Monday, August 22, 2011

Lies, damned lies, and statistics

This particular bit of propaganda, especially the claim about high speed rail ticket prices per mile, has been making its way around lately, including by State Assemblyman Jerry Hill. It provides an excellent example of the value of cherry-picking one's data to support the conclusion.

Using empirical evidence from analyzing fares on high-speed train routes in Europe and Japan, it appears the CHSRA’s high-speed rail per mile rate should be about $0.44/mile to recover operating and construction costs; 80% higher than their presently-used $0.24/mile. Setting aside for a moment the fact that all but two of the world’s high-speed rail routes are subsidized, and assuming they at least break even, the analyzed per mile rate would make a one-way SF to LA ticket cost about $190.5 Therefore, if the CHSRA’s assumed private operator must charge enough to break even, four tickets for a LA/SF round trip would cost at least $1,520.

First the lie: The California High Speed Rail Authority is not required to recover construction costs, only operating costs and any revenue bonds sold on the basis of HSR ticket revenue. It is therefore, irrelevant, to cite what may be needed to recover construction costs.

With regards to the damned lie, as I have already shown before, every high speed rail operator recovers operating costs based on passenger revenue. The claim that "all but two of the world’s high-speed rail routes are subsidized" is a falsehood based on which routes have fully paid off their construction bonds, an assumption that does not necessarily work if, as SNCF has chosen to do, profits are used to invest in expansion rather than paying off the extant loans. It is a rather foolish person who considers a business unprofitable simply because the original loan is not yet paid off.

Lastly, we come to the statistics. Their empirical evidence fares in Europe are peak hour next day 2nd class fares upon Paris-Lyon, the most congested and hence profitable line within France. As any economics student could easily point out, when lines are operating at or near maximum capacity and there is still substantial demand, the fares will, absent outside regulation, reflect the highest price the market will bear. At 425 kilometers and 86.4 euros peak, this reflects a price of €0.20/km or $0.47 per mile. However, if we look at the Paris-Marseille segment, which covers 783 kilometers in three hours, we have next day fares ranging from €54 to €97, or $0.16-0.29/mile. Advanced fares drop as low as €39.90 for two weeks in advance, twelve cents per mile. Seen in this light, if anything, the 24¢/mile fare proposed for CAHSR is actually too little.

Similarly, the Japanese fares referred to appear to be from JR East's Tokaido Shinkansen who charge a flat fare of 14,050 yen on the Nozomi express trains between Tokyo and Shin-Osaka. That equates to a hefty 53¢ per mile, but again, this is not a sign of actual operational need rather than profit-taking. The congestion and demand on the Tokaido Shinkansen is sufficiently high that JR East is planning to build a ¥9 trillion maglev line to supplement it (a line which will be entirely privately funded and expected to be operational). A longer journey from Tokyo to Hakata, again using the Nozomi express along the congested Tokaido Shinkansen, is ¥22,320, 40¢ per mile. It is, of course, notable that that fare level does not impair traffic, contrary to the assertions of the Community Coalition on High Speed Rail.

A more proper response by CC-HSR would have been to try plugging in numbers for themselves and seeing if the numbers worked. Then, and only then, would it be appropriate to criticize the 24¢/mile figure if the numbers did not pan out. Since they failed to do so appropriately, I'll make some effort at doing so.

1. The trainset will resemble a Siemens Velaro E with a cafe car and a total of 405 seats in three classes (however, for the purpose of this analysis, all seats will be treated as coach).
2. The notional average trainset will travel an express route from Los Angeles to San Francisco with no stops.
3. Attributed maintenance costs will be equal to $33.74 per mile, in accordance with the tolls SNCF pays along Paris-Lyon.
4. Electricity use shall be 35.41 kilowatt hours per mile, similar to the TGV Atlantique (page 74), and the price per kilowatt hour 8.46 cents.
5. Three employees (engineer, conductor, and cafe attendant) whose total compensation per hour is 40, 30, and 25 dollars per hour respectively, with an assumed 3 hours of paid time per notional average trip.
6. The average train will have a capacity factor of only 50%.

These direct marginal operating costs come then to a total of $16,154.82 for the 432 mile journey. At 24 cents per mile, the 202 passengers paid a total of $20,943.36 ($21,210 if we round up to the $105 ticket cost) in fares, meaning that the average train should cover its costs handily. Amtrak's experience is that the cafe car adds an additional 7% to the revenue, boosting total revenue to $22,694.70, an operating profit of $6,539.88. Provided that these numbers truly represent the average train, the California high speed rail system should not face any difficulty in raising sufficient operational revenue to run an operational profit at the 24 cents per mile fare level.

1 comment:

  1. I remember reading somewhere - no link, sorry - that maintaining the Tokaido Shinkansen involves 6,000 employees every night. At 6 hours a night, $50 per hour, 500 km, and 300 trains a day, it's $12 per train-km. The lower figure presumably comes from the fact that RFF tolls include depreciation on fixed plant and not just maintenance, whereas this is just maintenance. In addition, Shinkansen trains are lighter than TGVs and have lower axle load.