United Rail Passenger Alliance

Andrew C. Selden, President, Minnesota Association of Railroad Passengers; to the

National Railway Historical Society,
Minneapolis, MN
July 2, 2004

Thank you, Dan, it’s a real pleasure for me to be here tonight, to share some recent historical perspective with the National Railway Historical Society, and use that as a foundation for a new vision for our nation’s intercity rail passenger service.

To do that, I will offer three radical ideas concerning Amtrak, and use them to show why rail passenger service in America is so poor, financially and qualitatively, and then show how it could be radically improved, at declining public cost.

This is not about social or political philosophy, rather, this represents a vision to put America’s rail passenger service on a sound economic footing.

As Dan said, all of this is an outgrowth of the Carter Administration’s “massacre” of the long distance system in 1979, when we lost five long distance routes, yet Amtrak’s costs kept climbing afterwards.

Several of us began to ask a basic question: “What would it take to immunize the national system from its political and fiscal vulnerability due to its reliance on public subsidy?”

Eventually, our delving into this question lead to two basic insights, and a surprisingly powerful mathematical tool developed by Dr. Adrian Herzog of the University of California, which we dubbed “matrix theory,” which explains mathematically how Amtrak’s network operates.

Combined with research done by Byron Nordberg, Dr. Bill Hamilton, former Amtrak board member, Joe MacDonald, me, and others concerning Amtrak’s business accounting, we made two powerful discoveries:

One, if one were to use the same measures of performance that are universally used in the business world to evaluate Amtrak’s performance, we discovered that Amtrak’s routes’ performance was the exact opposite of what Amtrak reported, and what most politicians believe; and

Two, no law of nature mandates that intercity rail passenger services must lose money.

The rail industry, American society, and the competitive environment have all changed profoundly since the 1960s, such that today, carefully developing the national network unavoidably leads to financial breakeven in that network, everywhere except in the NEC.

Now, we all understood that these discoveries fly in the face of two generations of “received wisdom” about intercity rail passenger service, and they also confound the foundation of Amtrak’s 40-year-old business model.

But, understanding the implications of these insights explains why Amtrak is such a chronic loser; why we have so little rail passenger service in this country for so much public cost; and, how we can turn it all around.

Let’s explore this with three radical propositions:

Amtrak’s strategies cause its results. That is to say, Amtrak’s results occur precisely because of, and not in spite of, the business strategy that Amtrak pursues. There is a corollary to this and it is: trains still cannot outrun 737s.

Make no mistake, Amtrak’s financial results are truly awful, relentlessly persistent, and they are getting worse.

In a typical year, Amtrak takes in about $2 billion in revenue, offset by more than $3 billion in costs, leaving a net operating loss of between $1.1 and $1.4 billion.

Amtrak’s labor costs alone exceed its gross ticket revenue. Amtrak’s labor productivity is the worst in the business; by some measures, trunk airlines have three and a half times the labor productivity that Amtrak does.

And as bad as these results are, the trend is negative: Amtrak’s net operating loss continues to rise steadily over the last five years, despite billions of free federal capital. The roughly $27 billion that Amtrak has been handed in the last 25 years is worth more than $50 billion, in constant 2004 dollars.

Now, it’s true that a billion dollars just isn’t what it used to be … but, what exactly does the nation get for all these billions of capital, and endless losses? Operationally and financially? The answer is: not much.

Financially: Amtrak’s net operating loss is climbing steadily; even after full deployment of Acela high speed service, in what Amtrak regards as rail’s “strongest market,” during a period when long distance services were steady to declining slowly, Amtrak’s financial results have worsened.

Amtrak’s subsidy “need” is climbing steadily, even post-Acela; there appears to be no end of the subsidy needs of the NEC, while the national system seems remarkably steady. In the last three years, the Bush Administration has nearly doubled federal subsidy to Amtrak.

Operationally: After tens of billions of federal investment, Amtrak is still trivially irrelevant in all of its markets, with the possible exception of the New York – Philadelphia local market, where Amtrak generates nearly half of all Northeast Corridor ridership.

Amtrak’s market share is less than 1% in all of the markets it serves, including the Northeast Corridor. This market share is roughly equivalent to the market share of motorcycles in this country.

Since 1985, total travel in the U.S. has quadrupled; air has doubled since 1990. In the same period, Amtrak’s output has been essentially flat. The short corridors appear to fluctuate generally with the state of the national economy; the long distance services continue to suffer a slow decay, from a decreasing fleet and declining capacity; a dis-integrating network; and increasingly hopeless train operations. Year-to-date, for example, Amtrak’s premier service, the Acela Express, operating over its own railroad, has been late more than 25% of the time, even by Amtrak’s loose definition of “on time”; and the Sunset Limited, out of 233 trainstarts, has been on time exactly 12 times.

One of the basic insights of life generally is that “it’s not how well you play the game, it’s who keeps score that matters.” Any of you who has been involved in any kind of business activity knows how critical it is to measure performance in ways that are both accurate and useful, and to do so requires the use of the rights tools. Amtrak doesn’t get this right at even the most elemental level.

Does everyone understand the distinction between ridership and revenue passenger miles? Ridership is only a crude measure of transaction volume. Revenue passenger miles measures output, and in an intercity passenger transportation business, it is the only valid measure of output.

These are very different yardsticks.

Airlines always report their output in revenue passenger miles; Amtrak uses ridership. Why is this?

Amtrak uses ridership, not output, precisely because it masks the results of its operations, and masking results is necessary to enable it to justify its preordained obsession with short corridors in general, and particularly the Northeast Corridor.

Amtrak misrepresents NEC performance both by reporting ridership, and modal share versus the airlines, rather than revenue passenger miles and market share, or even load factor.

Load factor is a measure of capital efficiency. Airlines always report load factors; Amtrak never does. Load factor is simply revenue passenger miles divided by available seat miles, yielding a percentage occupancy value.

So, why are Amtrak’s financial and operational results so consistently awful? Why is it that I say that Amtrak’s strategies cause its results? Isn’t Amtrak’s problem that it receives “too little federal money,” according to Mr. Gunn and the National Association of Railroad Passengers?

Again, consider that since 1975 more than $27 billion in free federal capital has gone into Amtrak. Over that period, 75% of that capital, and management attention, has gone to the NEC; more than 90% has gone to all of Amtrak’s short corridors; and less than 10% has gone to the long distance routes.

What is the justification for this allocation? These reflect, after all, management decision-making. The announced justification is “pursuit of ridership” and the belief that high speed rail in particular will be the salvation for Amtrak.

Do you recall about five years ago how high speed rail was “going to save Amtrak”?

Acela Express was launched in late 2001, at a federal cost of more than $3 billion. By the way, none of that $3 billion is being amortized and charged against Acela’s revenues in Amtrak’s reports of Acela results.

Acela reached equilibrium ridership by about the second quarter of 2003, and that stable ridership level is surprisingly weak. Amtrak only discloses long cycle aggregate ridership numbers for Acela, never segment growth numbers or direct before-and-after comparisons.

Close analysis of Amtrak’s ridership numbers in the Northeast, however, show that there is almost no incrementality due to Acela’s higher speeds. The slightly positive ridership results in the Northeast in the last two fiscal years reflect network phenomena (mostly, more frequencies in New England and more frequencies operated through New York City), and, perhaps more importantly, correspond to the strong national economic recovery in the last 15 months.

But load factors remain very low: typical Acela Express trips in New England frequently operate with fewer than 100 passengers accommodated per trip.

In fact, since the full deployment of Acela Express, Amtrak’s Northeast Corridor and total net operating loss has increased, and continues to increase. We learn from Amtrak’s FY04 budget that the net operating loss in the Northeast Corridor will total $468 million this year.

From this data, it is clear that Acela Express has produced a negative return on investment of that $3 billion in federal capital, as well as an increased net operating loss and a declining market share.

We know that Acela caused these results, because the long distance operations were largely unchanged in the same period and the only other changes in Amtrak’s operations were slight improvements in California and Washington regional corridors.

The conclusion therefore is inescapable: 30 years of relentlessly consistent financial results based on a 1960s business strategy show irrefutably that these results do not come about despite Amtrak’s efforts, rather, the activities caused the results.

What a radical idea!

Let’s look now at my second radical proposition: Amtrak’s short corridors are the least productive of Amtrak’s routes. I also have a corollary to this: high speed short corridors are worse. There’s even a second corollary: the Northeast Corridor, despite its saturated service levels, relative to its potential is the worst served of all Amtrak markets.

First, consider the Chicago hub markets. Please follow on these graphs.

These graphs illustrate the results of operations of three distinct groups of trains that radiate out from the Chicago hub. I invite you to decide which group appears to be the most “successful”, and which should be getting more investment and attention.

Here are the three groups, ranked by ridership.

Here are the same three, however, ranked by output.

This graph shows why output matters a lot more than transaction volume: this is where the money is. This is the revenues of the three groups.

Finally, here are the three groups measured by Amtrak’s unique expression of load factor, again illustrating capital efficiency of the three groups.

This slide shows all of this data. Which group do you think is the strongest? If you were CEO of Amtrak, which would you be developing?

Group A represents all of the short corridors radiating out from Chicago, the Milwaukee services, St. Louis, Indianapolis and Detroit, and the downstate Illinois trains.

Group B is the same number of routes, but long distance services coming out from Chicago, and Group B-1 represents all of the Chicago-based long distance routes.

These results are not limited to Chicago. Here are national data by segment. Again, please follow on the graphs.

This graph shows the relationship between output and federal subsidy, by segment.

This graph is simple arithmetic: the total annual subsidy divided by output, by segment, showing a federal cost per unit of output in each of Amtrak’s three segments.

This graph shows the same data, but in a different display showing the segment relationships.

This graph, showing load factor by segment, shows where capacity is mismatched to demand: the short corridors as a group achieve about 33% load factors; the Northeast Corridor as a whole has about a 38% load factor, largely on the strength of the Philadelphia – New York local market, but some NEC services have load factors as low as the high 20s; Acela Express achieves a load factor of 49%, again on the strength of New York – Philadelphia traffic. The long distance trains, as a group, achieve load factors of 55 to 70%, and please keep in mind that a long distance train that is about two-thirds full is functionally sold out because of the large number of en route boardings and alightings.

The load factor data alone shows plainly that Amtrak is significantly over-invested in the Northeast Corridor and the other short corridors, and under-invested in long distance markets. That’s not social philosophy or political policy; it’s hard economic data.

In the West, the Coast Starlight usually outgrosses all of the Pacific Surfliners combined. Do you think that that one train costs more to operate than all of the frantic activity involved with all of the Pacific Surfliners operations, and all the labor costs?

And the Starlight is generally regarded as one of Amtrak’s strongest long distance trains because it carries a large number of riders. The Starlight carries more passengers, for example, than the Empire Builder, but the Empire Builder consistently produces 20 to 30% more output – revenue passenger miles – and revenue than the Starlight. In fact, the Empire Builder is Amtrak’s highest grossing, highest output, single passenger train.

Data like this shows that across the country, train for train, as the Chicago hub graphs illustrated, the long distance trains outperform all of the short corridors by factors of five to seven times in output and revenue.

That is to say, every federal dollar invested in long distance markets yields five to seven times the transportation output, and revenue, of a federal dollar invested in any short corridor, especially including the Northeast Corridor.

Moreover, the effects of distance appear to be scaled: the longer regional corridors consistently outperform the short ones, as happened in southern California when the San Diegan Corridor was extended north of Los Angeles to Santa Barbara and San Luis Obispo. Amtrak, by the way, was dragged kicking and screaming into operating that extension. It never occurred to Amtrak that anyone would want to travel north of Los Angeles by train.

What Amtrak regards as its “best market,” the Northeast Corridor, is not and cannot be profitable, measured by generally accepted accounting principles universally used elsewhere in American business, including all the airlines. Mr. Gunn said as much last year.

The NEC consumes about a billion dollars a year to achieve the weak load factors, rising net operating loss, and negative ROI that we detailed earlier.

More money is not going to turn that around. The last $21 billion didn’t change anything, in the same passenger markets that bankrupted the New Haven and Pennsylvania Railroads.

I’m reminded of some folk wisdom often mis-attributed to Albert Einstein, that “insanity consists of repeating the same actions and expecting different results.”

Amtrak’s own data on output, load factors and return on investment show clearly that the short corridors as a group represent a highly unproductive use of capital. Per dollar invested, Amtrak’s greatest share of revenue and output, hence its greatest return on investment, comes from the long distance markets, where the load factor data alone show that the long distance trains have the highest efficiency, and the highest unmet latent demand, of any of Amtrak’s markets.

Which brings me to my third radical idea: Amtrak gets too much federal subsidy. The corollary to this is: “It’s not what you’ve got, it’s how you use it, that matters.”

Currently, Amtrak begins each year with about $1.3 billion in free federal subsidy. Mr. Gunn, incidentally, wants $1.8 billion, but that’s just his standard method of operation: he’s been making the “Chicken Little” threat that the railroad will collapse if he doesn’t get all the money he demands in every job he’s held in the last 20 years.

But, what is it exactly that $1.3 billion in free capital “isn’t enough” for? We know where part of the money goes: $300 million goes to cover the fully allocated costs of the national system (the direct, cash costs of the national system by the way are closer to $200 million); and according to Amtrak’s own budget for this year, about $468 million will go to cover the NEC’s operating loss.

That leaves $500 million left over, for management’s discretionary use and investment. Where does all that cash go?

I’ll give you a hint, it’s not propping up the Sunset Limited; we already paid for that train. By now, it should be clear: it’s going to the NEC, and in fact, Amtrak’s FY04 budget shows that the NEC is scheduled to receive about 90% of that discretionary half billion, despite it’s continuing negative ROI on previous investments.

Having that much available money is what enables Amtrak to perpetuate its obsolete and foolish business strategy. That’s why I say that a half billion a year in free federal capital is too much, not too little money.

On the other hand, since Congress seems disposed to spend that much, what could we expect to accomplish with $500 million a year?

Well, what if we reexamined the assumptions? What if we changed the business model? What if we, for the first time, focused Amtrak’s investment capital FIRST, in markets with the greatest output, and the greatest potential incremental revenue passenger miles, and the segments with the highest load factors, which is to say, the greatest unmet demand?

What if we put the available capital into applications where we could expect a positive return on investment? Come to think of it, that’s a pretty radical idea, too.

How would we know where that is?

Well, you remember I mentioned at the beginning a “powerful mathematical tool”? Let’s apply it now to prioritize the application of available capital. By the way, those of you who might be interested in following the math that lies behind what we called “matrix theory” can find an explanation at the URPA website, www.unitedrail.org.

Since 1970, Amtrak has been crippled by an “endpoint” mentality and a focus, and obsessive attention, on discrete routes, usually the short corridors. Amtrak has paid lip service only to network connectivity, and in fact many connections have been deliberately severed. Take a look at their timetable: why is it that the northbound train going up into Vermont leaves Springfield just minutes before the arrival of the Lakeshore Limited, or why the northbound Talgo train for Vancouver leaves Everett, Washington just minutes before the arrival of the Empire Builder from the east?

What the mathematics underlying matrix theory demonstrates is that additive incremental growth in the network of origin/destination pairs drives not additive, but multiplicative, growth in the density of flow through the entire network, even at constant levels of market penetration.

In interregional long distance markets, the effects of scale are enormous. We use the term “utility” as an expression of the traffic potential, i.e., the flow and flow density, in a network where we used “n” to represent the number of stations in the network. The utility of a network is nearly the square of the size of the network. For examples with which you are all familiar, think about the telephone system, urban freeway grids, or the Internet. All of the flow density through the network is an exponential function of network scale.

For example, connecting two long distance routes does not increase the network by the sum of the origin destination pairs added, it more than doubles the number of origin destination pairs, and appears to well more than double the output of the network.

Extending or adding a branch to a long distance route does the same thing. One early model that we analyzed of the Southwest Transcontinental Corridor (the Chicago-Albuquerque-Los Angeles route) showed that if we modified that route by the simple splitting of the train westbound at Barstow, to send a section over Tehachapi and up the valley, in effect, re-inventing the San Francisco Chief, and on the east end splitting the train at Kansas City, with one section operating on its current route, and the other to Chicago via St. Louis, and adding a section – really just a thru coach and sleeper – to drop off the train at Flagstaff to operate down to Phoenix and Tucson, resulted in a prediction that ridership on that train would increase by a factor of six, assuming constant levels of market penetration in the new markets opened up by those changes.

Increasing ridership by a factor of six is a real problem: no Superliner train can handle that many passengers.

So we asked, what if that is wildly optimistic, and the actual yield of those changes was only half what the computer model forecasted? Well even at the level of tripling the existing ridership (remember that this train has very high load factors already) leaves us with the same problem: it’s almost impossible to handle 600 customers at any one time on a western Superliner train. That means right away you need a second frequency just to handle the ridership generated by the new markets served by the simple adjustments I mentioned. But that creates a problem, too, because doubling the frequencies increases the utility of the network, driving even further increases in ridership. It’s a real serious problem: an upward spiral in ridership, output and revenue.

Even the short corridors are sensitive to the effects of scale. I mentioned the southern California example, where output, average trip lengths and revenue shot up disproportionately with the expansion north of Los Angeles to Santa Barbara.

One objection we often hear is that it’s unrealistic to make a blanket assertion that increasing the scale of the network will drive exponential yields in ridership, output and revenue, because the size of the various stations varies considerably. What we have found is that at national or transcontinental scales, these effects wash out: the differences are not significant, because the large number of small stations tends to balance out the very small number of large stations.

Matrix mathematics also shows us that operating two existing routes to connect with one another versus not adds nothing or almost nothing to operating costs, but can drive usage through the ceiling. Let me give you an example.

This graph shows Amtrak’s current operation of two trains that come out from New York every morning and follow each other up the Hudson River Valley as far as Albany and then Schenectady, where one goes west to Buffalo and Toronto, while the other, the Adirondack, turns north to Montreal. In Amtrak’s operation, these trains do not interact with each other at all – there is no synergism beyond doubling the frequencies between New York and Schenectady.

But what if we were to take one of those trains – I don’t think it matters which one, but let’s arbitrarily say the Montreal train – and have it originate in Boston instead of New York, and operate on a schedule that would enable it to be in Albany at the same time as the Maple Leaf, allowing cross-platform transfers between the two trains. Yes, Boston is almost 60 miles farther from Albany than New York, so there will be increased operating expenses: an hour of crew labor, and a small increment of fuel costs and track rent. But in the scale of operation of these routes, those increases are almost immeasurably small.

Matrix mathematics suggests that making this one simple extremely low-cost change in operation, by enabling potential passengers to get from anywhere east and south of Albany to anywhere north and west of Albany, and vice versa in the afternoon, will more than double the usage of the two trains making up this very small network. It would open up for the first time in many years a thru service directly from Boston to Montreal, and also allow further connections at Springfield up and down the central New England route.

Thus, as illustrated by the Albany hub example, Amtrak’s greatest shortcoming is not a shortage of free federal cash, but a lack of network intraconnectivity, a lack of carrying capacity in its highest output markets, and a lack of vision.

We know the consistent results of the Downs/Warrington/Gunn strategy: financial ruin. The national system has shrunk: we have lost routes, cities (this is the gang that can’t figure out how to tap into the thirty million people who visit Las Vegas every year), network capacity, and especially connectivity.

There has been no capital support to the national system: the newest Superliner rail cars are now older than the youngest cars Amtrak inherited were in 1980. Many of you understand that many Viewliner sleeping car lines were annulled this winter due to lack of roadworthy cars.

Yet, despite all this, Amtrak’s five-year capital spending plans call for no new passenger cars whatsoever. Only a handful of Superliner and Viewliner cars are being grudgingly repaired under Mr. Gunn’s budgets, and even those using only “last dollars.”

What would my priorities be for the free half billion a year that Congress seems willing to offer Amtrak?

First, I would use significant operational reform (designed to drive up load factor without sacrificing output; i.e. match capacity to demand) and asset sales in the Northeast Corridor to raise $150 to $200 million a year. With that, plus about a hundred million from the five hundred we could fund the capital needs of the NEC at a steady rate of about $250 to $300 million a year. Remember, this is over and above the NEC’s annual net operating loss, which is now approaching a half billion a year by itself.

I would use the rest of the available discretionary capital to focus on markets with real growth potential. First, I would invest – and it will cost as much as $40 to $50 million – in a totally new management information system. Amtrak’s Route Profitability System is worse than useless.

Second, I would undertake an urgent rehabilitation of all rescuable Superliner and Viewliner rail cars.

Third, I would put out a semi-urgent solicitation of two proposals from the worldwide rail vehicle industry:

First, a hundred new Viewliner sleepers in “bed and breakfast” configuration (sacrificing one standard bedroom for a small galley facility; remember, all of these cars will be used in one-night markets), and about a hundred new long distance single-level coaches and associated lounge-type food service cars.

Second, I would negotiate an open-ended Superliner production line designed to produce cars at a rate of about one a week, which we could accelerate later, and I would begin with sleeping cars. In total, the network expansion throughout the national system would require a fleet of as many as 1,500 new Superliner rail cars.

Fourth, I would begin to put available federal capital into national system infrastructure, investments which I would leverage through tax credit-financed joint ventures with host railroads, and with state partners, to fix terminals, to create route intraconnectivity, and to make minor route extensions. We would ask states to invest only in capital projects, not subsidizing train operations.

Network density creates flow density, and it will grow rapidly … even at constant levels of market penetration. Growing train lengths will absorb all available new cars in the first few years, and demand will drive frequencies next. Then, and only then, and this is probably five to ten years out, we would look to add carefully selected routes, such as Washington, D.C. to Atlanta to Jacksonville to Miami, or restoring the Desert Wind.

Other, easy, near-term network enhancements, the real low hanging fruit, would include such things as a daily Sunset Limited; the Albany hub; sending the Silver Star to Montreal – do you realize that nearly 10 million Canadian visitors go to Florida every year? Amtrak’s share of that market is also zero; extending the California Zephyr to run down the coast line overnight to Los Angeles, partly to provide a second thru frequency on that route, but mainly to consolidate western Superliner maintenance at Los Angeles; I’d fix the terminal trackage at such locations as San Antonio, Minneapolis-St. Paul – if you have a chance, you really should go out and watch the Empire Builder tiptoe through the class 1 track leading into Midway Station – Phoenix, which as I said I would reach from the north from Flagstaff, not necessarily using the Sunset, and Portland; remember when Amtrak ran the thru sleeping car from New Orleans to Kansas City? Why wasn’t that extended 200 miles farther north to Omaha, where the train would have connected perfectly with the California Zephyr? That is networking.

Another possibility with excellent network implications would be to operate a train from St. Paul to St. Louis on BN via Galesburg, which we could call the Train of the Saints, which would interconnect all four of the major western transcontinental routes, taking all of the connecting pressure off of Chicago. Or, we could operate the Ann Rutledge to Dallas via Newton, Kansas.

With this kind of vision, and an investment strategy focused on real business results and real growth, in as little as five years Amtrak will be at a genuine operational breakeven in its national system precisely because, for the first time in its history, it would be operating a large, robust, and growing national network of intercity passenger services.

Thank you, and we can now take a few questions.