Collusive Bidding, Competition Law, and Welfare

1 Introduction

We analyze why competition authorities often do not challenge collusive bidding for patents and why courts employ the “rule of reason” to assess the effects on competition from patents sold or transferred in a bundle to a group formed by competing firms.^[1] Such actions can also create anticompetitive concerns and implications for competition policy.

We observe that competing firms often come together and form collective entities to buy patents instead of competing with each other in a bidding process (Meurer et al. 2012). One example of such an entity is the Rockstar consortium (e.g., see Chien 2010).^[2] In the bidding process for the Nortel Network’s patent portfolio, following its bankruptcy in 2011, large technology companies including Apple and Microsoft formed a cartel that outbid other competing bidders.^[3] After a period of patent litigation and licensing, the Rockstar consortium sold approximately 4000 of the approximately 6000 acquired patents to the Rational Patent Exchange (RPX) in 2014.^[4] Adopting a catch and release patent strategy—the acquisition of patents with the aim of licensing them first to the cartel members and then selling the rest—the Rockstar Consortium distributed the remaining 2000 most valuable patents to the consortium members.

Another cartel of buyers performing a similar catch and release patent strategy is Allied Security Trust (AST).^[5] The trust was created to defend producing firms against the patent holdup problem caused by non-practicing entities (NPEs).^[6] Wang (2010) describes AST as a defensive patent aggregator. AST searches the market for patents that could be of interest to its members. Each member of AST, which has paid an initial subscription fee, is alerted by AST about patents of potential interest. A member interested in the acquisition of specific patents sets aside an amount in an escrow account. After the acquisition is accomplished, the patents are licensed to all members who had previously declared their interest in the patents. The members of the trust that have not declared their interest in advance but want the patents after the acquisition has been made can also obtain them by paying a license fee based on the bidding cost. AST differs from other defensive and offensive patent aggregators in that it is only engaged in licensing and selling patents but not in enforcing them (Cosandier et al. 2014).

We examine the welfare implications of such a catch and release patent strategy performed by NPEs that have been set up by incumbent manufacturing firms.^[7] In the way the above transactions in the market for ideas happen (Gans and Stern 2010), we identify two market distortions. The first distortion is the government sanctioned monopoly granted to patent holders allowing monopoly rents to be extracted in exchange for revealing the new technology early instead of keeping it a secret and thereby providing other firms with incentives to innovate. This tradeoff becomes nontrivial considering the complexity of the patent system (e.g., Allison and Lemley 2002; Hall and Harhoff 2012; Harhoff et al. 2007; Nordhaus 1969; Romer 1990; Scotchmer 2004), which originates from the fact that weak intellectual property (IP) protection may lead to underprovision of new ideas, while strong IP protection may create a monopoly distortion. The second distortion is that such cartels are created before entering a patent bidding process. There is a “per se” legal argument that such collusive behavior will distort competition because it enables a group of buyers to collectively dictate the pricing of suppliers (e.g., see Tudor 2012; Sidak 2009; Maurer and Scotchmer 2006).^[8] The literature in line with this argument extends from economics (e.g., Bajari and Ye 2003; Carstensen 2017), including bidder collusion in auctions (e.g., Cramton and Schwartz 2000; Wang and Chen 2016) and in posted-price markets (e.g., Marshall and Meurer 2004; Marshall et al. 2014), to law and antitrust jurisprudence on buyer alliances (e.g., Ezrachi 2012; Jones 2012).

However, competition authorities both in the United States (U.S.) and within the European Union (EU) allow, and sometimes even encourage, the exchange of information among potential patent buyers and users before the bidding process. The Department of Justice’s Antitrust Division and the Federal Trade Commission in the U.S. have jointly stated that when buying alliances of patented inventions are part of an industry standard, they should be under the scrutiny of the “rule of reason” rather than challenged as “per se” illegal (see also Sidak 2009). The “rule of reason” doctrine was applied by the Department of Justice’s Antitrust Division to close the investigation on three high-profile acquisitions, including the previously mentioned Rockstar consortium.^[9] The argument for using the “rule of reason” model of analysis was to protect downstream users from the threat of holdup or injunctions undertaken by patent holders. It can be argued, however, that balancing the procompetitive and anticompetitive effects employed by the “rule of reason”, while solving the holdup problem by allowing collusive behavior among buyers (see Sidak 2009), does not necessarily cause welfare gains. The standard view of the competition authorities in the U.S. and EU that mergers in oligopolistic markets lead to increased consumer surplus on the basis of the “Cournot effect” is not always evident (Masson et al. 2014).^[10] This cause of action lacks a theoretical foundation, and neither the competition authorities nor the report and recommendations of the Antitrust Modernization Commission have produced any theoretical foundation for why the “rule of reason” argument should be applied.

We analyze collusive bidding for patents. We go one step beyond the DOJ’s observations and suggest that collusive bidding on patents held by NPEs, even when the patents are not essential for a standard (SEPs), may actually be beneficial in terms of social welfare. We consider ideas that not only create additional value but also destroy the value of existing assets held by producers that do not implement the new idea. The Rockstar consortium and AST were formed by incumbents, i.e., firms with assets in place used for producing in an existing market. New technologies, once commercialized, are important ingredients in economic and social growth, and when they add to the value of existing assets, it is beneficial for incumbents to implement them. However, if new technologies create value for new assets at the expense of existing assets, often referred to as drastic innovations or technology shifts (Arrow 1962), incumbents will be less inclined to introduce them. The new technology brings what Schumpeter (1975) called “creative destruction”. There are two ways a new technology can destroy the value of existing assets. First, it can make the existing economic activity of the patent holder redundant (or inefficient compared to what can be done with the new technology). Second, it can destroy the asset value of competing firms that were not able to obtain the new technology. This often leads to aggressive bidding for the invention by the incumbent producers.

In the presence of destructive costs, the bidding of incumbents against each other not only increases the patent price to the incremental value gained by the winning firm but also actually results in an even higher price to prevent a competitor from acquiring the new technology. The competitive winning bid is such that all firms end up with a value less than what they were making when the idea was not present. In other words, the value of the losing firms decreases because they are not the firms to obtain and implement the new superior technology. The new technology destroys the existing asset value of these incumbent firms. The value of the winning firm decreases because of the high price paid to win the bidding process. Collusive bidding may avoid this problem for all incumbent firms. Specifically, if firms have a choice to coordinate behavior (cooperate) before entering the bidding process, they may increase their profits by implementing the new technology.

Let us assume that the firms can anticipate the competitive outcome of the patent bidding and thus will stop the process. That is, the potential buyers will be simply waiting for another buyer to start bidding. In this case, it is optimal that no firm starts the bidding process. Therefore, the patent will not be sold, and no firm will implement the new superior technology. If, however, a possible cartel of buyers formed by the producers can negotiate with the patent holder, then the patent will be sold, and both consumer surplus and producer surplus will increase.^[11] We show how to avoid the expected result of the standard arms’ length pricing models that leave patents unsold and, hence, unused, even when they can improve social welfare.

We suggest that competing firms may prefer to collectively negotiate with the patent holder. This may raise anticompetitive concerns since firms are not bidding against each other for the patent, a necessary input for the superior technology to be implemented, but colluding with each other in the bidding process. We argue that if we do not allow the producers to collude, in practice, the patented technology may not be used at all, and hence, the increase in consumer surplus—a main objective of any competition policy—will not be realized.

Therefore, we develop a theoretical foundation for defensive patent aggregation or collusive bidding implementing a catch and release strategy.^[12] We show that a cartel of incumbents employing a catch and release patent strategy will always outbid a potential entrant, or a patent assertion entity over a technology that can destroy the value of the incumbents’ existing assets. Cosandier et al. (2014) find that auctioned patents are acquired by both defensive NPEs and offensive NPEs, or patent assertion entities with a strategy to enforce patents through litigation, depending on the use of the catch and release strategy. More important, we find that defensive patent aggregation increases welfare in terms of total surplus, i.e., the sum of consumer surplus and producer surplus, and the fee extracted by the patent holder is higher under collusive bidding than in the case where collusion is not allowed.

In Section 2, we present the problem. In Section 3, we develop a model of collusive bidding. We show that when a cartel of buyers formed by producers negotiates with the patent holder of a new superior technology, the patent will be sold, and both consumer surplus and producer surplus will increase. We also consider the case of a drastic technological innovation and the threat of potential entry and find that collusive bidding for patents held by NPEs is welfare improving. Section 4 concludes the paper. All proofs are given in the Appendix.

2 Problem Statement

Consider a group of identical oligopolistic firms, i.e., n firms for n ≥ 2, initially using an unpatented technology to produce a homogeneous product in the same industry. The firms are indexed by i and earn profit π _i,0 = π ₀ for i = 1, 2, …, n. This is the profit of a firm in the initial stage of the game or the status quo of the market. Then, a new superior patented technology appears. The new technology is superior in that it reduces the cost of production and thus increases the profit of the firm that uses it. The patented technology is owned by a non-practicing entity (NPE), i.e., a patent holder that is not any of the n incumbent firms and does not practice or use the patent to produce. If the new technology is acquired and used by firm i only, the (gross) profit of firm i becomes π i i , 1 , i.e., the profit of firm i for i = 1, 2, …, n, when only one firm has implemented the new technology, firm i. The profit of the other firms (in the presence of firm i with the new patented technology) is then π j i , 1 for i ≠ j, i, j = 1, 2, …, n.^[13] In addition, let the competing firms that do not use the new technology experience a drop in their status quo profits.

Assumption 1 states that any firm in the industry using the new patented technology will increase its (gross) profit at the expense of the other firms. The gross profit for a winning firm i is π i i , 1 , while the net (nonnegative) profit is Π i i , 1 = π i i , 1 − F , where F is the price paid by firm i to the patent holder for the new technology. An obvious way to determine who receives the patented technology is an (open) auction. The competing firms, therefore, will bid for the patent to prevent a rival firm from obtaining it.

Consider a price F > 0 at which firm i is willing to buy the patent. If firm i succeeds, it will earn a net profit Π i i , 1 = π i i , 1 − F , and the other firms will earn a net profit Π j i , 1 = π j i , 1 for i ≠ j, i, j = 1, 2, …, n. The patent price F is endogenous and needs to be calculated. Suppose that F < π i i , 1 − π i j , 1 . Then, there exists a premium φ > 0 such that F + φ ≤ π i i , 1 − π i j , 1 . This suggests that it is favorable for firm j to bid F + φ to prevent firm i from obtaining the patented technology at price F.^[14] This further implies that the equilibrium winning bid, F*, must satisfy F * = π i i , 1 − π i j , 1 . It follows that the winning firm i earns a net profit Π i i , 1 = π i i , 1 − π i i , 1 − π i j , 1 = π i j , 1 , and each of the firms that could not obtain the patent earns Π j i , 1 = π j i , 1 . In other words, a firm bidding for the new patented technology is willing to pay more than the standalone value of the patent. It will also pay a premium to avoid the loss it will incur if another firm in the market buys the patent. Since π j i , 1 = π i j , 1 , it follows from Assumption 1 that all firms earn less than what they used to earn when the patented technology was not implemented, i.e., at the status quo of the market. As a result, the incumbent firms lose if a competitive auction for the patent starts. The Nash equilibrium (Nash 1950) of this game is for the firms to compete in the bidding process, which will result in lower profits. Moreover, the bidding process is such that the NPE extracts the entire surplus and a premium from the use of the new patented technology. This may be a reason for the rising animosity between NPEs (that have often been defined as patent trolls (e.g., see Fales 2014; Ferrill 2005)) and patent practicing entities.

Clearly, the firms can avoid the loss incurred from the implementation of the new technology by preventing the auction. If the firms have the choice to coordinate behavior, then no firm will initiate the bidding process, the auction will not start, and there will be no winner. In this case, the patent will not be used in production, and the incumbent firms will continue to earn the initial status quo profits. This outcome is different from the Nash equilibrium of the noncooperative game, but it may in fact occur if the firms are allowed to cooperate and any firm can anticipate the other firms’ strategic choices. Note, however, that even this outcome is inefficient since the new and available, but patented, technology will not be implemented.

In an open auction, every firm can make a higher bid until no firm is willing to offer more than the most recent bid. We have already argued that this will result in lower profits for any firm. Consider also the possibility that the patent holder may make a secret deal with an incumbent firm to sell the patent for a price less than F*. The patent holder cannot, however, make a credible commitment to keep the offer price a secret if it is less than F*. This is because once the secret offer is made, it is in the interest of the patent holder to inform the other incumbents about the deal. The participation of the other firms in the bidding process will secure the patent holder a price F*. Therefore, private deals are not credible unless the price is F*, while an open auction will not start. In the next section, we develop a model that shows how such an impasse can be resolved in a way that actually improves welfare.

3 The Model

There are n identical firms in the product market in the initial stage of the game, as described in Section 2. Let Q ₀ and P ₀ be the initial total quantity of the industry and market price (i.e., the status quo industry quantity and market price), respectively. Additionally, let q _i,0 be the initial quantity supplied by firm i for i = 1, 2, …, n, i.e., the quantity output produced by firm i when the patented technology is not implemented. For purposes of exposition, we assume that these n firms are Cournot oligopolists (Cournot 1838), facing an inverse market demand curve given by^[15]

where a and b are positive parameters that characterize the demand for the product, i.e., a, b > 0. Let each firm have a constant unit cost of production given by c _i ≡ c for i = 1, 2, …, n. The new superior technology reduces the (unit) cost of production to c′, i.e., 0 ≤ c′ < c. We will first consider a nondrastic technological innovation.^[16]

Assumption 2, where Δc denotes the innovation size, ensures that each incumbent firm makes a positive profit and, hence, operates in the market even when only one firm has the new technology. In this context, we have the following expressions for the individual equilibrium (Nash-Cournot) quantities and (gross) profits in the product market when all firms use the status quo technology:

Since all firms begin with the same technology, Equations (2) and (3) give the quantity and profit expressions for all firms at the status quo of the market, i.e., q _i,0 = q _j,0 and π _i,0 = π _j,0∀i, j = 1, 2, …, n. The equilibrium total quantity of the industry and market price when all firms use the status quo technology are:^[17]

The equilibrium (gross) profit when only one firm uses the new technology and the rest of the firms continue to use the status quo technology is

for i, j = 1, 2, …, n, while the equilibrium (gross) profit of a firm that continues to use the status quo technology when one of the firms uses the new technology is

for i ≠ j, i, j = 1, 2, …, n. It follows that the equilibrium net profit of firm i for i, j = 1, 2, …, n, that has successfully bid for the patented technology is equal to the firm’s product market profit given by Equation (6) net of the price F paid to obtain the patent,

The net profit for the firms that have not acquired the patent is, of course, equal to the (gross) profit they earn from the product market. Note that Equations (3), (6), and (7) satisfy both Assumptions 1 and 2.

Let us also derive the social welfare of the industry when the patented technology is not implemented and thus all firms produce using the status quo technology. The total surplus at the initial stage of the game is defined by S 0 ≡ S 0 C + S 0 P , where S 0 C is the initial consumer surplus and S 0 P is the initial producer surplus. It can be shown to be:

Recall that consumer surplus is the area of the triangle formed by the vertical price axis, the (linear) demand curve, and the market price, i.e., S 0 C = ( 1 / 2 ) a − P 0 Q 0 , while producer surplus is simply the total profit of all the incumbent firms in the industry.

Consider a hypothetical situation where m ≥ 1 firms have implemented the patented technology by paying a fee f, i.e., let f be a positive fee paid by each firm to use the patent, and the remaining n − m firms continue to operate with the status quo technology. In other words, the patent users produce with a unit cost c′, while the other incumbents have a unit cost of production c for 0 ≤ c′ < c.^[18] As will be evident from Equation (17) below, the following assumption is sufficient to guarantee this:

Similar to the case when the new technology was acquired and used by only one firm, we can determine the expressions that characterize the equilibrium in this case. Let M be the set of firms that have implemented the new technology, and M′ its complement, i.e., the set of firms that continue to operate with the status quo technology. In our notation, m is the cardinality of M, and thus, n − m is the cardinality of M′. We then have:

for i ∈ M, j ∈ M′. Additionally, q i i , m is the equilibrium quantity of firm i with the patented technology when m such firms use the patent, while q j i , m is the equilibrium quantity of firm j without the patented technology when m other firms use the patent. The gross and net profits in equilibrium of these firms can be shown to be:

for i ∈ M, j ∈ M′. Note that to derive the net profit of a firm with a patented technology when a total of m firms use the patented technology, the fee f paid to the patent holder is deducted from the gross profit. Additionally, the net profit of a firm with the status quo technology when m other firms use the patented technology is equal to the (gross) product market profit. The total quantity of the industry and market price in equilibrium are:^[19]

Now, we can apply the same welfare analysis as for the basic case and show that consumer surplus and producer surplus when m firms use the patented technology are:

In this case, consumer surplus is the triangular area formed by the vertical price axis, the demand curve, and the market price when m firms use the patent, while producer surplus is the sum of profits made by the m firms using the new patented technology and the n − m firms using the status quo technology. It follows that social surplus measured by the sum of the consumer surplus, the producer surplus, and the patent fees, i.e., S m = S C m + S P m + m f , is

In Section 2, we showed that the incumbent firms lose if a competitive bidding process starts, in which case the NPE extracts the entire surplus and a premium from the use of the patent. Here, we show that consumer surplus increases as more firms use the new superior technology from Equation (18), which also improves social welfare from Equation (20) (see the proof in Appendix). It can also be shown that P n < P 0 , confirming that consumer surplus improves when all firms use the superior technology. Hence, the competition authorities should not challenge cooperation in patent bidding that results in outcomes where all firms use patented cost-efficient technologies. This is beneficial for consumers and producers. Note, however, that this will only happen if it is in the interest of the incumbent firms to bid collectively rather than to bid individually.

Finally, to show the main result from Proposition 1, consider the case of collusive bidding. Then, from Equation (14), if the share of the patent cost (price) per firm is f, we have

for i = 1, 2, …, n. Note that π i i , n > π i , 0 and, hence, ∑ i = 1 n π i i , n > ∑ i = 1 n π i , 0 , implying that the incumbent firms can generate additional surplus by using the patented technology. The incumbents and the patent holder can now negotiate on how to divide this surplus. The outcome of the negotiation process will determine the patent price F, where F = nf.^[20] The price and quantity in the market are, however, independent of the distribution of the profits (surplus) made with the new technology between the incumbent firms and the patent holder.

4 Conclusions

Collusive bidding is a practice often allowed, if not encouraged, by the competition authorities in the U.S. and EU, although it may create concerns for competition. In this paper, we provide a theoretical foundation that explains why authorities often do not challenge collusive bidding for patents and why courts employ the “rule of reason” to analyze the procompetitive and anticompetitive effects arising from collusive behavior among buyers.

We explain the incentives that competing firms have to form collective entities to buy patents from other entities, especially from non-practicing entities (NPEs), and follow a catch and release patent strategy. We argue that if collusive bidding is not an alternative, then the market for ideas (technology) may not emerge (Gans and Stern 2010), particularly when firms can coordinate behavior. More important, we show that collusive bidding increases consumer surplus, particularly when the technological innovation is drastic and also improves social welfare. It may even be used to prevent the emergence of monopolies with sole ownership of drastic technological innovations, which may otherwise hinder innovation and slow economic growth. These are important results considering the large literature arguing that collusion hampers the maximization of social welfare (e.g., Marshall and Meurer 2004; Wang and Chen 2016).

We analyze a technological innovation owned by an NPE that destroys the value of an existing (status quo) technology used by incumbents (producing entities), and incumbents alone can generate value from the new superior technology. In this context, cooperation in the bidding process (collusive bidding) among incumbents on a patent held by the NPE can generate value (surplus) that competitive bidding among incumbents cannot. This has implications for markets (e.g., emerging markets) where competition policy advocates (and institutions) tend to object to such collusive (or collective) behavior among buyers of new technology by extending the findings from the literature on the inefficiencies of collusion in general markets. We suggest that if such collusion is challenged by competition authorities, then the benefits from new technology may not accrue to the consumers of goods produced by implementing this technology. We also show that a potential entrant can never outbid an incumbent in patent bidding since the entrant does not have existing assets in place that will lose value if entry does not occur, i.e., the entrant loses the bidding process. It is the reduction in the value of the existing assets (the destructive feature of the new technology) that makes the incumbent willing to pay above the market value of the new technology and thus outbid the entrant. Moreover, potential entry increases the number of firms in the market, reducing the incumbent’s profit, particularly if the entrant wins the bidding process.

Finally, we address public policy concerns arguing that although applying the “rule of reason”, for example, to analyze agreements employing cooperative behavior among a cartel of buyers, may solve the holdup problem (Sidak 2009), it does not necessarily create additional value or welfare gains. Applying the “rule of reason” arguments to allow collusive behavior lacks theoretical or empirical support. Therefore, we provide a theoretical foundation to analyze such agreements under competition law.