Car scheduling: Why bother?

A. Scheffer Lang
Paul O. Roberts, Jr.
John P. Sammon

Over the last 25 years or so it has been an article of faith in the railroad industry that car-for-car scheduling systems would improve the quality of carload service and restore some of the markets lost to trucks. Our experience now suggests otherwise, and our enhanced abilities to analyze logistics costs suggest that car movement reliability has, in any case, become less important than it once was.

Thus, we argue that the railroad industry could just as well abandon its complex and expensive real-time car scheduling systems and focus on other approaches to service quality improvement.

Unfortunately, early aspirations for car scheduling systems were not well thought through because railroads do not operate in a way that allows these systems to work. Car scheduling implicitly assumes that cars can be handled one at a time, but field operating personnel do not handle cars one at a time. They handle them in cuts or blocks or trains-that is, always with a lot of other cars. Thus, out on the railroad it is usually impractical to act on the detailed information about carload schedule compliance that scheduling systems provide. The experience of more than one railroad bears this out.

At the same time, it is unclear that the improvements in trip time reliability sought by car scheduling offer as much value in reducing shipper non-transport logistics costs (specifically, inventory and stockout costs) as they once might have. The market of the 1990s and beyond is different from the market of the 1950s. Then, boxcars of merchandise were the dominant carload freight, and shipment weights in the 20- to 40-ton range were common. Today, rail carloads are in the 60- to 100-ton range, and fewer and fewer of them involve high-value commodities. That is a different business, one where transport costs are more important than the "non-transport logistics" costs associated with inventory carrying costs and stockout protection.

One of us has developed an analytical method of evaluating shipper costs and has applied a "shipper cost model" to the question of what carload movement reliability is worth. The work shows that the costs associated with movement unreliability have become relatively unimportant in today's heavy-loading, high-use-rate carload markets. Truckload service has become cost competitive with rail in the 20- to 40-ton range, moreover, and with their inherently better service quality, trucks have walked away with almost all of that light-loading, low-use-rate business. Improving carload movement reliability is not going to get that sort of business back, and the large-shipment market has a different cost structure.

A finding that movement reliability has become relatively unimportant may seem at odds with shipper experience during the "meltdown" in railroad service associated with large mergers. In fact, most of today's shippers would not have been greatly disadvantaged by those circumstances if they could have adjusted their safety stocks in advance and had known they were going to have to do so.

We are not suggesting that railroads ignore the reliability problem. We are suggesting that the economics of the problem have changed. Based on experience, we are also suggesting that efforts to improve performance should take a different form because the perfection sought by car scheduling systems is simply unachievable. So, we think railroads should stop trying to make real-time car scheduling systems work and should direct their attention instead to continuous improvement in shipment cycle times. They can do this by measuring and improving the components of that cycle time-pickup, classification, linehaul, delivery-and by rewarding field managers for improvement. Costs will go down. Service quality will improve. Business will go up.

In addition, railroads and their customers should focus on how best to configure supply chains to generate the most value for both parties. Linehaul railroad service is not a logical part of fractured, low-use-rate supply chains. When effective shipping rates are concentrated in the proper part of the supply chain, however, rail can produce the lowest combination of non-transport logistics and transport costs. This can be achieved by cooperative purchasing, transfer and transload distribution facilities, or changing pre-existing logistics systems.

We are not suggesting that railroads should stop scheduling their operations. Train scheduling is essential. So is scheduling connections for cars at yards. We are arguing, rather, that both the railroads and the Association of American Railroads should stop spending money on real-time car scheduling systems. We think there is a better way to run a railroad.

A. Scheffer Lang was the first Federal Railroad Administrator, during the Johnson Administration. Paul O. Roberts, Jr., is a former vice president of SAIC. John P. Sammon is a former senior vice president of Conrail. Their paper on car scheduling has been published by the TRB.