Game Theory Applied to Flight Overbooking Auctions
http://chicagotonight.wttw.com/2017/05/08/flight-overbooked-use-game-theory-get-biggest-payout
This article and video clip from Chicago Tonight talk about game theory to analyze auctions involved in flight overbooking. I didn’t really know this previously, but the idea is that most airlines intentionally overbook the majority of their flights, betting that some passengers won’t show up. Without many refund opportunities (aside from pricey flight insurance policies that most people don’t purchase), airlines are easily able to make extra money simply by overbooking. When everybody shows up, however, things get a little more interesting as passengers are offered compensation to give up their seats.
The article focuses on two airlines and their overbooking policies – United Airlines and American Airlines. When these companies only need to buy back one seat, the two airlines work the same. They set a price and see if anybody takes it. If somebody does, then the game is over. If nobody does, then they raise the price and continue on. But when there’s more than one seat to buy back, United Airlines and American Airlines differ in their policies. United uses what’s known as a Uniform Price Auction. To explain this, let’s say that there are two seats to buy back. If one passenger agrees to an offer of, for example, $200 in flight credits, and then nobody else agrees to an offer until $600 is reached, then both passengers will be given the higher price, $600. Conversely, with American Airlines’ Discriminatory Price Auctions, the same scenario would mean the first passenger would get only $200 and the second would get $600. A game theory concept known as the Revenue Equivalence Theorem states that United and American will actually end up paying out roughly the same amount (on average, across all customers) despite their differing policies. This sounds just like the comparison between First Price and Second Price auctions that we discussed in class, wherein the expected revenue to the seller is exactly the same for these two auctions (as First Price Auctions involve bidding below your value but selling at the highest bid, and Second Price Auctions involve bidding at your value but selling at the second highest bid). It seems that the Revenue Equivalence Theorem spans across many different games.
Another factor involved in airline overbooking is known as the Endowment Effect, a psychological concept. It essentially states that the price you’re willing to pay to give up something changes depending on whether you think you already own it or not. This concept implies that airlines end up paying out a lot more money when the auctions take place on the plane as opposed to taking place before passengers have boarded.
It’s unfair for passengers to be asked to play these games without knowing the rules, yet that’s what generally happens. The majority of customers are unaware of how to play and therefore are unable to properly decide on strategies. It’s tough because passengers with United may think that they should hold out for more money, when in reality their dominant strategy would be accepting the offer as soon as they’re comfortable with it. At that point, their payoff can only increase.
I witnessed one of overbooking situations firsthand before boarding a plane to go back to Cornell after a break last year. The woman at the gate was offering flight credits, and nobody wanted them for a while – so the offer just kept increasing. I recall the number getting up to something ridiculously high, perhaps $800, before somebody decided to accept the offer. I kept debating if I should just accept one of her offers even if it meant missing a bit of school (since it would have meant getting on a much later flight), but I decided to hold out. There surely would have been a point at which I would have accepted the offer, deciding that the money outweighed the consequences, but I couldn’t even decide on a number before the offer was taken. That’s another factor – these things move fast, and it’s tough to make good decisions that quickly. Anyway, though, it was fun to see some of what we’ve learned in class applied to an example that many people don’t know much about. I’ll now be more ready to maximize my payout if I ever find myself in one of these airline “bidding wars,” so to speak. Thanks game theory!