As Congress considers ways to stimulate the economy in the wake of coronavirus, many have called for major infrastructure investments, because it purportedly drives long-term economic growth. The World Economic Forum writes, “infrastructure creates economic growth.” Paul Krugman argues that “if the investment is productive, it will expand the economy’s productive capacity in the long run. This is obviously true for physical infrastructure.” Investment manager Steve Rattner, writes, “An infrastructure initiative would not provide quick relief, but it would support stronger growth in the future.”
But is this really true, especially when compared to other kinds of public investments that would also provide short-term stimulus? To be sure, spending money on pretty much anything will spur growth in a recession if it is debt-financed. As Keynes famously said, putting Treasury notes in old wine bottles and burying them in abandoned mines would create jobs as people dug up the bottles.
But assuming that the “shovel readiness” of investments are the same, policymakers are better off investing in areas that also boost long-term economic growth.
It has been an article of faith for decades that traditional physical infrastructure—concrete and steel—boosts long-term growth, but evidence suggests that the growth benefits are limited when compared to other areas, especially 21st century digital infrastructure.
Any infrastructure package is likely to invest heavily in roads and bridges. A widely-cited 1989 paper found that increasing traditional infrastructure investments in projects such as roads, transmission lines, and bridges by 1% increases productivity by 0.23%. But as the U.S. interstate system has become built out, the efficiency of these investments have declined. One study found that U.S. highway investments generated annual total economic returns of 18% in the 1970s, 5% in the 1980s, and just 1% in the 1990s. Similarly, a study of European transportation investment from 1995 to 2009 found insignificant returns to investment in motorways and a negative effect on GDP of transportation maintenance spending, even controlling for government quality. Most starkly, a study found “trivial” benefits of $850,000 per mile of roadway lane over 20 years in the United States.
In contrast, investments in digital infrastructure can generate greater overall economic returns. These include both dedicated digital infrastructure (infrastructure that is innately digital, such as broadband, 5G, cloud computing centers) and hybrid infrastructure (adding digital components to traditional infrastructure, such as smart meters, smart grid, and smart cities). For example, researchers estimate that if the rest of the European Union built out its digital infrastructure to the level Norway achieved in 2011, it would increase GDP by $315 billion.
Many benefits of information technology (IT) are intangible and not captured by standard statistics, with one study estimating that productivity is 16 percent higher than officially reported due to intangible gains from IT investments. Despite that, a meta-analysis last year found that 52 of 64 studies on the firm-level effects of IT investments since 2000 showed significant positive impacts on productivity. For example, researchers found that U.S. multinationals increased their productivity by 5 percent for every doubling of IT capital.
At the municipal level, digital infrastructure will form the foundation of smart cities, allowing services to be provided more broadly and cheaply. Equipping garbage cans with sensors backed by machine learning improved fuel efficiency by 46% and collection time by 18%. Smart lighting can reduce power consumption of streetlights by 38 to 60%.
Why does physical infrastructure then continue to get such attention? One reason is the inability to consider other kinds of public investment. Government can build roads, so government should build roads, goes the thinking, but this should not prevent consideration of other opportunities.
Many believe that infrastructure spending provides good jobs for non-college educated workers. But while 80 percent of workers in the construction industry do not have a college degree, according to the Department of Labor, the median annual salary is only $38,835. In contrast, the median annual salary for non-college educated workers in the wired telecommunications carriers (broadband companies) is $57,266—47 percent higher.
This does not imply that physical infrastructure should be ignored. There are potential projects that can have big payoffs, but a stimulus bill should avoid the notion that massive investment in infrastructure will pay long-term economic dividends. It is also worth remembering why our roads and bridges have gotten so bad. Congress has refused to increase the gas tax since 1993. If Congress wants to fix roads, the best way, as the National Surface Transportation Infrastructure Financing Commission recommended, is to raise the gas tax, enable more tolling, and move to a vehicle miles traveled payment system.
So, as Congress considers an infrastructure investment package, here are several ideas that warrant attention on the digital front:
- Fund a one-time, large-scale injection of capital for broadband infrastructure in areas of the country where it is too costly for private providers to serve, and attempt to transition away from recurring annual support.
- Provide a tax credit for all capital expenditures directly related to 5G investment between now and the end of 2021.
- Fund the FCC’s Lifeline program to expand and improve subsidized broadband options, including tablets and computers, for low-income users.
- Fund a nationwide smart cities program to help cities and towns use digital technologies to improve operations and improve quality of life.
Caleb Foote is a research assistant at the Information Technology and Innovation Foundation (ITIF), a nonprofit, nonpartisan think tank, and a graduate of Brown University with a concentration in economics.
Robert D. Atkinson is founder president of ITIF.
This article originally appeared in the ITIF blog. It is used with permission..
