What do economists believe about infrastructure?

Support for massive investments in transportation infrastructure, possibly with a change in the share of spending on transit, seems widespread. Such proposals are often motivated by the belief that our infrastructure is crumbling, that infrastructure causes economic growth, that current funding regimes disadvantage rural drivers at the expense of urban public transit, or that capacity expansions will reduce congestion. In fact, most US transportation infrastructure is not deteriorating and the existing scientific literature and does not show that infrastructure creates growth or reduces congestion. However, current annual expenditure on public transit buses exceeds that on interstate construction and maintenance. The evidence suggests the importance of an examination of how funding is allocated across modes but not of massive new expenditures.

That is from a new NBER working paper by Matthew Turner, Gilles Duranton, and Geetika Nagpal.


Unblock the paper.

From the abstract this is closed surface, matching travel patterns to modes of traffic. Places that do not match have congestion. Over the whole set, count along in a granular unit of payment, aggregated. Unblock the paper.

Try https://www.brown.edu/Departments/Economics/Faculty/Matthew_Turner/research.htm

"To sum up, the evidence that infrastructure has important implications on how economic activity is organized is compelling. However, most of this evidence points to the importance of infrastructure as a determinant of where economic activity occurs. The evidence that infrastructure affects the level of economic activity is mixed, is sensitive to econometric technique, and there is no clear basis for preferring one technique to another."

Yeah, the economic activity is concentrated on US Interstates, US class I railroad line, not in South America or Africa, or in Europe on their rail lines, river/canal, and highways, or in China on century old waterways or rail and highways built in the past 30 years.

If public infrastructure doesn't matter, than many nations should attract all the growth due to lower taxes, etc, lower over all costs, and governments more willing to allow industry to whatever it damn wants.

@mulp - I'm breaking my voluntary exile from the comment section of MR to say that economists are divided about whether infrastructure helped drive the American economy, or vice versa, it's a sort of chicken-and-egg problem. For example, many economists say the railroads were NOT important in early to mid 19th century America compared to barge traffic (canals, Mississippi river), except for 'time sensitive' freight, see: Walton & Rockoff, History of the American Economy, 10th ed. (a really good history of economics textbook).

Bonus trivia: a lot of people probably don't know Pittsburgh was linked by canal to the Mississippi river (you have to go west, young man, and higher in elevation), nor that south France was linked to the Atlantic ocean by 17th century canal builders (Pierre-Paul Riquet, Baron de Bonrepos), nor that US freight is about the only mode of transport not subsidized by the state (unlike air, road travel), nor that toll roads are historically not that popular with the public, anymore than paying market rates for water is (Walton et al. say perhaps not even one was that profitable out of around 2000 by 1811; how's that Dulles airport toll road working out?) nor that from 1815-1843, US state governments operated 75% of canals (they still do, rather badly, see www.sjsu.edu/faculty/watkins/tombigbee.htm), after which they became obsolete due to railroad building, nor ...ah, talking to myself again. Bye.

@Ray barge traffic would still require extensive amounts of dredging, locks, etc. so you’re describing a chicken-or-this-other-chicken problem. The hard part is “knowing” what type of infrastructure will be “needed”, and the required specs for it to be effective/competitive/productive.

From an economic stand point, this pandemic seems like a once and a generation episode of creative destruction. We should be building the infrastructure that can facilitate growth for the next 30-40 years.

What infrastructure has been creatively destroyed. There may be some in the restaurant sector, but most people will go right back to work and infrastructure will be intact. OK, except maybe subways.

Are those same economists the kind of people that will argue whether the Internet created a digital economy?

Probably, because we know ideologues will, and some set of economists really are just that.

Cut to the argument that the TCP/IP packets sent when I hit "Submit" have nothing to do with government infrastructure.

“Some people are saying”

“Some economists” “some kind of people” “ideologues” “some people really are just that”


Interesting book here on among other things the impact of GPS infrastructure on archeology:


"....the existing scientific literature and does not show that infrastructure creates growth or reduces congestion."

should be without the extra *and*:

"....the existing scientific literature does not show that infrastructure creates growth or reduces congestion."

Australian TV Crew Struck by Police in Washington

"And you really saw how they dealt with my cover team. They are quite violent. And they don't care who they are targeting at the moment."

Way to welcome our Aussie friends to our dysfunctional, can't-do-anything-right government.


On the other hand, as loyal readers of MR know, Australians hate America.

Google “mostly peaceful “, for an idea of how uniformly this is being framed.

Then you get a better idea of how professionally the police have been acting - often on-air, as the results of their marksmanship are broadcast live.

And it is a variety of journalists from a variety of countries in a variety of American cities getting a first hand taste of how they are supposed to be dominated, to keep Trump free to walk the streets - after the tear gas clouds disperse.

Maybe somebody's already at work on a novel of the period - "Mostly Peaceful and Incredibly Close."

"Capacity expansions typically meet with offsetting expansions in travel demand and do little to increase the speed of travel." This is a common fallacy. Demand curves slope down. If the quantity demanded rises that means the price must have gone down. And you add many new users. So consumer surplus goes up a lot when you add new roads, ESPECIALLY if they are very crowded afterwards.

True. They measure the cars on a highway, for an example, and claim the increase of numbers are due to new traffic created, but reality is that the new, better highway siphons off traffic from other routes making them better for commute too.

The biggest problem is they use aggregates, eg, averages, of "objective" easily tallied factors. But NYC subways are suffering high rates of delays which can't be overcome by individual effort due to aging infrastructure.

It does not matter if maintenance or track keeps it smooth, or new cars have been bought, if the signaling system is broken. This is caused by this:

"Where to begin? For one thing, there’s the old—so very, very old—infrastructure. “In fact it’s so old that the MTA can no longer buy replacement parts from the manufacturer,” James Somers wrote in a 2015 essay for The Atlantic. “It has to refurbish them itself.” "

That is true for the NYC and Boston and other century old systems. NYC is still suffering from the damage from Sandy that flooded a llot of ancient copper wire, plus weakened structures.

The systems build a century ago haven't seen much expansion in half a century while traffic has doubled. DC was under built from the start. So while lines could be taken out of service for extended maintenance 50 years ago, today doing so cuts allowed passengers dramatically, or in the case of DC, extended service on any line cripples the system.

For similar surface rail, installing signaling to meet mandated safety standards has taken far longer than expected because so much equipment decades old with various patches to support new signaling gear. ATC requires pretty much everything be reasonably modern, like after 1990, or 1980. Subways are simpler organizationally because the are owned by a government but rail is often owned by multiple private railroads plus government but ATC must be implemented seamlessly across all.

This problem is caused by the same argument the authors used when it came to upgrading government data systems. Hey, the IRS and CMS and SS are working ok, so they don't need tens of billions in computer infrastructure investment.

So, Congress wrote the CARES act to make it easy to provide aid to all the people screwed by being told not to go to work. But even easy changes required modifying COBOL programs written in the 50s, 60s, 70s when it was standard and all processing was batch. Interactive interfaces have been kludged on a batch system written in COBOL.

With these legacy systems, either major investments are required, or it goes bankrupt and is abandoned. But it's hard to argue for abandoning NYC or the US because replacing legacy systems is very costly, but only a few percent of GDP.

I always wondered if buildings roads that go deep into the exurbs is a solution to the NIMBYs of the cities or if that just delays the inevitable at an expensive price.

Another thought is to use that infrastructure money and go Romer by building a charter city somewhere in the middle of the country. It doesn't solve NIMBY but it channels energy in the right direction.

'In fact, most US transportation infrastructure is not deteriorating' - of course it is, as is the infrastructure of everybody else. This is why maintenance is so important,

Adjective, not the verb.

Civil engineer here. The idea that “infrastructure” is a general employment booster is a mistake. First, the people who work on such projects are, by and large, highly skilled professionals. There is no surplus of such people. Yes, on the margin, there will be additional hiring and training but those will be gradual. In addition, major infrastructure projects are highly regulated and typically face significant public acceptance issues. Thus it can easily take ten years or more, and tens of millions of dollars in studies and litigation, to go from conception to groundbreaking. Remember Obama and “shovel ready projects?” They didn’t exist and don’t exist now.

And yet, something like the Iron Mountain Road was built by CCC members in the 30s.

Sure everybody was better at just about everything back then, but you've got to start somewhere.

If there's any thought that recurs the last few days, it's that a whole lot of working-age people don't have enough to do.

"And yet, something like the Iron Mountain Road was built by CCC members in the 30s."

It's not hard to physically build the infrastructure, it's just slow when you carefully follow all of the regulations and procedures.

Currently, if I want to use a standard 120V plug in an industrial panel I am legally required to suit up in an Arc Flash protection coverall with faceshield and gloves because it is higher than 50 volts. That's the same type of plug to which your computer is hooked up.

Now that rule is so patently ridiculous that it's routinely ignored, but no one can find a loop hole that would actually allow us to treat a standard plug as a standard plug.

There are a thousands of rules to be followed. Everyone of them slows things down just a bit.

Depends what you think Obama was shovelling.

I’m probably the only here who is old enough to remember COVID-19. Mass transit kills.

The NYC subway system didn't just encounter this problem yesterday. The present situation is the result of a long series of bad decisions, year after year after year.

They chose, every year, to defer maintenance and investment. They chose their fare structure, employee compensation plans, etc.

The percentage availability of the New York City subway system has been very high over the last hundred years, has it not?

OK, things are great. Either way, no Federal funds justified.

There are separate issues about which infrastructure the federal government should support, but on the engineering end we should stick to formal metrics on reliability.

And Boomers from San Fernando should refrain from talking about the subway like they know something.

Or at least use metrics like “on-time percentage” and “$/mile construction”

The percentage availability is likely pretty high. They have been putting off projects to make it more efficient, like new signalling systems and automation, though so not to kill any of the jobs involved. No doubt it moves a lot of people, but it's often crowded where it could run more cars, and faster, if improvements were made.

It strikes me as intellectually inconsistent that behavioral economist Richard Thaler deserves a Nobel for his work on "nudges" to promote certain kinds of preferred behavior, but using infrastructure to promote a more efficient city structure is beyond the pale. The structure of cities in America came about arbitrarily, the sprawl attributable more to timing than logic, the automobile the main contributing factor but also the growing middle class of blue and white collar workers who could afford a detached single family house far from the urban center and congestion, crime, and the "other". No sentient person would actually design a city like the cities that dominate the American landscape. I suppose Cowen would argue that the "market" chose the design of American cities: it's what consumers wanted or the design, the sprawl, would not have happened. The "market" also produced segregation, smog, and global warming.

Tabarrok supported the bill in California that would free up areas near transportation (infrastructure) hubs to high density development. The bill came very close to passage but, alas, it failed. This time. Economists are wary of "planning", which I understand: much of the "planning" is to accommodate the economic interests that benefit most from the "plan".

Here's a historical anecdote: my great uncle was a pioneer in what is known today as "landscape design". Today, we think of "landscape design" as what wealthy people do with their yards. But that's not what the original landscape designers did: they helped design cities to make them more livable. Frederick Law Olmsted was the father of landscape design. My great uncle studied under his son Frederick Law Olmsted Jr. at Harvard and spent over half his career as the "landscape designer" for a western city, a city known today for it's many public gardens and green spaces. Imagine that!

>The "market" also produced segregation, smog, and global warming.

Why did you leave out Joe Biden?

For what it's worth, this has been one of my long-held beliefs.

Basically the rate of failure of infrastructure is very low. Especially if you look at it exactly as that, as a percentage of the in-use infrastructure which fails in a year.

I'm sure there is room for improvement, but you do it by analysis of failures and searches for the same at-risk conditions, rather than wholesale replacement of just old stuff.

What do you mean by failure?

Because if you mean like bridges collapsing, that would be a weird metric. The failure rate is almost entirely dependent on maintenance investment.

I'd think you could find a lot fewer failures than only catostrophic.

Maintenance of infrastructure has become more expensive than the initial construction, perhaps in part because of the obsession with controlled inflation. Highway construction and maintenance would seem to be an exception since a completed highway is little different now than it was a century and a half ago except for far less labor being required. Most US roads are either concrete, a product developed by the pre-Christian Romans or bituminous, essentially the same as it was when first adopted in modern times in 1822. These roads, especially in northern climes, begin to deteriorate immediately after being put in place. In fact, engineers design roads to last for just 25 years, the same length of time it takes to get a child out of the house. Maybe interested parties are perfectly happy with travel summers that include long lines of cars and neighborhood detours that are the result of failed road surfaces which also take a toll of automobile and truck tires and suspensions. That may be why little research has been done on advanced forms of highways except this small study from the Netherlands.

i have a feeling that once a few commenters get past the paywall, they are going to shred this paper’s blanket assertions.

Infrastructure is a term that is much broader than assumed by this study.

Having experienced a few transit systems around the world, I can say that Chicago and NY systems have quite easily been under invested in (the paper says 'maybe' subways). There is a lot of roundabout in the paper and not much depth to their arguments. I'd go with the civil engineers on this one.

Comments for this post are closed