There’s been a great back and forth over the past few days on high speed rail, most of it stemming from Ed Glaeser’s flawed cost-benefit analysis, and Yonah Freemark’s counter-analysis.

First, Tyler Cowen weighs in on Yonah’s HSR analysis.

Don’t do it.  Given irreversible investment, lock-in effects, and required hurdle rates of return, this still falls into the “no” category.  And that’s an estimate from an advocate writing a polemic on behalf of the idea.  I’m not even considering the likelihood of inflation on the cost side or the public choice problems with getting a good rather than a bad version of the project.  How well has the Northeast corridor been run?

…

General remark: It’s not about population density per se.  It’s about how many independent, hard-to-connect nodes the system has and that is why high-speed rail on the whole works better in Europe or Japan than in many other locales.  To give an example from a slightly different realm, I live right near the Metro in a high-density suburban area.  Yet I don’t take the Metro to my Arlington office, which is about two minutes from a Metro stop.  I’d rather do the 37-minute drive.  Why?  Because I stop at the supermarket and the public library on my way home at least half of the time or maybe I stop to eat at Thai Thai.  If those conveniences were right next to my house I’d consider the Metro but they’re not.

Ryan Avent responds, asking why libertarians hate trains.  Matt Yglesias chimes in on that, as well – adding a critique of zoning laws to the equation.  Zoning is likely the reason Mr. Cowen doesn’t have Thai Thai near his local Metro stop (and why there isn’t more stuff around that Metro in the first place).

At this time, Will Wilkinson enters the fray – saying that the reasons libertarians don’t care about zoning (and don’t put up a fight to right the wrongs of past highway subsidies, thus holding new HSR systems to a higher standard) is basically a status quo bias – that we’ve set out on a dependent path.

What makes this issue so tricky for me is that the status quo pattern of settlement and transportation certainly does reflect systematic regulatory mandates, but it’s not clear how worthwhile it is to try to back out of this pattern once it has been established — even if those mandates were stupid. The way we live is indeed very much a function of choices made by government some time ago and reinforced by its ongoing decisions to maintain the established system. I think the case for the proposition that many of these choices were big mistakes — that we’d have an overall better pattern of settlement and transportation had government made different choices — is pretty compelling. Yet it remains that whole cities have formed around the suboptimal status quo system and many tens of millions of people have invested in goods like houses and cars taking for granted the structure of the status quo system.

I love the smell of the the sunk cost fallacy in the morning.  While what Wilkinson writes here is true, it’s also largely irrelevant.  Ryan notes the reasons why this fallacy isn’t worth perpetuating:

Libertarians, for some reason I haven’t yet grasped, seem to view the world as remarkably static. In their world, population is not growing. New entrants to the workforce aren’t choosing where and how to live. Demographics are etched in stone; the population isn’t getting older and embracing smaller family sizes. And people never, ever move house. Libertarians also seem to like the “newspaper commenter” critique of urbanist arguments: “Why do you want to make everyone live in Manhattan?” There’s no such thing as a shift at the margin in this view.

In practice, the US is far from done building. Tens of millions of new homes will be built in the coming decades. Hundreds of billions of dollars will be spent on transportation infrastructure. The current built environment has, as a result of decades of government policy, taken on a rather suburban, auto-centric tilt. So what? No one is suggesting that we tear down all of that and replace it with something entirely new. I, and others, are suggesting that making it easier (or, you know, legal) to build in a denser, more walkable fashion would be advantageous. Similarly, given the burden of maintaining such a large and costly road infrastructure, it might be wise to devote a larger share of dollars for new construction to substitute technologies.

Market Urbanism doesn’t agree, or doesn’t seem bothered by the sunk cost fallacy:

So, in other words, building either of the options, roads or rail both require “a large environmental sacrifice”, but all other options must be kept off the table, so let’s just sweep that under the rug.  Yet, there is an other option to consider for those who really think something should be done about carbon: STOP WASTING MATERIAL AND ENERGY ON CONSTRUCTION OF INFRASTRUCTURE BOONDOGGLES THAT SUBSIDIZE TRANSPORTATION!  That still goes double for roads and airports, where congestion and carbon emissions could be reduced through revenue-generating measures such as congestion tolling.

I appreciate the job Market Urbanism does, but his assertion that we should just stop subsidizing transportation completely ignores the political realities of the day.  This is one of the primary reasons many folks go insane when arguing with Libertarians.  At least Will Wilkinson acknowledges the follies of path dependence – Market Urbanism seems fit to ignore the political realities we live with.

Perhaps the biggest thing to note came in Avent’s comment section: commenter ‘jack lecou’ notes that the change won’t be transformational – at least not in the immediate future.  The changes that got us to our current transportation system weren’t transformational, either – they were incremental.  The key in getting our transportation system back in balance is to make those kinds of incremental changes in both our built environment as well as our political environment.

This is essentially the only politically possible path to choose.   It’s also the most realistic path to implement change.

Following up on an idea from previous posts on HSR…

Several HSR skeptics have raised the point that we should invest more in urban mass transit than in intercity rail service.  My gut response was “do both!”  Yonah Freemark articulates things a bit more coherently:

High-speed rail is convenient to people living or working in center-city locations, not true for air travel. Meanwhile, car travel, while “convenient,” since one can drive directly from one’s driveway, encourages sprawl in a way not true of high-speed rail if stations are positioned only in inner-city cores, as they should be.

Arguing that improving urban transit should be prioritized over high-speed rail is acceptable, but ignoring the needs of long-distance travel is not. The United States has a serious need to invest in both intercity and intracity travel, and for trips of between 200 and 600 miles between large cities, high-speed rail is usually the most appropriate investment. In the pursuit of better transit within a city, we cannot forget that we also need to get between cities.

I think there’s a tendency to lump HSR and mass transit together, since they’re both seen as ‘transit’.  The reality is, as Yonah points out, that there are huge needs amongst the various travel corridors and distances.  Lumping these two disparate tasks together because they both use trains isn’t a helpful distinction.

It’s also worth noting that airspace isn’t infinite.  Airspace around major airports (Chicago and New York in particular) is very crowded and there’s not a lot of room for more flights.  Add in the fact that airplanes have no particularly viable fuel alternative to petroleum, and so many of those cheap, short-hop flights we enjoy today may well go the way of the Do-Do.

It’s all about using the right tools for the job.  Just as we’ve applied the car to all out urban transportation needs when it’s not the best tool, we’ve used the airplane for all our intercity needs when it’s not the best tool.

Ed Glaeser, what have you started?

First, you post a series of articles over with the New York Times on how the math for High Speed Rail doesn’t add up.   The poor assumptions force guys like Ryan Avent to rip these articles to shreds.  Any back of the envelope calculation will involve a lot of assumptions, but the kinds of assumptions Glaeser makes defy credulity.

Now, you’ve inspired copycats.   Avent again:

Today, the Washington Post’s lame excuse for an economics columnist, Robert Samuelson, used numbers from Glaeser’s analysis in writing an extremely regrettable piece arguing that investments in high-speed rail are misguided. But this is no honest entry into the discussion of how best to invest in transportation infrastructure. It’s a hack job, plain and simple.

From an urbanism perspective, the most egregious error Samuelson makes (as Avent points out) is his treatment of American space and density.

Samuleson:

What works in Europe and Asia won’t in the United States. Even abroad, passenger trains are subsidized. But the subsidies are more justifiable because geography and energy policies differ.

Densities are much higher, and high densities favor rail with direct connections between heavily populated city centers and business districts. In Japan, density is 880 people per square mile; it’s 653 in Britain, 611 in Germany and 259 in France. By contrast, plentiful land in the United States has led to suburbanized homes, offices and factories. Density is 86 people per square mile. Trains can’t pick up most people where they live and work and take them to where they want to go. Cars can.

Of course, stating that passenger rail is ‘subsidized’ without putting that into the context of all transportation subsidies is somewhat useless.  But that error can be forgiven in a persuasive piece.  The density argument cannot.

The US is a big place, but we’ve got a whole bunch of nothing out in the western portions of the country.  In fact, most of our population has clustered in several key mega-regions that all contain the proper distances and densities for high speed rail travel.   No one’s arguing for high speed trains between San Diego and Portland, Maine – but that’s exactly what Samuelson’s average density figure would imply.

(I can’t recall where I saw this, but blog comments in reaction to this piece somewhere likened Samuelson to the tale of the statistician who drowned trying to wade across the river.  Shouldn’t have been a problem, as the average depth was only 3 feet…)

Thankfully, Ryan Avent’s done a masterful job at debunking the faulty assumptions in these articles (as have several others).  Now, Yonah Freemark has put together a competing back of the envelope plan to rival Glaeser’s.  Yonah’s plan is available on Infrastructurist, and unsurprisingly reaches a different conclusion:

That may even to have been Glaeser’s intent in writing the series. The problem is that–through a sorry mix of omission, oversimplification, distortion, and deficiency–his calculations bear no relation the effects he is claiming to consider. So it’s important to show that “the numbers” do not at all undermine the viability of HSR in the US, even outside the northeast and California. In fact, the tend to support it.

By populating his model with a better set of assumptions, we hope to show how badly the economist missed the mark even on his handpicked example of an HSR link between Houston and Dallas. In reality, a well-designed high speed intercity rail project between the two largest cities in Lone Star State would likely produce a net economic benefit–nothing at all like the white elephant Glaeser conjures up. In this more comprehensive model that takes into account trivialities like regional population growth and a reality-based route, the total annual benefits $840 million compared with construction and maintenance costs of $810 million. Which is to say, our numbers show that HSR pays for itself rather handily.

Yonah’s analysis is worth a read.  It’s still a simplistic back of the envelope calculation, but based on a far more realistic set of assumptions.  Thank you for interjecting some sanity into the discussion.