One of the more interesting parts of the November 29 DOJ/FTC hearing on loyalty discounts (where I presented these remarks) was the panelists’ discussion of a number of “propositions” advanced, for purposes of discussion only, by the agencies. Unfortunately, we didn’t have time to discuss all the propositions. I’ve reproduced them below the fold, along with my own thoughts on whether they’re sound. (Please note the agencies’ insistence that “[t]hese propositions are solely for the purpose of discussion and do not necessarily represent the agencies’ views.”)
The propositions are:
Single product discounts should be per se lawful if the overall price (for all units) exceeds cost.
I agree with this proposition. A single-product loyalty discount could be matched by any equally efficient rival. Now, some folks (Einer Elhauge, most prominently) have argued that loyalty discounts may actually impair a rival’s efficiency vis-a-vis that of the discounter. The argument here is that the discount may be so effective at winning business for the discounter (and taking it from rivals) that rivals must cut back their production so much that they lose economies of scale. While that’s a theoretical possibility, I don’t think it justifies an exception to the Brooke Group rule (which says that below-cost pricing is necessary for Section 2 liability based on low single-product prices). First, any rule aimed at protecting rivals’ economies of scale would be too unwieldy to manage and would therefore chill pro-competitive discounting. How is a court (or, heaven forbid, a jury) to determine what constitutes minimum efficient scale? And how is a court to know what level of discounting is “too much” because it threatens to deny rivals such scale? Moreover, couldn’t a rival who found itself losing business that denied it economies of scale raise the capital needed to operate at such scale? Firms incur start-up costs all the time in order to get to minimum efficient scale. They’re able to do so, even though their first units are priced at an unprofitable level, because investors know that once they achieve minimum efficient scale, they’ll be strong enough rivals to earn at least competitive rates of return. In a nutshell, then, I’d endorse this proposition because (1) I don’t think discount-driven denial of scale is that big a problem, and (2) any alternative rule would be too difficult to administer and would therefore chill discounting.
The LePage’s decision’s vagueness is likely to chill pricing behavior that enhances consumer welfare.
The LePage’s court did not announce a clear rule and provided no safe harbors. LePage’s was not required to prove that it could not match 3M’s discounts. Nor was it required to prove that it was as efficient a producer of transparent tape as 3M (indeed, its expert economist admitted that it was not as efficient producer). All it was required to show was that 3M’s bundle contained products LePage’s did not make (so LePage’s couldn’t replicate the bundle) and that this fact made it difficult to compete with 3M. The burden then shifted to 3M to prove a “business reasons justification” for its bundled discounting, and the court suggested that such justification would require 3M to prove cost-savings approaching the amount of any discounts. When 3M could not make such a showing, it was hit with a $68 million judgment.
Bundled discounters beware. Be careful when you respond to those RFPs that ask you to quote customers a bundled price.
A bundled rebate or discount can exclude an equally efficient single-product competitor, even if the post-discount price for the bundle as a whole is above cost.
I agree with this proposition, as long as efficiency is defined in terms of the costs of producing the single product. [NOTE: Both Janusz Ordover and Tim Muris suggested that this definition is too cramped. They suggested that a firm might be inefficient with respect to scope even if it could produce the competitive product as cheaply as the discounter.]
An above-cost bundled discount may have the effect of excluding from the market rivals that (1) are more efficient at producing the products that compete with the discounterâ€™s but (2) produce a less extensive product line than the discounter. The Ortho Diagnostic court offered an example of how this anticompetitive exclusion could occur. Assume that you have two sellers: A sells both shampoo and conditioner, and B sells shampoo only. Assume also that customers normally use both products in cleaning their hair. Now, suppose that B is the more efficient shampoo producer â€“ its average variable cost (AVC) of producing a bottle of shampoo is $1.25, while Aâ€™s is $1.50. Aâ€™s AVC of producing conditioner is $2.50. If both products are sold separately, A charges $2.00 for shampoo and $4.00 for conditioner (a total of $6.00), but if a consumer purchases both shampoo and conditioner, A will sell the combination for $5.00, a price thatâ€™s still $1.00 greater than Aâ€™s cost. Under these circumstances, B could stay in the market only if it charged no more than $1.00 for shampoo (so that a consumerâ€™s total price of Bâ€™s shampoo and Aâ€™s conditioner would not exceed $5.00, Aâ€™s package price). Of course, B couldnâ€™t do so, given that its cost of production is $1.25. Aâ€™s bundled discounting would therefore seem to eliminate B as a competitor even though B is the more efficient producer and even though A charges a price greater than the AVC of its shampoo/conditioner combination. This example illustrates the primary competitive concern that has troubled courts considering the legality of bundled discounts â€“ namely, that a monopolist who sells in multiple product markets will use such discounts to exclude equally efficient rivals that do not sell as broad a line of products (and thus have fewer products on which to give up margin).
Now, admitting that above-cost bundled discounts may lead to the exclusion of equally efficient rivals does not imply a conclusion that such discounts should be regulated. If it’s impossible to deter discounts that exclude in this fashion without also deterring non-exclusionary discounts, then it may be best to adopt a hard-and-fast rule that above-cost bundled discounts are per se legal. (That’s Tim Muris’s position.)
A loyalty discount that allows a competitor to operate profitably at some scale can never be harmful to consumers.
I think I agree with this proposition. At a minimum, I agree with the proposition that a single-product loyalty discount that results in an above-cost price and allows a competitor to operate profitably at some scale should be per se legal. My reasoning is as stated above: An equally efficient rival could match a single-product loyalty discount. Moreover, any rival able to operate profitably post-discount ought to be able to raise the capital necessary to expand its scale to the degree necessary to achieve all available scale economies. (It might need to raise money and engage in some below-cost pricing for a while to reach minimum efficient scale.)
Because lower prices immediately benefit consumers, we should be extremely careful not to adopt legal rules that can result in false positives–that is, condemn legitimate price-cutting.
This proposition seems so obviously correct that I need not comment on it.
Loyalty discounts (either single product or bundled) should never be condemned without applying some kind of price-cost test.
I agree. For bundled discounts, a price-cost test (i.e., a showing that less diversified rivals could not match the discount without pricing below cost on their narrower product line) is a necessary, but not sufficient, showing for establishing the sort of rival exclusion that could lead to consumer harm. Any rival that could not make this showing could stay in business by lowering its prices to competitive levels, and the legal rules should encourage it to do so. (Dan Crane makes this point in his written remarks.) For single-product loyalty discounts, any above-cost discounted price could be matched by an equally efficient rival. While Elhauge and others have argued that single-product loyalty discounts may actually contribute to rivals’ relative inefficiency, a rule designed to protect rivals’ ability to achieve economies of scale is simply too difficult to administer and will likely have a chilling effect on discounting behavior.
In a loyalty discount case, intent is relevant to proving monopolization.
There was a good deal of discussion of this proposition at the hearing. At first, pretty much everyone said something to the effect of, “We disagree. Everyone intends to beat competitors. Thatâ€™s the essence of competition.” The moderator then asked (cleverly), “Should intent evidence be relevant to a defense? For example, suppose the defendant engaged in some conduct intending to achieve some procompetitive end but actually ended up excluding an equally efficient rival without achieving the end at issue.” Folks then seemed to think the intent evidence should be relevant.
On the one hand, this looks like a “heads I win, tails you lose” approach that only an antitrust defense lawyer could love. On the other hand, I think itâ€™s right. The courts have long recognized that intent evidence may be probative of effect and thus (and for that reason only) may be relevant in antitrust cases. If we view intent evidence this way, then the asymmetric view seems sensible. Evidence of an intent to hurt rivals is not particularly probative of whether the action being done is procompetitive (everyone intends to hurt/defeat their rivals, even when theyâ€™re taking output-enhancing, procompetitive actions). Itâ€™s also highly prejudicial. By contrast, evidence of an intent to pursue some procompetitive end is highly probative of competitive effect. Most actions undertaken in order to pursue a procompetitive end are, in fact, procompetitive. Evidence of intent helps us see exactly what those procompetitive effects are. (Sometimes theyâ€™re not immediately obvious.) At the same time, such evidence is not particularly prejudicial. If a plaintiff shows that a defendant did something truly predatory that raised the plaintiffâ€™s cost of business, foreclosed it from the market, etc., a factfinder likely would not be overly persuaded by internal documents showing that the defendant had a procompetitive intent. Thus, asymmetric treatment of intent evidence can be defended on basic Rule 403 grounds: The “probative value versus prejudicial effect” calculation is very different for evidence of anticompetitive intent than for evidence of procompetitive intent.
Loyalty discounts can produce anti-competitive effects similar to those of exclusive dealing or tying.
I agree with this proposition when we’re talking about targeted loyalty discounts offered to resellers (e.g., the loyalty and bundled discounts 3M tailored for Wal-Mart, K-Mart, etc.). With these sorts of discounts, the relevant competitive concern seems to be foreclosure from channels of distribution. For this reason, I believe Josh is right to call for a showing of substantial foreclosure before condemning a discount. As I noted in my comment on Josh’s post, though, some bundled and loyalty discounts are not targeted at particular resellers but are instead generally available (e.g., package pricing for end consumers). Those discounts, I believe, are better analogized to predatory pricing.
Loyalty discounts involving a single product and loyalty discounts involving multiple products can have similar competitive effects.
Yes, they can have similar competitive effects, but bundled discounts are probably more troubling. Both forms of discounts can have the effect of raising rivals’ costs by denying them economies of scale or relegating them to inferior channels of distribution. For example, a successful single-product loyalty discount could usurp so much business from rivals that their production falls below minimum efficient scale. [For reasons explained above, I do not believe this possibility justifies an exception to the Brooke Group rule for single-product loyalty discounts. But it is a possibility.] Similarly, if a manufacturer offers bundled discounts to most of the major retailers, less diversified rivals that cannot meet those discounts may have to find other, less desirable outlets for their products.
Bundled discounts are probably more troubling, though, because they can exclude equally efficient, less diversified rivals even if they do not impair those rivals’ efficiencies. (See, e.g., the shampoo/conditioner example offered above.) A single-product loyalty discount could always be met by an equally efficient rival whose efficiency had not been impaired by the discount.