Origins of Airline Oversales Auction System

Travelers being involuntarily bumped from flights for which they possessed a reservation and paid-for ticket was long a difficult problem in the airline industry. Until 1978, each airline had arbitrarily removed excess passengers with any selection policy it chose. For example, United Air Lines instructed its staff to bump old people and armed services personnel, on the assumption that they would be least likely to complain. But this policy produced trouble and outrage for many involuntarily bumped passengers.

The airlines publicly proclaimed that they never intentionally oversold a flight. But at the same time, they circulated among their employees elaborate instructions about how to handle oversales and bumping incidents. Here are some excerpts from the American Airlines Operations Manual (various 1974 and 1975 dates): "American Airlines never deliberately causes a passenger to be oversold. We tolerate a limited number of oversold and inconvenienced passengers only because we must allow some margin for error in our operation. . . . Reservations will select for removal the most recently sold locally boarding passenger, whenever good judgment dictates that this passenger will be less inconvenienced than some other passenger, except when undue hardship will be incurred. . . . If none of the alternatives provide a satisfactory solution and there are too many passengers on board, advise all passengers that their flight cannot be operated with more than the authorized number of passengers and request the proper number to deplane. . . . The passengers who voluntarily deplane should be given special treatment, appropriate interrupted trip expenses, and alternate transportation as soon as possible.

"In extreme cases it may be necessary to actually cancel the flight until all passengers have deplaned, after which, the flight can be reoriginated as an extra section and conformed to capacity. This action should be used as a last resort."

The manual included this precaution: "Never give an oversold passenger anything in writing which admits an error on the part of American Airlines." Marvin Rothstein, a member of American's operations research group later wrote about their overbooking system that "the scheme was never spoken of openly as overbooking; instead, we employed the euphemism 'revenue coordination."'

The Eastern Airlines procedures included seeking a volunteer to deplane (though without recompense). If that ploy did not work, passengers who had not yet actually sat down were the first to be bumped. And the employees were instructed to "never use the words 'oversale' or 'overbooked' in conversation with those denied boarding or within hearing distance of anyone except company employees involved."

The oversales problem fed on itself. The larger the probability of being bumped, the greater the passenger's incentive to make multiple reservations on different flights under a variety of names to insure getting on at least one. And the greater the extent of this practice, the greater the need for the airlines to overbook to ensure a decent load factor. The vicious cycle got worse and worse.

In 1978, the Civil Aeronautics Board (CAB) mandated an auction plan to reduce the number of involuntary bumpings. Since then, whenever there is an oversale, airlines have been required to ask for volunteers to wait for a later flight, using whichever incentive system the airlines choose. This article is about the history of how that regulatory reform came about, and its results. I hope that the reader will pardon a bit of pride as I tell the story.

I learned in the late I 950s about the United Air Lines bumping policy from a friend who had been a stewardess, Betty Glad. The conversation lay dormant in my memory until sometime around 1965 or 1966 when I heard at a party a sad saga of how someone had been summarily kicked off a plane and forced to endure a costly and unpleasant delay. The next day when shaving it occurred to me that there must be a better way; indeed, a market could solve the problem by finding those people who least mind waiting for the next flight. The airline flight personnel would simply need to ask each ticket holder the lowest amount he or she would be willing to accept to wait for the next plane, and then select the necessary number of low bidders. The practical details fell into place before the shave was complete.

The scheme is simply a reverse auction, asking each ticket holder to write down the lowest amount she or he would be happy to accept in return for waiting for the next flight. In case of an oversale, the airline agent would simply proceed from lowest bidder upwards until the required number of bumpees is achieved. Low bidders would be given the amounts they bid (or the amount that the highest-bidding bumpee bid, if that version is deemed more equitable and attractive) and they take the next flight, happy about it. All other passengers fly as scheduled, also happy. The airlines can overbook more, which aids them, too. Literally everyone is better off except the railroads and bus companies who compete in the transportation market. A cruder version is for the airline to cry a price and to ask for takers.

In 1966 and 1967 I wrote to all the airlines suggesting the scheme. The responses ranged from polite brushoffs, to denials that they overbooked, to assertions that the scheme could not work, to derision.

"(Seriously, it was refreshing to have an objective view of airline reservations, and I would like to discuss it further with you over lunch next time you visit New York!)"

The lash of competition in the airline industry (as regulated by the CAB) simply was not severe enough to move anyone to bother with the scheme, even though it could be adopted unilaterally and with the possibility of a promotional coup.

I described the idea in a professional journal article just two pages long entitled "An Almost-Practical Scheme to Solve the Airline Overbooking Problem." The article noted that in extreme circumstances airlines had occasionally sought volunteers with satisfactory results.

The scheme was labeled "almost practical" partly as an (unsuccessful) eyecatcher, but more importantly because-as the article explained- its chances of adoption were almost nil despite its practicality and ease of implementation and operation. I speculated that the lash of competition in the airline industry (as regulated by the CAB) simply was not severe enough to move anyone to bother with the scheme, even though it could be adopted unilaterally and with the possibility of a promotional coup.

Former airline executive Blame Cooke later offered another explanation in a letter to me: the NIH factor ("not invented here"). "I greatly fear that your overbooking auction plan suffers from a flawed premise and a fatal defect. The flawed premise is that you assume that airline management and regulation is a rational exercise. It is not; it is more accurately described as an exercise in applied insanity. The defect is your plan offers a market-sensitive and sensible solution to a real problem but a solution not conceived by an airline. Accordingly, the idea must be disallowed since it is well established in airline marketing that only ideas which originate within the airlines are permissible.

"Somewhat more seriously, your plan can't possibly be tried since the CAB would have to provide an exception from its denied boarding policy-a process which theoretically could happen but practically would probably take five years.

The scheme did not find publication easily in a professional journal. One never knows why an article does not catch referees' interest. Possible drawbacks in this case were brevity (only two pages), simplicity (equated with unimportance by many), absence of mathematics, and lack of connection to any ongoing work tradition among economists. The only apparent advantages were some novelty in the form of the auction, the scheme's rarity as a policy that constitutes a "Pareto improvement,' (in economists' jargon, a situation in which everyone benefits) and brevity and simplicity (an advantage to my mind, at least). Eventually in 1968 it found its way into print in the Journal of Transport Economics and Policy.

The scheme was received mostly with derision, both by the academy and by industry. Even the table of contents in the Journal of Transport Economics and Policy referred to the scheme as "lighthearted." With engaging frankness, Rothstein of American Airlines later described his reactions at the time: "In 1970 . . . I discovered a short article in an obscure journal. that contained a remarkable proposal for solving the overbooking problem: if too many reserved passengers show up at flight time, the airline agents should conduct an auction among them. . . Simon claimed that his solution would satisfy everybody: those left behind, those boarded, and the air carriers too. Nevertheless, he stated that 'of course this scheme will not be taken up by the airlines' because it will 'not seem decorous' or they will allege 'administrative difficulties.'... Amused, I imagined that the world would hear nothing more of the 'almost practical solution,' as he termed it."

Over the years, whenever there were CAB hearings on oversales and bumpings, I wrote to anyone who I thought might have some influence. Before the hearings scheduled in 1977 or 1978, I systematically wrote to a great many more persons-legislators, trade associations, financial analysts, Ralph Nader's Aviation Consumers Action Program, and so on. Still, no one else supported the idea.

In all my discussions on the idea, I insisted that one should not decide about it in the abstract, or even on the basis of hypothetical experimental data, but instead should conduct an actual experiment. But I was unable to persuade any airline (or the CAB) to conduct an experiment for even one day on a single airline at a single airport at a single boarding gate-an experiment that I believed would be sufficient, even with the inevitable breakdowns in any new activity. Rather, the industry and the bureaucrats preferred to insist on the basis of their "logic" alone that the scheme could not work.

I tried to convince eminent economists of the merits of the idea. I asked all the ex-presidents of the American Economic Association and the current members of the Council of Economic Advisers to write an endorsement to the CAB. Several did indeed write, including some very forceful statements.

But the reactions of the man I consider the greatest economist now alive (and the greatest spirit), and another of those economists whose work I honor most, were unusual and therefore particularly interesting. The latter, George Stigler, wrote that the scheme would not work because the passengers would form cartels and hold up the airlines for very high prices.

"Since your scheme strikes me as intellectually admirable and administratively impossible, I shall not write to the CAB. You should explore the possibilities of collusion by a group of 40 unemployed people," Stigler wrote.

Milton Friedman wrote as follows: "If the plan is as good as you and I think it is, I am utterly baffled by the unwillingness of one or more of the airlines to experiment with it. I conclude that we must be overlooking something. I realize that you have tested this quite exhaustively, and I have no reason to question your results; yet I find it even harder to believe that opportunities for large increments of profit are being rejected for wholly irrational reasons."

Surely nothing would have come of the scheme except for an extraordinary happening: For the first time in history, an economist, Alfred Kahn, was appointed to head the CAB.

Upon hearing that he was a candidate, I wrote to Kahn, and before his appointment he wrote me that the idea made obvious economic sense. Kahn announced something like the scheme in his first press conference. He also had the great persuasive skill to repackage it as a "voluntary" bumping plan, and at the same time to increase the penalties that airlines must pay to involuntary bumpees, a nice carrot-and-stick combination.

The voluntary bumping plan has now been in operation for 14 years, with excellent results. Everyone is delighted with the various versions of the plan that the airlines use. The people who care least about waiting for the next plane select themselves to get a payoff that they prefer to flying as scheduled. Neither arbitrary airline agents nor bureaucratic policy decide who gets thrown off. And the airlines have increased their efficiency by being able to safely overbook to a much greater extent than before, and therefore they now fly with fewer empty seats.

Year by year since 1978, the airlines have been increasing the extent of overbooking, from 6.4 oversales per 100,000 passengers in 1978 to 15.1 in 1991. This practice-which the law and the Naderite reformers condemned-has enabled the airlines to raise the capacity level at which planes fly, hence increasing their efficiency and ultimately lowering costs. Yet with the volunteer system the rate of involuntary bumpings per 100,000 passengers has fallen sharply- from 6.4 per 100,000 in 1978 to 1.1 per 100,000. That is the most important measure of the improvement in airline service from the point of view of passengers. To put it in absolute terms, as total volume almost doubled, the number of people booted from planes against their will has dropped from 155,507 to 46,879, and the number who happily accepted a windfall they never expected rose from zero in 1977 and 15,733 in 1978 (the first year of operation) to 599,205 in 1991. Data showing the complete results of the volunteer scheme-especially the numbers of involuntary and voluntary bumpees each year since 1978-are displayed and analyzed in a forthcoming article in the Journal of Transport Economics and Policy.

Not a single flaw in the system has turned up, to my knowledge. No one has claimed to be damaged by it. Even Ralph Nader praises it now, though he opposed it before adoption.

There have been interesting differences in the speed of adoption among the individual airlines. American embraced the system from the start, publicly stated that the vo1unteer system is an improvement in airline service (see quotations above), and quickly reached a high ratio of voluntary/involuntary boardings. In contrast, United was very slow to reduce involuntary bumpings; it took until 1985 for United to reach the rate of involuntary bumpings that American reached in 1980-81. Other airlines were spread across that spectrum.

After the scheme had been working successfully, I wrote to Milton Friedman roughly as follows (the letter has disappeared): "About the airline oversales scheme: I'm not given to 1-told-you-so's, but... I told you so." And he replied (in a fashion that will surprise no one who knows Friedman in person or by correspondence): "Many thanks for your 'I told you so; you are entitled to it."

The volunteer system for handling airline over-sales exemplifies how markets can improve life for all concerned parties. In case of an oversale, the airline agent proceeds from lowest bidder upwards until the required number of bumpees is achieved. Low bidders take the next flight, happy about it. All other passengers fly as scheduled, also happy. The airlines can overbook more, making them happy too. Literally everyone is better off except the railroads and bus companies who compete in the transportation market-a true Pareto improvement, which is rare indeed. As theory and empirical study predicted, the rate of oversales has increased since the system was mandated in 1978, while involuntary bumpings have greatly diminished, and capacity utilization has increased which results in cheaper fares to passengers.

Selected Readings

Simon, Julian L. "An Almost-Practical Solution to Airline Overbooking." Journal of Transport Economics and Policy. Vol. II, May, 1968.

Simon, Julian L. "The Airline Oversales Auction Plan: The Results." Journal of Transport Economics and Policy. Forthcoming.

Simon, Julian L. and Visvabharathy, Ganesan. "The Auction Solution to Airline Overbooking:The Data Fit the Theory." Journal of Transport Economics and Policy. Vol. III, September, 1977.

Origins of the Airline Oversales Auction System

Julian L. Simon