The Paradox of Infrastructure Investment: Can a Productive Good Reduce Productivity?

In 1999, federal, state, and local governments together spent nearly $250 billion on nondefense infrastructure ranging from computers for office workers to land for new streets. Given this substantial public capital investment, it seems worthwhile to consider what effect it might have on national well-being, especially in metropolitan areas, where most Americans live and work. Perhaps surprisingly, there is little academic consensus on this fundamental question, despite more than a decade of close scrutiny.

The potential benefits of public capital investments are of two sorts. First, the investments might directly or indirectly make us richer. By raising productivity, they could raise the income we receive from working or investing. The second potential benefit does not involve markets. We may, for example, value the time we save when our town upgrades a congested intersection, or our kids may enjoy visiting the new playground in our neighborhood. These nonmarket benefits are surely important, but we know little about how big they are or whether what we might call “quality of life” public investments are worth their cost.

Economists have focused on the productivity benefits of public investment for several reasons. First, measuring productivity is what economists do best, and data and methods for measuring state-level productivity effects are readily available. Second, some early research on public capital’s national effects seemed to show that it was extremely productive. Some authors even claimed that the slowdown in public investment in the early 1970s, when the interstate highway system was nearing completion, helped contribute to the national productivity growth slowdown that began about the same time. If they were correct, then the nation was in desperate need of a large new capital investment plan.

What Have We Learned?

Much of the more recent work on the productivity benefits of public investment has focused on the states rather than on the nation as a whole. One reason for narrowing the analysis is pragmatic research design. Data limitations and statistical problems make it hard to identify the separate effect of infrastructure investment on economic growth at the national level, and many observers have been skeptical of the size of the infrastructure effects claimed by some authors. States offer a veritable cornucopia of infrastructure and economic growth data that allow for more careful statistical analysis. In addition, most of the nation’s infrastructure is owned by state and local governments. The current trend toward decentralizing public responsibilities would probably give these governments a central role in any plan to boost public investment.

Analysis of the effect of state and local government investment on state-level economic growth has provided a range of estimates, and debate about the exact magnitude of the effect continues. But most authors now seem to agree that modest increases in state public capital stocks would not dramatically raise state economic growth?in other words, that increasing public investment will not add much to a given state’s ability to create jobs and wealth.

Economists have been hesitant to give the policy advice, “Don’t do it,” because they recognize that some projects may have a beneficial economic effect even if the average one doesn’t. They also realize that the productivity evidence does not take into account the direct household benefit of having a better infrastructure stock. The household sector has received relatively little attention in the infrastructure literature, in part because there is no widely accepted way of measuring benefits received directly by households. History and casual observation, however, provide at least anecdotal evidence that households value highly some elements of the infrastructure stock. Consider, as examples, the widespread suburbanization that was surely spurred by the development of urban arterial roadways, or the clustering of high-income residents around New York’s Central Park. In both cases, and a multitude of others, households express the value they place on public infrastructure by relocating to get access to it. This process bids up the value of land near new public works relative to land that gets no new investment. We can infer something about how households and, by the same logic, firms feel about public works by watching how they evaluate places with lots of infrastructure as against places with little. In some ways this is the question that economists have been asking: do states that devote a lot of resources to public investment capture more growth than those that don’t?

But research on the effects of infrastructure investment on firms’ and households’ location decisions tends to find that the benefits of new investments are localized?they are highest near where the investments are placed. This finding makes sense, particularly for public works like playgrounds, parks, or public buildings: the farther someone has to travel to use the facility, the less its net value. The geographic extent of infrastructure benefits, of course, depends on the size, type, and location of new public works. City playgrounds are most valuable to those who live closest to them and can use them frequently. Large, modern sewer systems provide benefits to homes and businesses that can hook up to them as well as to others whose local environment is made cleaner. And, importantly, centrally located or easily accessible facilities would be expected to provide more widespread benefits.

Network systems like highways are a special case, because improving any part of the system provides benefits that may extend throughout it. But most current highway improvements, especially in urban areas, are designed to relieve traffic hotspots or to provide other fairly localized benefits. A resurfaced freeway lane in San Diego is probably of little note to residents of San Francisco, even though it may be valuable to those who use the roadway every day.

One implication of the research finding that the benefits of public works generally decline with distance is that the effects of most modern infrastructure projects usually won’t be even roughly uniform over geographic areas as large as a state. Thus, studies that focus on states may miss some of public investment’s most important effects.

Taken together, the finding that the benefits of public works tend to be localized and the consensus that infrastructure’s state-level effect is small lead to a more general understanding of the effect of public investment. In particular, they suggest that infrastructure investments are valuable and that their primary effect is to rearrange the spatial distribution of jobs and residences within states. Firms seeking high-quality road connections may be able to find them in newly developed areas of their home state. Households tired of congested inner-ring parks may move to the fringes of metropolitan areas, where open space is more plentiful and from which they can commute to the same jobs, perhaps along the same roads that firms find valuable.

Where Does the Money Go?

Data limitations make it hard to nail down exactly where new public works investments, especially those funded by state governments, are made. The scant evidence available is that most new public works are put in place in suburbs rather than in central cities. During 1987?91, for example, Philadelphia’s Pennsylvania suburbs received about $641 million more than the city in new highway investments, much of it paid for by the state, even though an unusual new limited-access roadway through the central city, the expensive Vine Street Expressway, was being built during that same period at a cost of more than $200 million. A highway differential this large may have cost Philadelphia as many as 20,000 jobs. And evidence for the nation as a whole implies that the phenomenon is widespread.

So What?

No one who has ever watched office parks spring up around new highway interchanges will be surprised to learn that places receiving new public works draw activity. It may also not be surprising that the job growth next to new infrastructure systems comes at the expense of jobs somewhere else. Few readers will be astonished to discover that new territory on the fringes of metropolitan areas gets more benefits than the older urban center or that suburban jurisdictions are systematically able to extract more benefits from the political process than are central cities. The notion that public works provide local benefits to the places that get them at the expense of places that don’t is as old and as well understood as the term “pork barrel.”

Many will be tempted to interpret this spatial redistribution as a relatively benign phenomenon, with few implications for overall social welfare. True, some residents benefit at the expense of others, but surely the total benefit averages out to be zero at worst, or maybe even slightly positive. Of course, given a $250 billion price tag, benefits ought to be more than slightly positive if the nation’s money is to be put to its best use. But I will argue here that the total benefit of all this infrastructure investment may well be far less than zero.

How can relocating a job from one county to another be anything other than a wash in total? Answering this question requires a brief detour into the growing academic literature on nonmarket interactions among businesses. These interactions have potentially large effects on both economic growth and equity.

Benefits from the spatial concentration of economic activities, especially jobs, have long been a central feature of urban economic models, and recent studies have revived interest in the role of density in aggregate economic growth. Whereas earlier work was concerned primarily with urban growth and decline, newer studies have focused on the macroeconomic effects of the spatial organization of activities. Much of the work emphasizes the role of external economies of scale, meaning that the larger an economic area is, the more productive will be the firms there, because they benefit from the so-called “agglomeration economies” in densely developed areas. Thus where firms locate affects the size of agglomeration benefits available in particular areas, which in turn affects the productivity and location choice of other firms. Empirical evidence from increasingly sophisticated statistical models has demonstrated that, in the words of John Quigley, of the University of California at Berkeley, “increased size of cities and their diversity are strongly associated with increased output, productivity, and growth.”

As Quigley notes, both the scale and the variety of activities in particular geographic areas have been found to increase productivity. And the gains may be of several sorts. Spatially concentrated employers may benefit by sharing access to a dense labor market. If one employee quits, another may be quickly found when many workers live within commuting distance. Likewise, the market for productive inputs will be more efficient when it is “thicker”?that is, when suppliers and demanders of a variety of intermediate goods and services are numerous. Firms may also benefit when they can share the cost of publicly provided inputs, especially big-ticket items like infrastructure. Finally, valuable information about markets, products, and productive processes may simply be “in the air” and more available in dense environments than in less developed places. Effects of this kind help explain the concentration of high-tech firms in Silicon Valley as well as the concentration of financial services jobs in central business districts around the country.

While agglomeration economies are thought to be important contributors to growth, less is known about the scale over which they operate. Some authors have focused on metropolitan areas, while others have looked at smaller geographic units. One recent paper in the American Economic Review, for example, analyzed aggregate state productivity as a function of employment density. The authors found that a doubling of their county density index (which weights the densest counties most heavily) is associated with roughly a 6 percent increase in aggregate state productivity. Together, these studies make a compelling economic case for fostering the development of our densest and most diverse employment centers?commonly known as cities.

But the interactions among firms arising from “agglomeration” are reciprocal and are not directly priced in markets. Individual firms will locate where their profits are highest, without regard to the effects that their move has on other firms. The fact that the primary function of state capital investment turns out to be subsidizing the spreading out of jobs thus means that it has the potential to reduce productivity. And a more dispersed employment pattern is closely associated with a more dispersed residential pattern, which also leads to the problems of socioeconomic segregation and costly suburban residential growth.

Thus infrastructure is a seeming contradiction: it is a productive good whose aggregate productivity may be negative. Because it is productive, it draws activities toward it. But because it is primarily placed in low-density areas, it undercuts density and thereby reduces overall productivity. The real culprit here is not public investment itself, but the political process that puts it where it is least valuable. Reduced national productivity growth makes us all worse off.

So even if infrastructure dollars were free, it would be costly to invest them the way we do. Because infrastructure is instead very expensive, it is even more clear that our public investment policies need reform. Indeed, from the point of view of most urban residents, it would be better to burn the money than to spend it the way we do. Luckily, these are not our only alternatives.

What Should We Do?

Were states to get out of the business of capital investment, cities’ job bases, and by extension national productivity, might be larger than they are under the current arrangement. But states are not likely to do so, especially because they have substantial revenue sources earmarked for particular kinds of investment. As an alternative, states might begin to consider more carefully how their investment policies affect the spatial distribution of jobs and whether the patterns that emerge are consistent with their economic growth objectives. States that wish to use their public capital money as part of a strategic economic development effort will direct more resources to central cities and other locales that have high concentrations of jobs and avoid building new projects in the greenfields on the edges of metro areas. The result will be a more densely developed state, with more moderate growth in the suburbs, lower taxes, and faster economic growth. That’s a record any governor would be happy to run on.