We’ve seen how firms with market power can use direct price discrimination to increase their producer surplus above the amount they could earn by charging only a single price. The key is to charge higher prices to customers with relatively inelastic demand and lower prices to those with more elastic demand. However, being able to directly observe a customer’s demand type before purchase (as required with direct price discrimination) is often difficult. A firm might know that its customers have different price sensitivities, but it may not be able to tell to which group any particular customer belongs.

396

Even without this knowledge, a firm can still earn extra producer surplus through price discrimination by using a pricing strategy called indirect price discrimination, also known as second-degree price discrimination. In this pricing strategy, a firm gives its customers various pricing choices and allows the customers to choose among them.

There are many different kinds of indirect price-discrimination techniques a company can use. The principle that underlies all of them, however, is the need to set up the pricing options to convince customers to pick the “right” choice; that is, to purchase the option meant for their group rather than another option for a different group. For example, airlines choose ticket rules and prices so that business travelers with inelastic demand pay more, on average, for their tickets than leisure travelers with relatively more elastic demand. At the same time, however, the airline wants to keep business travelers from deciding that tickets meant for them are too expensive and instead buying up cheaper tickets intended for leisure travelers.

The most basic type of indirect price discrimination is the quantity discount, a pricing strategy in which customers who buy larger quantities of a good pay a lower per-unit price. For quantity discounting to work, customers who purchase larger quantities of a product need to have relatively more elastic demands than consumers who buy smaller quantities. If the consumers in the market do not have these elasticity characteristics, the firm would be trying to find a way to raise prices on the people who buy greater quantities, the opposite of a quantity discount.

To illustrate the idea, let’s say there are two types of customers of the online brokerage house E*TRADE. One type of customer is not very interested in trading stocks. Because of this, these customers don’t have a big incentive to shop across different online brokers in search of lower commission rates (the fees they pay a brokerage firm to facilitate a trade). Thus, their demands are relatively inelastic with respect to the commission charged. The demand curve for uninterested traders is D_u in panel a of Figure 10.6. The other type of customer is obsessed with trading stocks. Because these individuals trade many times each day, they are very sensitive to the commission rate. Thus, their demands are relatively elastic with respect to the commission. The demand curve for these obsessed traders is shown as D_o in panel b. The marginal revenue curves for each group are MR_u and MR_o, respectively. The marginal cost is the same for both groups.

E*TRADE would like to charge higher commissions to the uninterested traders with an inelastic demand than it charges the obsessed traders with the more elastic demand. This third-degree price discrimination (segmenting) would bring E*TRADE more producer surplus, but the company cannot pursue this strategy because it cannot tell which type of trader each person is when she signs up for an account. What E*TRADE does know, however, is what the demand curves of the two groups look like, even if it can’t identify to which group any given trader belongs. Based on the demand curve D_u, for example, E*TRADE would want to set its standard profit-maximizing quantity and price (commission per trade) for uninterested traders where MR_u equals MC: For Q_u trades per month, E*TRADE would charge uninterested traders $30 per trade. For obsessed traders, E*TRADE would like to follow the same procedure and charge them a price of $9 per trade; at that commission, the obsessed traders would make Q_o trades per month.

If E*TRADE could segment the market, it would charge each group P_u and P_o per trade, and at those prices, each group would make Q_u and Q_o trades per month. However, E*TRADE can’t directly assign different commission rates to different traders. And it can’t just offer new customers a choice of whether to pay $30 or $9 commissions no matter how much or little they trade because every customer would choose the cheaper option. What can E*TRADE do to take as much of each trader’s surplus for itself? Rather than offer all customers a $9 per trade commission, E*TRADE can tie that commission rate to a requirement that the customer make at least Q_o trades per month. For customers who do not want to make at least Q_o trades per month, E*TRADE can offer a $30 per trade commission plan that allows them to trade as little or as much as they’d like in one month.

397

The idea behind this strategy is that an obsessed trader, who demands a high quantity of trades and has a more elastic demand, will choose the $9 plan that requires a purchase of at least Q_o trades each month. An uninterested trader, on the other hand, will choose the $30 per trade plan. In other words, traders from both groups will sort themselves into the price and quantity combinations designed for them, even though E*TRADE cannot directly identify either type. This is the essence of any kind of successful indirect price-discrimination strategy: The firm must set its prices so that a customer doesn’t try to fake her demand type and buy the package meant for another customer type. We discuss this requirement for the successful implementation of all indirect price discrimination (including quantity discounts) next.

Incentive Compatibility The plan to charge uninterested traders a higher commission than obsessed traders is logical, but for such a plan to work well and allow E*TRADE to reap the maximum producer surplus available to it, E*TRADE needs to make sure that the uninterested trader won’t want to switch from her $30/Q_u package to the $9/Q_o package designed for the obsessed traders. That is, the $9 commission deal can’t be so good that the uninterested trader will make extra trades just to obtain the lower price. E*TRADE has to be sure that the uninterested trader’s consumer surplus is bigger with the $30 per trade package than with the $9 package that requires a purchase of at least Q_o trades. The offers need to be internally consistent so that each type of buyer actually chooses the offer designed for it.

398

Economists have a term for this type of internal consistency: incentive–compatibility. In this example, the two packages are incentive-compatible if:

Let’s see whether this set of offers is incentive-compatible. First, we need to show that the uninterested trader’s consumer surplus from trades at $30 each is greater than her surplus from making Q_o trades at $9 each. Finding the consumer surplus from the first offer is familiar territory. As shown in Figure 10.7, at a price of $30 per trade, an uninterested trader makes quantity Q_u trades, and the consumer surplus is the area under the uninterested trader’s demand curve and above the $30 price. This is triangle A in panel a.

Finding the uninterested trader’s consumer surplus for the $9 package offer is a bit trickier. The first thing we need to do is put the $9 package’s price and quantity combination in the diagram showing the demand for trades of an uninterested trader. Call this point X, as shown in panel a. Notice that point X lies above the uninterested trader’s demand curve. That means if an uninterested trader were to make trade number Q_o (at a commission of $9), she would actually lose consumer surplus by doing so. At a price of $9, an uninterested trader really only wishes to purchase Q_max trades, the quantity demanded at that price.

399

The fact that Q_max is less than Q_o implies that the uninterested trader’s willingness to pay for the trades between Q_max and Q_o is lower than the $9 she would have to pay for them. In fact, all trades for which her demand curve (which indicates her willingness to pay) lies below $9 will result in a loss of consumer surplus. In panel a, these surplus-destroying trades are those between Q_max and Q_o, and the total consumer surplus lost is the area labeled L. (The demand curve just runs along the horizontal axis once it hits the axis, because willingness to pay for higher quantities is zero.) That area is the downside for an uninterested trader accepting the lower-commission offer. There is an upside, however. The first Q_max trades she conducts create consumer surplus, area A + B in the figure. This consumer surplus is quite a bit larger than her surplus under the $30 per trade offer (area A) because the price is so much lower. The net consumer surplus an uninterested trader gets from taking the $9 package offer is therefore area A + area B – area L.

Comparing the uninterested trader’s consumer surpluses from the two offers, we can now see that she will choose the $30 per trade offer over the $9 package offer if

Area A > area A + area B – area L

0 > area B – area L

Area B < area L

That is, an uninterested trader will take the offer designed for her ($30 per trade) if the extra consumer surplus she would obtain from the lower commission rate (area B) is smaller than the loss she suffers from having to buy a larger quantity than she would have otherwise at the lower offered price (area L).

For uninterested traders, we have outlined under what conditions the offers are incentive-compatible. Will an obsessed trader choose the $9 package meant for her? We know that at a commission rate of $9 per trade, an obsessed trader earns consumer surplus on every trade up to Q_o; she is happy to trade that much at that price. Taking the $30 offer would require her to make a smaller quantity of trades than Q_o at a higher price per unit. Having to consume a smaller quantity even holding the price fixed at $9 per trade would make an obsessed trader worse off, because it would eliminate surplus-creating trades she would have made otherwise at that price. Even worse, however, would be that the trader would have to pay $30 instead of $9 for each of the trades she did make. Both the quantity restriction and the increase in price reduce the obsessed trader’s consumer surplus. Thus, the $9 package offer is better for obsessed traders.

We saw that an uninterested trader also faces a higher price and lower quantity if she takes the $30 per trade offer instead of the $9 package. So, why isn’t an uninterested trader automatically worse off by taking the $30 offer, as is an obsessed trader? The reason is that if an uninterested trader faced a price of $9 per trade but got to choose how many trades she made, she would never choose to make Q_o trades. She would only choose to make Q_max, the quantity of trades demanded at a price of $9 per trade. Any trades between Q_max and Q_o destroy consumer surplus for an uninterested trader because the price is higher than her willingness to pay. It is the potential consumer surplus-destroying trades tied to the $9 package that make it likely that an uninterested trader would prefer the $30 offer.

400

Airline tickets are a classic example of what we call versioning—offering a range of products that are all varieties of the same core product. Airlines have a group of business travel customers who are not very sensitive to prices and a group of leisure travelers who are highly sensitive to price. Airlines want to charge different prices to the two passenger groups, but they can’t tell who is flying on business when a customer buys a ticket. So, the airlines instead offer different versions of the product (tickets on a given flight) available at different prices. The cheaper version, with many restrictions, is intended for leisure travelers who buy generally well in advance of the travel date, stay over a Saturday night, and book a round-trip flight. The more expensive version has fewer restrictions and is intended for business travelers who generally don’t like spending a weekend away from home, often need to buy their tickets at the last minute, and may choose to purchase a one-way flight for each segment to provide them with added flexibility. By offering two versions of tickets for a given flight, the airline attempts to make the two types of customers sort themselves (and by doing so, the airline captures more producer surplus).

For this scheme to work, the airlines need to make sure the prices of each version are incentive-compatible. If the airline sets the prices for each group based on the markup formula it would use with direct price discrimination, the restricted-travel version might be too cheap relative to the ticket with fewer restrictions. In this case, business travelers might actually bite the bullet and start planning trips earlier or stay at their destination over the weekend. In some cases, business travelers might try to skirt the rules altogether. For example, it’s possible to avoid the Saturday stay requirement by buying what is known as “back-to-back” tickets. For example, a business traveler wishing to fly from Philadelphia to Orlando and back for a Wednesday meeting might buy one Philadelphia-Orlando round-trip with a departure on Wednesday morning and a return on Sunday, and an Orlando-Philadelphia round-trip with a Wednesday evening departure and a Sunday return. The traveler would only use the first leg of each trip. As you might expect, this kind of behavior is intensely hated by the airlines, which try to forbid it in every way they can, but, basically, it’s just the market’s normal response to indirect price discrimination.

402

Versioning and Price-Cost Margins With versioning, the different versions’ marginal costs do not need to be the same. All that is necessary for versioning to work is for the markup of price over marginal cost to be bigger for the versions bought by customers with less elastic demand.

Consider the example of an automaker like Toyota, which sells a lot of midsize sedans. Some of Toyota’s buyers in this segment will not be very price-sensitive. Maybe they are status-conscious, or just have a particular taste for cars with many fancy features. Others will be more price-sensitive. If Toyota could tell which type of customer was which when he or she walked through the door, Toyota could just use direct price discrimination and charge different prices according to the strategy we discussed in Section 10.3. In reality, however, it’s not always easy to tell what type of customer comes through the door at any particular time. So, Toyota uses indirect price discrimination and designs different versions of the car that it can sell at different markups, hoping to induce buyers to segment themselves based on their sensitivity to price and tastes for features.

For example, Toyota makes the Camry, one of the highest-selling cars in the world. It sells, nicely equipped, for about $26,000 in the United States. But Toyota also makes the Lexus ES, which is built on the same platform and in the same plant as the Camry. It is similar in many ways to the Camry but is more luxurious. Think of the ES as being a Camry but with a bigger engine, fancier seat controls, halogen headlights, and a theft-deterrent system. The ES sells for around $38,000.

While those extra options raise Toyota’s marginal cost of producing an ES, it’s unlikely that this increase in marginal cost would amount to $12,000 per car. Toyota charges more than the cost difference because the different versions split its customers into groups based on their price sensitivities. The Lexus group has less elastic demand, so Toyota’s markup over marginal cost can be higher, just like the Saturday-night stay splits an airline’s customers into leisure and business travelers.

To be incentive-compatible, Toyota can’t make the deal for the cheap version so good that it convinces the luxury customers to purchase Camrys instead. Quantitatively, think of it the following way. Suppose there are just two types of customers whose willingness to pay for each car is listed in Table 10.1.

Notice that both consumers believe the Lexus is worth more than the Camry. It’s not that Toyota has made a version that one group likes and the other doesn’t. The budget consumers value a Lexus more than a Toyota, but not very much more: $31,000 versus $28,000. The luxury consumers, however, value the ES a lot more than the Camry: $43,000 versus $30,000.

If Toyota prices the Camry at $26,000 and the Lexus ES at $38,000, the budget consumers get $2,000 of consumer surplus from buying the Camry and –$7,000 from buying the Lexus (it costs more than they value it), so they will buy the Camry. The luxury consumers get $4,000 of surplus from buying the Camry and $5,000 from the Lexus, so they go with the Lexus. Each group chooses the version designed to take advantage of the nature of their demand curves. That means these prices are incentive-compatible.

If Toyota priced the Lexus at $40,000 rather than $38,000, the budget consumers would still buy the Camry. The status consumers, however, would get more consumer surplus from buying the Camry ($4,000) than from buying the Lexus ($3,000), so they would also decide to buy the Camry. That $2,000 price increase for the Lexus would cause Toyota to lose $12,000 (losing a sale of a $38,000 Lexus at the old price for a $26,000 Camry instead) for each luxury consumer. By charging the group with the less elastic demand too high a price, Toyota would not be setting incentive-compatible prices and its attempt at indirect price discrimination would fail.

403

One detail that is important to note is that it is not the mere existence of customers with inelastic demand that allows Toyota (or any other firm) to indirectly price-discriminate with versioning. What is required is that differences exist in demand elasticities across customer groups. If different consumer groups had the same price elasticities of demand, even if relatively inelastic, then designing versions specifically for each group will not help a firm price-discriminate. Automakers offer cars with different paint colors, for example, but there is rarely price discrimination based on paint color because the price sensitivities of people who like blue cars and silver cars are no different.

There is virtually no limit to the kinds of versioning a company can implement to get its customers to self-select into groups based on their price sensitivities. Now that you understand this kind of price discrimination, you will see it everywhere you look. Some firms offer “enhanced” features, such as the way Intuit does with its TurboTax software. It has a bare-bones version that is actually free online, versions with special Q&A features, and a small business application package that includes the ability to handle more complex structures like partnerships. The marginal cost difference between editions is trivial, but by offering “bells and whistles” versions, Intuit is able to get the less price-sensitive business customers to pay more.

Coupons are also a form of indirect price discrimination. Retailers would like to charge shoppers who have less elastic demands more for products while setting a lower price for consumers who are more sensitive to price. Again, however, they have no way of directly identifying and separating these different groups when they buy, so they have to get the groups to do it themselves. Coupons are a device they use to do so.

The key to the way coupons work is that the people who go to the trouble of using coupons—who search for the right site or deal online, comb through junk mail, or search through newspaper inserts—are more likely to have more elastic demand. Because both the willingness to do the work clipping the coupons and the willingness to shop around for cheaper groceries are determined by the consumers’ perceived value of time, coupon clipping and the price elasticity of demand are likely to be correlated. That way, the people who actually end up getting a price discount from a coupon are those consumers with more elastic demand—exactly the group to whom the retailers would like to offer lower prices. The shoppers who are less sensitive to price end up paying the higher, undiscounted price.

That’s why coupons usually aren’t right next to (or especially already attached to) the items to which they apply. If they were, it would be easy for even the shoppers with less elastic demand to use them, and everyone would receive the discount. The fact that firms require consumers to expend a little effort to use a coupon is not coincidence; it is exactly the point. Mail-in rebates work on the same principle: Only those consumers willing to go through the trouble of filling out the form and sending it in—presumably the most price-sensitive ones—will receive the discount.6

404