Friday, August 12, 2011

America's Infrastructure Problem: Engineering vs. Economic Perspectives

Once every four years, the American Society of Civil Engineers publishes a "Report Card for America’s
Infrastructure," which gives typically low letter grades to various categories of U.S. infrastructure--and thus helps make the case for hiring more members of the American Society of Civil Engineers. Now ASCE has published a study of the economic costs of underinvestment in infrastructure (available here with free registration), called Failure to Act: The Economic Impact of Current Investment Trends in Surface Transportation Infrastructure.

My quick summary of the ASCE work in this area: Love the engineering approach to putting costs on U.S. infrastructure issues, but suspicious about the economic analysis.

Here are some examples of what I mean by the engineering approach of putting costs on infrastructure, from the report: "Investment of roughly $220 billion annually (2010 dollars) is needed from 2010 to 2040, based on unit costs, minimum tolerable conditions,  and data sources consistent with current application of federal highway, bridge, and transit investment models. This breaks down to an average investment of approximately $196 billion per year for highway pavements and bridges, including $161 billion for congestion mitigation and $35 billion for preservation of existing facilities. In addition, $25 billion per year in transit capital infrastructure investment (including rolling stock as well as trackage, terminals, and roadways for access) is needed. Approximately 37% of this highway and bridge investment and 25% of this transit investment will be needed simply to resolve existing deficiencies of almost $74 billion that are already affecting the U .S. economy. The remainder is needed to prevent deficiencies from recurring or getting worse over time....If present trends continue, the funding gap for rail and bus transit, seen as 41% in 2010, is expected to increase to 55% by 2040. The expected gap in highway funding, 48% in 2010, is expected to increase to 54% by 2040."

The report has useful estimates of what share of American surface travel happens on unimproved roads, or the costs for bridges that need repairing. " The United States carries a backlog of $3 trillion in unfunded
surface transportation needs, including a $2.2 trillion backlog for highways and bridges and $86 billion in unfunded transit capital infrastructure needs. This backlog does not include the cost of providing access to transit to the significant number of Americans who do not currently have access to fixed route transit
and/or demand response transit. Approximately 15% of transit revenue miles occurring in 2010 are on vehicles with a state of good repair of “fair” or “poor.” In addition, 31% of passenger car vehicle miles of travel occurred on roadways with less than minimum tolerable pavement conditions and 18% of passenger
car trips occurred on congested roads. By 2040, the proportion of transit revenue miles occurring on less than “good” vehicles will rise to 33%, and the 18% of passenger car VMT traveled in congested conditions will rise to 36%."

There are also estimates of some of the costs of infrastructure underinvestment: "In 2010, it was estimated that deficiencies in America’s surface transportation systems cost households and businesses nearly $130 billion. This included approximately $97 billion in vehicle operating costs, $32 billion in travel time delays, $1.2 billion in safety costs and $590 million in environmental costs. ... If present trends continue, by 2020 the annual costs imposed on the U.S. economy by deteriorating infrastructure will increase by 82% to $210 billion, and by 2040 the costs will have increased by 351% to $520 billion."

So far, so good. I don't know the underlying calculations or data sources here in any detail, but these are roughly the kinds of numbers one sees in various reports. But when the ASCE turns to longer-term economic projections, things get a little odd. Here are three examples:

1) A few days after I'd downloaded my copy of the report, I got a follow-up e-mail, which read in part: "The original report dramatically underreported the negative effect of Americans' personal income due to failing transportation infrastructure. ... Our original release projected that Americans' personal income would drop by $930 billion by 2020 but recover slightly in 2040.  The data clearly show that the effects will be dramatically more negative, with $3.1 trillion in personal income losses by 2040.  The negative effects on American GDP will also expand dramatically over time, with a near-term loss of $897 billion and a near-tripling of that loss to $2.6 trillion by 2040." I know nothing about what might have gone on behind the scenes here. Maybe someone just made an enormous error in calculations and no one else caught it. But my quick cynical reaction is that someone wasn't happy that the report didn't show big enough losses by 2040, and so the numbers got tweaked.

2) The report estimates that better infrastructure will add 400,000 jobs to the total level of U.S. employment in 2040. Every credible model I know of unemployment in the very long-term suggests that the economy, in one way or another, rebounds back to a "natural" level of unemployment. One can make a solid case that appropriate infrastructure investment helps productivity and thus wages. But I don't know of any theory which suggests that infrastructure makes the total number of jobs higher in the long-run.

3) The report has some weirdly detailed aspects, like how about 150,000 new jobs could be created in the entertainment industry by 2040 with better infrastructure, or how poor infrastructure by 2040 could reduce exports of meat products by $1.2 billion. Nobody has a crystal ball for predictions 30 years in the future, of course, but this kind of odd detail feels artificial and untrustworthy.

But the biggest problem I have with the economics of these ASCE reports is that the "costs" they estimate mostly seem to assume, as you might expect from a group of civil engineers, that the answer to infrastructure problems is always a matter of pouring more concrete. For example, consider the estimate above that we  need to spend $161 billion per year on infrastructure for "congestion mitigation." To an economist, the immediate reaction is to think about congestion pricing or tolls as at least a partial solution to this issue. Or consider the costs of environmental pollution from transportation. To an economist, an obvious answer is to impose pollution fees or some kind of cap-and-trade structure as a tool for reducing that pollution. Or on the issue of road wear-and-tear, an obvious answer for economists is to point out that heavy trucks and buses impose by far the most of the wear-and-tear costs, and charging taxes based on axle weights would be a natural way to provide incentives to spread out these loads while raising money for road repair.

Thus, I agree with the need for substantially greater investment in infrastructure. I live near Minneapolis, where the most-travelled bridge in the state collapsed just four years ago in 2007. For example, I would favor a major effort to improve right-of-ways for freight rail, with depots outside major metro areas for transshipping to trucks. I favor infrastructure spending targeted at the worst bottleneck sites for traffic jams, to clear up those local messes. But I would start my infrastructure planning by using congestion charges, pollution charges, and axle-weight charges, along with getting freight out of trucks and on to rail, and then see what else is needed. I suspect that the level and shape of infrastructure spending that comes out of this approach could turn out to be substantially different than the ASCE approach.

For some alternative takes on the economics of infrastructure issues, often complementary to the ASCE approach, here are some starting points: