An enormous number of pricing decisions in the modern economy, not only in financial markets but in many on-line markets in general, are facilitated by computer algorithms. Indeed, a market may include the interaction of several different computer algorithms from buyers, sellers, and market-makers. Maureen K. Ohlhausen, who is Acting Chairman of the U.S. Federal Trade Commission, asks \”Should We Fear The Things That Go Beep In the Night?Some Initial Thoughts on the Intersectionof Antitrust Law and Algorithmic Pricing,\” in a speech given at the Concurrences Antitrust in the Financial Sector Conference on May 23, 2017. As as setting for her discussion of issue raised by algorithmic pricing, she offers a hypothetical example of small-town gas stations, which goes like this:
\”I want you to imagine a small, rural town somewhere in the desert southwest. This little town is modest and utterly unremarkable in every way. But in the center of town, clustered around the only stoplight, are three gas stations. These three stations have the only gas for sale within 150 miles in every direction. The date is 1970, so this is a time with no internet, no personal computers, no cell phones, and certainly no algorithmic pricing. All three gas stations currently charge exactly the same price for a gallon of gas. Prices may go up and down as the wholesale price of gasoline moves, but all three stations generally charge identical prices, and have been charging essentially identical prices for years.
\”All of that is about to change. At 6:00AM one bright clear Monday morning, the owner of the first gas station gets out his ladder and leans it against the big price sign out front. He then climbs up the ladder and changes the price, making it five cents a gallon more expensive. Then he takes his ladder down, walks over to a lawn chair in the shade and sits down to have a cup of coffee. At 10:00AM, he gets the ladder back out and lowers his price back down five cents so his price is now the same as everybody else’s price. He repeats that same pattern of behavior every Monday morning. He never directly talks to his competitors about the prices he is charging or why he is doing what he did.
\”Has he violated the antitrust laws just by changing his price for four hours? If that is all he has done, the answer is no. Generally, firms are free to set whatever prices they choose, as long as they act independently. Nor would it be unlawful for one of the other gas stations to decide, on his own, to follow the lead of our analog-era friend with the ladder and start raising his own prices on Monday mornings. Even if all the stations in the town ultimately decide to follow the lead of the first station and raise prices five cents, then keep those high prices in place, the antitrust laws do not condemn this behavior.
\”So why don’t we enforcers take action in this situation to prevent conscious parallelism? The simple reason is that there is no sensible remedy here. In a free market, individual actors are free to set their prices on the basis of all the information legally available to them. It is axiomatic that we cannot tell firms to ignore the public behavior of their rivals when they set prices without deleting the “free” in free market. Enjoining this kind of behavior would inevitably lead to price regulation, which is completely inimical to the underlying purposes of the antitrust laws. Because we cannot police this sort of behavior directly, instead we try to make sure, primarily through our merger enforcement program, that the conditions that allow this kind of behavior to take place generally don’t arise in the first place. We also prohibit explicit agreements to set prices collusively and exchanges of competitively sensitive, non-public information between competitors.
\”Fortunately for all of us, there are many, many reasons why this kind of informal pricing interdependency frequently fails or breaks down in the real world. For example, when the products are highly differentiated, or the market participants have different cost structures, or transactions are relatively infrequent, it is very difficult to maintain stable, interdependent pricing just by watching the behavior of your rivals. So the specific facts of my gas station example are very important.
\”Everybody is selling the exact same commodity product. Transactions happen frequently, with each individual transaction relatively small and unimportant, so there is little risk of major losses associated with price leadership. Also, there are few participants in the market, making it easier to get to a point of tacit consensus. Finally, and most critically, there is complete price transparency because everybody can see the prices everyone else charges just by looking at those big signs. If we take away any one of those facts, the whole thing will generally fall apart on its own. For example, if firms could somehow secretly discount and steal market share from their rivals, they have a significant incentive to do that and so on. …
\”So while our friend with the ladder may eventually, informally lead everyone’s prices higher, things look a lot different from a legal perspective if he walks over to one of his competitors and starts talking to him about prices. Suddenly we now have conduct that has nothing to do with independently setting prices and reacting to market conditions. The policy considerations that tolerate unilateral but interdependent pricing no longer apply. Once competitors reach an agreement setting price or output, they are engaged in behavior with no social utility and an enforcement response by the government is warranted. So there is a critical legal difference between concerted behavior among competitors aimed at influencing prices and unilateral decision-making in light of observed market conditions. Setting prices together is illegal, while observing the market and making independent decisions is not.\”
Ohlhausen uses the small-town gas station example to argue that the issues raised by algorithmic pricing are reasonably familiar, that similar issues have come up in well-known cases in the past, and so the antitrust authorities will have power to act if needed.
For example, one concern about algorithmic pricing is that the players in the market may use it as a way of colluding with each other behind the scenes, in a way that isn\’t easily visible in market prices. Ohlhausen points out that more than 20 years ago, back in 1993, airlines were using the information on-line reservation systems as a way of limiting competition, but the antitrust authorities and the courts had no trouble understanding the problem and addressing it. She writes: \”This is because the type of technology used to communicate with competitors is wholly irrelevant to the legal analysis. Whether it is phone calls, text messages, algorithms or Morse code, the underlying legal rule is the same – agreements to set prices among competitors are always unlawful.\”
Ohlhausen proposes the \”guy named Bob\” rule: \” \”Everywhere the word `algorithm\’ appears, please just insert the words `a guy named Bob\’. … If it isn’t ok for a guy named Bob to do it, then it probably isn’t ok for an algorithm to do it either.\”
My sense is that Ohlhausen is surely correct about the underlying law here: that is, doing something with an algorithm does not offer any protection or immunity against antitrust rules. I also like the small-town gas station metaphor and the \”guy named Bob\” rule, as nice concrete ways of illustrating these issues. But I do worry that while algorithmic pricing doesn\’t alter the rules of antitrust, it may make collusion easier.
As one example, a few years back there was a scandal about the LIBOR benchmark interest rate (that is, London Interbank Offered Rate). The rate was set by having a number of big banks send in the interest rates they were being charged for borrowing. But it turned out that the individual actually submitting the rates were not sending actual numbers, but instead shading them up or down. As a result, the LIBOR would be a little higher or lower than it should have been. The effect was only a small amount for a short time, but LIBOR is linked to something like $300 trillion of financial instruments around the world, so a trader who knew that the rate was artificially high or low could find a way to cash in. This manipulation went on for years, and led to prosecutions at 15 major global financial institutions.
Yes, this particular case was eventually detected and prosecuted. But having algorithms ready to pounce on small and short-term movements in LIBOR clearly helped to facilitate the entire scheme. One wonders about the possibility of other cases, perhaps using market benchmarks or interest rates that are less prominent than LIBOR and involving far fewer participants, which are not detected. The problem here is not specifically algorithms, but rather that the internet offers lots of possibilities for those who wish to collude, and algorithms can make it easier to cash in on such collusion.
There are other cases where the algorithms themselves may become problematic. Algorithms are starting to offer the possibility of \”personalized pricing,\” which refers to presenting online shoppers with a carefully designed series of options and prices and \”if-you-buy-now\” sales that–in combination with their past buying patterns and well-known behavioral biases–push that person toward a certain choice.
In other cases, as we move deeper into the world of big data and artificial intelligence, algorithms for buying and selling and setting prices are increasingly moving beyond basic rules, like buying only when a price falls below a certain level. Algorithms are now analyzing past patterns of players in the market, including how those player have reacted in the past, and then planning and implementing a potentially multi-step strategy. My guess is that some point there will be an antitrust case featuring the \”algorithm defense,\” which basically says: \”Hey, I just set up the smart learning algorithm and let it run. How could I know that it would interact with other smart learning algorithms in a way that led to collusion?\” And the antitrust authorities (or other law enforcement) will need to argue that when a guy named Bob sets up and signs off on an algorithm, Bob needs to be personally responsible for what that algorithm does.