The Welfare Effects of Dynamic Pricing: Evidence from Airline Markets
onlinelibrary.wiley.com
onlinelibrary.wiley.com
- Ungated (working paper) versions are here[1] and here[2]
- This was published in Econometrica, one of the top journals in economics. That doesn't mean that the results are true, but it means that the the math and methodology are probably sound. If you want to attack the results, the weaknesses are probably in the (implicit) model assumptions.
- In economic lingo, "welfare" means consumer surplus (buyer gets a good deal) + producer surplus (seller makes a profit). Standard academic economics analysis is indifferent wether the producer or consumer gets the profits (i.e. between producers selling at costs and consumers buying cheaper that they'd be willing to pay, and producers maximizing profit while consumers maximum they're willing to pay). "Higher welfare" in dynamic pricing vs uniform pricing means that the sum is higher in dynamic pricing. In this context it basically means that more flights are flown using dynamic pricing than would have been in uniform pricing. Negative externalities from more flights are usually out of scope for a paper like this.
- Airlines can adjust prices continuously for the same flight as a response to demand changes, or price discrimination (charge more to those willing to pay more, i.e. business travelers). The paper claims as a key contribution the disentanglement of these motives. I haven't gone through it to see if it's sound, but when since it's published in Econometrica, it will probably take quite some time to fully digest.
[1] https://www.nber.org/system/files/working_papers/w28989/w289...
[2] https://cowles.yale.edu/sites/default/files/files/pub/d21/d2...
This is the kind of paper that goes a far way to establish tenure. Good job to the author if that is their stage in their career.
Naturally in this situation only the seller will benefit from the transaction. Everyone else is +-0. Utility/"welfare" is maximized but consumers get none of it.
And of course, in the end even talking about utility assumes the price someone is willing to pay reflects their utility gain. This is the most fundamental assumption in economics, but it's pretty clear that in the real world this is not the case. Especially for luxury goods such as air travel.
Why? I would expect luxury goods to reflect utility very accurately since they are not necessary purchases.
If you are paying $1k to go travel to Disney World, then that seems like a pretty good signal that the flight to Disney World gave you at least $1k worth of utility.
Or the utility a 80th percentile household in the US could buy for the same amount.
https://en.wikipedia.org/wiki/Utility
> a utility function that represents a single consumer's preference ordering over a choice set but is not comparable across consumers. This concept of utility is personal and based on choice
It does. It reflects their utility gain relative to the utility they gain by spending that money somewhere else, or just keeping the money for the future. People not made of straw do not generally believe that the price someone is willing to pay reflects somebody else's utility. Nor are they unaware of the tendency for marginal utility to decrease.
I do agree that it's unremarkable that price discrimination results in higher total welfare than uniform prices though, and all the other conclusions are completely in line with expectations. Its also the case that the airline's own price discrimination models will be more sophisticated with access to better (commercially sensitive) data than this stylised version
Think about the study of inflation. I’m sure lots of work has been done to show how low interest rates have impacted inflation. I won’t be surprised when they show how these interest rates have led to increased inflation, but I’ll be interested in the mechanics and model leading to the unsurprising result.
that's an objective-sounding position to take, but it's certainly not, even from a purely detached, academic perspective. in most situations, consumer surplus is better for a market and for the overall economy, because it provides more choice points for more efficient allocation. in the unfortunately rare case of a relatively fair and competitive market, producer surplus can be a net good, making the overall market and economy more efficient, but in practice, companies have shown time and again to be pretty poor allocators of capital because they necessarily take fewer bets with larger amassed asset pools. however, economists generally (implicitly) assume a priori that markets are relatively fair and competitive in their analyses (akin to assuming frictionless infinite planes).
all that's to say, size does matter, negatively beyond some optimal point on the economies of scale/scope curves. as such, markets are at their best when populated primarily by small- to medium-sized companies, with no artificial barriers to new entrants (e.g., not the airline industry).
no matter the 'prestigiousness' of the journal, seemingly small (implicit) assumptions like this that can have huge implications on results necessitates skepticism toward academic economics.
It mostly just means you need to be very careful in interpreting the results, or be careful when someone uses it to motivate policy prescriptions as they themselves may have a questionable interpretation — like with any academic finding.
Amazon tried this years ago. They changed pricing on the same sku based on who was shopping for it.
The blowback was tremendous. Amazon had to actually back down.
My take is that consumers understand and are willing to put up with dynamic pricing for airplane seats because the algorithm is presumably understood.
I'm willing to accept that the person sitting next to me on an airplane may have paid 1/3 than the price I paid on my ticket. Maybe the person next to me booked earlier, or booked when the airline had a promo going. Sounds reasonable.
But what happens when airlines quietly change algos...to raise prices because of my browsing habits, IP, or CC used?
I bet you this is coming
How would this work when most people get their flights through aggregators? I search for flights on Google Flights. I imagine Google Flights just hits a database somewhere to get times, prices, number of stops, etc.. rather than share my information with every airline.
Then you hit issues, delayed flights etc. and every single helpdesk' first question is: did you book with us or through reseller? If by reseller, please contact them, goodbye and have a nice day. If by us, here is completely brand new ticket for free at best possible available time.
Life if too short and I am too poor to deal with such crap when traveling (which is in my case only for vacations, no point ruining them by trying to save few bucks). And quite often direct selling is same price as the other sources.
So either go with agency, pay the premium for them to sort it out if possible. Or then directly to airline. The results might vary, but per location there might be perks with either.
I ring through to Bravo and ask them to reschedule a the flight as is required under EU law. No problem sir we are into it. I think nothing of it. 5 days pass. No email and my booking has not changed.
I ring again. Why is this taking so long. I was rudely informed to be patient as the "relevant department" ( note this term ) is processing it and it takes some time.
10 minutes later I get an email from bravo. All my flights are cancelled.no to information is provided about the reschedule.
I find their webchat app and get in touch.
"How can I help you sir?" Starts the conversation which quickly goes downhill. The man informs me that Bravo is not responsible for rescheduling the flight. My contract is with Singapore airlines. "Have a nice day sir"
No no no and no....I argue with him for over an hour. Eventually he agrees to escalate and then hangs up on me.
I try to contact Singapore airlines but they tell me to go back to Bravo. It's their ticket booking.
A day later I call back to Bravo and ask them the status. They just stonewall me. Everything is handled by the "relevant department" and I will get an "email as soon as possible" They refuse to divulge any information.
They start pressuring.me to cancel. They promise me a full refund. I say no I want a reschedule. "Ok sure but you will have to wait for the RELEVANT DEPARTMENT."
This is nine days before I fly. I've waited now 6 days for any actionable information from Bravo. Nothing comes. Eventually I give in..I book a new flight direct with Singapore airlines.
I contact the bravo chat again and ask them to cancel. "Ok sir but we can only refund you the part of the ticket that was cancelled". That is 1/4.
I explode..I post the chat log I'd recorded from yesterday showing that I'd been promised a full refund. "Yes sir ... Of course sir.but there will be a 35 euro handling fee"
No the fuck there won't be I responded and posted the recorded log where I'd explicitly the day before asked about handling fees and were promised there would be none.
I asked when I should expect the refund. "I've notified the RELEVANT DEPARTMENT and they will contact you as soon as possible"
At this point I realise their business strategy. They can offer cheaper flights because when everything goes well the customer gets the flight. If however the airline cancels one flight Bravo can then dick the customer around and attempt to find excuses not to pay up.
I somehow doubt I will get much of the 1000 euros out of them they owe me.
But yes that's exactly why we buy directly. Compare it to our trip to Bolivia via Miami few years ago. Come to the airport, do you have US ESTA visa? Nope what is that? We dont' travel to US so don't know its policies, just want to fly to Bolivia. Yeah but you go via Miami, its not secured airport unlike most airports anywhere, you need full tourist US ESTA visa. OK we go to airlines helpdesk (British airways) by their recommendation, they do quick ESTA for us based on our passports.
All good, we fly to London, but then flight to Miami is a no-go at the gate, can't find our ESTA code. Some US personnel on the airport in Heathrow found out the person who made ESTA selected incorrect country in our application. 30 minutes to departure. Sorry no-go. Straight on the airport within 20 minutes we got free tickets for same flight next day and other remaining leg and did ESTA correctly ourselves.
I have few similar stories, but they all point to the same conclusion - its not worth the extra risk, especially if there are more legs and multiple airlines are involved. It all looks nice till first cancellation happens. The amount of frustration and stress can easily ruin any mental benefits of (usually quite costly due to 4 of us) vacations.
When you book with Bravofly and equivalent, they book separate tickets for each leg with no promise of connection. That's how they get the cheapest possible fares. In the event of a cancellation of one of the flights the airline is obliged to offer you a rescheduled ticket between two destinations, but has no obligation to ensure the passenger gets from the start point to the end point of the original itinerary. So none of the alternative tickets SIA was obliged to offer Bravofly were any use to you. Bravofly don't have any obligation to reroute you either and even if they cared about goodwill I doubt their margins are big enough for them to volunteer alternative flights at their own expense.
Or put another way, I booked a cheap flight back to London via Zurich on their airBerlin's website. Couple in front of me booked the same route via Bravofly (as two separate tickets) for about £5 less. When airBerlin decided to cancel the flight to Zurich without telling some of their passengers until they arrived at the airport I got back to London around when I'd originally expected at airBerlin's expense on a better airline on a different route. The other couple got told they could have a flight to Zurich later that day if they still wanted it, but they'd miss the other flight back to London without compensation regardless. Or they could ring Bravofly and see if they'd help...
No Incognito or VPN needed
What I do still see changing based on different metrics like IP are car rental prices and hotel prices
All US car rental companies give much cheaper prices to foreigners. I don't know the reasoning but its absurd, A rental car in Manhattan for Easter weekend will cost you $300/day on kayak.com and $70/day at the same exact branch on kayak.com.hk (this also works on the companies official websites when changing "country of residence")
I think most of us are okay paying a surge rate to Uber when it’s raining at the ballpark (market circumstances), but not just because you are on the way to the hospital (personal circumstances).
This seems to hold in other contexts. People tend to tolerate price discrimination at an aggregate level (e.g. senior/student discounts) but loathe individual-level price discrimination (e.g. college tuition). I wonder if it's the unfairness or uncertainty
That is, an empty seat has value until the plane takes off, at which point the value goes to zero. On the flip side, a traveler has an opportunity to be in that seat until the plane takes off, at which point its value is zero. People tend to understand that those values continuously vary with time, and so the price will vary with time.
To contrast, books (in specific) and consumer durables (in general) are not perishable goods. The notable exceptions tend to be time-based; try getting chocolates in the heart-shaped box right before Valentine's Day in the US... and then the day after. Since the value of the book doesn't continuously vary with time over the short-run, people don't expect that the price will vary.
The GP was trying to explain why airline tickets, but not consumer goods, are dynamically priced when sold by the original vendors.
It's at least possibly bad for long-tern business. And in the case of small businesses, the owners may just not want to be that person.
For events, "list prices" are often gotten around by (in addition to ticket brokers being a thing) by various types of VIP seating, extras, and so forth which raise prices to closer to market levels while leaving normal ticket list prices--assuming you can get one.
[0]: https://help.ticketmaster.com/s/article/What-are-Official-Pl...
Popular concerts sell out very fast and are immediately being sold on stubhub. Not all of these tickets sold on stubhub are actual resales by consumers or scalpers, but are typically mass purchases negotiated by ticketmaster ahead of time.
Superfans glued to artist social media will get special promo codes which unlock tickets that are not otherwise available. This makes sense because they are the most likely to spend the difference on merch.
I'm not sure dynamic pricing is necessarily pareto improving (I think this might be what you mean by "is a good thing"? For folks that are learning this lingo, pareto improving -- at least one party better off and none worse off). I'd have to think through it a bit.
Or higher prices because you have Prime, or pricing either intentionally or inadvertently based on race, sex, religion, or sexual orientation?
I think consumers put up with dynamic pricing for airline seats because they don't have another option available, and because it's a long-standing practice that didn't change in an era where that change would have generated blowback.
> But what happens when airlines quietly change algos...to raise prices because of my browsing habits, IP, or CC used?
This has already happened: https://business.time.com/2012/06/26/orbitz-shows-higher-pri...
Orbitz showed higher prices to Mac users because they figured (probably correctly) that Mac use was correlated with willingness to pay more. They stopped that practice when it became widely known, but it wouldn't surprise me at all if others do the same and don't get caught.
This is an industry that has a history of ever-increasing executive pay rates with gradual reductions in service quality, gradual increases in incidental fees, frequent bankruptcies, and occasional governmental bail-outs.
The solution is to self-insure by having $1000, or whatever, ready for if you need it for increased travel costs or other emergencies. And try to avoid feeling either anger or regret if you have to change your plans. If you don't have a spare few hundred bucks then this method isn't available to you.
But keep in mind consoles are all network effects, and the real money is on subscriptions and commissions on games. So, Sony gets a one time pop in revenue, fewer units lead to fewer exclusives, leads to more people selecting XBox in the future.
I think MS was willing to lose a billion on the first Xbox to get sufficient traction. If we discount inflation entirely and assume that was a reasonable value, Sony would have to sell 6.67 million units to be equal to the cost of losing a console war. And they could always end up pulling a Sega.