Photo: Oliver Holzbauer/Flickr (Creative Commons)
Airline overbooking isn’t evil. In fact, if done properly, it’s good for airlines, good for customers, and good for the environment. Sold seats are clearly good for airlines and their shareholders. High utilization is good for customers because it reduces seat costs for airlines which normally operate at single digit profit margins. In the consumer airline industry, customers typically do enjoy the gains from increased operator efficiency. Full seats are good for the environment since the fuel burned and the emissions produced are less per airline traveler. By overbooking “done properly” I mean there have to be controls in the system to ensure customers get what they bought and that airlines make decisions that are in the best interests of the environment, customers, and their shareholders.
The airline seat overbooking system as currently operating in the United States puts all the control in the hands of the airline and they get most of the gain rather than customers. United Airlines, in what appears to have been a truly exceptional entry into the corporate Darwin Awards, has been heavily covered in recent news for forcibly dragging a paying customer off a plane. It’s hard to read the news and not see numerous mentions of this event but here’s a good example: United Airlines Offers Refunds as Outrage at a Violent Removal Continues. In this event a 69-year-old physician was forcibly removed from the United Airlines flight. The customer was injured, will take legal action, and even though the customer is just about guaranteed to win a large settlement, it’ll be dwarfed by the negative publicity of this event on the airline.
Digging deeper into the United removal of this paying passenger, the details are actually a bit worse than the explanation that the plane was over-booked. United reports the plane wasn’t overbooked but they needed to transport crew so they elected to forcibly remove paying passengers. After the event, United Chief Executive Office, Munoz, is reported to have said in an internal memo that the passenger, a physician was “belligerent” (United Airlines says controversial flight was not overbooked; CEO apologizes again). The CEO has since reversed this statement and now says “no one should ever be mistreated this way.”
Clearly United has serious issues starting at the very top of the company that won’t likely be solved without management changes but that’s not my focus here. What I’m arguing is that commercial airline seat overbooking isn’t actually a bad way to manage expensive resources and it needn’t be customer unfriendly. I’m going to argue that lawmakers need to make changes to make the overbooking system work properly, but the natural inclination to ban overbooking because of the clumsy handling of this event by United isn’t necessarily the best approach.
Flying commercial aircraft is expensive and whenever an industry needs to sell a capital-intensive resource, utilization becomes one of the most important factors to the long term success of the industry. I work in cloud computing where data centers run several hundred million dollars each and the most successful providers are spending billions on capital expenses each year. Every basis point is huge and a few percentage points difference in utilization could be the difference between excellent profit margins or operating negative. Airlines are one of the prototypical cases of capital-intensive industries.
Airlines make their money moving people and cargo. The cargo hold and the cabin have fixed divisions but the overall plane is weight-limited, so the cargo sales and passenger sales are not actually fully independent. However, selling all the passenger seats is the goal of every passenger-carrying airline and successful airlines operate at very high seat-utilization levels. This is done using a combination of yield management and overbooking. Yield management is a pricing strategy based upon computer models of customer behavior on specific routes. All successful airlines do it and the goal is to sell all seats and to sell them at the highest possible profit.
These algorithmic pricing systems do a remarkably good job of selling all the seats and getting competitive prices from every seat sold. However there are latencies in the sales channel and other issues that prevent yield management from solving the problem entirely and, even if they did, some customers simply don’t show up. Since an empty seat represents a substantial source of lost revenue, no airline wants to fly with an empty seat. The next tactic after yield management to ensure the maximum possible load of paying customers is to oversell seats. In this approach, the airline sells more seats than they have available knowing the statistical likelihood of customers not showing up.
Overselling sounds customer-unfriendly – why allow an airline to sell a seat that it doesn’t actually have available? It’s a great question and the United incident mentioned above clearly underlines that poor leadership certainly can yield very poor results but that’s a management constant rather than specific to overbooking. But, overbooking as currently practiced, does have problems.
What’s broken here is the airlines are allowed to simply fail to honor their commitment and forcibly remove a paying passenger from the airplane. The system would work fine if the airline had to pay passengers to accept alternative transport and if there were no bounds on what this might cost. This puts an economic safeguard on the airline to only sell that which they are highly likely to have. If they don’t have sufficient seats, they are obligated to offer alternative transport on other airlines, pay customers whatever it takes to get off a plane or, in the worst case, charter transport for the remaining passengers. Certainly there is always a price that would get me off a plane. Recently the US President said the same (David Dao, Doctor Dragged Off Plane, Files Court Papers Demanding Airlines Preserve Evidence). I suspect he is a bit more expensive than I am, but it turns out that in the vast majority of times, a few hundred dollars will open up the needed seats. Sometimes it might cost more but that’s the cost of over selling
What about the customer that don’t show up? Doesn’t there have to be some factor to prevent them from booking seats and not showing up since they are adding considerable risk to the airline? I agree. Customers should have to pay a no-show penalty. It’s a competitive market and the airline should set it to whatever they believe covers their overbooking costs or encourages the desired behavior. Some airlines might chose to set this “no show” cost at free for some tickets, some passengers, or even all passengers. But it would be the airline’s choice to set the cost of a no show. Currently they mostly don’t charge now but, again, that’s their choice.
In this proposed model airlines have to move all paying passengers or make arrangements with some of the passengers to accept compensation for different travel arrangements. They are free to charge no-show passengers a fee if they so choose to limit the airline passenger no-show risk. In this model, there is economic pressure on the airline to fly full but to also satisfy the needs of their customers. There is also economic pressure on passengers to show up for the flights they book. It’s a good model that helps airlines and travelers. Airlines can overbook and customers can no-show but both have to pay the cost of that action. The current model where an airline is not obligated to fulfill their commitment on selling a ticket by either transporting the customer or offering them compensation acceptable to the customer doesn’t work, and the United handling of this most recently situation really underlines that failing.
Airlines seat overbooking isn’t evil but allowing airlines to cap the costs and be able to just refuse a serve a paying customer clearly doesn’t work.
Update: On the topic of customer centricity, Jeff Bezos, annual letter to shareholders is an excellent read: 2016 Amazon Letter to Shareholders.
Update: Some of the right things are starting to happen in the airline industry:
From: LA Times
- United Airlines will offer up to $10,000 when a traveler voluntarily gives up a seat on an oversold flight
From: Washington Post
- Delta Air Lines said gate agents can offer up to $2,000, up from a previous maximum of $800, and supervisors can offer up to $9,950, up from $1,350
- American Airlines updated its Conditions of Carriage after the United incident, stating that it “will not involuntarily remove a revenue passenger, who has already boarded.” The airline does not set a cap on compensation for passengers, but said gate agents work with customers to ensure they set compensation amounts properly to obtain the correct number of volunteers.
Complete rubbish. Overbooking is vile evil greed that exists only to lard the pockets of airline CEOs. If a passenger does not turn up, the seat still belongs to them as they paid for it. Supporting the inhumane act of forcing a passenger off a flight is a bit like saying, let’s shoot every fifth passenger dead and book the freed seats up for profit, because, after all, if done properly, it’s good for airlines, good for customers, and good for the environment. Seats freed up by capping passengers are clearly good for airlines and their shareholders.
The point I make in the article is over-booking is fine (in my view) if airlines have to pay people to get off the plane and keep offering more until they get the seats needed by those they have agreed to fly. In that model, over-booking is self-policing and self-correcting. In the model rarely practiced in North America where an airline can refuse to transport someone with a purchased ticket puts all the power in the hands of the airline and the model is open to the kinds of profit seeking abuse that concerns you. I agree that model doesn’t work.
Same way, power oversubscription is also used in datacenter field, that’s a good way for datacenter to utilize the power and lower the operation cost of datacenter.
Yes, same principle. Good analogy.
I disagreed with your original central point that “there is nothing wrong with overbooking”. However, you clarified this in the comments. Your point now seems to be that: there is nothing wrong with overbooking, as long as airlines are not allowed to cap (limit) the buy-out cost if they overbook.
I wholeheartedly agree with this version. Airlines should be required to pay whatever cost is required by the group of denied/inconvenienced passenger(s). At the time the contract was signed (ticket was purchased), both parties promised to do their part (to transport the person+luggage at a specific date/time from place A to place B, in exchange for the already paid fare). If the airline, for whatever reason, cannot do their part, they must offer whatever the other party (to that contract) requires. This is basic business law, falls under the fairness doctrine, and is minimally expected of customer-facing entities.
Anything less is unconscionable! This is one reason why regulations are important. You may remember, about a decade ago we had several incidents where passengers were kept inside a grounded plane on the tarmac, for up to 8 hours, without food, water, air conditioning, or the ability to use clean restrooms. Government regulations put a stop to such behaviors. Similarly, we now need regulations to bring-back fairness into the pricing (and penalty/compensation) for overbooked seats. Airlines will not solve the problem — as we saw when they kept limits and conditions on their “penalty” costs. For example, if someone is denied a seat today, and then given a “free” replacement ticket, that “free” ticket includes the bonus privilege of their being selected AGAIN (on their replacement flight) to be denied again… and again… and again….. ad infinatum. Legally, such a passenger will have no recourse!
One way to do overselling better would be to predict the likelihood of a flight being very popular and important for people – like your Christmas Eve flight example – and then reduce or eliminate overselling for such flights. That would again reduce the chance of customer aggravation and the amount that might need to be paid to repurchase the ticket.
Good idea David. The major airlines do this today as part of the algorithms to figure out likelihood of people showing up for a given seat on a specific flight. Certainly there are flights where it costs more to get someone interested in accepting compensation and alternative transport.
2017.05.02 Update:
*United Airlines will offer up to $10,000 when a traveler voluntarily gives up a seat on an oversold flight
*Delta Air Lines said gate agents can offer up to $2,000, up from a previous maximum of $800, and supervisors can offer up to $9,950, up from $1,350
*American Airlines updated its Conditions of Carriage after the United incident, stating that it “will not involuntarily remove a revenue passenger, who has already boarded.” The airline does not set a cap on compensation for passengers, but said gate agents work with customers to ensure they set compensation amounts properly to obtain the correct number of volunteers.
Hi James,
Very interesting post as always, thanks for sharing your thoughts with us!
Regarding the no-show penalty, though, I’m unconvinced that it would be financially interesting for the carrier, regardless of any image or non-financial aspect.
On a flight with 100 seats, priced 100$ per seat, without overbooking the airline can generate 10k$ in revenue. Say 10% of seats are unoccupied on average, this means that just overbooking these 10% the airline can sell an extra 11 tickets, and generate 11.1k$ in revenue.
Now, if the carrier decides to introduce a no-show penalty, more doubtful passengers would chose not to buy the ticket. We don’t know the elasticity of this relation, but let’s start at a 1:1 relation in percentage. If the penalty is 10% of the initial price, no-shows decrease by 10%, and so forth. A 10$ penalty would mean only 9% no-show rate, meaning only 9 overbooked ticket sales worth 10.9k$, and 90$ in penalty fees… Which is less revenue than without the penalty fees!
In fact, the relationship would have to be pretty flat for it to make sense in this case, with a ratio of penalty% to no-show% over 2:1…
Furthermore, if the “certain” passengers are indifferent to the price difference, the “doubtful” passengers aren’t. Carriers with a penalty are going to push these “doubtful” passengers away (decreasing their no-show rate), but this is a huge opportunity for other carriers to increase their revenue by selling tickets that these doubtful passengers won’t use! A competitor to the example in the previous paragraph would see his no-show rate rise from 10% to, say, 11%, meaning they can sell 1 extra ticket and generate 100$ incremental revenue without even doing anything, whilst the competitor has decreased revenues!
Of course, in the real world spikes in occupancy and other factors (including image) are very real and complex, but the case for no-show penalties doesn’t seem intuitively interesting. In fact, it looks like the only rational business decision would be to set this penalty to 0$, as the carriers are in a sort of “prisoner’s dilemma”…
I agree. I’m allowing that some carriers may elect to charge for “no show” passengers. I think it should be 100% there decision but I think there is a good chance that many airlines would elect not to charge. From the original article: “Some airlines might chose to set this “no show” cost at free for some tickets, some passengers, or even all passengers. But it would be the airline’s choice to set the cost of a no show. Currently they mostly don’t charge now but, again, that’s their choice.”
Hi James. I am a new reader to this site. I enjoyed both of your articles that I’ve read so far, and consider your writings to be thoughtful and informed.
Though, I must respectfully disagree on some point(s) in your article and related comments. While I agree the dynamic model you portray is correct, of an industry with high-amortization costs for capital equipment/assets, where corporate/industry profits are directly correlated to utilization efficiency. To your credit, you do say that not always do airlines, and not all airlines, pass on/translate gains in utilization into reduced prices for their customers. The U.S. airline industry is now an oligopoly, and they tightly control what has become a basic necessity (akin to an essential utility) — long-distance transportation. Unfortunately their control favors their own profits, over the discomfort and onerous/unfair contract terms that they make their customers endure. In the case of contract terms (conditions of carriage), most airlines now won’t even honor their commitment to provide the transportation for which we have paid — allowing them to force-ably leave passengers stranded overnight in a far-away city, with little if any compensation.
I disagree with your general comments that overbooking is not a problem. The recent pronouncements by several airlines, that they now will permit up to $10K in paybacks to “denied” passengers, we can safely conclude that their economic model allows them to pay such “high” costs — for multiple seats per flight. This lets us safely assume that major carriers can easily absorb $50K to even $100K in “lost revenue” per plane, though only occasionally. I notice that they limit these paybacks to only “fare-paying” passengers — which is another game they play, whereby they exclude “frequent flyer” passengers who have more than paid for each “free” seat through their multiple prior purchases; such “previously paid” passengers can be denied even under their new rules.
The airlines’ financial model provides significant profit per seat, including for empty seats. The only seats on which they may lose revenue/profit are those they sell through the secondary market (such as Priceline’s “make your offer” pricing). Though, this is now quite rare due to their propensity to overbook each flight with up to 5% of the plane’s seat capacity.
I also disagree with your saying that “Some airlines might cho[o]se to set this ‘no show’ cost […] of a no show. Currently they mostly don’t charge now….” I must disagree on this. In my experience, there are two factors related to the “no show” charge:
(a) most airlines actually do have a “no show” charge which they implement through a “no refunds” policy (which effectively forfeits the ticket price if someone is a “no show”) and through “rebooking fees” they charge in case there is a valid reason for reissuing a ticket.
(b) the base price for tickets includes a “risk” component, that allows refunds or reissuance. Tickets purchased by corporate customers are always fully refundable and thus are usually 3x-5x the cost of an equivalent refundable seat purchased at the same time. This increased margin (which is full profit for the airline) acts as a “risk eliminator” that the seat will go empty. If a corporate customer has bought the seat, and at the last minute it is refunded — leaving the seat empty, the airline had already earned their normal cost + a 2x-4x profit on that seat. The fact that it was refunded has no effect to the airline. However, if a seat is purchased as non-refundable, it is sold at a discounted fare. If that seat is abandoned (the passenger is a “no show”), again the airline loses nothing. The airline’s pricing model incorporates the risk of a “no show”, and already minimizes/eliminates any revenue/profit loss due to the seat becoming a “no show”. When we also include the additional fees they charge for rebooking/reissuing tickets, their revenue model actually gives them a benefit from “no shows” — they make money on it.
I believe your claim that “most [airlines] don’t charge now” for “no show”, is incorrect. Major carriers easily recoup the cost of each seat, whether the seat was for a refundable ticket that is refunded at the last minute, for a “no show” passenger (who will lose their full fare), or for any other reason.
Sanjay said “The recent pronouncements by several airlines, that they now will permit up to $10K in paybacks to “denied” passengers, we can safely conclude that their economic model allows them to pay such “high” costs” — I don’t follow your disagreement. This approach either pays passengers or, if the airlines were to find it excessively expensive, they would overbook less. Self corrected.
Generally no where close to $10k will be required to find a passenger willing to accept the compensation. I like any system that that allows profit from overbooking but puts the cost of getting it wrong with the same party that is profiting from the practice, the airline. With costs and profits aligned, the system will work.
There is overbooking/oversubscription everywhere. Money comes to mind – fractional reserve banking is based on the premise that not everyone will turn up at the same time at the bank (aka bank run). Electric grids assume that not everyone will plug in at the same time. All applications of efficient use of queueing theory.
In the case of airlines, I agree that airlines control the market for seats too tightly. There should be a secondary market where someone can offload their tickets for a haircut/gainto either third parties or back to the airline. Sounds like a version of scalping, but scalping is a very useful market clearing mechanism though it gets a bad rap.
You hit the key point: in any market with very high capital costs, using the resources purchased becomes the difference between profitability and massive losses. You’ve hit on the key reason I’m so interested in the airline industry. It’s not because I fly frequently although I do — my core interest is in studying the airline industry as an example of a mature capital intensive industry. I work in cloud computing and we face some similar challenges: data centers cost 10x what an airplane costs and, just as in the airline industry, the difference between respectable profits and massive losses is a knife edge where tiny changes in resource monetization and utilization can make or break a business. Any industry facing this sort of challenge is super interesting to me.
In fact, I’m sufficiently interested in the problems faced by the airlines that I would actually enjoy spending a couple of years working at one. I’m convinced there exist algorithmic solutions that would allow current levels of seat monetization, and in the cases of many airlines, higher but without unhappy customers. Even accepting my proposed rule that all customer sold a seat should get timely passage or compensation they prefer with later passage, we can have 100% happy customers and still better monetize each flight.
There are lots of alternatives available. You mentioned opening up a secondary seat market. I mentioned investing more deeply in yield management systems but only doing it under the premise that all customers get passage or compensation they want more. There is lots of room to improve current systems and these improvements don’t have to come at the cost of customer satisfaction.
> Since an empty seat represents a substantial source of lost revenue, no airline wants to fly with an empty seat.
> What about the customer that don’t show up? Doesn’t there have to be some factor to prevent them from booking seats and not showing up since they are adding considerable risk to the airline? I agree. Customers should have to pay a no-show penalty.
Until now, I never understood what people in the US meant about an empty seat being a source of lost revenue. In New Zealand, with lower class fares, you book the ticket and whether you get on or not, the money is paid and you can’t get it back. https://airnz.custhelp.com/app/answers/detail/a_id/2480/~/what-happens-if-i-no-show-for-my-flight%3F
Adding a penalty for no-shows seems like a very sensible start to remove the incentive for Airlines to overbook, and put the penalty for missing a flight with the person that missed the flight.
Even though it is counter intuitive, an empty seat represents lost revenue even in New Zealand. Sure, all seats are sold but if one seat is unsold, it is still possible to get more revenue from that trip by selling the empty seats.
One approach is to give it to a stand by passenger. In this approach, a passenger gets a high discounted ticket but runs risk that they may not get on the flight. It’s a bit more revenue for the carrier so better than flying empty. Another approach is to over-book in which case the person filling the empty seat is a potentially a full fair passenger. Much better for airline revenue but it carries a higher risk that some passenger with a purchased seat will arrive and not find the airline with a seat available. Overbooking is higher revenue for the airline but increased downside risk for the customer.The usual approach to resolving the fairly common seat shortage is the airline offers compensation sufficient to convince some customer prefer the compensation rather than the seat.
This is the system working. What doesn’t work is when the airline doesn’t offer sufficient compensation, no passengers take the offer, and they are allowed to deny a fully paid customer their seat. If the airlines where not allowed to cap the offered compensation, the system would work fine. All passengers would get the seat or compensation they prefer. The airline would get the additional revenue from selling all seats and even selling some seats twice. But they would have to weigh that against the cost of compensation. This is the back-pressure that keeps the system working properly. Passenger compensation caps are what causes these problems.
Great post and discussion — one element to be aware of is the airline has to take into account seat resources for employee transport both for positioning employees for work shifts (mentioned in the post) and the not so minor employee benefit of “free” space available personal and vacation travel (which can include friend and family passes). I think the current system works pretty well in the US currently, although there should definitely be a US version of EU comp reg 261 – https://en.wikipedia.org/wiki/Flight_Compensation_Regulation_261/2004 . Wish my wife and I luck on our trip to Australia in July, courtesy of my retiree Dad’s passes!
EU Regulation 261/2004 caps the max compensation that an airline has to offer if they don’t have the seat capacity that they sold available. Airlines and the airline lobby argue that they can’t be held responsible if equipment breaks, if weather prevents a trip, or if overbooking or employee movement needs change. I agree that weather conditions should be excluded. A reasonable argument can be made on equipment problems also being excluded. After that, what we need is a simple rule: if transit is sold it must be provided in the same time frame or acceptable compensation must be agreed upon. Most of the airlines do this most of the time. When there is trouble is when they don’t offer sufficient compensation to interest the needed number of passengers and simply refuse to board passengers or force them off.
When the airline caps the amount offered when oversold and the jurisdictional authority allows capping, the airlines are allowed to reap the benefit of overbooking without paying the true cost.
When those single digit profit margins at United add up to billions ( ! ) of dollars per year there are certain expectations ( and smaller airlines have hit 20%+ profit margins ). The main expectation being that if I buy a ticket to get on a plane, I should be able to do so, barring a problem that causes the whole plane to be unavailable. If the perception of your service becomes one where I can spend $1,000+ for a ticket and still not get on the flight, then your business is likely not going to be sustainable.
I agree with your expectations Joseph. They seem perfectly reasonable. On your expectation that any airline that consistently treats customers poorly should not expect to survive, that too seems reasonable but United has been struggling by that measure for years and continues to have have customers. Here’s another example from the LA Times of United threatening to put a passenger in hand cuffs: http://www.latimes.com/business/lazarus/la-fi-lazarus-united-low-priority-passenger-20170412-story.html
Airlines should be forced to execute a bump auction at the gate in these cases. The passenger who surrenders the ticket is given next available transport to final destination AND cash. In essence the airline is re-purchasing the ticket from the holder. No ceilings on the price. As the price climbs, someone will sell. This market solution would eliminate the need for a regulatory morass. Corporations and businesses love regulations which reduce the rights of their customers; but conversely, hate those which limit their own actions.
The auction results would even feed into the yield management systems.
Might proceed as following:
Attention all passengers on ORD to LGA gate C11. We need to repurchase a ticket. See the gate agent if you are interested in this auction. Winning bidder has seating on the ORD to LGA fight departing in 2 hours.
All gathered auction is starting now.
Raise hand if you wish to accept.
We are offering this future seat and $50.
and $100,
and $150,
and ….
A some point someone WILL accept.
Auction based systems, when the market is efficient and there are enough buyers, work well. However, the day before Christmas, the auction may not work well with very few customers willing to accept. It may be much more difficult to find a solution to a the problem. If the auction is required to be uncapped then, yes, I agree your system will work fine and this is basically what I’m advocating. The airline should have to either deliver that which they sell or buy it back at some price acceptable to the customer they are buying from.
Any system that faithfully delivers on these tenants is fine with me:
*If an airline sells a seat, they are obligated to either deliver the goods or buy it back at a price acceptable to the customer (uncapped overselling costs)
*If a customer buys a seat and doesn’t show up, they forfeit the ticket price. Smart airlines will over flexible tickets and refundable tickets at extra costs but that decision should be 100% theirs based upon their capability, there costs, competition, and what they think customers most value.
As you point out, there is always some cost that will get a seat back into circulation. Some days it’ll be higher than others but there is a always a price that will win. They only need to pay the lower price set by any customer but they need to accept responsibility for selling more resources than they have an pay this lower cost rather than offering $400, finding no takers, then throwing a paying customer off the plane.
I disagree with the no show policy. If I pay for a refundable ticket, I am paying for the contract right to use this for passage at another time or to have the money back. There are legitimate reasons for this, say I want to visit my child who is expecting; just don’t know when.
I will grant that the airline might not allow boarding unless you check-in within a reasonable window, like 20-30 min, before the scheduled departure. But if refused boarding, a refundable ticket holder should always be refunded at the full amount.
I would grant the airline the right to use that seat if you fail to check-in within a reasonable window.
Airlines should also compensate you for wasted time in the airport or notify you of scheduling changes, even those due to weather. I should not have to sit in the airport nor on flightaware to find out if my plane has even left is previous departure point nor penalized for making efficient use of time by not showing up until the departing equipment does.
We live in a world where it is easy and cheap to disseminate this information. Airlines can cheaply and quickly notify their customers of changes to flight schedules. With this information we can make better use of our time.
During weather incidents when seats can become exceedingly rare; the airline can offer its customers the ability to electronically defer passage; thereby, taking the strain off of the system. This would allow passengers with flexible schedules the ability you yield their seats to fliers who must travel.
Gerald said “I disagree with the no show policy. If I pay for a refundable ticket, I am paying for the contract right to use this for passage at another time or to have the money back.”
Sure, absolutely. What I said originally said appears compatible with your position: “Customers should have to pay a no-show penalty. It’s a competitive market and the airline should set it to whatever they believe covers their overbooking costs or encourages the desired behavior. Some airlines might chose to set this “no show” cost at free for some tickets, some passengers, or even all passengers. But it would be the airline’s choice to set the cost of a no show. Currently they mostly don’t charge now but, again, that’s their choice.”
If tickets are entirely refundable, then that actually makes this situation more likely as well. If you only got the refund if you notified the airline within X hours of flight (x does not have to be an integer) to get the refund then passengers who can’t show have an incentive to get that information to the airline early. That gives more clarity on how much each flight is, or isn’t oversold.
Any system needs to ballance risk with gain. Currently airlines can seriously overbook with the only downside risk being the capped compensation they offer. The current max compensations keep the problems down in single digit percentiles but the caps are sufficiently low that it remains a problem. My recommendation is to remove the caps so that a every sold seat has either a transported customer or a customer happy with the compensation they accepted.
In this model, airlines are free to overbook to boost revenues but they still have to fulfill the the commitment they made when they sold a seat. This model also rewards those with better seat utilization predictive modelling systems. Adding accountability through lifting compensation caps will allow the best to continue to extract more revenue from their fleet, will force the worst to over-sell less, and it ensures that customers commitments are fulfilled.
> Since an empty seat represents a substantial source of lost revenue, no airline wants to fly with an empty seat.
You’ve overlooked one item here. The vast majority of tickets sold are “non-refundable”. Which means the entire plane could be all no-shows, and the airlines have already sold (and get to keep the revenue) for every seat on the plane.
The reality is that the airlines are using overbooking as a way to double sell those seats that fall into that percentage of no-shows. So the reality is they are trying to sell 105 seats on a plane with only 100 seats, every time the plane flys. This is why the incentives are not aligned properly. There is way too much incentive to overbook, and so they overbook too much.
I don’t think I’ve overlooked that Anon. I understand that many seats are non-refundable so it may be the case that an empty seat is still a sold seat. I’m arguing that the model where an airline sells all the seats but some customers don’t show up, flying with those seats empty still isn’t the best thing for the business. If they find a way to sell the unused seat, the can earn yet more revenue. If the resource is otherwise unused, it makes sense to find a buyer for it. This clearly better for the airline in that it brings in more revenue. It’s also better for the environment because it reduces the fuel burn per actual customer. It’s also better for the customer because, in a highly competitive industry where margins are thin, reductions in cost get partially passed on to customers in reductions in price. Selling an empty seat again actually is better for everyone.
There are many ways to sell a seat a second time. One is the standby program where customers willing to fly later can show up and attempt to fly but not be guaranteed a seat. These systems are inefficient in that most customers won’t fly standby but they work and better than never selling any empty seat. This gives buyers willing to miss a flight the ability to potentailly fly very inexpensively. The downside of these programs is the seats sell very inexpensively. The airline takes less risk but gets less potential value. Overselling is another way to ensure that the plane flys with people in all seats. The upside of overselling is the the plane typically flys closer to 100% occupancy. The downside of this approach is there is more risk. Either some customer gets disaapointed or the airline ends up with an uncapped liability.
I’m arguing that the decision to oversell is a decision an airline could chose to make. Their upside is obvious. So, it’s important that the holder of the upside, the airline, also be holder of the downside risk. If there ends up being a shortage of seats, they need to find a price such they can buy back a seat. Essentially, the airline need to provide the resources they sold and it should not be possible to say they will that but only up to a downside cost of $400 per seat. If they chose to play the game and economics says they should, they also need to accept the downside risk and ensure that everyone flys or ends up with compensation they want more.
Large successful airlines in the EU such as Ryanair operate a no overbooking policy along with a no refund policy for no shows.
In many ways this is much fairer since your seat is going to be there.
The low cost airlines such as Ryanair have had some PR challenges but nothing like this incident.
I’m fine with that policy as well. Certainly it’s a simple solution to simply not oversell and not, therefore not have to resolve the resource problems that come from overselling.
My take is that there is value in overselling. Some customers, for whatever reason, will not show up. And, yes, all the seats may be sold but, if not all the seats are occupied, there are unused assets on the plane. If a previously sold seat can be sold again because the original customer didn’t’ show up, that’s great. More revenue for the airline and possibly lower prices for customers due to the combination of lower for the airlines and competition.
My suspicion is the eventual outcome will be successful airlines will find ways to get all the seats occupied since this is the more efficient solution. The key point is that they should not be able to say “whoops, me bad” and force customers off the plane. If they oversell, then it’s their obligation to find an alternative in the same time fame or find some customer willing to accept financial compensation to give up their seats.
I like the Ryanair model fine — certainly a lot more than the United model — but argue that overselling with a safety net is even better. In this model, the airline oversells seats slightly to ensure that all seats are occupied which increases their revenue and potentially reduces seat prices due to the combination of lower costs and competition. All customers either fly on the plain they purchased a seat on or they have accepted compensation that they valued more. Good for the airlines and good for all the customers.
An empty seat, even if paid for, is not ideal for customers, the airline, nor the environment. But, given the choice between that and the United incident, I would happily accept the slight reduction of efficiency that comes with the Ryanair model.
I think you bit off more than you can chew on this one, James.
“I’m going to argue that lawmakers need to make changes to make the overbooking system work properly…”
I’d argue against that, though I can understand your perspective. This is a complicated issue.
First, the airline (UA) should have upped their offer more, but did not. This is a problem with management at UA, who have decided to remove power from the gate agents’ hands and place it with the computer system (read: management). Management has effectively decided that they know better than any given gate agent in every situation. That’s a problem, in my view.
Second, the computer decided who to remove. Again, management making decisions for everyone. Yay.
Third, UA processes do not have a way of backing down to a customer’s refusal to leave the plane. They could have gone back and offered more cash, though perhaps once a decision has been made it is best to see it through. This is another management decision.
Fourth, those processes involve calling the aviation police and using state violence to enforce commercial contracts. Probably not a huge deal in itself, though something that we should all be highly cautious about condoning.
Fifth, the police are trained to escalate issues. UA did not beat this man or bloody him. This is where I personally feel there’s a real issue. There was no reason to escalate to violence as seen in the numerous videos. This is a social failing, in my view, and something that troubles all police forces, not merely those in airports. There had to be a way of working this out without resorting to violence.
Sixth, while UA is claiming this a “denied boarding” event, the truth is that the man was already boarded. I’m certain that UA lawyers will argue otherwise, that somehow having boarded the plane isn’t “boarding”, but the reality is there’s a good case to be made that the doctor was “refused transportation.” There are reasons UA can do this in their contract of carriage, which we all agree to when purchasing a ticket. None of the reasons listed cover this situation, at all. It’s my opinion that UA violated their contract of carriage and should be penalized for that. In fact, if you insist on passing a law related to this, provide *stronger protections* for those who are “refused transport” because, right now, they have none. The irony is that if UA had involuntarily denied boarding to the doctor, he would be compensated 4x his fare up to $1350, but because he was “refused transportation” he has no federal protections beyond civil court actions, which are likely limited in terms of the damages they can award (IANAL).
James, there is already substantial disincentive to overbook thanks to the 4x fare penalty, capped at $1350, that must be paid to pax who are involuntarily denied boarding (IDB). UA screwed up here in their policies and implementation (and worst of all, their PR). Legislatively, removing the cap would help (I’ve flown $1600 transcons in economy, so $1350 is too low a cap in my book considering how pricey international tickets can be). I believe we agree on that point, but I think the difference it will make will be small, at the margins of an already marginal event (roughly 1 in 20k pax are IDB’d). I would say we should consider adding protections for those “refused transport” before or at the same time as removing the compensation cap.
Where we might disagree is that I strongly believe this affects a tiny and already highly affluent portion of the population that is far better equipped to defend themselves legally than those most at risk in our society. Surely, there are more pressing legislative issues to address.
As to your major idea that airlines should penalize those who no-show, they already do. Your fare is forfeited if you do not reach the airport within a specific time. This is a generalization, obviously, and they have a window for what is sometimes called the “flat tire rule” (with different rules for this depending on the airline) but otherwise the value from discounted tickets is completely forfeited. Or perhaps I misunderstood and you are suggesting that we should pay *more* than our airfare for missing a flight. In any case, I’d say this problem has already been solved.
All in all, I agree with the headline, I hope no ill-considered laws are passed, I pray that no legislative time is wasted on such unimportant, marginal issues, and maybe one day police will stop brutalizing citizens in enforcing corporate contracts. Oh, and UA broke their contract by “refusing transport” to the doctor and deserves to lose in court.
PS – Word on the street is that Munoz is too nice of a guy to have released the initial statements; those with more knowledge than I claim that legal (or possibly some really awful PR folks) must have “cleaned up” his communications. Could be wrong on that, of course.
Thanks for the contribution James. I appreciate you putting it all down.
On one of your points you covered what I would like to see. You said “First, the airline (UA) should have upped their offer more, but did not. This is a problem with management at UA, who have decided to remove power from the gate agents’ hands and place it with the computer system (read: management).”
That is essentially what I would like to see: if a company chooses to overbook then they should be required to pay whatever it takes to resolve the problem. Allowing airlines to cap the fees spent is a mistake. I’m fine with them overbooking but, if they decide to reap the gain from overbooking, they need to also be held responsible for the cost. There should be no cap on the cost of getting all paid passengers on the plain, on a different plain, or compensated to the point that the customer is willing to accept the deal. Essentially, if an airline decide to reap the benefit over booking, they should also sign up to resolve any resource shortages that follow from the over booking.
On the customer front, I feel much the same way. If you don’t show, I’m fine with a customer being stuck with outcome. As you point out, many tickets today are non-refundable. I have needed more schedule flexibility in many trips I’ve done but I’m also fine with that not being the default offering. It’s an airline decision on how to compete and whether they they need to offer business travelers more flexibility. Most airlines to this today.
The key point from my perspective is there is nothing wrong with overbooking. What’s wrong is allowing airlines to cap the responsibility that follows from running short on resources. They should not be able to cap the buy out cost if they overbook.
Thanks for your thoughts.