The Pricing Wars

January 22, 2023

A great economic war is on the horizon. Few know, and fewer are ready.

Part 1: Price Discrimination

Traditional (consumer) economic theory is based on the premise that the same product will be sold at a fixed price to an arbitrary consumer. Optimal prices which maximize profit for the business are well understood. The classical approach, illustrated below in figure 1, is to sort consumers by how much they value the product (with this value on the Y-axis) and draw a horizontal line which represents the cost to produce the product. Then, for a given price (horizontal line), all consumers who value the product above that price will purchase it, and those who value it less, won't. The surplus value to society is the area between the consumers who value the product above the price, and the line representing the cost of production. The profit captured by the supplier is the area bounded above by the price, and the value captured by the consumer is the remaining area bounded below by the price. An idealized company is expected to consider the whole range of prices and select the one which maximizes their profit, which usually leaves some value on the table for the consumers, too.

Figure 1: classical supply and demand. In this scenario, the widget costs $1 to make, and the company can maximize its profits by setting the price to $2.75, which causes 30% of consumers to buy the product. The company captures about 70% of the realized surplus and the consumers the remaining 30%. However, 25% of the achievable surplus is not realized because consumers who value the widget between $1 and $2.75 are priced out.

There are two things I want you to notice.

  1. For a fixed price, the overall surplus value to society is maximized when the product is sold at-cost. However, this provides the company with precisely zero profit.
  2. If the price is not fixed, then the same total surplus is realized so long as each consumer is offered a price above the true cost but below their own value.

This leads naturally to the notion of price discrimination. I want to emphasize that price discrimination, despite its name, need not be a bad thing! As we've just seen, it is in fact the only way to maximize the surplus value to society while also allowing the company makes a nonzero profit. The mechanism by which it does this is to directly transfer some of the buyer's surplus value into the seller's profit.

I suspect that one of the dominant economic trends of our time is the increased capacity for price discrimination. What do I mean by this? Take the humble bazaar so often the focus of economic analysis. For a given shop, they have very little information about the customer who walks by, so they have no choice but to assume the consumer is average. Perhaps they can infer a little bit from the dress, accent, and mannerisms of the consumer -- indeed, one should expect to be charged a higher price by a street-seller when wearing a suit than a T-shirt! But the seller's capacity to make these kinds of judgements is very limited. Furthermore, if the items are to be labeled with their prices (to lower the friction of a consumer deciding to buy it), it becomes even more difficult to adjust them to the consumer. So, the economic model of a fixed-price product works relatively well in the bazaar.

Beyond the bazaar, there are many forms of price discrimination in practice today. Brick and mortar retail stores certainly price their products differently for different locations. Part of this might reflect that the costs are different in these different places, but the more important effect is that consumers in some areas are less price-sensitive than others. Two of my favorite examples of price discrimination are student discounts and grocery store coupons. Regarding student discounts: having a student ID is a pretty good proxy for not having much disposable income, and thus greater price-sensitivity, so charging lower prices makes sense. Regarding coupons: the main fact of couponing is it requires extra time and effort, and thus a lawyer who makes $300/hr won't bother, and will just buy whatever she wants, whereas for a cashier making minimum wage it is well worth their time. In this way, the store is able to sell the same goods to different people for different prices. Note that the only reason the coupon works in this capacity is because it's a pain! Making couponing easier would defeat the economic purpose of the coupon.

The internet changes the story dramatically. Amazon has detailed data on their customers, and can likely predict quite accurately how much each one would be willing to pay for a given product (at least, certainly better than the bazaar shopkeeper!) Right now, I suspect they are holding back: culture still mandates that when you send your friend a product link, they should be shown the same price as you, or else you and your friend will feel cheated and refuse to buy the product. But even without such overt discrimination, there are many knobs Amazon can turn to achieve a similar outcome. For example, they might give free shipping to some zip codes but not others. Perhaps they know that different kinds of consumers buy items at different times, and adjust their prices accordingly. (This is also related to how sunscreen tends to be cheaper in the winter.) Or, they might simply tune their search results so that consumers with less price sensitivity are only shown more expensive, high-margin items which have essentially equivalent functionality.

While I don't know of any in-depth studies of the internet's effect on price discrimination, there is at least some evidence for the theories I put forth above. Target has bragged about being able to adjust their prices on a minute-by-minute basis.

I also want to spend a moment on the feeling of price discrimination. The feeling of price discrimination, as the consumer, is of being just barely willing to buy something. When the price causes you to lose your enthusiasm for the purchase, and yet you just barely decide to buy it anyways, then the seller has played the game perfectly. An overly enthusiastic purchase is (at least in the short-term) a failure for the seller.

I believe that these trends will continue to accelerate. First, I think the culture of expecting fixed prices is slowly eroding on the internet; I'm not sure I'd be surprised (or care all that much) if my friend were offered a lower price by Amazon, even though I'd be furious if it happened in person. Second, the low-hanging fruit for pricing is being picked, and soon only more aggressive strategies will be able to improve margins. Third, the continuing growth of data, and improvement of models, means consumers' values, and thus optimal prices, can be ever better predicted.

Zooming out, this paints a somewhat bleak picture. While price discrimination, if used in moderation, has the capacity to increase the total surpus to society while reducing inequality (the grocery store couponing example represents this well) in its limiting behavior it captures essentially all of the surpus for the small number of oligarchs who own and run the markets.

In principle, competition should save us. After all, if a company is charging the highest possible prices, then surely another can come in and lower prices, thus capturing greater market share and greater profits. In practice, I am skeptical of this for two reasons. The first is that many industries today are quasi-monopolies which have highly defensible value propositions or economies of scale and thus relatively little competition. The second is that even when there is competition, companies often find clever (but currently legal) ways to coordinate. The quintessential example of this is the “price-match guarantee,” which seems pro-consumer on the surface but in fact allows sellers to create a sort of cartel by automatically punishing any competitor which tries to lower their prices. Thus can many separate stores simultaneously converge on identical (and profitable!) prices for many different products.

The situation is not yet so dire as to necessitate a coordinated response, but I believe we are close. When that happens, so will begin:

Part 2: The Pricing Wars

I believe that, to fight this sort of pricing discrimination, consumers (or more likely, organizations on their behalf) will coordinate behaviors to achieve better prices and thus a bigger piece of the value pie. We don't know yet what this will look like, but I want to give a microscopic example of how this might work. I'll focus on airlines, even though airlines are not particularly high-margin businesses. (Their profits suffer quite heavily from intense competition.) Nonetheless, they are particularly good at price discrimination.

There are many ways they engage in the practice, but one of the simplest is by changing the price of tickets over time. Usually, tickets get more expensive over time, rewarding customers who book early -- but not always! For a flight with many empty seats and price-sensitive customers (perhaps to a destination more known for tourism than business), it may make sense for the airline to drop prices more precipitously near the departure date, in an effort to drum up more demand. After all, since the flight was scheduled anyways, each new seat that they sell is essentially pure profit.

As an individual consumer, it is basically impossible to predict these trends. Tools like Google Flights can give you some idea of how ticket prices compare to “normal,” prices, but no consumer wants to spend the time or effort to carefully analyze the expected price fluctuations over time, and then schedule when they are planning to buy the ticket. They'll look at the price, and decide whether to buy it or not.

Now imagine a broker company which tries to beat the airlines at the pricing game. This company can collect lots of data for different airlines, routes, flights, and so on, and predict how a given route will be priced on a given day. Then, here's what might happen.

  1. The consumer requests a flight with a set of parameters. For example, the consumer might specify that they want to fly a certain route on a particular date with no more than 1 stop, which takes no more than 7 hours total, on a relatively modern airplane.
  2. The broker decides that for this particular route, it can save 5% on average by waiting for the right moment to by the flight. It offers to satisfy the consumer's request for 3% less than the best current place. The broker takes the 2% as its own margin. For certain queries, it might be able to save 10%, and for others it might conclude that prices will only go up the broker can offer no discount.
  3. Now, this doesn't work particularly well if flight prices only monotonically increase -- then the broker is always best off buying the same ticket the consumer would have bought. But when prices are expected to fall, the broker's arbitrage pressures the price to decrease earlier and increase later. The overall effect is that this particular avenue of price discrimination by the airline is defeated, subject to the success of the broker in wooing consumers.

Another benefit of this approach is that once enough consumers use the broker to buy their flights, the broker can collectively bargain on behalf of these consumers. This gives the broker semi-monopsonistic powers, which further leads to lowered prices.

I have no idea whether this broker would work in practice. Perhaps consumers want more certainty in terms of which flight they'll take, and a few percent discount isn't enough to woo them. Maybe loyalty programs create sufficient lock-in to counter the broker's discounts. But in any case, it illustrates in principle how a centralized entity could leverage both data and scale to fight to lower prices on behalf of consumers, just as the tech world has leveraged data and scale to increase margins.

I also like this example because it illustrates an approach that I believe will become more important as the pricing wars escalate: intentional purchasing. The consumer need not ask the airline broker to book them a specific flight. Rather, the consumer communicated their desires to the broker, and left the implementation to the broker so that it could better leverage its data and scale. While this is relatively straightforward with airline tickets -- a red-eye from LAX to JFK is an almost identical product regardless of whether it's operated by Delta or American -- there will be increasing incentives to implement intentional purchasing in other domains too.

Of course, if this broker existed, there is no reason to think the airlines would take it lying down. They might find ways to suss out data about consumers' requests to the broker, or coordinate a response and refuse to do business with the broker, or something more clever. But this is why I call the coming fight the pricing wars! They will be fought not on the single battlefield of airlines but on thousands ranging from bookstores to car dealerships to carbon offsets, with calculated moves and responses, each player vying to determine who gets consumers' dollars and how many.

I'm optimistic about the impact of the pricing wars. Much will depend on how each of these individual battles play out, but complete dominance of one side over the other is, in either case, probably a bad thing. If the sellers win, consumers receive too little surplus and wealth aggregates to the sellers. If the buyers win, there is too little profit to be invested as capital, and progress stalls. Right now the pricing wars are a one-sided invasion. It's about time for defensive action.