Bob Poole is the Director of Transportation Studies at the Reason Foundation. Here is a bit from his latest newsletter:

The last time this country had a major debate over the role of infrastructure in the economy was in the late 1980s and early 1990s. One of the main voices in that debate was David Aschauer, who made the controversial claim that public capital investment was more economically productive than capital investment by the private sector (e.g., in new facilities, computers, etc.). Many scholars took issue with his thesis, and indeed the decade of the 1990s witnessed two key trends: an explosion in private-sector productivity thanks to the full integration of computers and the internet into business practice and the continued decline in the productivity of highway investment.
A paper from the end of that era was cited recently in congressional testimony by the National Association of Manufacturers. In 1996 Ishaq Nadiri and Theofanis Mamuneas did econometric modeling to assess the “Contribution of Highway Capital to Industry and National Productivity Growth,” funded by the Federal Highway Administration. Their September 1996 paper found that the level of highway capital (and especially of National Highway System capital) does contribute significantly to economic growth and productivity—but at a much smaller level than in previous studies (such as Aschauer’s). And they found that the impact of private capital investment (in industry) is about four times that of highway capital investment, again, contra Aschauer. But what I found especially interesting was their finding about the marginal benefit of highway investment. Here, they found that while it is positive overall, its main benefit is to manufacturing industries (and is actually negative for some service industries). Moreover, like most of the literature, they found that the net social rate of return on highway capital investment was about 35% in the 1950s and 1960s, when the Interstate system was being constructed, but declined to around 10% in the 1980s and since then (up to the mid-1990s, when they were writing this) has converged to about the long-term rate of interest.
As I’ve written elsewhere about similar findings, I don’t agree that this means there is no case for further investment in highway infrastructure. What today’s single-digit return on investment says is that much of what we spend on highways is not targeted to projects that can make a big difference—such as widening truck-clogged Interstate corridors and adding congestion-priced express lanes to gridlocked freeways. Instead, most of it funds lots of small stuff that keeps the wheels turning (and people employed) but doesn’t add much, if anything, to economic productivity.
And that is why I can’t get excited about transportation stimulus bills. Or about significantly increasing the size of the federal program in the next reauthorization bill—without fundamentally reforming how we decide to invest those funds.