June 5, 2003 Subcommittee on Surface Transportation and Merchant Marine

The National Association of Railroad Passengers is a non-partisan organization funded by dues and contributions from approximately 16,000 individual members. We have worked since 1967 to support improvement and expansion of passenger rail, particularly intercity passenger rail.

The Association's central point about a new source of funding begins at page 5 of the statement I filed in the record of the full committee's April 29, 2003, hearing -- a Railroad Finance and Development Corporation that issues tax credit bonds. I will not repeat that here.

I. Operating Grant Requirements for National Network (Long-Distance) Trains

We believe that this should remain a 100% federal responsibility. Some have suggested that Amtrak should be treated the same as transit commuter systems, requiring states or localities to pick up all the operating costs.  Transit systems are local and started by local initiative. Amtrak began because of a federal initiative. It provides interstate service, and the federal government should have the responsibility for any operating costs not covered by revenues.

Our view does not change in light of new analysis showing remarkably low, short-term costs for some of these trains. As previously indicated, it is a challenging task to maintain existing state funding programs for intercity passenger rail -- programs focused on short-distance trains which were initially added solely because a state or states agreed to offer partial financial support for them. States now are being pressed to provide full funding for those trains.

It is not realistic to superimpose on that pressure the requirement that states also fund long-distance trains. The likelihood of getting every state along any given route to provide funding is small, as is the likelihood of getting states to agree on a proper cost allocation method or, in some cases, the right schedule. The most economic schedule for such trains must be set around providing attractive times at the endpoints. That includes the need to take into consideration proper connections with other long- and short-distance trains at those endpoints. Yet, by definition, some points must be served in the middle of the night.

When Secretary Mineta many months ago was asked, "what if a state refused to pay," he answered that the train could run closed-door through that state. That is not practical. Suppose every state but Colorado agrees to fund the Chicago-Bay Area California Zephyr. Having the train skip Denver (as well as Glenwood Springs and Grand Junction) simultaneously would make the train vastly less useful to the other states along the route, and -- as a consequence -- drive up the amount of operating losses they would be required to pay. Or imagine Illinois not agreeing to pay for any long-distance trains. The system could not function even if Illinois stood alone.

It makes much more sense to press Amtrak to operate the most efficient national system possible, and to allow states to proceed with what many of them have wanted to do for a long time -- improve short-distance corridors. Those improvements, incidentally, will benefit the long-distance trains as well. The most dramatic, recent example of this is the rail/rail grade separation in Los Angeles used by Metrolink commuter trains, Amtrak's San Diego line, and Amtrak's Southwest Chief. But almost every long-distance train stands to benefit from one or more corridor projects -- and these projects often address the most congested segments of the long routes.

It is, however, reasonable to expect states and localities to pay for stations.

II. On-Time Performance of Amtrak Trains

It will be difficult for new legislation to produce better on-time performance if the track owner acts like it doesn't care. In any event, any legislation should take into account the subject’s complexities.

First, it should be noted that On Time Performance (OTP) on the California corridors is far from ideal. To be sure, lateness normally is not calculated in terms of hours. On the other hand, riders who think they are taking a trip whose total duration is a few hours have much higher OTP expectations. On June 9, 10 and 11, I rode three Capitol Corridor trains and two San Joaquins with the following results:

• June 9:  The 3:45 pm San Joaquin from Bakersfield on June 10 originated only two minutes late but departed Stockton (where I got off) 34 minutes late.
• June 10:  The 12:30 pm Capitol Corridor train from San Jose left on time but departed from Emeryville (50 miles later, where I got off) 33 minutes late.
• Train 538 (3:30 Oakland-to-Sacramento) left Emeryville 3 minutes late, Martinez 7 minutes late -- the only satisfactory trip out of the five.
• The 4:05 pm from San Jose departed Martinez 47 minutes late and arrived Sacramento 39 minutes late.
• June 11: The 6:35 am San Joaquin departed Sacramento two minutes late and arrived Bakersfield one hour 48 minutes late. The bulk of the delay was due to vandalism -- the train struck some debris on the tracks.  However, the train was 31 minutes late leaving Wasco -- the last station before hitting the debris (and the last station before Bakersfield).

My understanding is that the two Capitol Corridor trains from San Jose were delayed by Union Pacific freight problems, including two trains with "outlawed crews," that is, trains which needed relief crews under the Hours of Service Act before they could proceed. On the San Joaquin Corridor, a tremendous growth in BNSF freight traffic followed -- and partly consumed -- state-funded capacity investments.

When Amtrak OTP percentage figures are reported, they generally indicate simply the percentage of runs that reached final destinations within x minutes of the scheduled time, where x = 30 minutes for long-distance trains, and lesser amounts (sliding scale based on miles traveled) for short-distance trains.

However, this percentage figure does not tell much about service quality. Consider these two situations, based on operating 14 trains per week (seven in each direction) on a long distance route.

EXAMPLE A
All trains operate between 31 and 45 minutes late--0% on-time performance.
Average minutes late per trip is roughly 38 minutes.

EXAMPLE B
12 trains operate either on-time to the minute or less than 30 minutes late; two trains are six hours late.The route has an 86% on-time performance.
Average minutes late per trip is roughly 60 minutes.

The two calculations produce opposite results. Notice that the average minutes late method more closely reflects passenger satisfaction levels, while the Amtrak calculation is at odds with those levels.

For Example A, the Amtrak calculation has the worst possible result (zero percent on time) even though, if that long distance route actually achieved that type of performance consistently, the on-time issue would barely be on anyone’s radar screen.

Related issues:

1. Incentives for railroad performance. With a simple yes-or-no on-time calculation, as soon as the train is so late as to make it fall outside the 30 minute window, the railroad loses any incentive to handle the train decently, and the train could get later still. When total minutes of delay are counted, the railroad has an incentive to continue to deliver the best possible performance no matter how late the train gets.
2. At how many points is on-time performance counted? Amtrak schedules (not unlike those for intercity buses and airlines) typically are strung with recovery time before major cities, to allow for a limited amount of delay while keeping actual performance closer to expectations for the majority of customers. Depending on the size of the recovery time, however, this can mean substantial lateness at intermediate points even when the train is on-time at major points.
3. Does the railroad respond to incentives? So far, some like BNSF do respond, UP does not. That BNSF cares is evident from their big operations center in Fort Worth. All the dispatchers face about six huge screens of data -- one of which is devoted to various measures of Amtrak trains' on-time performance on BNSF. That is one of many ways BNSF top management tells its people that it cares about earning the incentives for good on-time performance that are built into Amtrak's contracts with most railroads.
4. The significance must be considered for different causes of delay. For example, in Amtrak's contracts freight railroads generally are not liable for minutes of delay associated with causes like malfunctioning Amtrak passenger cars and locomotives. Other causes are treated differently, including the vandalism problem cited above, where arguably neither Amtrak nor the railroad is to blame.
5. The suggestion that a route be "open for competitive bid" if a certain level of OTP is not achieved implies that most causes are Amtrak's fault and that things will get better if another operator is assigned. A top priority of the freight railroads is to insure that Amtrak's track access rights are not transferred to anyone else. Certainly, in the unlikely case that the railroads are defeated on this issue, it's not clear why Union Pacific would provide better service to a different operator. In fact, it may be worse.
6. Evaluating true causes of dispatching delays. The following issues impact the quality of handling Amtrak trains get from dispatchers:
• (a) Relationship of track capacity to total traffic on the line
• (b) Attitude of host railroad's top management towards passenger trains
• (c) Attitude of dispatchers towards passenger trains
• (d) Skills of dispatchers
7. The analysis of capital investment needs for chokepoints should be independent and should consider the factors in #6. It would be unfortunate if a publicly funded investment program rewarded railroads that did a poor job of handling Amtrak trains and penalized railroads who did a good job.