Anyone who has ever spent hour after aimless hour in an airport lounge waiting for a flight to be announced knows how frustrating airline delays can be. Back in November 2000, the U.S. Department of Transportation reckoned that the 11 largest U.S. carriers had a quarter of all flights arriving more than 15 minutes late. Though fewer people have been flying after the terrorist attacks on the World Trade Center and the Pentagon last September, the situation with delays has hardly improved. The Wall Street Journal reported in March that the U.S. Department of Transportation still gets 10,000 complaints a year, mostly about airline delays and similar problems.


Does that dark cloud have a silver lining? Possibly: A research paper by Christopher Mayer and Todd Sinai, professors in Wharton’s real estate department, debunks the notion that air traffic delays are an unmitigated scourge. Their research shows that most of the delays due to air-traffic congestion are evidence of trade-offs made by an air travel system in which passengers get something in return for congestion – more frequent service to a greater number of destinations.


Conventional wisdom on flight delays maintains that they result from congestion, which is a result of lack of capacity at airports. “The prevailing thought on the delays is that there is only so much capacity, and every carrier is vying for runway space,” says Sinai. “Everyone wants this limited resource, and the airlines over-schedule flights as a result.”


But the conventional wisdom – as is often the case – is misleading. While congestion or absence of capacity may explain why flights get delayed at airports like New York’s LaGuardia or JFK, which are not dominated by a single airline carrier, it doesn’t indicate why airports such as Philadelphia, where a single carrier (US Airways) dominates, suffer from some of the most aggravating delays. If the conventional notion were true, such airports should have the least congestion problems.


Mayer and Sinai offer an alternative explanation: hubbing. When an airline carrier dominates at an airport, it uses the airport as a hub and operates flight routes along so-called spokes to outlying cities, and this has certain implications. “In order to operate a hub efficiently, you have to allow as many connections as possible for people who are going in and out of the hub airport,” says Mayer. “You have to bunch the flights together in order to allow people relatively short connecting times with a large variety of destinations.” That bunching of departures and arrivals leads to peaks of air traffic that causes congestion and delays. The professors’ research notes that airport congestion due to multiple airlines over-scheduling service only accounts for a modest level of airline delays, while hubbing acts as the main contributor to delays overall.


For their analysis, the authors used U.S. Department of Transportation information “on flights from 1988 to 2000 by all major air carriers with more than a 1% U.S. market share.” This data resulted in the pair analyzing more than 66 million flights.


The authors measure congestion based on the extra travel time a passenger spends on a given flight compared to how long that flight would take in the absence of other air traffic. They prefer this method to the common practice of defining an “on time” arrival as being no more than 15 minutes later than scheduled. The latter definition remains a rather arbitrary measurement, as airlines may simply adjust their scheduled travel time to reflect the additional congestion. “Sure your flight arrived on time,” says Sinai, “but it still took you 30 minutes longer to get there because the airline padded its schedule.” 


Their results show that flights starting at large hub airports needed on average 7.2 minutes more to reach their destination than flights originating at a non-hub airport. Planes flying to a hub airport are delayed up to 4.5 more minutes, on average.


Mayer and Sinai also found that most of the congestion at hub airports is incurred by the airline that hubs at that airport. Airlines that do not hub at that particular airport experience little, if any, congestion-related delays there. Says Sinai: “The hub airline has no choice but to have all its flights arrive and depart in a short span of time, so their passengers can connect. In a sense, their planes get in the way of their other planes, not their competitors’ planes.” Since non-hub airlines do not have any connecting passengers at another airline’s hub airport, they generally avoid scheduling flights during the hub airline’s peak activity.


While weary airline travelers might find the idea of coming to terms with increased travel time maddening, Mayer and Sinai suggest that congestion remains a necessary outcome of more convenient and far-reaching airline service. Says Sinai: “Airlines make a trade-off between passengers’ desire for connecting service to more destinations and the greater delays the additional hub traffic would create.”


This argument, which suggests a compensating benefit to air traffic congestion, distinguishes the authors’ conclusions from prior research on the subject. Mayer points out, “Hub airlines inflict delays upon themselves and are fully aware of the consequences. This is significantly different than the view that, because they do not take into account the effect of their scheduling on other carrier’s flights, airlines create an inefficiency that the government must correct.”


Mayer cautions that public policy initiatives that ignore the reason behind hubs’ contributions to delays could have the perverse effect of making the air-traffic system less tuned to consumers. While the authors find some support for very limited government intervention to decrease airline delays, such as a small “congestion tax” intended to make airlines realize the effect of their flights on other airlines, Mayer points out that too high a tax could cause airline carriers to reduce the hub operations that benefit many travelers.


Today, Sinai says he has “come to accept some of the delays as a way in which the system must work.” In essence, the need for airline service to an ever-increasing number of destinations outweighs the burden of delays. The authors note, “Delays are not necessarily evidence of a socially inefficient outcome, but might instead reflect the optimal use of scarce runway capacity.” Sinai adds that, “government attempts to reduce delays at hubs should not necessarily be viewed as correcting a problem in the market.”


While Sinai supports increasing airport capacity as a way to improve efficiency, he points out that delays still might not disappear. “Suppose the federal government builds additional runways. Then, the hub airline will increase the number of possible connections at the hub. Consumers will have more choices of where to fly, but there may still be delays at connection times.”


Strangely enough, the possibility for increased delays could result from additional capacity. “You could imagine an airline adding a new hub into an airport with the increased capacity,” says Mayer. Delays associated with the hub activity would soon follow.


Mayer and Sinai note that airlines have no real incentive to provide information on their predictable delays in arrival and take-off of planes. In fact, airlines do not fully adjust their schedules for increased congestion due to hubbing, a problem that has been getting worse over time. “That passengers purchase tickets based on scheduled travel time rather than actual expected travel time might contribute to an airline’s unwillingness to fully account for expected delays in their schedules,” the paper concludes.


The authors’ forthcoming research takes a look at the impact of industry competition, publication of the rankings of on-time airlines, and Internet access to actual flight data as incentives for airline carriers to present more reliable scheduling data. Meanwhile, those who are forced to cool their heels at airport terminals might console themselves with this thought: Not all delays are evil. As the poet Milton said, they also serve who only stand and wait.