Does Highway Spending Really Pay Off?
Highway spending is meant to be a public investment in the nation's infrastructure that pays off for everyone. Does it?
The New York Times, "Economic Scene" , May 19, 2004
Congress and the White House are wrangling over how much the federal government should spend on highways.
The White House proposed a $256 billion budget for roads, bridges and transit over the next six years. (The previous six-year measure, which expired last year, cost $218 billion.) The House passed a $284 billion version, while the Senate further upped the ante to $318 billion. The White House threatens to veto those amounts as too lavish.
To political analysts, the highway bill is a popular program that has fallen prey to partisan, and intraparty, bickering. Lost in the dispute is the economic question: What are we getting for our money?
Unlike such transfer programs as Medicare or Social Security, which redistribute existing income, the money the government spends on infrastructure isn't supposed to simply shift resources from one group to another. It's meant as public investment: spending that enhances productivity and thus increases the country's future income, making everyone better off.
The returns on those investments don't come back to the Treasury directly, the way private-sector returns do to investors. But in theory, infrastructure investments benefit taxpayers indirectly by increasing the nation's wealth.
How effective is the investment?
In an article in the March issue of The Journal of Urban Economics, two economists look at exactly how highway spending increases productivity -- by lowering businesses' inventory and logistics costs -- and calculate how the returns on highway spending have changed over time.
To make that calculation, Chad Shirley of the RAND Corporation and Clifford Winston of the Brookings Institution used census data on the inventory levels at 50,000 to 75,000 individual plants from 1973 to 1996. They looked at how infrastructure investment, both within each plant's state and across state lines, affected those costs, holding constant other influences like interest rates and changing inventory practices.
The results are striking. Infrastructure spending does indeed lower inventory and logistics costs, increasing productivity. But the rate of return plummeted over time, from more than 15 percent in the 1970's to less than 5 percent in the 1980's and 1990's. (These figures are corrected for inflation.)
There's a logical reason for those diminishing returns.
''By the late 1970's, the Interstate highway system was substantially completed,'' the economists write. ''During the past two decades, the primary objective of highway spending has shifted from expanding the nation's capital stock to maintaining it. Undoubtedly, the improvement in costs and service from such investments and the concomitant reduction in plants' inventories cannot compare with those produced by the construction of thousands of miles of new roads.''
The changing value of new infrastructure investment, Dr. Winston explains in an interview, helps to reconcile two distinct lines of economic research.
Macroeconomists have emphasized the returns of government infrastructure spending, arguing that public investment is important to spur economic productivity.
Transportation economists, meanwhile, have looked at the specific details of the system: how roads are paid for, where they are built, what trade-offs are made between upfront construction costs and maintenance, whether road users pay the full costs they incur, and so forth.
This research has ''consistently found very poor performance -- lots of inefficiencies,'' Dr. Winston said. ''The stuff is mispriced. The stuff is improperly built. There's a huge amount of waste.''
How could infrastructure spending be both productive and inefficient?
One possible explanation is that as the system has matured, the inefficiencies have become more important, dragging down the returns on new spending.
Consider the choice between the immediate cost of building thicker roads in the first place and the long-term cost of repairing thinner roads as they wear down. An economic calculation would have suggested much thicker Interstate highways, even ignoring the cost of disrupting traffic with repairs.
But the Interstates were built relatively thin, in part because of political pressures to get the system spread everywhere as quickly as possible. Now they're wearing out, and money must go for repairs. That investment, while necessary, is inefficient and yields a relatively low return.
At the same time, Dr. Winston notes, the absolute amount of highway spending has gone up. Even good returns tend to diminish from the first dollar to the last.
''Even if the possibility for getting returns had not changed since the 70's, you're going to get lower rates of return just because you're pumping more money into the system,'' he said.
Of course, drivers might not care much about economic returns if highway spending reduced congestion. Wasted time isn't counted in government productivity statistics, but it's still costly.
Unfortunately, a new unpublished working paper by Dr. Winston and a Brookings colleague, Ashley Langer, suggests that highway spending isn't doing much to reduce congestion costs either.
On average, they write, ''one dollar of annual highway spending reduces the annual congestion costs to road users only 8 cents.'' This is not a return on a one-time investment but a continuing expense; we have to keep spending that dollar to get the 8 cents.
Dr. Winston cautions that the specific number may change as the working paper is revised, but that the general point -- a very small effect -- seems certain.
While the papers attack the problem in different ways, Dr. Winston said: ''The touchstone of the whole enterprise is trying to say, What are we getting for this spending? The returns on this are really low.''
But, he acknowledges, neither Congress nor the White House seems terribly interested in the economic questions. ''It's tough to get the debate centered on the issues we're talking about,'' he said.