When funding infrastructure, federal elected officials do not start by asking, “What projects need to be done?” Instead, they ask, “How much money should we spend?”
Then, through a process that amounts to little more than a parlor game, they decide on a giant number — say, 550 billion. They put a dollar sign in front of it and divvy it up in legislative text. (You Are Here.) They pass it and have a great big signing ceremony where they pat each other on the back for all the great infrastructure they have “created.”
Now, state transportation officials — who are the most knowledgeable on what is actually needed and have heretofore been taking the back seat in this process — have to figure out which projects get funding. Again, the incentive is to throw every project at the wall and see what sticks. Forecasts are part of this process. So if you’re a consulting company putting together a forecast, and you want a project to get built, what do you say? You say that the project is essential because tons of people would use it in ten years. There are no downsides to that approach. The project is more likely to get built. If your forecast is right, you look smart. If your forecast is wrong, nobody will blame you because it was just a forecast, and predictions are hard.
It’s not all self-interest. Making predictions about what transportation networks people will use in the future is genuinely difficult work that is always going to be prone to large errors. But if it were simply a matter of error, you’d expect there to be just as many underestimates as overestimates. That’s not what the Wall Street Journal reports:
A Federal Transit Administration survey of 27 recent public-transit projects that opened between 2007 and 2015 found that the average one overestimated ridership by about 21% two years after opening. That was an improvement over previous years. Projects that opened between 1990 and 2002 overestimated ridership by an average 77%, the FTA found. . . .
Public road projects overestimated use by about 6%, according to a study of roughly 1,300 projects by the National Academies of Sciences, Engineering and Medicine that was conducted before the pandemic. And traffic counts on a sample of toll roads around the world were roughly 77% of what had been forecast.
They’re not just overestimates; they’re not even close, and they’re worse for transit projects than they are for roads. Decisions based on those forecasts waste taxpayer money.
There are many efforts under way in the transportation industry to get forecasts to be more accurate, but there’s one that would make a big difference: Get the incentives right. One proposal in the Journal story from a University of Kentucky engineering professor does that:
“There is some incentive for forecasts to be high if they make [a project] more likely to get built,” he said. “If we were to systematically evaluate every forecast we do and issue a report and say, ‘This is how accurate we were after the fact,’ it takes away that incentive.”
Aside from making forecasters own their predictions, it would help to pick the projects first, then fund them. The goal of infrastructure projects should be to complete infrastructure projects, not to spend a legislatively specified amount of money. State transportation officials know more about what projects need to be completed than members of Congress do. The decision-making process should start with them, and it should stay within the states as much as possible.