Abstract
This article addresses one aspect of patent pools that has not received much attention—a patent pool’s role in stabilizing a cartel of downstream producers. This article first reviews the problem of cartel cheating. Any potential mechanisms that a cartel can use to increase its stability face three main challenges: (1) the cost of management, (2) agency costs, and (3) the requirement of secrecy. This article argues that the vertical licensor–licensee relationship inherent to a patent pool would contribute to more effective monitoring of compliance with cartel agreements by licensees. It also argues that the aggregation of patents in a patent pool would better effectuate the punishment of cartel cheating. The article’s main finding that a patent pool is uniquely suited to manage a cartel is a reminder that an overly permissive view of patent pools can invite anticompetitive hazards.
I. Introduction
Over the years, U.S. courts have condemned several patent pools for violating Section 1 of the Sherman Act because they facilitated price-fixing. 1 For example, the U.S. Supreme Court in Standard Sanitary held that a patent pool, which was composed of patents on the enameling process of sanitary ironware, facilitated a cartel of sanitary ironware manufacturers. 2 In analyzing patent pools, courts and antitrust enforcement agencies have focused predominantly on the ex-post effects of patent pools on the output and price in the product markets. 3 However, they have given little attention to the question: What, from an institutional perspective, makes a patent pool a superior entity to abet collusion in the downstream product markets.
The answer to this inquiry starts from an understanding that a cartel is unstable because cartel members have strong incentives to cheat on their agreements. 4 Thus, a cartel needs some mechanisms to strengthen its stability. 5 This article will argue that a patent pool is an ideal candidate to manage a cartel because of two reasons: (1) the vertical licensor–licensee relationship with downstream producers 6 and (2) its potential market power through the aggregation of patents. 7 This article is composed of four sections. Section II focuses on basic economic principles underlying cartels’ instability. Section III examines the functions of a cartel manager. Section IV argues why a patent pool can be an ideal cartel manager. Section V further illuminates arguments in Section IV by analyzing how Hartford Empire, one of the largest patent pools in the U.S. history, managed a cartel of glass container producers in the 1930s.
II. Cartels as Unstable Institutions
A cartel is an association of independent firms established to reduce competition through price-fixing, output reduction, or other restrictive practices. 8 However, economic models have shown that cartel members have strong incentives to cheat on their agreements. This section explains the following: (A) why firms form a cartel and (B) after the cartel formation, why firms cheat on the cartel agreements.
A. Why Firms Form a Cartel
Firms collude with each other to obtain market power so that they can raise prices profitably. 9 Collusion is by no means the only way to achieve market power. History is abundant with examples of a single firm attempting to gain market power through exclusionary conducts. For example, the American Can company in the early 1900s attempted to obtain a monopoly in the can market by denying competitors can-making machineries and raw materials. 10 However, excluding competing firms from the market is a time-consuming and expensive undertaking. 11 A faster approach to obtain market power is for firms in the same market to commit to collectively reducing output or raising prices. 12 If each firm honors its commitment, each will obtain a larger profit than it would under competition. Collusion most likely takes place in a concentrated market with sufficient barriers to entry, where a few firms’ decisions to collude can affect the market as a whole. 13 On the contrary, in a fragmented market, the output reduction by a group of colluding firms will not suppress the overall market output significantly. 14 In addition, the elevated price due to collusion may lure firms previously outside the market into the market without sufficient barrier to entry. 15 Thus, the market price will eventually drop to the marginal cost and all firms will make zero economic profit, which is the minimum level of profit needed for a company to stay in the market.
B. Why Cartel Members Cheat
Evidence has shown that a large portion of cartels is short lived due to cartel cheating. 16 Under the Cournot oligopoly model, if each member in a ten-firm cartel cheats by equating its individual residual marginal revenue with its marginal cost, the market price will be at the ten-firm oligopoly level, which is roughly 10% above the competitive level. 17 Cartel cheating is attributable to at least three reasons. First, there is heterogeneity in the cartel members’ cost structures. 18 Firms in a cartel may not have the same productive efficiency due to using different technologies or operating at different scales, leading to varying profit-maximizing prices across a cartel. 19 Generally speaking, a firm with high marginal cost would prefer a higher cartel price. 20 Thus, any agreed-upon price or output will favor some over others, inducing cartel cheating. 21 Second, firms’ motives of self-enrichment often lead to opportunistic behavior. If a cartel member knows that other members will follow a cartel agreement, that member stands to obtain an immediate windfall through secret cheating. 22 Hypothetically, two widget makers, Firm 1 and Firm 2, form a widget cartel. We assume that the joint profit-maximizing price for the widgets is $2 per unit for the total output of 200 units and the marginal cost is constant at $0.4 per unit. We also assume if Firm 2 produces 100 widgets as agreed and Firm 1 breaches the agreement by making 150 widgets, the increase in the total output will reduce the unit market price to $1.50. Therefore, Firm 1 can make an extra profit of $5 by breaching the agreement than complying with the cartel agreement. 23 The third source of cartel cheating is unexpected changes of market conditions, such as a sudden drop in market demand, which requires constant adjustment of the optimal collusive equilibrium. 24 Because high market volatility compels a cartel to be in a frequent readjusting mode, when readjustments are lagging, cartel members may find it more lucrative to deviate from the cartel agreement. 25 Empirical studies have shown that cartels are more likely to last in stable markets with predictable fluctuations in demand. 26
III. Cartel Manager
A cartel manager acts as an agent of a cartel intended to increase its stability. 27 In face of opportunistic cheating, a cartel manager is generally tasked with monitoring cartel members’ compliance with the cartel agreement and punishing cheating member. This section, which is composed of four subsections, explains a cartel manager’s functions and limitations. Subsection A examines generally how a cartel manager can reduce members’ incentives to cheat. Subsection B analyzes constraints that a cartel manager faces. Subsection C explains why, given the above-mentioned constraints, it is beneficial to choose a vertically related dominant firm as a cartel manager. Subsection D illustrates Subsection C’s arguments by describing how Standard Oil managed a cartel of railroads in the late 1890s.
A. Reducing Cartel Cheating
The incentive frame for a cartel member can be illustrated with a simple inequality: a cartel member will choose to breach the cartel agreement if the benefits of doing so are larger than the cost due to punishment: Profitbreach > Losspunishment. 28 Profitbreach depends on multiple factors, such as the number of firms in the cartel, each firm’s production capacity, the degree of homogeneity of the products, the cheating strategy, etc. Theoretically, a cheating member who faces no capacity restraint in a homogenous market can capture all cartel profits by pricing its products slightly below the cartel price. However, both capacity restraint and product differentiation limit the profits of cheating. When a cheating firm faces constrained capacity, it would not be able to increase its output to the level of cartel output to capture the whole cartel profits. Similarly, when products are differentiated, a cheating member will have difficulty alluring customers away from other members simply by lowering its price. 29
Profitbreach is accumulated during the temporal gap between the commencement of cheating and the detection of cheating. 30 Thus, the sooner cheating is discovered, the smaller Profitbreach is. In order to detect breach of cartel agreements as soon as possible, a cartel manager often requires cartel members to submit relevant data on production or sales. Empirical evidence has shown that the level of specificity of sales data that was collected varied greatly across cartels. 31 In addition, because many cartels rely on members’ self-reporting, there are obvious concerns about underreporting as a firm may intentionally or inadvertently underreport to avoid detection of cheating. 32 Thus, to aid a cartel manager’s oversight burden, cartels may adjust their collusive schemes. For example, cheating detection under the territorial allocation scheme is relatively easier than under the price-fixing scheme. When a cartel divides a market based on geographical locations and assigns each member a “home” territory, unauthorized sales by a member can be inferred from an unexplained sales drop in other members’ home markets.
On the other side of the inequality, Losspunishment depends on the punishment strategies that cartels implement. For example, competitive pricing—selling at the price equal to cost—is a common strategy against price chiseling, under which Losspunishment for a cheating member equals an allocated portion of cartel profits. 33 In choosing punishment strategies, cartels need to ensure not only that Losspunishment is large enough to induce compliance with cartel agreements but also that nonbreaching firms will actually implement the punishment under a chosen strategy. Implementation of punishment potentially suffers a freeriding problem as some nonbreaching members may rely on others to execute the punishment. For instance, a cartel may agree to use predatory pricing—selling at the price lower than cost—to penalize cartel cheating. Under the predatory pricing scheme, each nonbreaching member is required to increase its output and sell its products at the price below cost. Because predatory pricing lowers penalizing firms’ profits, some firms may choose to secretly reduce their output to minimize loss during the period of discipline. 34 Therefore, in order to prevent freeriding, a cartel manager must deprive individual members of their discretions over the execution of punishment. 35
B. Constraints on a Cartel Manager
A cartel manager faces constraints in performing its functions to reduce cartel instability. These constraints play a significant role in determining what entity is chosen as a cartel manager. There are three main constraints: the cost of managing a cartel, the agency costs, and the secrecy requirement. First, a cartel manager must be capable of managing a cartel, including monitoring compliance and punishing cartel cheating, in a cost-effective way because the costs of managing a cartel, which is normally shared by cartel members, diminish the overall benefits of collusion. 36 Second, utilizing a cartel manager may create agency costs. 37 Agency costs arise when an agent, who is expected to act in the best interests of its principal, instead acts to advance its own interests. 38 The orthodox principal-agent literature prescribes that agency costs can be mitigated by alignment of conflicting interests. In the context of corporate management, the alignment of interests among managers and shareholders is achieved by tying the managers’ compensation to shareholders’ financial gain. 39 With regard to the cartel management, a cartel manager’s self-interests must be aligned with a cartel’s goal of maximizing joint profits. Third, because a cartel is an illegal association, a cartel manager’s task must be performed in secrecy and be kept off the radar of antitrust enforcement agencies. 40 The secrecy requirement compels a cartel manager to resort to covert practices, which likely makes cartel management even more costly.
C. A Vertical Firm as a Cartel Manager
Given the constraints facing a cartel manager, firms, who are vertically related to cartels in the upstream or downstream markets, have been used as cartel managers in an increasing number of cases. 41 In economics, two entities are in a vertical relationship when a product or service provided by one is an essential input to the other’s service or product. 42 For example, a cartel manager can be an input supplier to a cartel of downstream manufacturers. A vertically related firm can be an effective cartel manager due to two attributes: (1) its vertical relationship with a cartel and (2) its control of assets for punishment.
1. Vertical Relationship
The vertical relationship reduces a cartel manager’s exposure to antitrust scrutiny. Direct communications between competitors in the same market would naturally invoke suspicion of collusion. 43 On the contrary, communications between vertically related firms, such as a manufacturer and an input supplier, are commonplace and less likely to invite antitrust scrutiny. 44 In addition, although vertical agreements are “contracts, combinations…conspiracies” within the purview of Section 1 of the Sherman Act, vertical agreements are generally analyzed under the rule of reason standard, which is more lenient than the per se standard commonly used on horizontal agreements among competitors. 45
The vertical relationship can also mitigate agency costs. A downstream cartel can elicit an input supplier/cartel manager’s faithfulness in policing the cartel by promising to protect the input supplier’s market position in quid pro quo. Specifically, a producer cartel can enter into exclusive dealing contracts with the supplier/cartel manager, which can raise the barrier to entry into the input market by foreclosing a substantial portion of customers from new input suppliers. 46
2. Control of Assets for Punishment
As discussed in Section III.A, there is a concern about freeriding in implementing punishment for cartel cheating. 47 When freeriding is widespread, punishment becomes ineffective. Because a vertically related firm, e.g., an input supplier, owns assets essential to cartel members’ production, such control of essential inputs places the input supplier at an advantageous position to punish cartel cheating. For example, an input supplier can unilaterally raise the price of inputs for cheating members. However, for such strategy to be effective, the input supplier needs to have market power in the upstream input market so that a cheating member cannot easily switch to alternative inputs when an input supplier raises input price as a punishment.
D. Standard Oil
One example of a vertically related firm acting as a cartel manager is Standard Oil Company in the early 20th century. Many economic historians concluded that Standard Oil managed a cartel of railroads for petroleum transportation from the late 1870s to the early 1910s. 48 The railroad petroleum transportation market during that time was characterized by high fixed costs of building rail tracks and low average variable cost. The high fixed costs created a formidable barrier to entry and the market was dominated by three major railroads: New York Central, Erie, and Pennsylvania. 49 Although the three railroads attempted multiple times to establish cartels in order to take advantage of the high market concentration which was conducive to collusion, cartels broke down quickly after their formation because the relatively low average variable cost facing each railroad made it very lucrative for them to cheat on cartel agreements by secretly providing shipments at lower prices. 50 Detection of cartel cheating was particularly difficult because railroads and their customers, oil refiners, dealt with each other customarily through individual negotiation. 51 Consequently, the petroleum transportation market experienced cycles of cartelization and intense competition due to cartel cheating. 52
Facing rampant cartel cheating, the three major railroads solicited Standard Oil, a refiner in Ohio, to be a cartel manager. 53 The refinery market, which is upstream to the petroleum transportation market, was fragmentary and occupied with many small-scale refiners. 54 Standard Oil as the largest refiner in Ohio accounted for only about 4% of the refinery market nationwide. 55 The railroad cartel knew that Standard Oil’s low market share would hinder its ability to punish cartel cheating because a cheating member could easily switch to other refiners. 56 Therefore, the three railroads started a scheme to increase Standard Oil’s market power by charging independent refiners, Standard Oil’s competitors, a very high rail rate while giving Standard Oil significant rebates. Because the railway cost accounted for a large fraction of refiners’ total cost—more than 40%, 57 independent refiners faced a significant price squeeze between a flat petroleum price and a high rail rate. 58 With the help of the railroad cartel, Standard Oil acquired many small and unprofitable refiners and expanded its market share precipitously to more than 90% within a decade. 59 Standard Oil then used its large market share in the refinery market to aggressively police the railroad cartel. For example, in 1876, the Pennsylvania attempted to vertically integrate into the refinery market by acquiring a mid-sized refinery. 60 The vertical integration would allow the Pennsylvania to charge a lower rail rate to the self-refined petroleum and circumvent the price-fixing agreement through internal transactions. 61 As a response, Standard Oil withdrew all the shipments from Pennsylvania and shut down its Pittsburgh refineries. Thus, oil transportation on the Pennsylvania dropped dramatically, causing the Pennsylvania to suffer significant financial losses in the short term. 62
IV. Patent Pools as Cartel Managers
Although courts and regulators in many cases have held that patent pools are procompetitive because they can reduce the transaction costs of patent licensing, an overly permissive treatment of patent pools would potentially invite anticompetitive hazards. 63 This section addresses a specific type of anticompetitive hazards: a patent pool can facilitate collusion in a downstream product market as a cartel manager. 64 Relying on the analysis from Section III, this section argues that a patent pool is at an advantageous position to manage a downstream cartel mainly due to two reasons: (1) the vertical licensor–licensee relationship between a patent pool and downstream producers, which is discussed in Subsection A; and (2) a patent pool’s control of patents for punishing cartel cheating, which is discussed in Subsection B.
A. Patent Pools as Input Suppliers
A patent pool, which can be a partnership or a limited liability corporation, is a common holding entity, into which patent owners transfer their patent rights. 65 In some cases, patent owners assigned ownership or granted exclusive licenses to a patent pool; 66 in others, patent owners only granted to a patent pool nonexclusive licenses. 67 In exchange, a patent pool will grant licenses to the portfolio of pooled patents to contributing patent owners or third parties. 68 The following diagram in Figure 1 illustrates the transactions between a patent pool and patent owners.
Diagram of transactions between a patent pool and patent holders.
As shown in the diagram, a patent pool is in a vertical licensor–licensee relationship with patent owners, who are often downstream product manufacturers. As explained in Section III.C, such a vertical relationship puts a patent pool in an advantageous position to police a cartel of downstream producers. A patent pool can impose vertical restraints on each cartel member through patent licenses in order to reduce horizontal competition among cartel members.
69
Patent laws, as interpreted by the courts, have afforded licensors a broad authority to restrain licensees’ activities when such activities are related to licensed patents, including limiting licensees’ outputs, allocating licensees’ territories,
70
or even fixing licensees’ sales prices.
71
In addition, restrictive patent licenses, though not immune from antitrust scrutiny, can be often defended as means to improve efficiency or exploit patent values rather than to impede competition, thus are less likely to raise suspicion of an antitrust violation.
72
First, price and output restrictions can incentivize licensees to choose a combination of product attributes most desired by consumers and enable a licensor to reallocate output between licensees to reduce overall production costs.
73
Second, courts have recognized that licensing restrictions are important for a licensor to explore patent values. For example, in the General Electric light bulb case in the 1920s, the Justice Department challenged General Electric’s grant of a patent license on incandescent filaments to Westinghouse because the patent license fixed the sales price of light bulbs made by the licensee Westinghouse under the license.
74
General Electric, who controlled 69% of the light bulb market at that time, argued that because its patent royalty was set as a percentage of Westinghouse’s revenue, the ability to regulate the price of licensed light bulb that Westinghouse could charge was essential to maximize General Electric’s return on its innovation.
75
The U.S. Supreme Court approved the price-fixing provision with the following reasoning: One of the valuable elements of the exclusive rights of a patentee is to acquire profit by the price at which the article is sold. The higher the price, the greater the profit, unless it is prohibitory. When the patentee licenses another to make and vend, and retains the right to continue to make and vend on his own account, the price at which his licensee will sell will necessarily affect the price at which he can sell his own patented goods. It would seem entirely reasonable that he should say to the licensee, “Yes, you may make and sell articles under my patent, but not so as to destroy the profit that I wish to obtain by making them and selling them myself.”
76
Furthermore, the vertical licensor–licensee relationship also gives a patent pool an advantage in detecting cartel cheating. A licensor can legally assert in a patent license broad rights to inspect a licensee’s production and distribution of goods made under the license, which can readily be used to detect cartel cheating. 78 Patent laws are largely silent on what kind of information about licensees’ productive activities a licensor can review. In addition, legitimate business reasons may justify a licensor’ broad inspection right. First, inspection allows a licensor to monitor licensees’ productive efficiency, which is directly related to a licensor’s profits. 79 Second, in many cases, licensees themselves would prefer a rigorous inspection regime by a licensor. For example, a licensee may have agreed to pay a higher royalty based on the promise that it would have the exclusive right to sell the patented products in a designated territory. As a result, that licensee would naturally desire the licensor to inspect other licensees’ sales to ensure its exclusive rights. 80
B. Punishing Cartel Cheating by Patent Pools
As discussed in Section III.C.2, an input supplier can use its control of inputs necessary to a downstream cartel’s production to punish cartel cheating if that input supplier has market power in the input market. 81 By the same token, if a patent pool has market power in the upstream technology market, it can use such power to prevent cartel cheating. Patent pooling agreements often entail the aggregation of patent rights in a patent pool. Whether the patent aggregation gives a patent pool market power in the upstream technology market depends both on the mechanism of transfer of patent rights and on the existence of comparable alternatives to the patented technology. 82 An assignment of patent ownership or a grant of exclusive rights to a patent pool is more likely to give the patent pool control over patented technology than a mere grant of nonexclusive rights. If the license that a patent pool receives is nonexclusive, then a downstream manufacturer who is denied a pool license can still try to obtain licenses to essential patents from individual licensor members. 83 Furthermore, if there are no alternatives to the patented technology, a license from a patent pool becomes a necessary input to downstream production, which gives the patent pool coercive power to police downstream producers. This subsection discusses three main tools at a patent pool’s disposal for using its market power in the upstream technology market to punish cartel cheating: (1) refusing to grant a patent license, (2) increasing royalties to raise a cheating member’s cost, and (3) holding royalty hostage.
1. Refusing to Grant a Patent License
A patent pool can refuse to grant or renew a license to a cheating member as a punishment. As discussed previously, this strategy’s success depends on whether a cheating member can obtain a patent license to a substituting technology or use an unpatented technology that provides a comparable result. Legally speaking, a refusal to grant or renew a patent license has a strong legal footing. The U.S. patent law unambiguously states that a refusal to grant a patent license cannot be a patent misuse. 84 In addition, U.S. antitrust law differentiates between a unilateral refusal to license and a concerted refusal to license. Although courts rarely deem a truly unilateral and independent refusal to license as an antitrust violation, a concerted refusal to license is more likely to face antitrust scrutiny because it more closely resembles a horizontal collusion. 85 Therefore, in executing the strategy of using a refusal to grant a patent license to police downstream cartel members, a patent pool may attempt to create an impression that its refusal is truly independent and lacks an element of horizontal coordination. 86 For example, in Bement, six manufacturers of float spring-tooth harrows assigned 85 patents on the mechanical structures of spring-tooth harrows to a patent pool, the National Harrow Company. 87 In return, each contributing manufacturer received a pool license to all the patents as well as an ownership stake in the patent pool. 88 When the patent pool refused to grant a pool license to cheating members and sued for patent infringement, the U.S. Supreme Court treated the refusal as a unilateral action between the patent pool and a licensee. 89 The court overlooked the fact that because the patent pool’s licensees were also its shareholders, the pool’s licensing decisions were subject to the other harrow manufacturers/licensees’ approval. 90 Therefore, it is fair to conclude that the patent pool in Bement succeeded in concealing a clear element of horizontal coordination regarding its refusal to grant a patent license.
2. Increasing Royalties to Raise a Cheating Member’s Cost
A less-severe punishment than a refusal to grant or renew a patent license is to charge a cheating member an elevated royalty in order to raise that member’s production cost. Raising a rival’s cost has long been analyzed as an exclusionary practice in an attempt to achieve a monopoly. 91 However, it can also be used to punish cartel cheating. 92 For example, also in Bement, patent licenses granted by the National Harrow Company required harrow manufacturers to pay a royalty of $1 for each float spring-tooth harrow sold pursuant to the license. 93 For each unit sold contrary to the restrictions of the license, a licensee was required to pay a royalty of $5. 94 Because a cheating member has to pay an inflated royalty for the units sold in violation of the cartel agreement, cartel cheating becomes unattractive and costly.
3. Holding Royalty Hostage
The hostage model to support a hard-to-enforce agreement was first developed by Nobel-winning economist, Oliver E. Williamson. 95 He pointed out that because enforcing an agreement can be prohibitively expensive or legally impossible due to the illegal nature of the agreement, parties to an agreement have incentives to find alternative enforcement mechanisms to prevent opportunistic breaching. 96 He argued that if parties to an agreement can exchange assets whose value can be unilaterally sabotaged, those assets can be used as a hostage against either party to disincentivize breaching. 97 In addition, Williamson emphasized that only assets that do not have a short-term alternative usage—in other words, “transaction specific”—can serve as a hostage to support a hard-to-enforce agreement. 98
Because a cartel agreement is not enforceable by courts, the hostage theory can be readily adapted to strengthen cartel commitments. 99 To apply the hostage model to cartel cheating, the first step is to identify, for every cartel member, transaction-specific assets that can be used as a hostage. 100 It turns out that a royalty rebate is an ideal candidate. A patent pool can first charge licensees a high royalty and then reimburse them a large fraction of the royalty if they fulfill their commitments. If a member is found to have cheated, a patent pool can reduce or delay the royalty rebate. For example, the patent pool in Standard Sanitary used royalty rebates as a hostage to deter cartel cheating. The patent licenses at issue required all the licensees to pay a royalty of $5 per unit. 101 Each licensee was entitled to an 80% royalty rebate if it followed the restrictions in the patent licenses. 102 Otherwise, the 80% royalty rebate would be forfeited. 103 The U.S. Supreme Court concluded that the threat of the forfeiture of the 80% royalty rebate increased cartel members’ fidelity to the cartel agreement. 104
V. The Hartford Empire Patent Pool
The U.S. Supreme Court in 1945 dissolved Hartford Empire, one of the largest patent pools in the U.S. history, after the Court concluded that it facilitated a collusion among glass container manufacturers. 105 The Hartford Empire case provides an insightful example of how a patent pool can cartelize a downstream product market. The meteoric rise of Hartford Empire, growing from a small research company to a gigantic patent pool controlling almost all the key patents in the glass making industry, is a close parallel to the Standard Oil’s rapid market share expansion in the 1870s. Like the Standard Oil, Hartford Empire’s dominance in glass making technology was unmistakably aided by glass container manufacturers who faced market uncertainties amid rapid technological change. 106 In return, Hartford Empire used its dominance to police the glass container cartel. This section is composed of three subsections: Subsection A describes the glass container market prior to Hartford Empire’s rise, Subsection B examines the history of Hartford Empire’s rise to market dominance, and Subsection C analyzes how Hartford Empire used its dominance to police the downstream glass container cartel.
A. Competition in the Glass Container Market Before Hartford Empire
It is useful to review the history of the U.S. glass container market in order to fully understand why the manufacturers decided to engage Hartford Empire to be a cartel manager. Until 1903, the glass container market in the U.S. was fragmented with many small firms, each making one or two specific types of glass containers. 107 The market fragmentation was likely attributable to the crude hand-feeding method of manufacturing, which, unlike mechanized methods, was an extremely inefficient process and required high specialization in the workers’ skills. 108 The appearance of the suction machine in 1903, the first fully automatic glass making machinery invented by Owens Bottle Company, revolutionized the industry. 109 The suction machine would draw molten glass by a vacuum into a mold and then the molten glass would be shaped by the mold into a desirable configuration. 110 Installing the suction machine required a high upfront cost; however, the automatic process reduced significantly the average variable production cost compared to the old hand-feeding method. 111 The combination of the high fixed cost and the low average variable cost incentivized manufacturers, who could afford the suction machine, into expanding their production, which led to a high concertation in the glass container market. After the introduction of the suction machine, five glass container manufacturers controlled more than 90% of the market. 112
The invention of the gob-feeding process in 1914 threatened to destabilize the glass container market again. The gob-feeding process delivers the molten glass on a feeder to the molds in viscous pasty gobs. 113 The gob-feeding machine was much cheaper than the suction machine. 114 It also enabled manufacturers to produce better quality glass containers with even lower average cost. The low upfront cost made the gob-feeding machine accessible to smaller manufacturers and reduced the barrier to entry into the glass container market. 115 Therefore, the invention of the gob-feeding machine threatened to increase competition in the market.
B. The Rise of Hartford Empire
Hartford Empire was formed in the early 1910s through the merger of Hartford Fairmont, a small research company with a capital of $120,000, and the Empire Machine Company, a small engineering firm. 116 After the merger, Hartford Empire owned a few patents on certain components of the gob-feeding system and began to offer patent licenses to glass container manufacturers. Legal conflicts quickly developed, which could be generally categorized into two types: patent interference proceedings to determine whether Hartford Empire was the first to invent the gob feeder and patent infringement disputes with several manufacturers. 117
The patent conflicts led to a network of licensing agreements between all the major manufacturers and Hartford Empire. 118 First, in 1922, Hartford Empire entered into a cross-licensing agreement with Corning Glass Works, a large producer of specialty ware such as electric bulbs and heat-resistant containers. Under the agreement, Hartford Empire was given an exclusive right to Corning’s patents for glass container production. In return, Hartford Empire promised not to license its patents to manufacturers who made specialty ware in competition with Corning. 119 The agreement essentially gave Hartford Empire control of Corning’s patents in exchange for Hartford Empire’s protection of Corning’s dominance in the specialty ware market. Records show that after the cross-license, Corning enjoyed a near monopoly in the specialty ware market for decades. 120 In 1924, Hartford Empire reached a cross-licensing agreement with Owen, the largest glass container maker of the time, under which Hartford Empire received an exclusive right to Owen’s patents subject to Owens’s veto on future grants of sublicenses. 121 In exchange, Owen received a nonexclusive license to use all of Hartford Empire’s patents for glass container production with an obligation to pay a stipulated royalty. Owen was also entitled to one-third of Hartford Empire’s overall licensing revenue. 122 In 1932, Hartford Empire struck an almost identical cross-licensing agreement with another large manufacturer Hazel-Atlas. Like Owens, Hazel-Atlas gave Hartford Empire an exclusive right to its large stock of patents and any future patentable inventions, and received a nonexclusive license to use Hartford Empire’s patents. 123
Because Corning, Owen, and Hazel-Atlas all granted Hartford Empire exclusive rights to their patents, by 1935, the Hartford Empire patent pool effectively controlled about 600 patents covering the furnace, the feeding machine, an annealing oven, and other essential components of the gob-feeding system. 124 As the gob-feeding process kept improving, the hand method and the suction process were no longer commercially feasible alternatives. 125 Thus, a patent license from the Hartford Empire became indispensable to the whole glass container industry. By 1938, Hartford Empire’s licensees made about 94% of total glass container production in the United States. In specific lines of containers such as milk jars and fruit jars, Hartford Empire’s licensees made up close to 100% of the total production. 126
C. Hartford Empire as a Cartel Manager
An internal memorandum showed that Hartford followed four principles in licensing: (1) licensing only to selected manufacturers, (2) refusing to license to price-cutters, (3) restricting licensees in certain fields of manufacture, and (4) limiting the length of licenses in order to retain more control over licensees. 127 In the 1930s, Hartford Empire granted a series of restrictive patent licenses to various manufacturers, which had the effects of dividing the glass container market based on product types. 128 For example, Thatcher Manufacturing Company obtained an exclusive right to manufacture milk bottles and Ball Brothers received an exclusive right to manufacture fruit jars. 129 In addition, records show that Hartford Empire served as a communication conduit to covertly facilitate output reduction among its licensees. 130 For instance, when Ball Brother was contemplating on seeking a license to make fruit jars, it was concerned that Owen and Hazel-Atlas would make too many fruit jars because their licenses were unrestricted. After Ball Brother conveyed its concern to Hartford Empire, Hartford Empire successfully persuaded Owen and Hazel-Atlas into capping their fruit jar production. 131 Moreover, Hartford Empire had frequently refused to grant a license to small manufacturers whom it thought would compete with existing licensees. For example, Hartford Empire refused to grant a license to a small firm in Michigan because Owen also had a factory in Michigan. It feared that if a license were to be granted, Owen would face competition in the Michigan market. 132 In general, Hartford Empire would grant a license only when it determined that the output level was significantly below the market demand. 133
Last but not the least, Hartford Empire rigorously used its monopoly in the gob-feeding technology to police the glass container cartel. In several instances, as soon as licensees complained to Hartford Empire about others’ cheating, Hartford Empire aggressively acted on these complaints. For example, when Ball Brothers, who had an exclusive right to make fruit jars, complained to Hartford Empire that Anchor Hocking Glass, who was given a license to make only packerwares, was knowingly selling packerwares to consumers for preserving fruit, Hartford Empire punished Anchor Hocking Glass by forcing it to surrender the residue right of its patent license for $100,000. 134 In another occasion, Hartford Empire refused to renew a license to Three River Glass after other licensees reported that it had been selling at a price below a “general price level.” 135
In summary, the invention of the gob-feeding machine threatened to lower the barrier to entry to the glass container market and increase market competition. It was in the interests of large existing firms to avoid what they feared as a potentially “ruinous” competition. Hartford Empire was then entrusted with control of a large patent portfolio through a series of cross-licensing agreements. In return, Hartford Empire used its monopoly in the gob-feeding technology to police the glass container cartel. Particularly, it punished cartel cheating by canceling or refusing to renew licenses to cheating members.
VI. Conclusion
I have argued in this article that a patent pool is an ideal candidate to be a cartel manager. A patent pool’s vertical licensor–licensee relationship with downstream producers reduces the cost of cartel management. In addition, the vertical relationship better shields the cartel operation from antitrust scrutiny. Furthermore, if the aggregation of patents gives a patent pool market power in the upstream technology market, it can use its control of patents to punish cartel cheating. This article does not contend that all patent pools are cartel managers; however, it does suggest that anticompetitive threats by patent pools should not be overlooked and a per se rule of legality to patent pools is unjustified. This article also recommends that when analyzing patent pools, antitrust regulators should not focus solely on the restrictive terms of patent licenses. Instead, regulators should also pay attention to how a patent pool monitors its licensees’ performance and deters its licensees from violating licensing terms because such information may be indicative of whether a patent pool is a cartel manager.
Footnotes
Acknowledgments
Special thanks to Herbert Hovenkamp, Jason Rantanen, and John Reitz for their help with this paper.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.