Measuring Infrastructure Investment in the United States

09/01/2020
Featured in print Digest
Working Paper Figure w27677

Real net investment per capita has drifted downward since the financial crisis and stands near its lowest level since 1983.

Infrastructure provides critical support for economic activity, and assessing its role requires reliable measures. In the working paper Measuring Infrastructure in BEA’s National Economic Accounts (NBER Working Paper 27446), researchers Jennifer Bennett, Robert Kornfeld, Daniel Sichel, and David Wasshausen assess trends in basic, social, and digital infrastructure in the national economic accounts and develop new prototype measures for maintenance expenditures and state-level highway investment.

Real gross infrastructure investment, measured in 2012 prices, peaked at $340 billion in 1968, declined slightly during the next decade and a half, and then grew to reach nearly $800 billion in 2017. These overall figures show the resources devoted to infrastructure each year, though they mask differences across categories. Investment in basic infrastructure, which includes the nation’s roads, bridges, power grids, dams and so on, follows a similar path until the early 2000s. Since then it has been largely flat, while the share of infrastructure investment in other areas has increased. These areas include social infrastructure, encompassing schools, hospitals, and public safety, as well as digital infrastructure.

Digital infrastructure, the fastest-growing area, includes cell towers, computers, communications equipment, software owned by companies in broadcast and telecom, data processing, internet publishing, and information services. Investment in this area rose about tenfold from the 1980s to 2017.

Real net infrastructure investment per capita, which adjusts gross investment for population growth and depreciation, highlights whether infrastructure investment is keeping up with wear and tear as well as population growth. Here, basic infrastructure has drifted down during the last decade and currently stands near its lowest level since 1983. Moreover, the remaining useful life of many types of basic infrastructure has moved lower in recent decades.

Accurate measurement of both net investment and the stock of infrastructure requires an estimate of how quickly a given piece of infrastructure capital wears out. The depreciation profiles that are used to calculate the net capital stock of infrastructure in the United States date back nearly 40 years and are well below those used in Canada and several European nations.

Another vexing challenge is distinguishing between major infrastructure upgrades, which count as investment, and maintenance, which doesn’t. New prototype measures for highways indicate that spending on maintenance represents about 15 percent of gross investment.

The researchers note that there is substantial variation in infrastructure investment across regions. For real gross highway investment per capita, for example, upper Midwest and North Central states are in the top quintile in the last decade, while many Western states such as Arizona, California, Colorado, Oregon, and Utah fall in the bottom 40 percent.

—Laurent Belsie