High speed rail in Brazil

  I spent last week frantically writing about imperialism.  I need to spend this week frantically writing about nuclear power.  Instead I'm here writing about ... well ... high speed rail.

As our regular readers know, high-speed rail is dead in Argentina.  But it seems as though Brazil plans to have a line running between Rio de Janeiro and São Paulo by the time the 2014 World Cup rolls around.   Now, it isn't immediately a bad idea.  It will take about 80 minutes to cover the 258 miles between the two cities on the express train, which is competitive with the 55 minute flight, especially given the São Paulo airport mess.  Congonhas is overcrowded (when I lived in Brazil, I lived right next to the airport, with planes roaring right overhead) and Guarulhos may as well be on Mars.  In addition, unlike Rosario, Rio de Janeiro is a major metropolis and the traffic between the two places is massive, about 7 million transits per year, with an additional 11 million going between São Paulo and Campinas. 

So even given all the other infrastructure needs that Brazil has, this project can't be dismissed out of hand.

The catch?  The cost keeps skyrocketing.  The initial estimate was $11 billion; a new more-detailed report just upped that to $15 billion, with a worst-case estimate heading over $20 billion.  The financing will be via a public-private partnership, but they typically involve very large explicit or implicit government outlays. 

What do you think about the idea?  Should a country like Brazil build an HST?

Comments

[Doug M.]

In adjusted terms, I suspect Brazil's pcGDP today is quite a bit higher than Japan's was when it built the shinkansen in the 1960s. (Of course, the definition of "high speed" has crawled since then -- the first generation of shinkansen cruised at "just" ~130 mph.)

Amusing factoid: the first plans for shinkansen were developed in the middle of WWII. Japan didn't really go in for Albert Speer-style infrastructure craziness, but this was a partial exception... there were serious proposals for a shinkansen network linking all the major cities of the Greater East Asia Co-Prosperity Sphere. Hey, it took 150 years to build the Chunnel...

Doug M.

[Peter]

Amusing factoid: the first plans for shinkansen were developed in the middle of WWII. Japan didn't really go in for Albert Speer-style infrastructure craziness, but this was a partial exception... there were serious proposals for a shinkansen network linking all the major cities of the Greater East Asia Co-Prosperity Sphere. Hey, it took 150 years to build the Chunnel...

Plans for the Second Avenue Subway in New York date back to the 1920's. Two perfectly serviceable elevated lines were demolished in 1940 and 1955 in anticipation of its construction. A 1960's bond issue paid for the line's construction, and a few tunnel segments were dug, but in the mid-1970's the city exploited a legal loophole and took away the bond funds in an attempt to avoid a politically unpopular fare increase (which happened anyway). Today, what amounts to about half of the originally planned line (the "Stubway") is in the site-preparation stages; no actual digging has begun, and it's unlikely in the extreme that anything actually will get built.

[Noel Maurer]

Hi, Peter,

Are you sure that work hasn't restarted on the Second Avenue Subway? I was just in New York, wandering around the Upper East Side, and the streets seemed pretty tore up in the low seventies. This website seems to indicate that construction is underway. Over here it looks like $1 billion has already been appropriated to finance work in 2009-11.

So why do you say that's its unlikely that anything will be built? Are you referring to the line outside the 63rd-96th street extension? In that case, I sadly agree with you. But the extension to 96th Street (along with the 7 train to 11th Avenue and the more important than both East Side Access) seems like a done deal.

No?

[Noel Maurer]

Doug: do you have the Penn World Tables on your computer? I'd be happy to send 'em along.

Let's check the conjecture.

I hate GDP per capita as a measure. It's too affected by the age structure of the population, and so doesn't pass the bullshit detector: economies with the same GDPC can look very different, and changes in the GDPC don't produce anything like what you'd expect from actual economic growth or decline.

So let's use GDP per worker instead; that's a measure I understand. Simple productivity, unadjusted for skills or capital. It's a number that usually matches what you'll see on the ground, certainly much more so than the near-useless GDPC figure.

Japan in 1964 in 2000 PPP-adjusted dollars: $13,309.

Brazil in 2003 in same: $15,462. Quick and dirty adjustment from the real GDP and labor force stats available from the IGBE (the work I do for you!) ups that to $17,103 in 2008.

Which is equivalent (inasmuch as these are at all comparable) to Japan in 1967.

So in that sense, you're right. The Brazilian economy today is more-or-less as productive as Japan's in 1964. And since the economy in the south is yet more productive still, then Brazil is "ready" for an HST between its two major cities and on into the S.P. industrial heartland.

That said, the context is different. Japan was growing a lot faster in the 1960s than Brazil is today. In the last few years Brazil has humped out growth around 4-5 percent per year, after a quarter-century of stagnation. Japan, however, was rocketing along around 9 in the 1960s, smack in the middle of what was going to turn out to be a quarter-century of the fastest growth in the world to that date. (And even when it decelerated in the 1970s, it continued to grow pretty fast for a rich country all the way until 1990.)

In addition, Brazil lacks for a lot of basic infrastructure --- road and rail cargo transport is a mess, and urban transport is even worse than in Argentina --- that AFAIK Japan had pretty much already built by the 1960s.

Doug, I ask you: do you think the different context makes a difference? Would you approve a bullet train if you were a Brazilian public official?

[Will Baird]

Slightly off kilter here:

Do you think that a HSR is a good idea for Cali?

I have to admit I voted against the bond measure that they put out for the rail system this past Nov.

The reasons being that it was underfunded (only about 30% through the measure, iirc). I strongly believe they underestimated the costs: even with the real estate crash and Cali a bit desperate for jobs, once they make it anywhere near midstate or close to the Bay Area they're gonna get sued to death, which will further delay the fscking project. It was going to take too long (30 years?!) Finally, I think they oversold themselves: $20 ticket from SF to LA? Riiiiiight.

I guess in the end, I think it's just too expensive at this point to afford for cali...for any time in the foreseeable future.

Anyways, what do you think?

Oh, to answer your question on my blog for a while ago, Lyuda and I are likely to use the evil Cold to Hot tunnel because the cost of us for two bart tickets is a little too high relative to gas. Even when gas is around $5/gallon. We did the numbers and driving into work for me and her hopping BART the rest of her trip actually works out cheapest. Managing the Cold to Hot is all about timing...leave work prior to 4:30 and you're fine. Leave between 5 and 6 and you're DOOOOOOOOMED.

[Noel Maurer]

Hi, Will.

I think you voted the wrong way.

The traffic projection are around 90 million per year. Assuming an $50 fare (which is low, I think) you have revenues of $4.5 billion a year, more than enough to justify a $40 billion project.

Considering as I think it would be competitive at twice that fare, given times and congestion, those numbers seem conservative.

You do need to believe that an HST can capture 40% of the current travel market, or 25% of 2020's. But that seems reasonable.

The idea seems to make a lot of sense, even before considering the environmental benefits.

California is a lot richer than Brazil.

Although I agree that a 30 year build time is crazy. Five is more reasonable. Ten on the outside.

[Bernard Guerrero]

Seems like a bad idea. I'll hang my hat on the following:

-relative lack of plain-vanilla infrastructure

-heavy dependence on raw material exports, which makes me think they have a few bad budget years coming

-HSR projects always remind me of retailers building/buying Manhattan locations. More of a monument than a going business propostion, and a huge "sell" signal. This is personal bias, though, to be trumped by IRR.

[Noel Maurer]

I agree about the IRR, but if I've read the numbers correctly (and I don't have access to the original report), it looks like the HST-builders in Brazil are estimating about $560 million in revenue every year from the Rio-SP portion. ($80 × 7 million.) That's an bad proposition for a project involving $15 billion in initial outlays; there is no interest rate at which the NPV of the cash flows is above zero.

Build it for the initial plan of $8 billion and get the 20 million passengers first projected, and the IRR is 16%.

Dicey, unless you assume that there is some big externality that the train will fix.

Problem is, if you go there then the project has to compete with alternative uses of public funds ... like the plain-vanilla cargo railroads and ports you speak of. Which for a big primary producer like Brazil, probably generates a much higher social return.

[Bernard Guerrero]

Isn't that 7 million figure high? The way I read your original post, it's the total current number of transits between SP and RJ.

[Noel Maurer]

Hmmm. I'm thinking about that California HST projection. The Acela gets only 3.2 million per year at an average fare of $126.

Here are more Cali projections. They give the same basic numbers. 28.9 million between LA and the Bay Area, including intermediate trips, with an additional 24.9 million using local trains on the same track alignment.

It looks reasonable, but still. That's a lot of people to get out of their cars. Consider these numbers.

[Noel Maurer]

Hey, Bernard,

I re-read the article, and it's projecting that the train will get 7 million riders per year.

The money paragraph was vague: "De acordo com os estudos técnicos, a demanda de passageiros é de 7 milhões de pessoas por ano no trecho entre as capitais paulista e carioca."

"According to the technical studies, the passenger demand is seven million people per year for the portion between the Paulista and Carioca capitals."

But on reading the article, it's pretty clear that it's the traffic projection for the train, not between the two cities in general.

[Noel Maurer]

Alright. Here are some numbers. In 2005, there were 19 million air trips between the three Bay Area airports and the six L.A. airports. That will rise to 32 million by 2020.

The HST plan assumes that you'll also pull a lot of people out of their cars. Half the air market (16 million) plus 12 million drivers.

The air market seems reasonable. The drivers, I don't know. At 16 million per year (roughly five times current Acela ridership, but that does seem reasonable) and $30 billion in capital costs, you'd need an average fare of at least $150 to get anything that ballparks a reasonable return on a government-guaranteed project, possibly more, depending on operating costs.

That, though, seems quite reasonable as a lowball worst-case scenario which assumes no federal funding and that all bonds are general-obligation.

Will, it still seems that you voted the wrong way on the project.

[Will Baird]

Still gotta disagree.

1. 30 /fscking/ years, Noel! 30 with a blink tag. That speaks of a stinky going to be unfinished project.

2. Cali budget. Big fubared whoopie cushion. Voters rejected the propositions. that means a massive hack and slash. We can't afford a big project like this until the budget's fixed.

3. I completely and utterly believe that they have underestimated the costs. *cough*bay Bridge West Span*cough* CalTrans has gotten really bad at estimating the ubersized projects. I am concerned about the Caldecott 4th bore's costs.

4. They sold it at $40 round trip to and from the Bay Area from LA. You just pointed out it'd take $150...mmmm...

As for how much they pull off of the planes and cars will depend on how much security theater they implement with it. Probably not as much as at airports, but nontrivially more than on the interstates. :)

So, I'll stand by my vote. A HSR may be a good idea, but this was a bad plan. IMNSHO. :)

[Noel Maurer]

Consider your own arguments. They are valid points, but they don't lead to the conclusion that the bond issue was a bad idea.

(1) The 30 year timeframe refers to the extensions to San Diego and Sacramento. According to the business plan, the main line will be done within eight years of the start of construction. That is entirely reasonable for a project of this magnitude.

(2) The California budget is a problem because California has not raised taxes and cannot meet its current obligations. More debt is utterly irrelevant --- either the state will raise taxes now, or it will cut services now, or it will default now. The bond issue had no bearing on these current problems. The funds from the HSR bond issue won't be committed until these issues have been resolved. If they are resolved by default, then the point will be moot. If they are resolved by the other two means, then there will be no economic problem from taking on additional debt, since that debt will produce a dedicated stream of revenues in the future.

(3) The per-mile costs are in-line with actual experiences on rail and highway projects inside California. Current California projects are coming in under budget. The BART extension came in on budget (although a case can be made that delays raised that to 1.6), and the Santa Monica rebuild came in under budget. One might not believe the figure, but that is an act of faith, not of logic.

(4) That is not true. The lowest estimate on the HSR webite assumed an average fare $40, with most of the trips occurring inside metropolitan areas along the HSR right-of-way. The highest estimate assumed an average fare of $76. More relevantly, you can't compare these figures with my back-of-envelop calculation, because I assumed that there would be no local use of the project. The reason I did that was to stress-test the projections. For what it's worth, the HSR Authority believes that a $108 LA-SF fare would be optimal.

(5) Your conjecture about security theater is at odds with the actual experience of the Acela on the east coast or high speed trains in Europe.

A reasonable point that you mentioned earlier is the possibility that private and federal funds won't come through. After all, right now the state of California will only be responsible for a third of the total cost. That, however, is also irrelevant to the initial bond issue. The reason is that if the federal government doesn't make a commitment to assume its share of the project, or private investors can't be interested, then the state bond funds won't be committed. Once again the entire issue will be moot.

All references to the business plan come from documents on the HSR website. The useful page is here.

I wish you'd talked to me before. It still seems as though you voted the wrong way.

(FWIW, I was registered in the State of Florida in 2004 (I had only just returned from Mexico) and I voted to repeal the HST mandate in that state, where the idea truly didn't seem to make any sense. A rail-buff I am not, as you know.)