Price Fixing: "it Was the Machines, Sarah!"

• Jurisdiction: United States,Federal
• Publication year: 2019
• Citation: Vol. 2 No. 2

Dante A. Stella and Howard B. Iwrey^*

This article explores the changing nature of e-commerce "markets" the role of pricing software, and practical steps to minimize inquiries from the authorities.

E-commerce, particularly in "marketplaces," has brought about a fiercely competitive environment in which sellers frequently use price-setting software to achieve optimum pricing given various market conditions. This software uses algorithms of varying complexity to adjust prices on an almost real-time basis. For in-house counsel charged with antitrust compliance, this technology presents novel issues—due to the rapidity and reliability of the software, the relative ease with which the government or private parties can detect its use, and the ways that like Skynet in the Terminator, it can go rogue. In some instances, use of pricing software has already resulted in criminal antitrust prosecution, even for small companies. This article will explore the changing nature of e-commerce "markets," the role of pricing software, and practical steps to minimize inquiries from the authorities.

How Has the Market Been Changing?

In classical economics, the "market price" of a good or service is defined by the intersection of a demand curve (defined by the preferences of all buyers) and a supply curve (to which all sellers similarly contribute). Prior to the internet and the rise of e-commerce, customers did not have access to perfect information on pricing across sellers competing for their business. At best, a purchaser of goods or services would have to consult competing price lists, make written inquiries, or make phone calls. Even when commerce was first moving to the Web, it was still necessary to visit multiple websites to compare prices. Price shopping was burdensome, and in many cases, where impracticalities of long-distance buying (like shipping costs and lead time) inhibited arbitrage, negotiation of

[Page 97]

prices was much more local and limited. For example, a customer may pay more for a cubic foot of Styrofoam packing fill locally because a less expensive product sold in another state might be too expensive to ship in. Likewise, a driver with a burned-out clutch may be more inclined to pay more for a new part available immediately from a dealership than wait a few days for an independent garage to obtain a cheaper part.

Likewise, a seller in the pre- or proto-internet environment had a limited ability to divine the prices of its competition. Older readers will recall that in that "analog" era, a seller could only gain information about competitors' pricing from publicly available printed catalogues or word of mouth. Catalogues were printed only at intervals, injecting a time lag, and even then they typically showed manufacturers' list prices—which in many industries bore little or no resemblance to the price—as discounted at various percentages to wholesalers/retailers (including via back-end rebates) and then marked up to the end user (the "street" price). Customers, on the other hand, could be expected to provide inaccurate, difficult-to-verify, or outright fabricated information about competitive prices. For example, a customer could walk into Beta's business and declare that "your price is higher than Acme's." For business reasons, if not also fear of an antitrust violation, Acme probably would not confirm its selling price to Beta.

The rise of e-commerce websites run by individual sellers, and later, massive "e-markets" like Amazon and eBay that present the products of multiple sellers simultaneously, has resulted in far more informed buyers and sellers. A buyer can now compare prices for a particular product on numerous websites in a matter of minutes using Google Shopping, flip over to eBay or Amazon to see competitive pricing in list format, or even use real estate tools like Zillow that attempt to show price trends and norms. With special rate deals on Priority Mail and SmartPost, freight forwarding like eBay's Global Shipment Program, and Amazon' newish vertically integrated delivery service, buyers can connect with, and buy from, even the smallest businesses on the other side of the country. Such developments break down the geographic barriers to buying from out-of-state sellers.

The information situation, however, works much better for buyers than sellers. On Amazon, for example, it does not behoove a seller to be more expensive than any other seller unless customer feedback or quicker delivery can support a higher price. In

[Page 98]

practical terms, perfect information for buyers can lead to a nearly textbook perfectly competitive market—one in which sellers can either match the going price (becoming price takers) or in which they can choose to forego sales. This can lead to relentless price changes in e-marketplaces, as often as several times an hour for some products.

Because few businesses have the personnel and resources to continuously monitor price changes on e-commerce websites, many have turned to algorithmic pricing software, sometimes called "repricing software." An algorithm, put simply, is "a process or set of rules to be followed in calculations or other problem-solving operations."¹ The algorithms for pricing software can be as simple as a rule telling the program to match the lowest price, or as complex as a computation taking into account both measurable market conditions (like sales prices) and external factors (such as net prices or commodity prices). Many sellers will choose off-the-shelf software, but others may commission their own.

How Does This Intersect With Antitrust Enforcement?

Civil damages for violations of the Sherman Act have their own simple algorithm: treble damages unless your company first obtained leniency on the criminal side.

Where the government—be it the Antitrust Division of the Department of Justice ("DOJ") or the Federal Trade Commission ("FTC")—investigates a group of sellers, the process often starts with a tip or some readily identifiable phenomenon. Following a further look, this would lead then to a grand jury subpoena. Even where there is no smoking gun in a party's documents, the government often collects the various sellers' transactional sales data separately, attempts with an economist to identify where and how prices moved, and then does its best to convince a grand jury to infer cause to prosecute a conspiracy. Due to resource and time limitations, much of this process relied on game theory: a seller would be highly incentivized to turn in a competitor, and even if a seller participated in a conspiracy, it could reduce its own civil liability and punish its competitor by reporting the conspiracy to the government. The competitors would stand to suffer harsher criminal penalties if convicted—and in any event have three times

[Page 99]

the potential liability in any civil suit. The DOJ used this technique to devastating effect In re Automotive Parts Antitrust Litigation to extract almost $3 billion in fines arising from guilty pleas made by Tier I and II suppliers (i.e., direct and first-order indirect suppliers to vehicle manufacturers). This game theory also benefits settling defendants by curtailing what could otherwise be a large and expensive document production.

The rise of e-markets—sites where multiple sellers offer the same product simultaneously—makes investigations considerably easier for the authorities. The government is able to observe the pricing in a marketplace (as would any customer or competitor), and it can instantly see coinciding price movement. In fact, this movement—and the use of automated software to implement those moves—is detectable via machine learning.² This opens the door to almost continuous scanning of e-markets for suspicious...