I just began to follow the Sports Analytics Blog on Twitter. I click through every now and again. Interestingly, they have a multi-part piece going up currently about dynamic pricing. However, I have some qualms with the apparent misunderstanding of business models in certain sports. Therefore, I figured I would go ahead and use this blog for something and write a short blurb about it.
First, let's talk about dynamic pricing. The term itself, I guess, is rather new and comes along with technology and data management improvements. However, the idea is not new, and much of the roots of its theory come from economics. I am going to talk about dynamic pricing here without using the words "analytics" or "moneyball", because those words just get thrown around all over the place.
Dyanmic pricing has its roots in economic theory on price discrimination and product differentiation. Price discrimination discusses the ability of a firm to charge different prices to different people based on their willingness to pay. However, this used to be a big problem for firms to do, as you can't just ask someone how much they want to pay for a product when they come up to the register (they'll tell you they value it at a penny, so the theory goes). This is what car salespeople try to do when you buy a car: get you to signal to them how much you are willing to pay. Once you do, you are toast.
But it is important to remember that price discrimination is not necessarily a bad thing (well, the European Union would like you to think so). From a neutral standpoint--where we as good little economists do not favor the consumer or the firm--price discrimination is more efficient. The firm no longer has to choose a single price to charge to maximize profits. It can charge more to those willing to pay more, and less to those willing to pay less. This probably isn't good for those who will pay more for the product, but now--at least down to marginal cost--those people that could not buy the product at the higher, single price can now purchase it. Just as with redistribution, some are worse off and some are better off. But if the cost of figuring out who is willing to pay what are low enough, this is much more efficient than taxation and redistribution.
Product differentiation is one way in which a firm can charge different prices to different types of consumers without asking them, but it technically differs in that it uses multiple products instead of different prices for a single product. In this case, the firm varies the type of product it sells. An easy story is IBM and its printers. Simply, IBM had two types of printers, each for a different price. One printed fast. For the other, IBM actually put a special chip in it specifically to slow down the printing speed, despite it being the same exact printer. They charge more the faster printers, and allow the consumers to sort themselves out by differentiating these products and letting the consumer decide how much printing they need. Yes, the slow down chip cost a few marginal pennies for IBM, and they still went out of their way to charge less for this printer (about half, from what I can gather). If they had sold the better printer to everyone for cheaper (so more people could buy them), they would not have maximized profits. Additionally, more people were happy, as they could buy the cheaper, lower level printer for their basic home needs.
Interestingly, there is built-in product differentiation when buying tickets to athletic events. Do you think the Yankees front row box seats are the same product as the upper level bleachers? What about a Yankees game against the Red Sox game versus a game against the Royals? Games on Saturday and games on Monday? Teams have been doing this for a while. Even season and single game tickets are different products. And the dichotomy of these latter two products are the key to various business models in different leagues.
One of the important aspects of sport is the fixed supply: you can't make more seats if people want them, and you can't reduce costs by taking them away. The stadium is built, the seats are there. If you are not going to sell out a game, you may as well give out your tickets for free at the last minute. If you are selling out, you can increase prices to suck up some surplus (or increase concession and merchandise prices inside the stadium--remember this is a joint maximization problem for the team). This way, you can get concession revenues and make the stadium full (which may be more fun for the fans anyway). Additionally, you can also charge more for in-arena/stadium advertising because you have more eyes on the ads. In fact, teams do this. The Detroit Pistons were handing out tickets in droves (to local youth groups, etc.) a day before game time in order to continue their sellout streak. Maybe they should have called it a "give out" streak. But it doesn't matter how they got into the stadium to those putting billboards inside, and the team's marginal cost of having more people in the stadium is essentially zero (give or take a few extra staff, maybe).
So where does the term "dynamic pricing" come from? Well, this simply comes from real-time price discrimination and product differentiation. The key here is "real-time." Without the real time inclusion, then we're back to boring old economics.
Hotels and airlines are notorious for real-time price changes. They know that, depending on the time of day and amount of time before the day of your stay/flight that you make your purchase, they can figure out your likely willingness to pay for a given room or seat. Of course, this is a rough estimate, but it turns out that they are very good at this. They do this with data they collect on all of their past sales.
More recently, teams have been delving into the use of these real time pricing models to--as the theory goes--simultaneously increase profits and allow more people into the stadium. Those that wouldn't spend $50 on a Yankees ticket now get to buy it for $40. Those that really wanted to attend that game bought it way in advance, and perhaps paid $60 for the same seat. The key to the MLB business model here is that its ticket sales depend heavily on single game tickets. The fact that there are many tickets to sell for each game--and they are not all sold preseason--allows MLB to do this. NBA and NHL also have the ability to do this to some extent, though probably not to that of MLB. I also heard that recently some college football teams are doing this (South Florida), but not the ones that sell pretty much all of their tickets preseason (i.e. Michigan and Florida--note that they do price discriminate through donations and student tickets).
OK, so back to Sports Analytics Blog. What irked me a bit was this article, which gives the impression that NFL is not currently using dynamic pricing and is therefore making a poor business decision. They use the example of the team losing a high profile player, which is fine: the Patriots games are arguably a very different product without Tom Brady on the field. So far so good. But if Brady is injured midseason, the team cannot suddenly change its prices! NFL is almost a completely season ticket league (or tickets for single games purchased preseason). Therefore, the real-time changes within the season aren't tractable, unless you are the Jacksonville Jaguars.
But, Brian, they could just hang on to them and sell them throughout the season (or keep prices super high early on), you say!?!?!
Well you are correct. But in a short season like NFL, with huge revenue dependencies on television contracts and selling out games, the uncertainty involved in doing that may outweigh any benefit they get from pricing. The short season allows for only 8 chances to get things right. They let fans take the risk on purchasing the tickets and possibly seeing a down season. That's a reasonable business decision, given the broadcasting structure and short season with lots of uncertainty (many times, pretty good teams don't make the playoffs when you have a 16 game season--think about who would make the playoffs if MLB was only 16 games). Fans can sell off the tickets on the secondary market later if they decide they won't make the game.
To be fair, the SAB article puts things in terms of losing a player to free agency, etc. But if that is the case, we're really not talking about anything "dynamic." We are back to pretty basic pricing decisions. And let's also remember this ignores economic theory in general. Economic theory on sports says that teams choose their talent level before the season begins (with short term adjustments), based on what they know they can charge to their fans for a team with X wins (i.e. their objective function, assumed to be profit maximization). Obviously these short term adjustments are important--like losing a player to free agency--but this isn't really dynamic. It's just pricing.*
Now with that said, I certainly agree that teams should be considering these short term changes in talent levels if they can do so at a minimum cost (this seems likely). But the timing of these decisions and the timing of ticket purchases in NFL as a whole would result in a relatively low use of a full-on dynamic pricing model. The blogger at SAB, who I am sure is a sharp business person, seems to have a slight misunderstanding of dynamic pricing as real-time pricing, and of the business structure of NFL. That doesn't mean teams shouldn't fully consider their business decisions with the information at hand. They absolutely should! But it doesn't make them poor businessmen for not being as quick to adopt these techniques.**
Lots can be said about sports pricing, and there is plenty of research to be done. For now, I'll leave you with some good reading on pricing (these are gated, sorry). Note that this is an extremely limited list of papers, and there is plenty more out there to be read (including basic texts on pricing in fixed supply industries like sports, entertainment, hotels, and airlines).
Berri, D. & Krautmann, A. (2007). Can we find it at the concessions? Understanding price elasticity in professional sports. Journal of Sports Economics, 8, 183-191.
Salaga, S. & Winfree, J. (2013). Determinants of secondary market sales prices for National Football League personal seat licenses and season ticket rights. Journal of Sports Economics, DOI: 10.1177/1527002513477662.
Fort, R. (2004). Sports pricing. Managerial and Decision Economics, 25, 87-94.
Soebbing, B. & Humphreys, B. (2012). A test of monopoly price dispersion under demand uncertainty. Economics Letters, 114, 304-307.
*In my brief experience working with a sports ticket sales department and their
analytics, there is plenty
of room for improvement here. I still can't understand how cold calling
people to purchase game tickets actually work. Has anyone ever gotten a call from a representative at their local pro sports team and been persuaded to go ahead and buy those ticket for this weekend's game?
**Now I do ignore a few interesting things about sports pricing
(or at least just glaze over them). Note that usually some sort of
monopolistic firm is required for this tactic. Otherwise, firms will
bid each other down to cost. At the very least, there would need to be
differentiation of these competing products for price discrimination to
happen. Both of these conditions likely hold in pro sports. The second
is that there is a secondary market for tickets. These are important
considerations for teams, as fans selling these to other fans for more
money could cut into some of the additional revenues that those fans
would otherwise spend inside the stadium. Thirdly, I ignore directly
addressing the inelastic pricing of tickets across many sports.
Remember that there is a joint maximization problem (parking,
merchandise, concessions, BEER) not just maximization of gate revenue--most likely dynamic pricing could be used in some sense for these other considerations in the NFL. I also ignore the use of price dispersion, which could be an
important tool for teams (especially in the data collection phase of
willingness to pay in given situations). Finally, there are interesting applications of luxury products and keeping prices high (i.e. Yankee front row box seats) and reference prices (prices that a consumer uses as a baseline for "high" or "low" price for a given product).