Of this (via MR):

It’s an auction conducted at the airport terminal.  In this auction you are a seller and you are bidding to sell your ticket back to the airline.

Optimists look at this and contemplate the efficiency gains:  this is a mechanism for appropriately allocating scarce space on the plane. Pessimists detect a nasty incentive:  now that the lowest bidder can be bought off the plane the airline has a stronger incentive to overbook.

The pessimists are right precisely because the optimists are right too.

Consider standard airline pricing with no overbooking.  You buy a ticket in advance for a flight next month.  Lots of uncertain details are resolved between now and then which determine your actual willingness to pay to fly on the departure date.  One month in advance you can only form an expectation of this and that expected value is your willingness to pay for a seat in advance.

This is inefficient.  Because, after the realization of uncertainty it could be that your value for flying is lower than somebody else who didn’t buy a ticket. Efficiency dictates that you should sell your ticket to him on the day of the flight.

One way to implement this is to hold an auction on the day of departure.  Put aside the issue that flyers want advance booking for planning reasons.  Even without that incentive, just-in-time auctions solve the inefficiency problem with conventional pricing but airlines would never use them.

The reason is that an auction leaves bidders with consumer surplus (or in the parlance of information economics, information rents.) As a simple example, suppose there is a single seat avaiable on the flight and two bidders are bidding for it.  An optimal auction is (revenue-equivalent to) a second-price auction so that the winning bidder’s price is equal to the willingness to pay of the second-highest bidder.  That is lower than the winner’s willingness to pay and the difference is his consumer’s surplus.

The airline would like to achieve the efficient allocation without leaving you this consumer’s surplus.  That is impossible in a spot-auction because the airline can never know exactly how much you are willing to pay and charge you that.

But a hybrid pricing mechanism can implement the efficient allocation and capture all the surplus it generates.  And this hybrid pricing mechanism entails overbooking followed by a departure-day auction to sell back excess tickets.

The basic idea is standard information economics.  The reason you get your information rents in the spot auction is that you have an informational advantage:  only you know your realized willingness to pay.  To remove that informational advantage the airline can charge you an entrance fee to participate in the auction before your willingness to pay is realized, i.e. a month in advance as in conventional pricing.

Here is how the scheme works in the simple example.  There is one seat available.  Instead of selling that single seat to a single passenger, the airline sells two tickets.  Then, on the day of departure an auction is held to sell back one ticket to the airline.  The person who “wins” this auction and makes the sale will be the person with the lowest realized value for flying.  The other person keeps their ticket and flies.  On auction day, the winner gets some surplus:  the price he will receive is the willingness to pay of the other guy which is by definition higher than his own.  (Delta is apparently using a first-price auction, but by revenue equivalence the surplus is the same.)

But in order to get the opportunity to compete in this auction you have to buy a ticket a month in advance.  And at that time you don’t know whether you are going to win the auction or fly.  The best you can do is calculate your expected surplus from participating in that auction and you are willing to pay the airline that much to buy a ticket. Your ticket is really your entrance pass to the auction. And the price of that ticket will be set to extract all of your expected surplus.

Note that the only way that the airline can achieve these efficiency gains and the accompanying increase in profits is by overbooking at the stage of ticketing.  So the pessimists are right.

(You can write down a literal model of all of the above.  The conclusion that all of your surplus is extracted would follow if travelers were ex ante symmetric:  they all have the same expected willingness to pay at the time of ticketing.  But the general conclusion doesn’t require this:  all of the efficiency gains from adding a departure-day sellback auction will be expropriated by the airline.  That follows from a beautiful paper by Eso and Szentes.  To the extent that fliers retain some consumer surplus it is due to ex ante differences in expected willingness to pay.  The two fliers with the highest expected surplus will buy tickets at a price equal to the third-highest expected surplus.  This consumer surplus is already present in conventional pricing.)