Advocates for passenger rail in America are excited. The stimulus bill provided $8 billion for high speed rail construction, California has passed a bond for nearly $10 billion to build a system in that state, and other projects from Florida to Chicago are moving forward. The Federal government is planning to issue the grants to “jump-start” development of a national system this fall. In this climate, Harvard Economist Edward Glaeser posted today the second in a series where he says he will attempt to conduct an economic analysis of high speed rail.

“Personally, I almost always prefer trains to driving,” Glaeser wrote in his introductory post to the series last week, but quickly added, “the public must be wary every time our leaders decide to spend billions of our tax dollars,” citing economists who have long argued passenger rail is rarely worth its cost.

The result of the first analysis is unsurprising: an unfavorable finding to a conventional cost-benefit analysis evaluation. Some will quibble with the assumptions or methodology, and they are very rough. (Ryan Avent posted this scathing critique on Streetsblog earlier today, the Transport Politic also evaluated the assumptions)

However, I’m not going to attack the math. I’m not overly worried about the outcome of a cost-benefit analysis calculation, because we almost never make big transportation infrastructure decisions through this kind of analysis. We built the interstate highway system because it was deemed a suitable national goal. Cities are building light rail transit because they want it for a variety of reasons. Certainly, economic analysis informs many specific aspects of the system design, such as weighing possible routes, deciding on service levels, and sometimes selecting the mode. However, experienced transportation planners know because of the future’s uncertainty, even the most rigorous analysis can be wrong. And on many of the big questions, such as what mode of transportation to invest in and where it should go, are decided through the political process regardless of whether they make economic or practical sense. Finally, even the most brilliant and accurate cost-benefit analysis is only meaningful if the actual cost is somewhere close to the estimate cost. It turns out creating accurate estimates — and ensuring the project is built for a similar price — is very difficult.

For large infrastructure projects like high speed rail, accurate cost estimates almost never happens. According to Bent Flyvbjerg’s Megaprojects and Risk, the cost overruns can truly be spectacular:

Project	Cost overrun (%)
Boston Central Artery/Tunnel Project (Big Dig)	196
Great Belt rail tunnel, Denmark	110
Shinkansen Joetsu rail line, Japan	100
Washington Metro, USA	85
Channel tunnel, UK, France	80
Paris-Auber-Nanterre rail line	60

That means the total actual cost of Boston’s Big Dig was nearly three times the original estimate. A 100 percent cost overrun means it was double. On average, out of the study of 258 total projects in the book, the actual costs were 45 percent higher than estimated.

The causes of what they call this “calamitous history” are many and diverse. For the list above, I selected a variety of projects, from different times and cultures, to show it’s not uniquely an American problem. They’re not even all government-led projects, most notably the Channel tunnel was a private initiative. Private ownership does help, but cost overruns for a selection of private transport projects in the book run from 80% for the Channel tunnel to 15% for France’s Pont de Normandie bridge.

Much of the rest of the book is a detailed discussion of the causes for error in the creation of estimates, and possible reforms to improve the efficiency of megaproject construction. On the cost estimate side, the authors argue project studies rarely incorporate sufficient accounting of the inevitable risk, and are often skewed to meet political needs.

On the construction efficiency side, the authors propose four basic “instruments of accountability”:

• The involvement of risk capital: At least one-third of the project’s budget should come from private investors with no sovereign guarantee – i.e., they would lose money if the project goes over budget or has lower revenue than anticipated.
• Explicit formulation of regulatory regime: identifying all associated costs, even non-obvious ones like access ramps and stations, creating independent environmental review and other necessary oversight committees, and define the project financial and decision-making structures. They suggest a state-owned enterprise approach or build-operate-transfer approach, which place the government at arms length and minimizes conflicts of interest where the government is working both as an active booster and trying to meet public-interest objectives.
• Performance specifications: Define performance targets instead of specifying the specific approach for technical and environmental goals.
• Transparency: Enhanced transparency and public involvement to scrutinized plans and minimize opposition.

Our conventional project planning system could stand to gain from some of these reforms, and the time is now to incorporate them into the selected high speed rail projects. I still believe it’s important to try to create accurate estimates of total costs and benefits, and bring them into the decision-making process. However, since cost-benefit models are limited by uncertainty and often disregarded by the political system, we must also focus our attention on how to complete projects in the most cost-effective way possible.

> Edward Glaeser series: Part 1, Part 2
> Flyvbjerg, Bruzelius, and Rothengatter: Megaprojects and Risk