Oligopolies ~= Monopolies

FIRST POST:

While this may be long for a first post, it’s actually a portion of a paper I wrote for one of my law school classes. Any thoughts on issues requiring review are welcome; should you see any market harm to consumers by an industry you believe is too big/too coordinated, feel free to share that as well. Note that I am extremely deferential to capitalism: any and all opinions and thoughts are structured solely in the context of having appropriate (and not onerous) regulations in the market. The father of the Invisible Hand, Adam Smith said: “When the regulation, therefore, is in favour of the workmen, it is always just and equitable; but it is sometimes otherwise when in favour of the masters.” There is a careful balance that must be found, as overly restricting the free market would be paramount to regulating blood out of the human body.

OLIGOPOLISTIC MARKETPLACES

Two foundational principles appear relevant to George Stigler’s groundbreaking work on understanding market participants in oligopolies. Game Theory, a mathematical calculation for analyzing potential outcomes based on actions of participants, forms one foundation of Stigler’s analysis. It is defined by Merriam-Webster’s Dictionary as “the analysis of a situation involving conflicting interests…in terms of gains or losses among opposing players.”[1] This foundation is a bedrock principle upon which Stigler relies. Stigler states that “it is a well-established proposition that if any member of the agreement can secretly violate it, he will gain larger profits than by conforming to it.” [2] He subsequently identified three problems relevant to cartels attempting to cooperate successfully based on this principle.

Stigler’s three tenets state that cooperative action requires a party to: 1) identify the terms of cooperative action, 2) detect a deviation from those terms, and 3) punish the deviation.[3] Quite simply, the more firms in the marketplace, the harder it is to identify cooperative deviating action and punish it. At some point, it becomes cheaper to simply compete with the firm rather than enforce a costly cartel pact.

Although Stigler’s analysis assumes a lack of competition law in the relevant marketplace, the methodology nonetheless holds true for marketplaces regulated by governments with unenforceable competition laws (and both single-action and repeated-action analysis). In instances where a sector is dominated by a few small firms, the assumption underlying those participants remains constant. Oligopolistic marketplaces with few participants allow production of goods and services whereby marginal revenue exceeds marginal cost. Incentives to cheat are premised on the ability to capture, in the short run, the excess revenue from competitors by lowering prices. This situation only occurs when Stigler’s three points are not met. Antitrust law serves as the roadblock to an entity seeking success in the three terms outlined above (either by dis-incentivizing these actions through economic fines, issuing injunctions, or introducing methodologies designed to increase competition).

The modern legal doctrines promulgated in the cases above offer significant deference to oligopolies. In them, whether under Section 1 or 2 of the Sherman Act, affirmative acts were required for liability regardless of the existence of conscious parallelism or a mere monopoly. Why, then, is Stigler’s model relevant for rejecting the rule of reason under Section 1 liability, and requiring application of a Per Se (No Fault) rule based exclusively on Market Share?

The crux of my argument centers on the pooling of information an offending oligopolist (or cartel) uses under Stigler’s analysis. In it, he notes that cartels or firms will identify price-cutting agreements based on a firm’s own old customers, the attraction of old customers of a competitor, and the behavior of new customers[4]. Stigler states that, with respect to a firms own old customers, “there are of course limits to such pooling of information: not only does it become expensive as the number of firms increases, but it also produces less reliable information, since one of the members of the pool may himself be secretly cutting price.[5]”

In 1964, this theory of price-cutting detection led to a widely-recognized conclusion that when the cost to cheat is too high, firms will return to competitive action. In essence, Stigler argued that Cartel Action occurs when:

Benefit of Cartel Action (B) > Monitoring Cost based on Stigler’s 3-part test (M1+M2+M3), or

B > (M1+M2+M3)

Stigler’s calculations assume a lack of regulation; it is thus prudent to add in the cost of Antitrust regulation. The cost of antitrust is made up of the cost of avoidance (A) and the cost of penalty upon detection (D) (these variables are not mutually exclusive). Additionally, there are two types of antitrust actions discussed above: explicit (action under Sherman Act 1 and 2 above), and implicit (where agreement is proved under doctrines like conscious parallelism). Both explicit and implicit antitrust are thus made up of avoidance and detection costs. The following formula results:

B > (M1+M2+M3) + Explicit Collusion + Implicit Collusion, or

B > (M1 + M2 +M3) + Ex(AExp + DExp) + Imp(AImp + DImp)

Again, the cost of avoidance (A) is independent of detection (D) (if detection occurs, it’s irrelevant what was paid to avoid detection). However, the variables are backwards-correlated: a higher level D may signify a lower quantity of A. Nonetheless, the calculations for these quantities and their relationship should be completed prior to entry into the formula (that is, they are not correlated in the formula itself, they may simply have some of the same inputs).

D may be quantified using Learned Hand’s test from United States v. Carroll Towing Co. That test stated that a burden (or cost) is a function of the magnitude of that injury times the severity of that injury[6]. In terms of the example above, D is a function of the magnitude of the penalty upon detection (Treble Damages) represented as G, multiplied by the probability of the detection occurring, or P. The formula now looks as follows[7]:

B > (M1 + M2 +M3) + Ex.(AEx +GEx*PEx) + Imp.(AImp+GImp*Pimp)[8]

Thus, when the costs of Monitoring (Part 1) and Antitrust (Parts 2 and 3) are smaller than the benefit, the firm will engage in anti-competitive practices. In other words, the firm will attempt to act against the open market when:

B > Monitoring + Antitrust

Some Examples will provide reference:

• Example #1: If a party chooses not to act (or does not act) in a capacity that would find them liable under explicitly defined violations or implicitly defined violations in the Sherman Act, then PImp and PEx would go to zero. Consequently, the party would not be required to spend funds avoiding detection. In this example, A would also likely be zero because a lack of violative action means the firm has no cost of avoidance.

• Example #2: If the party colludes in an explicit manner, PEx increases significantly (we assume that explicit collusion, i.e. formal agreements, are easier for regulators to catch). As a result, antitrust costs increase dramatically due to the collusive behavior. The benefit (if our penalties are large enough in this jurisdiction) should lower than the cost of antitrust, and the firm will likely choose not to collude.

• Example #3: If the party colludes in a provable implicit manner (that is, there is no direct evidence, but conscious parallelism with plus factors is proved), then the example operates like Example #2, as the action is provable. Thus, because Pimp acts like Pex, the result is the same[9].

• Example 4: If the party colludes in an improvable implicit manner (that is, there is no direct evidence, but conscious parallelism without plus factors is proved). In that instance, the likelihood of detection, that is Pimp, is near to zero. This is a direct result of the courts deference to naked conscious parallelism sans plus factors. Courts allow circumstantial evidence to establish agreements as required by Section 1. However, additional circumstantial evidence beyond parallel movement is needed (see discussion of plus factors, supra). This makes proving agreement even with conscious parallelism very difficult, and makes catching this type of collusion more difficult.[10] Here, because Antitrust costs are near to zero (no explicit collusion and almost no implicit collusion), the only roadblock to collusion is Monitoring costs (Stigler’s theory). Thus, if B > M1 + M2 + M3, the firm will collude. Historically, monitoring costs were so high, that it effectively deterred collusion.

Thus, without explicit collusion, courts have deferred to the marketplace. This deference to lies in that of the parties monitoring costs that traditionally deterred firms from colluding.

THE DECLINE OF MONITORING AND DATA COLLECTION COSTS IN BUSINESS ENTERPRISES

The Twenty-First century has witnessed an information explosion never before seen in history. The Library Congress estimates that its’ 147 million items would account for roughly 208 terabytes of information[11]. At a cost of approximately $110.00 per terabyte[12] , digitizing and storing all the records in the Library of Congress would cost approximately $22,880.

According to Moore’s Law, computing power doubles approximately every two years[13]. Regardless of the cost incurred at year one, within ten years, utilizing that same cost (all of other factors being equal), computing power will have increased by a factor of 524,288 (X20) over twenty years. If the processing of a computer at the time when Stigler promulgated his thesis (1969) is X, then processing power today is X times 2.20 trillion (X42). The result is processing speed and power that was never available to previous generations.

This exponential drop in pricing of information leads us to one conclusion: the costs of gathering information and processing data (the “transaction costs” per piece of information collection) are so small the costs become irrelevant. The creation of a global networked internet infrastructure now allows corporations to obtain competitor price information and customer decision information in near real time. This deluge of information drastically affected the traditional analyses for oligopolistic price deterrence.

In the formula presented above, monitoring costs under Stigler’s theory of oligopoly were the essential deterrent to collective and collusive action. In essence, the monitoring costs described above are exclusively related to how much money an entity may spend attempting to collect information on its competitor’s price practices. The advent of the internet, of automatic technology systems and real-time price monitoring and analytical tools has successfully driven monitoring costs to near nothing.

Accordingly, the only deterrents left in the formula proscribed above are the costs, or penalties, of Antitrust Regulation. As with any rational and capitalist economic analysis, when the benefit is greater than the cost, the act will occur. As monitoring costs have descended towards zero, firms realized that the benefit derived from acting as a cartel was higher than the costs.

We must be careful not to confuse cause and effect: the cost of avoiding implicit collusion (conscious parallelism with plus factors) means that detection will not occur. Avoiding collusion is an input in the formula, not an output. Thus, we can change the “Antitrust” variable. What exactly should one change about this variable to reduce the benefit derived by an oligopolistic for colluding?

Even more damningly, the courts have held consistently that sharing of information constituted agreement, and was thus actionable under Section 1 of the Sherman Act[14]. To hold this in today’s society would be catastrophic to business enterprises. It is impossible to dispute the benefits society enjoys when retailers and put their prices and products online. Simple software programs written by a firm can easily analyze competitor price data and action on a near real-time basis. Thus, if the exchange of information about pricing in an oligopolistic market were all that was required for liability, as American Column and Lumber would suggest, liability would become ubiquitous. This would be a naïve conclusion, and is accordingly dismissed. However, it does not solve the problem that monitoring costs have decreased to such an infinitesimally small amount they are no longer a relevant variable in the equation. The equation has thus become:

B > Ex.(AEx + GEx * PEx) + Imp.(AImp + GImp * Pimp)

Given the increased likelihood that the blatant nature of explicit collusion would still be caught by society, it does not become a concern. However, given the low probability of catching implicit collusion (with no plus factors), it would seem inevitable that firms begin to collude.

In cases like American Column & Lumber Co. v. United States[15], the distribution plan that contained market letters and sales reports are now pieces of information that firms can, at least in part, aggregate on their own using internet-based tools and algorithms. Extending this doctrine would make liability for collusion universal; judicial doctrines based on aggregation of information must then be soundly rejected. Instead, we must focus on what constitutes agreement in order to find liability for collusion under Section 1.

RECOMMENDATIONS

The formulas above, while novel in presentation, are simply recreations of cost-benefit analysis and decision-making on a slightly more granular level. This analysis is keenly focused on firm action. However, these conclusions still appear to produce desirable outcomes from a societal perspective: if cheating occurs, then price inefficiencies exist, and dead-weight-loss to society is not disputed, but inevitable. Simply put, this is undesirable for society.

Let me be clear in summary: the decisions and policies put in place first by our legislature, and later interpreted by our Judiciary have unintentionally led to this dilemma. The technology revolution has stripped our laws of their novelty and efficacy. A few tweaks to the system may once again restore balance and disincentivize firms to collude.

The legal question turns on one of intent. As discussed above, conscious parallelism may be a form of agreement under Section 1. In this situation, an affirmative act (or plus factor) is required to find an entity liable. The court, in United States v. Container Corp.[16] referenced market structure as a component of its analysis under what constituted implicit agreement under Section 1.

My first suggestion involves the removal of oligopolistic market structure as a factor for analysis under what tends to look like a Rule of Reason review for conscious parallelism and agreement. It would be difficult to simply state that the existence of an oligopolistic market structure should be a distinguishing factor between applying the Rule of Reason and Per Se violations. A middle ground is available.

The court in FTC v. Indiana Federation of Dentists noted that a restraint could be unreasonable because it was “Per Se unreasonable, or because it violate[d]…the ‘Rule of Reason.’ [17] However, instead of adopting the traditional Per Se rule against boycotts, the court deferred and held for the insurers on the middle ground that “no elaborate industry analysis is required to demonstrate the anticompetitive character of such an agreement.[18]” Most importantly, the court stated that “Absent some countervailing precompetitive virtue…such an agreement limiting consumer choice impeding the ‘ordinary give and take of the market place’[19] cannot be sustained under the Rule of Reason.

The Federal Trade Commission issued guidance, suggesting that analysis ranging from the Rule of Reason to Per Se application is best referenced as a continuum, and that some firm actions that fall in between should be evaluated under a ”Quick Look,” doctrine[20]. In these cases, the burden of proof normally given to the plaintiff is “satisfied through the presumption of competitive harm flowing from the nature of the conduct…[21]”. Accordingly, the Commission shifted the burden to the defendant to prove the pro-competitive nature of its activity for issues of agreements to restrain trade.

The suggestion follows: upon the existence of an oligopoly (according to standard jurimetrics), the burden should shift to the defendant for offering pro-competitive reasoning for the activity. This would increase the cost of Antitrust with respect to the equation above by increase the cost of being penalized. In short, burden-shifting would result in more liability by potentially colluding parties.

The second potential resolution to the issue of conscious parallelism and implicit agreement is to qualify price-tracking itself as a doctrine under which liability exists. In the event one or more firms are price-tracking a market leader in real time, a correlation coefficient of price-tracking[22] can be cross-referenced with a traditional Herfindahl-Hirschman Index number[23] (so as to normalize depending on market concentration). Again, jurimetrics analysis can provide historical context as to if the tracking based on that market structure is irregular.

This methodology is directly analogous with that presented by Learned Hand in his discussion of market share as a determinant of monopoly for Sherman Act Section 2 liability. According to this logic, market share, even in oligopolies, can be a powerful tool in enforcing potential problematic activity. Much like in ALCOA[24], where Justice Hand found that “90 percent is enough to constitute a monopoly,” a market share understanding may develop overtime as the theory is tested across jurisdictions. Given the strength of the existing case law on analyzing monopolization according to market share, there may be sufficient justification for establishing a similar test for collusion in oligopolistic markets.

The third suggestion is the embracing of the oft-discussed theory of signaling as a plus factor for unconscious parallelism. Signaling, according to federal agencies, is defined as a process “in which corporate executives signal to each other about the need for or expected actions on market price increases. These signals can be in the form of publicly talking about a price increase, or making an advance announcement in the media of a coming increase.[25]”

Lastly, I suggest embracing of the use of Section Five of the Federal Trade Commission Act. This section allows the Federal Trade Commission to prevent persons “from using unfair methods of competition in or affecting commerce and unfair or deceptive acts or practices in or affecting commerce.[26]” As Chairman Leibowitz and Commissioners Kovacic and Rosch opined in In the Matter of U-Haul Int’l, Inc and AMERCO, “In contrast to conspiracy claims that would violate Section 1, invitations to collude [Under Section 5 of the Federal Trade Commission Act] do not require proof of an agreement; nor do they require proof of an anticompetitive effect.” Moving the burden lower would allow for easier prosecution.

CONCLUSION

The spectrum of the economic marketplace has changed as technology has grown faster and smarter. No longer is it easy to police the distribution of monopolies and oligopolies in a field of competitive markets. There is no doubt that the increase of technology has shifted market structure classification to that of extremes. The availability of price-checking software and mobile computing combined with real-time analytical updates puts much of the power in the hands of the consumer. This creates marketplaces that resemble those of the perfectly competitive ideologue. Conversely, this same technology is also allowing oligopolists to implicitly collude through a variety of actions, in effect allowing them to act like monopolists. This pressure on the opposite end of the spectrum cannot be regulated by existing statutory or judicial antitrust schemes. The tools exist to combat the propensity of market participants to collude. By understanding the incentives of potentially collusive market participants, small system tweaks can be implemented in order create a more orderly and efficient marketplace to the betterment of society.

[1] Merriam Webster, Game Theory. [2] A Theory of Oligopoly. George P. Stigler. The Journal of Political Economy. Vol. 72, No. 1. (Feb. 1964), 44-61. [3] Two Sherman Act Section 1 dilemmas: parallel pricing, the oligopoly problem, and contemporary economic theory. Jonathan B. Baker. The Antitrust Bulletin, Spring 1993. Pg. 143. [4] Stigler at 51. [5] Id. [6] Carroll Towing. 159 F.2d 169 (1947). [7] In the formula, Ex = Explicit Collusion, and Imp = Implicit Collusion. A is equal to the cost of avoiding each type of collusion, and D is the penalty cost upon detection (quantified using Learned Hand’s formula from Carroll Towing). [8] The concept of explicit antitrust vs. implicit antitrust with respect to agreement means formal agreements or unconscious parallelism. See discussion of Section 1 of Sherman Act, supra. [9] If a party chose to collude consciously and explicitly, then the variables would increase dramatically on the right side of the equation. The result would be that it would be economically inefficient for the firm to engage in collusion, or that it would get caught. Either is an acceptable outcome to society. [10] Plus Factors and Agreement in Antitrust Law, http://capcp.psu.edu/papers/2011/plusfactors.pdf. [11] Library of Congress. http://www.loc.gov/about/facts.html, from http://www.readwriteweb.com/enterprise/2011/11/infographic-data-deluge—8-ze.php. [12] Based on the cost of $875.00 per eight terabyte hard drive on Amazon.com, December 19, 2011. [13] See Supra. [14] See discussion of American Column & Lumber Co. v. United States, supra. [15] 257 U.S. 377 (1921). See supra. [16] Infra. [17] 476 U.S. 447, 458. In that case, a group of dentists disputed insurance practices and signed a pledge to deny insurers copies of x-rays, effectively instituting a boycott. [18] Id at 459. [19] Quoting National Soc. Of Professional Engineers v. U.S., 435 U.S. 679, 692. [20] Federal Trade Commission, The Truncated or “Quick Look,” Rule of Reason. http://www.ftc.gov/opp/jointvent/3Persepap.shtm. [21] Id. [22] Simply expressed as a percentage of how often a price drop by one market participant leads to a price drop by another market participant. [23] The Herfindahl-Hirschman Index. http://www.justice.gov/atr/public/testimony/hhi.htm. [24] Infra. [25] The Provocative Practice of Price Signaling: Collusion Versus Cooperation. Larry L. Miller, Steven P. Schnaars, Valerie L. Vaccaro. Business Horizons, July-August 1993. [26] 15 U.S.C. 45(a)(2).