Airline overbooking — which inevitably leads to bumping some passengers — is one of the classic problems to cover in an operations class. It comes down to a simple newsvendor problem. Given a forecasted distribution of how many people will not show up for a flight, the firm has to balance the cost of bumping a passenger with the possibility of flying an empty seat. Now the New York Times says that airlines have gotten ever better at managing the overbooking process (Airlines Look to Limit Bumping, Apr 6), and are leaving fewer and fewer behind:
Even as airlines have substantially cut capacity — meaning fewer but more crowded flights — they have so refined their computer tools over the last decade that they can pretty much predict which passengers will show up for a flight and which ones won’t. In the process, they have gotten better at the science of overbooking a flight.
Last year, 13 out of every 10,000 passengers were bumped on domestic flights — or 762,422 out of a total of 582 million. That was down from over 20 per 10,000 passengers in 1999, according to the Department of Transportation. In over 90 percent of cases, airlines found volunteers to give up their seats in exchange for some compensation.
Here is a graphic take on that data:So why are fewer people getting bumped? I accept that airlines have gotten better at modeling no shows and thus have fewer surprises. That just makes sense. With more tickets being sold on-line (and a decent share of them on the airline’s website), they should have better demographic information and such that should lead to better forecasts.
I wonder, however, what else is in play. The article mentions that over time the industry has imposed additional terms that have forced consumers to commit. For example, you have to ticket a reservation withing 24 hours of making it, so some demand is locked in. I also suspect that increased change fees matter. If you know you will lose a large share of the ticket value by not making the flight, you will put more effort in to making the flight. I also wonder to what extent the airlines have shifted the mix of what they sell. If they are now selling fewer fully refundable tickets, that would also lower the no-show rate.
A final point is that as the airlines have shifted their operations, the cost of bumping a passenger has increased. That is, the pitch of the story amounts to “even though the airlines have decreased capacity in order to have fuller flights, they are bumping fewer people.” But if you routinely have fuller flights, it is much harder to accommodate bumped fliers. That is, the basic economics of the newsvendor problem would get you less overbooking even if your forecasts haven’t improved.