I’ve been reading the papers for the FTC RPM Workshops, though I cannot attend.  On the procompetitive side, I especially recommend Ben Klein’s explanation of how RPM facilitates the supply of promotional services in the absence of dealer free-riding.  Critics of RPM, in my view, generally do not understand the fundamental economic point that retailer competition alone is not sufficient to guarantee the supply of promotional services because of incentive conflicts between manufacturers and retailers.  Klein and Wright (JLE, 2007) explains this incentive conflict in great detail, and how fixed per unit time payments (slotting contracts) can be used to solve this common incentive problem and are part of the normal competitive process.  Klein’s newest RPM explains how RPM contracts can be used to achieve the same effect, that is, solving the incentive conflict between manufacturers and retailers to facilitate the supply of efficient promotional services.

The most common argument raised by defenders of the Dr. Miles rule, including Justice Breyer in Leegin, is that Telser’s (1960) classic discount dealer free-riding story for RPM (RPM solves the problem of consumption of promotional services at the full service retailer before buying the product at the discounter — and thus unraveling the supply of services in equilibrium) does not apply to a number of products where we observe RPM used.  This is where Klein & Murphy’s (1988) seminal explanation comes into play, documenting how the incentive conflict is a real economic problem solved by these vertical restraints, and part of the normal competitive process.  The Klein’s RPM Workshop piece builds on and updates that analysis.

One of the other issues that I’ve been keeping my eye on during the hearings is the evaluation of the current empirical evidence.  I’ve written before that in my own evaluation of the evidence, “the evidence overwhelmingly shows (see also here) [that RPM agreements] are highly likely to make consumers better off in practice.”  I also wrote that I hoped the RPM Workshops would take a hands on and rigorous approach to evaluating the state of evidence in order to design appropriate antitrust enforcement approaches to RPM and vertical restraints generally:

In my view, while there is still a lot to learn about precisely how RPM works, when and by whom it is adopted, and to what effect, there is simply no empirical evidence that its effects warrant per se illegality…. Its my sincere hope that the policy debate to be had on RPM with the pending legislation and FTC Workshops upcoming will be fought on this margin rather than on marketing.

In this light, it caught my eye that Patrick Rey’s slides and paper, which offers yet another possibility theorem of how RPM “could” result in anticompetitive outcomes.   The possibility theorem paper is nothing new in the sense that there are a ton of these around.  But the claim of empirical support is.  Indeed, my views on this matter are well known that there is not much empirical support at all for the anticompetitive theories of vertical restraints including RPM (see also the Lafontaine & Slade and Cooper et al literatuer surveys).  So I did some digging.  Here’s the claim from Rey’s paper with Thibaud Verge:

Our analysis supports this claim and shows that RPM can actually eliminate competition, not only among competing fascias, but also among competing brands. This possibility has been validated by recent empirical studies. Using data about retail prices of food products in French retail chains during the period 1994-1999, Biscourp, Boutin and Vergé (2008) find that the correlation between retail prices and the concentration of local retail markets was important before 1997 and no longer significant after that date. This suggests that the price increases that occurred after 1997 were indeed due to the impact of the new legislation on intrabrand competition.

So what does the Biscourp et al. study actually analyze?  You might think from the context that BBV (2008) studies Minimum RPM contracts.  But you would be wrong.  What did they actually study?  Get this: a set of French laws that make it illegal for retailers to sell “below cost.”  The Loi Galland came into force in 1997 and clarified a pre-existing ban on below-cost sales.  In other words, the Loi Galland set mandatory government enforced minimum price floors that apply to all retailers.  Boutin & Guerrero provide some details on the 1996 Loi Galland:

The Loi Galland gave a simple, precise definition of the actual purchase price and hence of the below-cost retail price floor: `The actual purchase price is the unit price stated on the invoice plus taxes on sales, specific taxes applied to the resale, and transportation costs.’  Since 1997, the definition of purchase price has thus been restricted to the price stated on the invoice, with no deductions such as year-end discounts. The Act also tightened official verification and raised fines. Only the margins formally applied at the invoice date and shown on the invoice the “upfront margins” can be passed on to final consumers through reductions in the final selling price. By contrast, all other discounts are described as “hidden margins” and therefore excluded from the below-cost retail price floor. Examples include margins linked to an annual sales volume, to the retailer’s display of the product on a minimum shelf length, to business cooperation, or simply to the respect of mutual commitments over a certain period. Consequently, hidden margins can in no way be passed on to consumers.

The law in other words, is similar to sales below costs laws i the US that are designed to anticompetitive raise prices, are binding on all sellers, and enforced by the government with significant fines.  Unsurprisingly, a law designed to prevent price-cutting achieves its intended effect and is found to have an anticompetitive effect.  But I’m frankly lost as to how Rey & Verge claim from this study of government imposed sales below cost laws that there is empirical support for the proposition that voluntary, privately negotiated RPM contracts are likely to be anticompetitive.  Note that Lafontaine & Slade’s leading survey of the literature argues that this distinction is quite important — concluding that mandatory restraints are far more likely to generate anticompetitive outcomes.

There is more.

Rey and Verge also claim that another paper from Bonnet and Dubois (2008) “supports his analysis of RPM.”  Curious, I took a closer look at this paper.  Does this second paper actually study RPM agreements?  Again, the answer is no.  B&D (2008) do something different: (1) they get retail prices and quantities for branded and unbranded bottles of water, (2) estimate a random coefficients logit model for the demand for bottled water (no data on costs or markups except for general input price indices), (3) infer wholesale and retail markups under a variety of assumptions about the nature of wholesale and retail competition which amount to 12 different models, and (4) from these estimated markups, estimate marginal costs (e.g. 12 different cost equations).  From these cost equations, they conduct a series of non-nested hypothesis tests, comparing the 12 different models against each other (table7, p. 34 at link above).  The authors conclude that “the results finally show that the best model appears to be model 10, that is the case where manufacturers use two part tariffs with resale price maintenance.”  From this series of assumptions and steps, the authors conclude that the branded water sellers are actually using RPM and two-part tariffs (“Our empirical analysis allows it to be concluded that manufacturers and retailers use nonlinear pricing
contracts and in particular two part tariff contracts with resale price maintenance.”)!  Further, the authors simulate the effect of moving from model 10 to other models of pricing and find that pricing is lower in other models and therefore conclude that RPM has anticompetitive effects.

There are some problems with this analysis.  First, suffice it to say that it is difficult to make confident policy statements about the effects of RPM contracts without studying actual RPM contracts.  There is no actual evidence that the water sellers are using RPM.  Indeed, RPM is illegal in France.  The inference is generated because the non-nested hypothesis test (which is notoriously weak) implies the model best fits the data.   Second, my understanding of merger simulations at the FTC is that when marginal costs are inferred from equilibrium conditions, attempts are made to verify the validity of the inference by comparing the inferred level to some actual measure (so that the model can be tossed if they don’t correspond to one another).

Evaluating the empirical evidence in favor of the various pro- and anti-competitive theories is an incredibly important step in the process of identifying the appropriate legal test to apply to resale price maintenance (and other vertical restraints). While the Rey & Verge piece offers an interesting theoretical effect of RPM, the key issue identified by Justice Breyer’s Leegin dissent is understanding how RPM contracts work in practice. In my view, Rey & Verge’s claim that there is empirical support for their model is unfounded. To the contrary, neither the BBV or BD papers add any empirical contribution to the debate over the appropriate antitrust treatment of privately and voluntarily adopted RPM contracts.