There’s some chance that this will be our final Infrastructure Week. But before this very-online joke passes into the ether, it seems natural to think about it in a proper historical context. Well, at least proper to me.
The importance of infrastructure in the development of the American economy is well-trod ground. Duranton and Turner (ReStud 2012) estimate large effects of highways on city-level employment growth from 1983 to 2003. Highways also helped integrate poorer and more isolated regions. Jaworski and Kitchens (ReStat 2019) show massive effects of the Appalachian Development Highway System on both Appalachian counties directly and the rest of the nation. Before that, railroads and canals spurred the settlement and development of much of the country.
But if highways and other types of transportation infrastructure are so valuable, why does it seem our current stock is depreciating so fast and we aren’t building a lot more? The short answer is that it seems infrastructure costs a lot. Much, much more than it “should” and much, much more than it did in the past. But why? Here, the political economy story takes center stage.
In a masterwork of data collection, Leah Brooks and Zachary Liscow study this question in the aptly-titled “Infrastructure Costs.” They begin by documenting that yes, conventional wisdom is right: infrastructure costs have been rising. Real spending per mile of highway interstate has increased more than 3x from the 1960s to the 1980s with a key turn up in the early 1970s. They dismiss a number of plausible theories for this change. It is not that inputs or labor costs have risen dramatically, though labor costs and staffing may play more of a role in America’s inability to cheaply build mass transit like subways. And, it is not about the selection of roads. (That is, states did not build the cheap roads first and the more expensive roads later.)
Instead, Brooks and Liscow argue the culprit is, of course, tortuosity.
Tortuosity, of course, is a measure of the wiggliness (or squiggliness, depending on your preferences) of a road. Starting in the 1970s, citizen’s voice became a much more important factor in infrastructure planning. And citizens used their voice to demand routes that did not run through certain (their) neighborhoods, as well as the use of other costly disamenity-shields like noise barriers. Brooks and Liscow also find a massive increase in miles of bridges per mile of interstate (from 0.02 in 1960 to 0.10 in the 1990s).
In “Freeway Revolts!”, Jeffreys Brinkman and Lin document the history of this episode of citizen’s voice and estimate the consequences of highways running through urban census tracts the revolutionaries wanted to avoid.
So what happens to a neighborhood after a highway runs through it? The hyperlocal consequences of the highway are generally negative. In a case study of Detroit, my former student Chelsea Carter shows tracts have short and long-run declines in property values and population density. Brinkman and Lin show freeways (or highways or parkways or interstates, depending on where you grew up) caused slower population growth and income growth and reduced land values as well, but only for center city tracts. Highways were generally a boon for outlying areas.
With all this opposition, where do highways ever get placed? In Detroit, Carter finds the key role of costs: interstates appear to be routed through poorer neighborhoods because low property values mean cheaper land acquisition costs and smaller future losses to the city’s property tax base. Brinkman and Lin also find that during the course of the freeway revolts, “better-educated and more-white neighborhoods were increasingly more successful at avoiding planned freeway construction over the 1960s.” On the extreme end, there’s the case of the Embarcadero Freeway.
Has this opposition to transportation infrastructure always been the case? In American economic history, the railroads are a natural precursor to the interstate highway system. But rather than railroad revolts, the political economy story appears to play out differently. (If you missed it, it was west and frontiers week here at Broadstreet a month ago.)
The railroads of the late 19th century, like the highways of the mid-20th, were not without opponents. An entire subgenre of westerns dramatizes the often violent battles between farmers, ranchers, and archetypal railroad tycoon villains. But here, the railroad is not usually fought because it is loud or polluting. Instead, the fight was over whose land the railroad would go through. Spoiler alert for decades-old films, but the savvy Brett McBain of Once Upon a Time in the West got himself killed because he bought up the land Morton’s railroad would have to travel through and designed a station and city around the crossing. In satire, Mel Brooks’ Blazing Saddles hits similar themes. Though the town residents don’t yet know it, the town of Rock Ridge is valuable because the railroad will run through it.
And we don’t have to take Sergio Leone and Mel Brooks’ word for it. The economics literature agrees. Donaldson and Hornbeck (QJE 2016) show that agricultural counties gaining access to markets via the railroads benefitted dramatically, gains that were capitalized into land values. And even with a canal planner as skilled as Robert Fogel, counterfactually expanded waterways could replace only one-fifth of the railroad benefits. In follow up work, Hornbeck and Rotemberg (NBER WP 26594) show that the estimated benefits of the railroads were actually larger as economic activity moved from less to more productive counties.
So what changed? Glaeser and Ponzetto (Economics of Transport 2018) offer a model that seems to explain some of the key differences between eras (though of course, with N=2 things are overdetermined). Politicians and their voters face a principal-agent problem. What kinds of projects will a politician undertake? If there are projects that voters misperceive as beneficial (they see only the benefits but not the long term costs), politicians will deliver more of these projects. But projects with diffuse benefits and concentrated costs are just the opposite and politicians will deliver fewer of these.
Large highways through the middle of nowhere, with maintenance costs paid by less salient future taxes, are clearly in the first category and we’ll get too much of them. But the highways that run through cities and destroy neighborhoods or upset neighbors with noise and pollution generate a vocal opposition and will be underinvested in. As density increases or as the political power and savvy of urban residents increases, we might expect the revolts Brinkman and Lin document and the increasing costs Brooks and Liscow document.
But what about the railroads in the 19th century? Here, Glaeser and Ponzetto’s theory offers a few more clues. Collective action against railroads may have been just less likely to succeed. For one, we know schooling attainment increased in the US for every birth cohort in the 19th and 20th centuries. Second, as Bazzi et al and other work demonstrate, the residents of the west, from Rock Ridge to Flagstone, were and are skeptical of collective action and communalism, the kinds of local cohesion that might be required to block powerful interests from running railroads through your town. Or maybe, the costs of railroads were just far outweighed by the benefits of market access in the low-density west?
A better comparison for the freeway revolts might be the construction of the urban elevated train system in the late 19th century. Before urban rapid transit went underground—a technological challenge illustrated by Doug Most in The Race Underground—cities ran railroad track through crowded neighborhoods, usually on an elevated track, in New York, in Chicago, and in many smaller cities. At the risk of ending on an unsatisfying note, I’m not sure what the political response to the elevated infrastructure was. But, this is exactly where I’m hoping to push some new work, and (I’ll admit now that we’re at the end) this post was a commitment device to get my train of thought down on paper. The train probably won’t run on time, but it’ll show up eventually.