European experts weigh in on Authority’s operating cost model
By law, the California High Speed Rail Authority is supposed to eventually offer train service without any public operating subsidies. CHSRA’s claims that it could do so even on an abbreviated route like Merced to the San Fernando Valley were questioned on both the basis of revenue/ ridership forecasts that seemed high and an operating cost structure that seemed too flexible and too cheap.
In response, the Legislature ordered the Authority to have the UIC, the industry group representing the world’s railways, review its operating cost model in time for the January 2, 2014 Business Plan.
The review is done and there is finally a copy online at the California High Speed Rail Peer Review Group’s website.
Response by Peer Review Group (also includes comments on other topics)
Bottom line: new Authority operating cost model a step in the right direction but still not there.
Our basic complaint about the previous model is assumed a way more flexible cost structure than existing high speed rail operations. The Authority’s model assumed that its operating costs would go up and down with ridership in a way that they could virtually never lose money. The reality is you have a certain cost of being in business and maintaining the infrastructure with economies of scale as you are able to use your infrastructure (trains and tracks) to provide more service.
The new model definitely moves in this direction, but it seems to be still underestimating the cost of maintaining its infrastructure and assuming that it can operate with perfect efficiency.
Things that we found interesting:
1) It is a LOT more expensive to run trains at 220 mph than 186 mph (300 kph). There is a lot more wear and tear on the trains and the infrastructure. Train maintenance costs are at least 20% higher, track maintenance costs are at least 40% higher and energy use is between 10-30% higher, depending on topography. This validates the Peer Review group’s concerns about the Authority’s comparison of costs to other rail lines that operate at slower speeds than the Authority’s planned speed.
The Authority’s cost estimates for the maintenance of infrastructure, while now higher than current European costs, still don’t seem to reflect the full magnitude of the wear and tear of going at 220 mph. This is even more a concern as recent findings that the Central Valley is subsiding at an alarming rate (almost a foot a year, according to multiple sources – URS report page 6 or cool NASA/ JPL data for south San Joaquin Valley, read this July 2013 report or google “Red Hat” for Madera/Merced information) over parts of its route suggest a need for ballasted track. Ballasted track is your traditional rails on top of a gravel bed. It is cheaper to build than the alternative fancy concrete rail beds, but much more expensive and noisy to maintain, even with the super cool specialized equipment used (here is one video ). It will drive any neighbors quickly bonkers, as the maintenance is done in the nightly window between midnight and 6 am.
The high cost of ultra high speed is the rationale behind one of the ideas UIC had to lower costs: run the trains that aren’t directly competing with airlines (everything but the non-stop SF/SJ to LA trains) at a lower speed.
2) While trying to stay clear of the landmine called “ridership forecasts”, the UIC hints strongly that the Authority is overstating its ability to pack each train and hence underestimating the costs per passenger. The Authority does not take the hint. Even if their ridership forecasts are accurate in terms of demand for train service, actual results are likely to fall well short of forecasts. The French try the hardest to pack each and every train. They get them about 70% full. The Germans have prioritized frequent and consistent service. Their trains are about 50% full. The Authority is assuming the frequency and consistency of the Germans with passenger load factors that make the French seem like amateurs. Outside of a couple of limited examples in Asia where the population density is so high that there are no tradeoffs between schedule and operating efficiency, there are always tradeoffs. The current route, with multiple stops and an effort to serve long distance commuters, is especially susceptible to these types of real world issues. If the Authority is lucky, forecasted load factors are only twice what actual ones will be as the Authority is assuming a German like approach to scheduling.
Here ( simplified) is the method the Authority used to figure out its passenger load factors.
The Authority specified a schedule and then forecast annual train demand based on this schedule. It took the number of passengers and assumed exactly 1/365 would want to travel each day and that of the passengers traveling each day, they would naturally divide themselves between all the different trains being run that day. This might seem like an okay approximation, but it is not. This is actually the best case scenario; real life will be worse, much worse.
Let’s start with a few concrete examples. The Wednesday before Thanksgiving vs today (“seasonality”). Commuter demand for a train from Palmdale to LA in the morning (a lot) vs in the evening (nothing).
Think about your own long distance “recreational” travel. How much of this is bunched around holidays and spring break?
The asymmetry of demand is one of the major reasons that commuter rail lines are expensive to operate. They are full in the morning going the city but empty in the morning going towards the suburbs. Operators either have to have a lot of extra trains and a place to park them near the downtown or be willing to run empty trains back to the suburbs.
This problem is even worse for high speed rail. For every seat sold Monday morning from Palmdale to LA, there is a seat available from San Francisco to Palmdale for which there is, approximately, zero demand. You might be able to sell 1 ticket SF – Fresno for every 10 commuter seats. The result: a lot of seats empty for most of the ride. The cost per person is all of a sudden much higher than expected.
The Japanese offered feedback last year on the Authority’s plans (JRTT and JR East). They highlighted that morning rush hour looks very different from evening rush hour. This presents other problems. Trains don’t just conjure themselves out of thin air. These inconsistencies have a way of stranding crews in the wrong place at the wrong time.
There are obviously ways to try and minimize these problems. The French do it by running fewer trains and nudging people through pricing to spread themselves out. Jacking prices for prime travel times can generate cash that soothes the pain, but it does so by purposely leaving NOT meeting all the demand at those times. Overnighting a train in commuter spots can help a little, but as the UIC reviewers point out, why are using the equivalent of a Rolls Royce when a Prius makes more sense for this purpose?
This is basically what Amtrak does. Sure you can take the Acela from New Haven to New York on Monday morning, but it will cost you$122, only slightly less than the cost of a ticket from Boston. Thanks, but you will just take MetroNorth and then a cab. Amtrak has solved its problem of an empty seat, but it also lost you as a customer.
If the Authority wants to make its model approximate reality, it needs to use the equivalent of $122 for a Palmdale to LA ticket for its commuter market. Which it hasn’t. Which it should. Which it should have done a decade ago.
While getting an operating model right at this stage of the game, years before anyone will be able to buy a ticket, may seem premature, this is the first exercise that should have been done. Understanding your business model is Business School 101.
The Authority would have realized the cost of adding route miles to the system. Each mile costs a lot to maintain. Extra miles also mean that the train must travel at ultra-high speed speeds to be competitive for markets like LA to San Francisco where the plane is the main alternative. At a minimum, the current route is 20% longer than necessary. Adding in the costs for 220 mph vs 186 mph, this means operating costs 50% higher than necessary. This is a dollar figure that might, just might, have injected some common sense into the incredible political mechanism of route design for this project.