Glaeser Goes Out With a Whimper

For those just tuning in, economist Ed Glaeser has been writing a four-part series on the potential costs and benefits of high-speed rail at the New York Times' Economix blog. He began three weeks ago with an introduction. The following week he addressed direct costs and benefits from a hypothetical line, and last week he attempted to gauge the environmental benefits of high-speed rail construction.

(Image: Treehugger.com)

The whole of the analysis has been highly flawed (see my earlier criticisms here and here). It is overly simplistic, excludes important variables, and relies on faulty assumptions.

Glaeser has quite nearly admitted as much, arguing that he chose the hypothetical Dallas-Houston route -- which is not in the administration's plan for a high-speed network -- in order "to avoid giving the impression that this
back-of-the-envelope calculation represents a complete evaluation of
any actual proposed route."

But it doesn't take a close read to see that Glaeser wishes to demonstrate that rail investments do not make economic sense. He has not been particularly charitable in acknowledging the shortcomings of his work, and he has therefore left his readers with a very misleading picture of the probable outcome of construction of a high-speed rail system.

I have continued to hold out hope that he'll improve his analysis along the way, but as of today we have the final chapter -- on high-speed rail's potential reshaping of the American economy -- and it, too, is embarrassingly bad.

In his earlier posts, Glaeser did not take population growth into account -- a rather large failing while analyzing a piece of infrastructure we can expect to last for decades. This time around he aims to defuse this criticism by writing:

It's worth recalling where those numbers come from. To arrive at 1.5 million, Glaeser took the current population of the Dallas and Houston metro areas and generated a ridership estimate using the population-to-ridership ratio for the Northeastern Corridor and then dividing by two (to take into account the lower density of the Texas cities).

Looking at the numbers that way, it becomes clear that you don't need anything crazy to happen to get to three million riders. Take expected population growth and partial convergence toward rates of transit ridership seen in the Northeast (or in California, for that matter) and you get there in no time.

Why might rates of transit ridership increase? Both Dallas and Houston are rapidly adding to their transit networks. One might also take into account demographic changes and expected changes in energy and congestion costs, but Glaeser pretends these matters are of no importance and doesn't bother to explain why he has opted to omit them from his analysis.

And as I mentioned before, Glaeser does a poor job estimating direct benefits to rail riders. Using a standard business airline fare and the time saved by flying rather than driving, one gets a more accurate picture of the value of a likely rider's time. Using that more accurate figure, rail's benefits roughly equal its costs even if one accepts Glaeser's 1.5 million rider estimate.

Continuing, Glaeser says high-speed rail investments probably won't make for good stimulus, given that construction wouldn't take place quickly. As I've pointed out before, that's all well and good if you believe that the American economy will be at full employment within a year or so. If you think, as Fed economists do, that full employment may not return for a decade or more, then the stimulus argument in favor of rail grows quite a bit stronger.

What about the potential for high-speed rail to revitalize struggling regions? Glaeser notes that there are good historical examples of infrastructure investment boosting regional economic fortunes, but, he says, these may not apply to the case of high-speed rail. Why not?

Well, because Buffalo is too far from New York to bring any significant competitive advantage over flying. And Philadelphia should be a beneficiary of the Acela, he says, but falling population in that city suggests that it isn't.

There are many things wrong with this argumentation. One is that Glaeser's own method for comparing rail versus flight times shows that rail from Buffalo to New York City still produces a nice time advantage. Take a 90 minute flight time, add the hour early one has to arrive at the airport and 36 minutes of travel time to and from airports, and you get a little over three hours for flying to two and a half for the train. That's not nothing.

Glaeser also waves off the utility of a Toronto to Buffalo line based on the inconvenience of crossing the border, but flyers and drivers must also deal with a border delay, so it's not clear why that's a big deal.

Meanwhile, the blithe use of population change in Philadelphia as a proxy for economic benefit is a little silly. For one thing, it would seem to ignore actual trends. Since 2000, the rate of population decline in the city of Philadelphia has sharply diminished.

From 2000 to 2001, the city's population declined by 15,000. From 2003 to 2004, by contrast, population fell by just over 7,000. And from 2007 to 2008, Philadelphia lost a mere 1,200 people.

Just using Glaeser's fly-by-night statistical methods, it seems as though the introduction of the Acela has in fact materially slowed population decline in Philadelphia. And obviously there are other variables which show that Philadelphia has enjoyed a serious economic rebound over the last decade.

A bigger failure is that Glaeser doesn't consider the value of bringing Buffalo closer together with other Midwestern metropolitan areas. The value of economic interactions between metropolitan economies is a function of the size of the urban economies and the distance between them.

Lowering the time and cost of intercity travel between struggling Midwestern cities should effectively allow those places to lever up the size of their metropolitan economies. Intercity commerce should increase, leading to improved economic conditions.

Glaeser next turns to whether rail investment might change land-use patterns within cities. He says that there is unlikely to be much in the way of a shift from suburb to city, citing this paper by Nathaniel Baum-Snow and Matthew Kahn.

They find that transit use declined from 1970 to 2000, largely due to suburbanization. Millions of people moved from places where there was transit to places where there wasn't transit, and so transit's share of trips declined.

What does this tell us about transit's ability to guide land use? I suppose it says that the presence of transit can't reverse a wave of depopulation driven forward by failing city finances, desegregation, urban riots, and so on. I suppose it says that if people are moving to suburbs and suburbs are overwhelmingly served by road infrastructure, then transit's mode share will fall.

But what it also suggests is that land use isn't simply an endogenous factor responding to changes in household tastes and wealth. Rather, it's a function of deliberate government policy.

If one builds a transit system and surrounds the stations with parking, then no, transit will not do very much to shift land uses. If one builds a rail line between cities that do not allow dense, mixed-use development patterns, well then those patterns won't emerge, it's safe to say.

So an important question to ask is how land use patterns have shifted after construction of rail or transit, in places where governments have facilitated a shift. And I think there can be little question that the answer to such a question is very different than the one arrived at by Mr Baum-Snow and Mr Kahn.

This is also a message that has resonated with many city governments in recent years. Increasingly, changes in land-use are being specifically targeted as a potential benefit from new infrastructure construction. Governments are learning. Economists may not be.

At any rate, Glaeser reveals the extent of his short-sightedness by concluding (charitably, in his view) that rail might lead 100,000 suburbanites to move to the city in both Dallas and Houston, generating some $30 million in gains.

But this is all wrong. The question is not whether high-speed rail will lead some number of suburban residents to pick up and move to the city (although demographic shifts suggest that is likely to happen).

Dallas and Houston will each add a million or more people in the next two decades. America will probably grow by 130 million people between now and 2050. These people will, in all likelihood, live somewhere.

Where they live will be a function of the infrastructure that we build. I suspect that even Glaeser will agree that had the United States developed the Eisenhower high-speed rail network rather than the Eisenhower interstate highway system, our urban geography would be significantly different today.

If we build no new transportation capacity, then existing infrastructure will rapidly become choked, adding to the potential benefits to be had from building rail.

If we instead build new highway and airport capacity, then that will influence future development patterns and mode share. I challenge Glaeser to demonstrate that that future is greener and better off economically than one in which rail is built.

This is the principle shortcoming of Glaeser's analysis -- that it fails to take into consideration the alternatives.

I believe that increasing metropolitan congestion, rising energy costs, changing demographics, and new transit investments will generate a shift in housing and transportation preferences in coming decades. I think it's wise to accommodate this shift by building high-speed rail.

Glaeser seems to believe that in coming decades congestion costs will cease rising; otherwise he'd build future increases into his model. He seems to think that the addition of over 100 million new Americans need not lead to any new infrastructure investment; otherwise he'd compare the economic benefits and life-cycle emissions of rail investments to alternative investment plans.

I think those beliefs are daft and indefensible. And four posts into his high-speed rail series, Glaeser hasn't given any of us reason to think that his analysis is worth taking seriously.