Einer Elhauge and Abraham Wickelgren, of Harvard and the University of Texas, respectively, have recently posted to SSRN a pair of provocative papers on loyalty discounts (price cuts conditioned on the buyer’s purchasing some amount, usually a percentage of its requirements, from the seller). Elhauge and Wickelgren take aim at the assertion by myself and others (e.g., Herb Hovenkamp) that loyalty discounts should be per se legal if they result in a discounted per-unit price that is above the seller’s incremental per-unit cost. E&W would cast the liability net further.
We advocates of per se legality for above-cost loyalty discounts base our position on the fact that such discounts generally cannot exclude aggressive rivals that are as efficient as the discounter. Suppose, for example, that widgets are normally sold for a dollar each but that a seller whose marginal cost is $.88/widget offers a 10% loyalty rebate to any buyer who purchases 80% of its widget requirements from the seller. Because the $.90 discounted price exceeds the discounter’s marginal cost, any equally efficient widget producer could compete with the discount by lowering its own price to a level above its cost.
But what if the loyalty rebate actually causes a rival to be less efficient than the discounter? Some have argued that this may occur, even with above-cost loyalty discounts, when scale economies are significant. Suppose that the market for tennis balls consists of two brands, Pinn and Willson, that current market shares, reflective of consumer demand, are 60% for the Pinn and 40% for Willson, and that retailers typically stock the two brands in those proportions. Assume also that it costs each manufacturer $.90 to produce a can of tennis balls, that each sells to retailers for $1 per can, and that minimum efficient scale in this market (the lowest production level at which all available scale economies are exploited) occurs at a level of production equal to 35% of market demand. Suppose that Pinn, the dominant manufacturer, offers retailers a 10% loyalty rebate on all purchases made within a year if they buy 70% of their requirements for the year from Pinn.
While the $.90 per unit discounted price is not below Pinn’s cost, it might have the effect of driving Willson, an equally efficient rival, from the market. Willson could avoid losing market share and thus falling below minimum efficient scale only if it matched the full dollar amount of Pinn’s discount on its smaller base of sales. It wouldn’t be able to do so, though, without pricing below its cost.
Consider, for example, a typical retailer that initially (before the rebate announcement) satisfied its requirements by purchasing sixty cans of Pinn for $60 and forty cans of Willson for $40. After implementation of the rebate plan, the retailer could save $7 on its 100-can tennis ball requirements by spending $63 to obtain seventy Pinn cans and $30 to obtain thirty Willson cans. The retailer and others like it would thus have a strong incentive to shift purchases from Willson to Pinn. To prevent a loss of market share that would drive it below minimum efficient scale, Willson would need to lower its price to provide retailers with the same total dollar discount, but on a smaller base of sales (40% of a typical retailer’s requirements rather than 60%). This would cause it to lower its price below its cost. For example, Willson could match Pinn’s $7 discount to the retailer described above only by reducing its $1 per-unit price by 17.5 cents ($7.00/40 = $.175), which would require it to price below its cost of $.90 per unit.
When one considers dynamic effects, examples like this don’t really undermine the case for a rule of per se legality for above-cost loyalty discounts. Had the nondominant rival (Willson) charged a price equal to its marginal cost prior to implementation of Pinn’s loyalty rebate, it would have enjoyed a price advantage and likely would have grown its market share to a point at which Pinn’s loyalty rebate strategy could not drive it below minimum efficient scale. Moreover, a strategy that would prevent a nondominant but equally efficient firm from being harmed by a dominant rival’s above-cost loyalty rebate would be for the non-dominant firm to give its own volume discounts from the outset, securing up-front commitments from enough buyers (in exchange for discounted prices) to ensure that its production stayed above minimum efficient scale. Such a strategy, which would obviously benefit consumers, would be encouraged by a rule that evaluated loyalty discounts under straightforward Brooke Group principles and thereby signaled to manufacturers that they must take steps to protect themselves from above-cost loyalty discounts. In the end, then, any equally efficient rival that is committed to engaging in vigorous price competition ought not to be excluded by a dominant seller’s above-cost loyalty rebate.
Moreover, even if a loyalty rebate could occasionally drive an aggressive, equally efficient rival from the market, a rule of per se legality for above-cost loyalty discounts would still be desirable on error cost grounds. An alternative rule subjecting above-cost loyalty discounts to potential treble damages liability would chill all sorts of non-exclusionary discounting practices, so that the social losses from reduced price competition would exceed any social gains from the elimination of those rare discounts that could exclude aggressive, efficient rivals. In short, the social costs resulting from potential false convinctions under a broader liability rule would overwhelm the social costs from false acquittals under the per se legality rule I have advocated.
The two new papers by Elhauge and Wickelgren contend that I and other per se legality advocates are missing a key anticompetitive threat posed by loyalty discounts even in the absence of scale economies: their potential to chill price competition.
The first E&W paper addresses loyalty discounts involving “buyer commitment”—i.e., a promise by buyers receiving the discount that they will purchase some percentage of their requirements from the discounter (not its rivals) in the future. According to E&W, the discounter who agrees to this sort of arrangement will be less likely to give discounts to uncommitted (“free”) buyers in the future. This is because, E&W say, the discounter knows that if it cuts prices to such buyers, it will have to reduce its prices to committed buyers by the agreed-upon discount percentage. The discounter’s rivals, knowing that the discounter won’t cut prices to attract free buyers, will similarly abstain from aggressive price competition. “The result,” E&W maintain, “is inflated prices to free buyers, which also means inflated prices to committed buyers because they are priced at a loyalty discount from those free buyer prices.” Despite these adverse consequences, E&W contend, buyers will agree to competition-reducing loyalty discounts because much of their cost is externalized: “[W]hen one buyer agrees to a loyalty discount, all buyers suffer from the higher prices that result from less aggressive competition,” so “an incumbent supplier need not compensate an individual buyer who agrees to a loyalty discount for the losses that all other buyers suffer.”
The second E&W paper contends that loyalty discounts may soften price competition and injure consumers even when they do not involve buyer commitment to purchase from the discounter in the future. According to E&W, “[b]ecause the loyalty discount requires the seller to charge loyal buyers less than buyer who are not covered by the loyalty discount, the seller cannot lower prices to uncovered buyers without also lowering prices to loyal buyers.” Given the increased cost of competing for uncovered buyers (i.e., any price concession will require further concessions to covered buyers), the seller is likely to cede uncovered buyers to its rival, which will reduce the rival’s incentive to compete aggressively for buyers covered by the loyalty discount. In short, E&W contend, the loyalty discount will facilitate a market division scheme between the discounter and its rival.
As is typical for an Elhauge paper, there’s some elaborate modeling and math in both of these papers. The analysis appears to be rigorous. It seems to me, though, that there’s a significant problem with both papers: Each assumes that loyalty discounts are structured so that the discounter promises to reduce the price from the amount collected in sales to others. While I’m reluctant to make sweeping claims about how loyalty discounts are typically structured, I don’t think loyalty discounts usually work this way.
Loyalty discounts could be structured many ways. The seller could offer a discount from a pre-determined price—e.g., “The price is $1 per widget, but if you purchase at least 80% of your widgets from me, I’ll charge you only $.90/widget.” Such a discount doesn’t create the incentive effect that underlies E&W’s theories of anticompetitive effect, for there’s no reason for the seller not to reduce others’ widget prices in the future. Alternatively, the seller could offer a discount off a list price that is subject to change—e.g., “If you purchase 80% of your requirements from me, I’ll charge you 10% less than the posted list price.” This sort of discount might discourage sellers from lowering list prices, but it shouldn’t dissuade them from also giving others a break from list prices. Indeed, in many industries hardly anyone pays list price. The only loyalty discounts that threaten the effects E&W fear are those where the discount is explicitly tied to the price charged to others—e.g., “If you purchase 80% of your requirements from me, I’ll charge you 10% less than the lowest price I’m charging others.”
This last sort of loyalty discount might have the effects E&W predict, but I’ve never seen such a discount. The loyalty discounts and rebates I encountered as an antitrust lawyer resembled the first two types discussed above: discounts off pre-determined prices or discounts off official list prices (from which price concessions were regularly granted to others). The loyalty discounts that E&W model really just look like souped-up “Most Favored Nations” clauses, where the seller promises not just to meet, but to beat, the price it offers to other favored buyers. It may make sense to police such clauses, but wouldn’t we do so using the standards governing MFN clauses rather than the rules and standards governing loyalty discounts? After all, it’s the seller’s promise to beat its other price concessions, not the buyer’s loyalty, that causes the purported anticompetitive harm.
UPDATE:
I just recalled that this is not the first time we at TOTM have addressed Prof. Elhauge’s models of loyalty discounts containing a Most Favored Nations-like provision. FTC Commissioner-Appointee Josh Wright made a similar point about a paper Elhauge produced before these two. If you found this post at all interesting, please read Josh’s earlier (and more rigorous) post. Sorry about that, Mr. Commish-to-be.