About the only thing Hillary Clinton and Donald Trump agree on in this election is the need for the federal government to spend a lot more on America’s infrastructure. Clinton has proposed $275 billion over five years for roads, transit, airports, and other public works, paid for with higher corporate taxes. Trump has vowed to “at least double” that.
Plenty of observers have applauded these proposals, arguing that vast new infrastructure spending is just the tonic America needs in the 21st century. But Harvard economist Edward Glaeser sees things differently. To him, there’s a very good chance that much of this new spending will simply be misdirected or wasted.
“My biggest fear is that we’re not actually going to fix the biggest problems we have,” he says of these proposals. “Instead we’ll just end up with more highways in North Dakota or more white elephant projects that deliver little value.”
It’s not that Glaeser is against infrastructure. America’s roads and airports are vital, and they really do need upgrades. But the way the federal government currently addresses these issues is badly flawed. Congress hands large checks to states, which often just build shiny new roads rather than fixing existing ones. Instead of targeting urgent needs, federal spending often goes toward questionable projects like Detroit’s little-used monorail or California’s troubled high-speed rail or roads and interchanges of dubious value. Infrastructure spending is rarely evidence-based; few projects go through rigorous cost-benefit analysis.
On top of that, politicians often tout infrastructure spending as a way of creating sorely needed jobs. But in a recent City Journal essay, Glaeser argues that this is misguided. The places with the worst unemployment problems often aren’t the places with the biggest infrastructure needs. What’s more, a look at ailing Japan, which has spent over $6.3 trillion since 1981 on truly impressive bridges and bullet trains, suggests infrastructure isn’t always a cure for economic woes.
I recently talked with Glaeser about what makes our approach to infrastructure is so dysfunctional — and what he thinks a better system would look like.
When you look at Clinton and Trump proposing to spend hundreds of billions of dollars on infrastructure, what’s your biggest worry here?
So it’s not that I’m opposed to infrastructure. I agree that we need to fix our potholes, that we have to do a better job with airports like JFK. What I’m worried about is that we tend to hand out federal funding in a way that’s not usually tied to specific benefits, that’s not particularly cost-effective or pragmatic.
What I’m worried about is that the money would be poorly targeted, not going to the areas that actually need it. That we’ll get too much spending on new roads, and not enough maintenance for existing roads. My biggest fear is that we’re not actually going to fix the biggest problems we have. Instead we’ll just end up with more highways in North Dakota or more white elephant projects that deliver little value.
And to be clear, I have no idea what the right level of spending is. I could well be convinced that you could spend $275 billion over five years in a way that’s efficient. But I’m worried that simply increasing Department of Transportation spending without a well-thought-out plan will lead to little but waste.
What are good past examples of waste?
I look at things like Detroit’s People Mover Monorail, which never would’ve been built without federal aid [and now struggles to fill capacity]. This is what happens when we tell ourselves that infrastructure can bring a city back, that it will magically generate economic benefits. We don’t think about the fact that this isn’t actually delivering services that people value. It’s already pretty easy to get around downtown Detroit — this is a city built for 1.8 million people that now has half that. Building a monorail to glide over essentially empty streets [is] not going to deliver value.
So I think it’s mistaken to see infrastructure as primarily a macro solution. The smart thing to do is focus on cost-benefit analysis for individual infrastructure projects, and ask whether they deliver enough of a payoff to justify their costs.
So how do we make sure our infrastructure investments are actually wise ones?
Partly it’s a question of who funds the infrastructure. The more that a piece of infrastructure can fund itself with user fees, whether it’s gas taxes or tolls or landing fees for airports, the more likely it is to be well-targeted. Whereas the more that a piece of infrastructure is just paid for in large chunks by Washington, the less likely it is to be well-targeted. If Washington’s picking up the tab, there’s no reason [for states and local governments] to worry too much about costs and benefits.
I still think fundamentally the people who use roads should pay for roads, rather than general taxpayers. [The latter approach is increasingly common with federal highway bills.] In a world in which we worry about energy overuse and carbon emissions, the case for having higher gas taxes seems stronger than ever. And I understand they’re unpopular, but part of the job of political economists is that we’re not in office and we should say what we think.
There’s also an issue of inequity and inequality here. Take JFK Airport. The average traveler in JFK has income that’s around 10 times higher than the American average. When you use federal money to pay for infrastructure that is mostly used by the rich, it exacerbates inequality. So the natural answer for me is that JFK needs to be self-funding with higher landing fees [which would, in turn, be passed on to fliers].
And I should say that it’s not as if my views are at all unusual among transportation economists. The consensus around having the federal government spend much more on infrastructure mainly exists among those who study the macroeconomy rather than those who actually study infrastructure.
Is it hard to do cost-benefit analysis on infrastructure — or is the issue that we rarely do it?
There are certainly nuances here, but the primary issue is that we just don’t do it. There’s a great line from the Office of Management and Budget that noted that during the Bush years, highway funding was “not based on need or performance.”
At best, sometimes we’ll rely on cost-benefit analysis provided by the company that’s going to build the project — that was the case with high-speed rail in California. That makes absolutely no sense. Now one of the things I would like the federal government to be doing is giving much more support to local governments to do cost-benefit analysis.
So when academics look at the costs and benefits of various types of infrastructure, what projects do they find most valuable?
There’s a strong consensus that maintaining existing infrastructure gives you much more bang for your buck. There have been diminishing returns to building new roads, particularly since we completed the National Highway System. Whereas if you have existing corridors with potholes, the returns to fixing that are very high. [See here for more.]
Another area of agreement among transportation economists is a profound enthusiasm for buses over trains. Bus rapid transit is considered a very high-return investment. These aren’t necessarily buses operating on crowded city streets; these are buses with dedicated lanes that can achieve almost the same speed as trains.
The beauty of buses, from a cost-benefit perspective, is you don’t need to lay down massive infrastructure that you’re stuck with forever. If a bus route doesn’t attract enough people, you switch the route. Or you stop running it. It’s flexible in a way that trains aren’t. And that’s tremendously valuable in a world of uncertainty.
Now, this is not about gutting the subway in New York or the Metro in Washington, DC. But for new stuff, investing in buses tends to make more sense given the modest densities of most American metropolitan areas.
Then why do states and local governments often tend to be more enthusiastic about trains and light rail than buses?
Part of the problem is that because buses tend to be transportation for the poor, we often allow their quality to be dismal. But they don’t have to be. Buses can have wifi on them; they can have nice stuff. So there’s an image problem. Trains are cool-looking. We have to do more to make buses cooler. That’s one reason I actually like the “Boris buses” in London, the souped-up versions of the red double-decker bus.
The other issue, though, is that the big spate of train building in the 1970s was specifically abetted by the Federal Highway Aid Act of 1973. Back then, support for more highway spending was petering out, and in order to get another big pass-through of highway money, Congress got urban representatives on board by saying, “We’ll throw in money for trains.” So that was part of it. I don’t think there are too many governments that would find building a light rail system with their own cash all that sensible.
In your City Journal piece, you argued that the federal government should have far less of a role in funding infrastructure and the states should have far more. The idea is that this would force states to be more disciplined about spending.
I want to push back on this a bit. Right now, there are a lot of things states obviously aren’t doing — like fixing existing roads. So if the federal government pulled back, isn’t one likely outcome that nothing would get done?
I would be wary of pulling back entirely right away. There is a branch of political economy that says if the federal government pulled back, states would be forced to do more and they’d make better decisions. I guess I can’t be that laissez-faire. I don’t think we should shut down the Department of Transportation tomorrow.
But the current process of having the federal government write large checks and just send them to states is not particularly effective. So the right answer is to rethink.
You start by devoting significant resources to figuring out where America has problems and why they aren’t currently being solved, and then seeing if fixing them requires changing the financing rules or working in combination with the states. Focus on where the problems are and where you can get significant improvements by changing funding formulas, changing how funding is allocated.
Now, I can still see a role for the federal government in coordinating local investments, or measuring safety, or providing support for cost-benefit analysis. I have no problem with using federal spending to nudge localities in the right direction. And I have no problem subsidizing certain things like bus transport for poor Americans. Or if you have areas that are too poor or depopulated to pay to maintain their water systems, that’s something we need to think more seriously about. But for most infrastructure, the right answer is for those who use it to pay for it.
We have seen a few efforts at reforming federal spending of late. The Obama administration’s TIGER grants were goal-oriented, aimed at improving environmental problems. And if you look at Clinton’s infrastructure plan, she seems to prioritize maintaining existing roads over new ones. Are there other changes you’d like to see?
One thing the Obama administration did that I really liked was its proposal to change the rules prohibiting tolls on interstate highways.* That [prohibition on tolls] is a crazy rule to have in the 21st century. I’d like to see the current restrictions on landing fees for airports tweaked as well. People should pay for the social cost of their flying. The TSA should be paid for by fliers.
The federal government has arguably had some big infrastructure successes in the past. Various canals in the 19th century. The National Highway System. The Hoover Dam. Those have had some major benefits. But are you saying the era of big, important infrastructure projects is over and we should just focus on maintaining what we have?
My answer would be I don’t know.
One aspect of this is that we may have hit diminishing returns for efforts to increase speed. In 1816, it cost as much to ship goods overland 30 miles as it did to ship goods across the entire Atlantic Ocean. So there were huge effective distances. But we’ve invested a lot over the past 200 years, and consequently it’s much easier to cover distances for people and goods. Transportation is now such a small share of the costs for most goods. And the marginal benefits of further reductions in speed is likely to be low.
That being said, investments related to new technologies could well have large impacts. Investments in broadband probably had major benefits. The other place to look is with autonomous vehicles.
What sorts of infrastructure would you need for autonomous vehicles?
I think the right infrastructure is still not clear. But one thing you do want to worry about is the behavioral response. One of my favorite papers in transportation in the last five years is by Gilles Duranton and Matthew Turner on “the fundamental law of road congestion.” They find that vehicle miles traveled increases roughly one to one with highway miles built. In other words, you can’t solve traffic problems just by building new roads, because more people decided to drive on those roads, or they decide to live farther away.
So you could see a similar effect with autonomous vehicles. If you reduce the cost of driving, that could create more demand for commuting by car, or commuting at car during peak hours, because people don’t mind sitting in traffic as much.
One way to get around this is with congestion pricing. And this gets interesting. One of the key political economy points we’ve learned is that it’s really difficult to slap a congestion price on a road that used to be free. But slapping a toll on a new road or bridge is much more feasible. So, since autonomous vehicles will be new, sticking a toll on them that’s connected with congestion might be something that you want to make part of the technology from the get-go.
Infrastructure spending often gets touted as a way to create jobs — particularly if the economy is underperforming. But you don’t think that’s necessarily the best way to think about this?
In a lot of these discussions, we often seem to have this image of the 1930s, where you take unskilled, unemployed workers and give them a shovel and get them building roads, and they’ve got a job and that’s great. But that’s just not who works in the transportation infrastructure business anymore. It’s an industry that requires skilled engineers. Whereas the vast majority of Americans without college degrees are now in the service economy. So it’s not at all obvious you’re going to get those workers.
It’s also true that the places with the worst unemployment are not the same places that need more infrastructure. And certainly when you continue to follow the Highway Trust Fund apportionment rules, the dollars going into low-density America were much higher on a per capita basis than the dollars going into high-density America. So there’s often a spatial mismatch.
It seems like there’s this inherent tension between using infrastructure spending to create jobs and using infrastructure spending to fund the projects that have the highest ratio of benefits to cost.
Absolutely. And if you are concentrating on getting money out of the door quickly, it almost ensures that the money will be spent poorly in terms of improving our quality of life. It is very hard to make investments as difficult as something like infrastructure quickly and wisely. With transportation, it’s too important to be done in a rush. We need to step back and figure out what investments matter and where we’re going to get bang for buck.
Are there lessons to learn from other countries here? People are always pointing to all the great high-speed rail that China and Spain are building.
I think both China and Spain have cautionary tales. Spain went on an infrastructure bonanza, and their economic troubles have been extremely significant. I’m not saying infrastructure caused it, but it sure as heck didn’t solve it. China is a little different. Parts of China have been urbanizing very quickly, and you actually do need a lot of infrastructure. But there’s a huge amount of waste there as well. It’s one thing to build more stuff around Shanghai, but you’ve got a lot of fancy rail in the interior of the country, where it’s hard to imagine the benefits exceed the cost.
That said, I think we can always learn from successes in other countries. The congestion pricing experience of Singapore, London, Stockholm, those are always worth emulating.
One of my favorite infrastructure financing stories is mass transit in Hong Kong. The system costs about $1 to ride, it’s a very functional profit-making enterprise, and the way they fund themselves is by building skyscrapers on top of subway stops. Essentially the skyscrapers are a way of capturing the value.
So we could learn to get more creative in financing. Things like tax increment financing, where you look at changes to property values as a result of infrastructure, seems like a reasonable thing to look at. I like the idea of allowing developers near new subway stops to be able to pay for a density bonus, where they can build much higher and pay an ongoing fee for that to the transit provider.
A related question is whether the US pays too much for its infrastructure relative to other countries. It sometimes seems like it’s just much more costly to build stuff here.
The first thing to say is that it’s always difficult and expensive to retrofit existing cities, particularly if you’re not willing to bulldoze neighborhoods, which we shouldn’t be willing to do. That makes it costly to do things like [Boston’s] Big Dig, which would be expensive in London or Berlin or even Istanbul — although labor costs might make things cheaper there. So I tend to take the cost of infrastructure as a relative given rather than thinking there’s some magic public policy lever we can pull and radically reduce the cost of things.
That said, there are specific things we can do to reduce costs. I’m supervising a student who’s working on the process in Massachusetts of bidding and revising cost estimates. Those rules differ by state, but I think there are technocratic improvements along those margins that could lower costs by maybe 10 percent if you were really smart about it. But I don’t think you’re going to chop 75 percent off. But to your original question, I’ve never done close cross-national comparisons here, so there could be other ideas out there.
Transcript has been lightly edited for length and clarity.