Public infrastructure investment, such as spending on roads, bridges, and other such projects, is one of the most advertised tools of anti-recessionary fiscal policy. Donald Trump, all through his candidacy to now the presidency, is looking to push for a gigantic $1.7 trillion infrastructure plan. Why? When the economy struggles, politicians and public economists call for greater infrastructure spending as a form of stimulus, especially when the spending takes place in their district or state. Despite the ubiquitous presence of infrastructure-as-stimulus policy proposals, there is little practical evidence that public infrastructure projects are a net positive to the economy, or that they even boost net employment figures. There appears to be a disconnect between political rhetoric, political theory, and economic reality.
Theory of Infrastructure Stimulus
Government stimulus spending, whether on infrastructure or other goods and services, is predicated on the Keynesian assumption that an underproductive economy can be spurred back to full output by using new public expenditures to boost aggregate demand. Specifically, as it relates to infrastructure, the belief is that involuntarily unemployed persons can be given public infrastructure jobs and receive an income which, to the extent that it is spent quickly, promotes even more growth.
Going further, Keynesian stimulus spending assumes little or zero opportunity costs if the deficit spending occurs during a period of higher-than-normal unemployment. In fact, John Maynard Keynes prophesied that public infrastructure deficit spending could produce a multiplier effect on economic growth. This especially should be true when real interest rates are low.
Problems With Theoretical Infrastructure Spending
One chief problem with the theory of infrastructure spending is that it ignores so-called "Cantillon effects" for the relative change in different prices as the result of new money entering the economy. Since new spending increases prices and demand in some areas faster and more deeply than in other areas, it has the side effect of misdirecting production away from areas where private citizens might voluntarily choose to dedicate their money. Essentially, the economy trades off a short-term reduction in unemployment for a long-term misallocation that produces higher unemployment.
Contrary to what the original theory stipulates, there are likely very large opportunity costs and implementation costs associated with infrastructure spending. Since governments do not produce anything with a calculable market value because their revenues, or taxes, are independent of consumer valuations and therefore blind to any real economic feedback, there is almost no way to know if general infrastructure spending is the best use of resources, let alone any specific project for a road, bridge or highway. It is far more likely that resources are put to more productive use if made through private voluntary transactions because of the effective feedback loop inherent to markets.
To the extent that infrastructure projects are financed through immediate taxes, the private economy immediately shrinks by at least a corresponding amount. If they are financed through government bonds, then current capital markets experience crowding-out effects and other financial assets become more or less expensive than they otherwise should be. Later on, when those government bonds are paid back through higher taxes or higher inflation, the private economy loses again.
Economics, as a science, struggles to produce convincing empirical results. It is difficult to find solid, demonstrable evidence about how effective changes in infrastructure spending have been. In a 2014 working paper for the International Monetary Fund (IMF), economist Andrew M. Warner found little evidence that global infrastructure projects produced economic gains. Even when projects received credit for growth, Warner found that the economy was already improving at a similar rate by the time construction began.
It should also be noted that the government generally is not excellent at managing money or roads. Federal spending for highways is as much a political tool as an economic one, and states that do not comply with federal mandates often have their infrastructure money held as ransom. Projects also tend to lose their "shovel-ready" status because of lengthy and expensive environmental and permitting reviews. Approvals for public infrastructure projects can take between five and 10 years to be implemented, all the while costing taxpayers as tedious approval processes play out.
President Trump has made no secret of his desire to address flagging infrastructure in America, and in January of 2017 told a gather of major city mayors that the admin plans to "invest around $1.7 trillion in infrastructure." One of Trump's main arguments on the 2016 campaign was that he would fix crumbling infrastructure, and it remains to be seen if his admin will be able to deliver of these lofty promises.
Other Practical Challenges
In 2013, the National Bureau of Economic Research (NBER) and the Federal Reserve Bank of San Francisco published a paper entitled, "Roads to Prosperity or Bridges to Nowhere? Theory and Evidence on the Impact of Public Infrastructure Investment." Therein, economists identified at least four challenges to standard infrastructure-as-stimulus theory: the endogeneity of public infrastructure spending to economic conditions, the decentralized nature of implementation, lags between approved spending decisions and actual project completion, and a high degree of public awareness leading to anticipatory effects.
There are other serious challenges built into the model used in the NBER/Fed paper. Consider the type of theoretical economy described during their analysis: "we consider a cashless national economy consisting of two regions" of "possibly different sizes" where "each region specializes in one type of tradable good" and "firms are monopolistic suppliers."
These are consistent themes in almost all macroeconomic forecasts. Much of what makes a real economy function is assumed away to simplify models enough to produce testable and predictable results. The original theory of public infrastructure spending was even less sophisticated than the NBER/Fed version. It should not be surprising that practical reality, so different than the parameters of macroeconomic models, produces different results.