterprise that ought to fund itself. Still, as we’ll see, even under the latter view, HSR makes sense for California.
It has the potential to create jobs while offering a cheaper, greener and faster form of travel. It could also be another nail in the state’s financial coffin.
As we know, critics have made the latter point quite often. But it’s not a point that makes sense. There’s no possible way that HSR would be “another nail in the state’s financial coffin.” The annual debt service that the $10 billion bond adds to the state budget is about $600 million. Our current budget deficit is about $20 billion. As we know, HSR will generate significant economic benefit to the state, including increased tax revenues. Will it be enough to offset the cost of the debt service? Maybe, maybe not. But it’s hard to see how the HSR bonds are at all risky to the state budget.
Rosenberg and Richards deserve credit for properly explaining why the HSR project’s cost estimate rose in 2009:
The Rail Authority’s estimate has risen mostly because when the measure went on the ballot, it didn’t account for inflation, which is expected to total 19 percent over five years of construction.
Keep in mind that’s merely an estimate. In fact, inflation right now is extremely low, and many rail projects have had initial bids come in well under budget as labor and materials costs have declined during the recession. While there’s an epic debate among economists about whether we’re going to see deflation or inflation over the next several years, the overall forces appear to favor at best cost stability, if not outright deflation, as high unemployment and low consumer spending will combine to make inflation unlikely.
The central piece of their article is an assessment of the project’s cost estimate, currently pegged at $42.6 billion:
A Bay Area News Group analysis of high-speed systems around the globe suggests that the project could cost less than the current estimate, as little as $38 billion. But it is most likely to cost more — up to $73 billion, even if built on time.
That analysis is based on the per-mile capital costs of high-speed rail systems built in Europe and Asia in the past decade, as outlined in a World Bank report released last month. The report said costs varied widely depending on terrain, the complexity of engineering work required, how many rail cars were needed and the extent to which routes passed through urban areas.
Unfortunately, the article does not explain the reasoning behind the figure of $73 billion. As such, it has to be considered a mere assertion, a figure that we should not take seriously until we see some explanatory detail.
The World Bank report that was cited in the article estimated that costs for HSR outside China “range from USD 35-70 million/km.” Considering that San Francisco to Anaheim is 748km according to the California High Speed Rail Authority, that produces a low end estimate of $26 billion and a high end estimate of $52 billion. I’m still not quite sure where the reporters’ $73 billion estimate comes from, so for now I don’t think we should treat it seriously, at least not without further explanation. (And it’s worth noting that the World Bank’s report doesn’t explain the cost range they provided either.)
The question of possible cost increases is important, though it has to be considered in context. I’ve always said that it’s possible, maybe even likely that the final price tag will be higher than $42.6 billion. I don’t think it’ll be very much higher, but it might well be. That wouldn’t represent a “boondoggle” because we still have to ask whether HSR is a better deal than the alternatives. As we’ll see, I firmly believe the answer is “yes.”
One of the primary reasons why costs rise is that the design changes thanks to political pressure. Here’s an example from the article:
Costs along some parts of the rail line zoomed up earlier this year, including one five-mile stretch on the Peninsula where estimates soared $135 million. The reason? Engineers had estimated the cost of building two tracks there when they actually needed four.
Of course, technically engineers originally thought they would indeed be building just two HSR-specific tracks. Then it was decided to instead build four tracks that would be shared with Caltrain. That’s a fundamental shift in HSR design. It’s also the right move, and well worth the extra $135 million (which isn’t a “soaring” estimate when you’re looking at a $42.6 billion project).
There is a growing tendency, however, to see megaproject cost increases as somehow being inevitable, as if they were a force of nature. And sure enough, as with any discussion of cost increases, the article trots out good old Bent Flyvberg to make that very point:
A research team led by Oxford University professor Bent Flyvbjerg has studied the estimates and final price tags of 258 megaprojects across 20 countries, including bridges, rail systems, tunnels and freeways.
What they found was astonishing: Nine out of every 10 projects finished over budget, and the average urban rail system ended up costing 45 percent more than projected.
Flyvbjerg says projects in California haven’t bucked the trend. Taxpayer-funded initiatives such as the Bay Bridge rebuild, BART’s extension to San Francisco International Airport and the Los Angeles Metro rail system all failed to meet estimates.
Of course, this excludes those projects that have met estimates, such as the Metro Gold Line Eastside Extension or the Seattle Central Link project – the latter of which was cut by 1/3 in 2001 when costs spiraled out of control.
As to the Bay Bridge, that project shows that a combination of design changes and unexpected inflation pressures can drive project costs higher (as can happen on any project, public or private). Mayors Willie Brown and Jerry Brown both demanded and won a “signature span” for the East Span replacement project, pushing back the timeline and driving up the cost. By the time bids were solicited in 2004, the global economic boom, especially in China, drove the cost of steel through the roof. Had the original plan proposed by Governor Pete Wilson been adhered to, the cost of the design change could have been avoided, but the cost of the steel price inflation might not have.
Flyvberg’s theory is familiar to many of us by now:
The researchers noticed two problems over and again: “optimism bias,” the tendency to focus on an idea’s potential while downplaying pitfalls; and “strategic misrepresentation,” which means officials lowballing costs to make a project look more attractive.
This is because projects aren’t proposed by accountants and engineers, but by politicians, who operate in an environment where taxes are seen as verboten and every cost is closely scrutinized, unless it’s for something that’s seen as politically mainstream like wars or freeways, in which case the sky’s the limit. In other words, the problem Flyvberg identifies isn’t with megaprojects, but with politics. And as I noted above, it doesn’t necessarily mean it is impossible to bring in a project on-budget.
Further, it’s interesting to see what is and isn’t considered a boondoggle. Here’s Wikipedia on the costs of the Interstate system:
The initial cost estimate for the system was $25 billion over 12 years; it ended up costing $114 billion (adjusted for inflation, $425 billion in 2006 dollars) and taking 35 years to complete.
That’s a rather stunning boondoggle – costs were over 4 times what was expected and it took almost three times as long to complete as was planned. But how many people, aside from some of us hardcore transit bloggers, would call the Interstates a boondoggle?
The point is that Americans have come to believe that Interstates were necessary and the costs justified. That was a political choice, just as were the numerous political choices in cities and states across the country that resulted in the higher costs and longer construction times.
As the article acknowledged, it is plausible that the HSR project could be brought in for the projected cost. But it is possible that political choices will be made to increase the cost. For example, the PCC might well prevail in their desire to have the tracks tunneled. That would represent a significant cost increase, but the political leadership could decide that’s justified. The point is that a choice would be made to increase the HSR project cost. It’s not inevitable, and a choice could be made that a tunnel won’t happen. And a choice could also be made that the Peninsula will themselves fund a tunnel, or that a federal earmark will be won, or that the Chinese will pay for it.
If those are the desired choices, than a higher project cost shouldn’t be seen as some example of a flawed project, but of a belief that it is worth spending more money. Too often, it’s argued that keeping project costs low is the overriding imperative for any project, and in a democratic political process, that’s not always the case. People might decide something else is more important.
These points notwithstanding, the current practice of the HSR project would seem to undermine the notion that Flyvberg’s theory applies to it. The CHSRA did not hide the fact that accounting for projected inflation would add nearly $10 billion to the project cost. They made this public as soon as they learned of it. And the Authority has been up front about the financial consequences of various design alternatives. They’ve been trying to keep the line on project costs, rejecting a tunnel for much of the Peninsula precisely because of the costs, to the outrage of some on the Peninsula.
As we know, this is an academic discussion if the CHSRA cannot fund the project as currently budgeted. The State Auditor noticed that not all of the project cost has been accounted for and thought it was a scandal, despite the fact that most infrastructure projects have to finish their planning before they get funded. The Mercury News article explored this:
The authority has $9 billion in state bond money and a $2.25 billion federal stimulus grant committed. It says the remaining three-fourths of the construction costs will come from government and private investors.
The state is banking on at least $15 billion more from Washington, D.C., by 2016 and has applied for up to $1 billion from this year’s federal budget. Even if it gets the full amount this year, the authority would need at least $2.3 billion each subsequent year from Uncle Sam, a plan the state’s nonpartisan legislative analyst recently characterized as “highly uncertain.”
The U.S. government’s entire high-speed rail budget this year was $2.3 billion, and that was its biggest such layout ever; the feds have tentatively sliced next year’s amount to $1.4 billion. California has the largest and most advanced project in the nation but must compete for the grants with 13 high-speed rail corridors across 31 states.
Rail officials argue that the competition actually increases the odds Congress will keep money flowing across the country.
“We’re not going it alone in D.C.,” said Barker of the Rail Authority. “It’s something we’re confident the federal government wants to do.”
It would have been useful had the article included mention of the $50 billion for HSR that is planned to be included in the upcoming Transportation Bill reauthorization, which would have suggested that the plans for federal funding aren’t far-fetched at all.
Of course, if no federal funds materialize, then the whole system doesn’t get built. That’s why the Prop 1A bond and the federal HSR stimulus require whatever money we do spend in the interim to have “independent utility” so that if we never do finish the project, the infrastructure that does get built can actually be used.
The authority plans to ask local cities, transportation agencies and developers to foot another $4 billion to $5 billion. The final $10 billion to $12 billion would come from foreign countries, companies and other investors. Specific plans for those two pots of funding have yet to be established.
San Jose Mayor Chuck Reed, for one, thinks the Rail Authority is dreaming. “I don’t think there is any hope local governments can come up with that kind of money,” he said. “I’m looking at my city’s own finances and resources, and we don’t have money to spare.”
One might ask the city of San José, then, whether they will insist on a “signature span” for the HSR viaduct just south of Diridon Station, given this concern. And in any case, the CHSRA does not appear to be expecting cities to spend money out of their general fund, but to find ways to leverage revenues from things like TOD and redevelopment districts.
The article then examines the question of operations, and here they would definitely have benefited from a comparison with other countries’ HSR experience, rather than relying on Flyvberg’s study:
Still another financial challenge for the authority is operating the line once it’s built. That, too, already looks harder than initial indications.
The high-speed rail line’s revenues depend on two factors: how many people ride it, and how much they pay.
The state predicts the line will carry 41 million passengers annually by 2035. UC Berkeley experts hired to validate those ridership numbers recently reported them to be unreliable.
And in the dozens of urban rail projects studied by Flyvbjerg’s team, the average train system produced half the number of riders that planners expected.
As we know, the UC Berkeley ITS report did not deem the ridership numbers to be unreliable. They instead said they disagreed with some of the methodological choices used by Cambridge Systematics in the HSR ridership study, and concluded that the ridership numbers could be correct – or they could not be.
Further, we have plenty of evidence from around the world that HSR projects – which is what the California HSR project ought to be compared to, not the nebulous “average train system” Flyvberg describes – that HSR is a clear success with riders. The Taiwan HSR system, which was plagued with financial problems relating to construction, has had dramatic success at gaining riders and recently began generating a profit. The Acela has over 50% of the travel market on the US Northeast Corridor. The Madrid-Barcelona AVE high speed train has been a dramatic success in just two years.
Of course, there are examples from the US of non-HSR systems that have had ridership success, such as Phoenix’s light rail project, which has exceeded expectations and the Seattle light rail project is on track to meet theirs – and both are projects that opened in the last two years.
The article also took a shot at the proposed higher fare:
Meanwhile, the authority has nearly doubled the estimated price of traveling from San Francisco to Los Angeles to $105, as measured in today’s dollars. The original number was based on the notion that a bullet train ticket would be half the price of a comparable air ticket; the authority changed that to 83 percent in hopes of generating more profit.
But the Rail Authority concedes pricier tickets will cost them millions of riders.
Two things here: The CHSRA has not “changed” the price of a ticket. They instead offered a second scenario where prices might be higher than originally planned. No final decisions have been made yet.
And that leads to the second point: even if pricier tickets will cost them millions of riders, the system is expected to make even more money, over $2 billion a year in free cash flow within a few years of operation. The whole point of the second scenario, with ticket prices set at 85% of airline prices is to show how the system could be priced to maximize profitability instead of maximizing ridership. Show me an interstate or airport that creates over $2 billion a year in free cash flow and we can talk about the comparable profitability of road or air versus high speed rail.
The article closes with a litany of examples of HSR projects that didn’t cover their construction costs, and a welcome acknowledgement that “most lines cover their operational costs.” As we know, the HSR project isn’t necessarily intended to cover its own construction cost – that goes back to the point I made at the outset about infrastructure being a public good, not necessarily something that pays back its construction cost.
But even the examples they give all had specific reasons for their problems:
When Italy first planned to run high-speed trains between Rome and Florence in 1977, officials estimated 60 percent of the $3 billion project would come from the private sector. But costs more than doubled and no private investor participated, leaving Italy to issue bonds to shoulder the entire burden.
Of course, there was a major global recession in the late 1970s and early 1980s, along with massive inflation, both of which explained that outcome. And Italy solved the problem, floating bonds to pay for the cost – another example of a political choice that HSR was worth paying for.
After tracks were extended to Milan in 1991, though, private companies enticed by ridership figures joined in and helped raise more than half the cash needed for extensions to other cities.
Meanwhile, France saw high-speed rail ridership soar from 12.5 million in 1980 to nearly 23 million in 1992; it’s now a moneymaker.
Rosenberg and Richards deserve credit for including these examples; too many other journalists would not have offered this evidence that proves HSR can attract riders and private investors.
But in Japan, where government funded the entire system, there were problems when some cities demanded more stations be built. The result was a $200 billion debt in 1987. To spare taxpayers that crushing burden, the government had to sell part of the system to private groups.
Here again, we see that higher costs were driven by political demands, not by any inherent flaw in the project. Further, Japan had a national debt crisis far beyond HSR by 1990, which it is still dealing with 20 years later. Privatizing the HSR system is one of the few success stories in Japan in that long era of economic difficulty.
Still, the overall point is that if HSR costs do rise, it will likely be because a political choice was made that they should go up because of a belief that what was being bought was worth the extra price. If that’s the decision that Californians come to, then that’s the choice they make, and it shouldn’t be seen as flawed – nor should it be seen as flawed if Californians make another choice, to hold down project costs and resist calls for more expensive HSR design alternatives.
More importantly, we must always consider the cost of doing nothing. It isn’t zero. The cost of widening Highway 99 alone in the San Joaquin Valley has been pegged at $25 billion – but HSR could help serve those transportation needs and those of other parts of the state for not that much more. The CHSRA has estimated the cost of expanding freeways and airports to meet the demand that HSR can handle to be between $80 and $150 billion – which makes HSR a savings even under the worst-case scenario.
The impact of rising oil prices, global warming, and the ongoing recession are all further costs that can be lowered with high speed rail. Those costs need to be included in the discussion. If there is a risk that the HSR project will go over the projected $42.6 billion budget, there’s a bigger risk that California will suffer significant costs by NOT building it.
In conclusion, any discussion of HSR costs has to include the context – that includes the cost of doing nothing, accurate comparisons to other countries’ experiences, and the political decisions that societies make all the time. I believe we can and should build HSR for the current projected price. I’m also willing to pay a bit more if it’s justified and if the additional costs are funded. We can be cost-effective and provide Californians the system they want.