Abstract
This paper derives necessary and sufficient conditions for compulsory licensing to increase consumer surplus and total welfare, taking into account both static (technology transfer) and dynamic (innovation) effects. When the risk-free rate is low, compulsory licensing is shown unambiguously to increase consumer surplus. Compulsory licensing has an ambiguous effect on total welfare, but is more likely to increase total welfare in industries that are naturally less competitive. Finally, compulsory licensing is shown to be an effective policy to protect competition per se. The paper also demonstrates the robustness of these results to alternative settings of R&D competition and discusses their significance for the debate on compulsory licensing in the context of standard-setting organisations and pharmaceutical trade.
This is a preview of subscription content, log in via an institution to check access.
Similar content being viewed by others
Notes
This might occur through horizontal or cross-licensing agreements, patent pools or patent settlements that facilitate product market collusion, through vertical licensing deals that facilitate resale price maintenance arrangements, as well as through tying of patented products and refusals to license. See Pate (2003) and Katsoulacos (2009).
This was the remuneration principle implemented in the European Microsoft case: “The requirement for the terms imposed by Microsoft to be reasonable and non-discriminatory applies in particular to any remuneration that Microsoft might charge for supply. For example, such remuneration should not reflect the strategic value stemming from Microsoft’s market power in the PC operating system market or in the work group server operating system market.” (European Commission 2007, 31. Emphasis added.) In this paper, we will follow the vast majority of the literature (e.g. Schmalensee 2009; Carlton and Shampine 2013) by not differentiating reasonable and non-discriminatory licensing terms from those that are fair, reasonable and non-discriminatory. Instead, we rely on the most commonly-used acronym, FRAND. We discuss the issue of FRAND licensing in more detail as part of the regulatory implications covered in Sect. 6.
CSU, LLC v. Xerox Corp., 203 F.3d 1322 (Fed. Cir. 2000). The US Patent Act, Section 271(d) states that “[n]o patent owner otherwise entitled to relief for infringement or contributory infringement of a patent shall be denied relief or deemed guilty of misuse or illegal extension of the patent right by reason of his having done one or more of the following: ... (4) refused to license or use any rights to the patent.”
Microsoft v. Commission, Case T201/04 (2007); IMS Health and NDC Health v. Commission, Case C418/01 (2004), ECR I-5039; and Magill ITP, BBC and RTE v. Commission, Cases C241/91 and C242/91 P (1995), ECR I-743.
It should be emphasised that, even in the US, the application of the per se legality legal standard has not been uniform. In Kodak, a “rebuttable presumption” or modified per se legality standard was employed. See Katsoulacos (2009) and Coco (2008) for details. Even in the US, therefore, the appropriate legal treatment of refusals to license IP is not a settled matter. Eastman Kodak Co. v. Image Technical Services Inc., 125 F.3d 1195 (9th Cir. 1997).
The move from a low-false-convictions “exceptional circumstances” rule in IMS Health and Magill towards a low-false-acquittals rule in Microsoft is seen by some as going along with the Commission’s move towards a more economics-based approach, and the related belief that refusals to license IP are likely to be harmful in a much wider range of circumstances than previously supposed, necessitating a more stringent approach (Katsoulacos 2009).
The welfare effects of compulsory licensing have previously been discussed in Chen (2014), Stavropoulou and Valletti (2015) and Bond and Saggi (2012). Our results extend Chen (2014), by treating innovation as a continuous rather than a binary variable. While the latter approach gives insights into the cases where innovation either falls to zero or remains unchanged following the imposition of compulsory licensing, our approach allows us to look at the important intermediate cases where innovation rates may fall somewhat in response to compulsory licensing. Stavropoulou and Valletti (2015) and Bond and Saggi (2012) investigate the welfare effects of compulsory licensing in the specific context of North–South pharmaceutical trade, while our model is general.
See, e.g., Segal and Whinston (2007), who emphasise the front-loading effect, according to which weaker protection for innovative entrants may increase R&D, since it increases the incentives to innovate in order to replace the (more protected) incumbent. In models of sequential innovation, the neck-and-neck effect predicts that firms will invest more in R&D when they are closer together, so that the bunching of firms that is brought about by compulsory licensing may promote R&D spending in subsequent stages of innovation. See Aghion et al. (2005). These effects are discussed in more detail in Vickers (2010).
Two exceptions are Canada and New Zealand, which follow a total welfare standard.
See, e.g., Acemoglu and Akcigit (2012). They show that a so-called trickle-down effect makes a staggering of compulsory licensing fees optimal, whereby firms that are furthest behind pay more for the licence. While this mitigates the fall in innovation rates compared to a uniform compulsory licensing policy, innovation rates do still fall relative to the full IP protection benchmark.
This paper also relates to the literature on the impact of downstream competition on upstream innovation decisions, as in Iskhakov et al. (2013). In particular, we consider the case in which competition is promoted via the imposition of a compulsory licence, which simultaneously weakens IP protection. See Gilbert and Shapiro (1996) and Kühn and Van Reenen (2008) for more general discussion of the R&D effects of compulsory licensing.
Rather, this results suggests that refusals to license should be viewed as strongly presumptively illegal if consumer surplus is the relevant welfare measure.
In terms of the wider differences between the US and European competition policy regimes, it is worth noting that in the US, antitrust claims are normally private, while in Europe they are generally public, with the Commission acting as prosecutor and judge, subject to the judicial review of the CFI and ECJ (Coco 2008). While this paper does not consider directly the optimal structure of competition law enforcement, our conclusion that the presumption of legality in refusal to license cases is very low does suggest that the degree of regulatory oversight associated with the European system more generally is, in this context, desirable.
This terminology follows Beath et al. (1987, 1995). (The persistent dominance scenario has alternatively been described as one of “increasing dominance” in the literature. See, e.g., Vickers 1986.) See also Carlton and Gertner (2002), who argue that most R&D-intensive industries, such as the IT, pharmaceutical and chemical industries, are characterised by action–reaction competition.
Another concern in the action–reaction setting is that dominant firms may refuse to license in an attempt to limit the innovative advantage that less efficient firms enjoy.
Our set-up is also consistent with certain representations of product quality. For example, in Häckner (2000) (assuming homogeneous products), demand for firm i takes the form \(p_i (Q)=\alpha _i -Q\), where Q is total output and \(\alpha _i \) is firm i’s quality parameter. Since costs are normalised to zero in that framework, this is equivalent to our demand environment in which costs vary and quality remains constant (see Sect. 2.3 below).
See, e.g., Ulph and Ulph (1998, 2001) and related literature (the term “competitive threat” is also used in Beath et al. 1989a, b). The reliance on the competitive threat as the sole determinant of firms’ innovation incentives may also be justified by the following scenario. Suppose that producing firms bid to acquire the innovation from an outside firm. In this auction game, if winning firms pay their own bids, the producing firms’ bidding incentives will be given exactly by their competitive threats, since one firm will inevitably acquire the patent rights. In this setting, the competitive threat represents an exact solution to the firms’ R&D investment problem. See, e.g., Reinganum (1983, p. 742).
See Appendix 2 for a more detailed derivation of these results.
See Kamien and Tauman (1986) and Kamien et al. (1992) for licensing based purely on royalties, and Sen and Tauman (2007) for the case of optimally combined fixed fees and royalties. Chen (2014) examines the particular case of a vertically integrated firm with a monopoly position in the upstream IP market that competes with a downstream rival. The key feature of the model is an exogenous probability of a follow-on innovation that makes the upstream IP market competitive. In that case, provided that the probability of follow-on innovation is sufficiently high, the integrated firm may refuse to license to the downstream rival, even with royalties, in order to exclude it and thereby maintain downstream dominance when the market for IP becomes competitive. We consider a more standard case with two vertically integrated firms that compete both in the IP and the product market, and with innovations that are, for the majority of the paper, non-drastic (non-essential). This allows us to focus more carefully on the implications of compulsory licensing for innovation (which we treat as a continuous rather than binary variable) and welfare.
Rey and Salant (2012) is another example of a paper that assumes fixed fee licensing in the context of abuse of dominance regulation involving IP.
This captures total welfare in an instantaneous, product market sense. An expression for present discounted total welfare, taking into account R&D costs incurred at Stage 1 of the model and firms’ innovation probabilities, will be derived in Sect. 4.2.
Because, in that simple case, aggregate output, price, industry revenue and consumer surplus are unaffected. See Salant and Shaffer (1999).
Février and Linnemer (2004) develop general results for marginal cost changes. Since we are concerned with discrete cost changes, these results apply here only in the sense of first-order approximations. While the nature of the results is unchanged under the approximation method, we focus on the linear demand set-up for tractability.
This measure of industry competitiveness is also used in Katsoulacos and Ulph (2013), for example. Measuring elasticity at the competitive equilibrium avoids the cellophane fallacy.
See also Vickers (1986). Budd et al. (1993) show in the context of a more general dynamic model that the industry tends to evolve in the direction where firms’ joint payoffs are higher. Harris and Vickers (1985) emphasise the role of asymmetrical patent races in the persistence of monopoly, while Reinganum (1985) explains the persistence of monopoly by a continuing process whereby successful entrants replace successive incumbents.
Instead of relying on firms’ competitive threats, we could alternatively rely on their profit incentives, defined as the difference between the profits from winning the innovation race and the status quo, to determine innovation incentives. This gives rise the replacement effect, which tends to favour innovation by the follower: since the incumbent earns more profits in the status quo, he faces (ceteris paribus) lower incentives to innovate. By focusing on the competitive threat (as has also been done in the existing literature, e.g. Ulph and Ulph 1998, 2001), we are implicitly emphasising the efficiency effect over the replacement effect. This simplification comes at no great cost, however, since our framework is still flexible enough to generate an action-reaction outcome, in which the follower is predicted to innovate successfully and overtake the incumbent (see Sect. 5). This will allow us to check the robustness of our main welfare results to a setting of action-reaction R&D competition.
In the context of our linear demand set-up, this condition is equivalent to the requirement that \(2\varepsilon <3g+8G\).
See Salant and Shaffer (1999). In the context of our linear demand example, it is straightforward to verify that a sufficient condition for the follower to license is that \(g<G\). This condition is far from being necessary, however. Indeed, the necessary and sufficient condition for the follower not to license is that \(3g>2\varepsilon +10G\). In other words, a necessary condition for the follower to refuse to license is that the gap opened up by the new discovery is more than three times as large as the initial gap of the leader.
This replicates the result of Katz and Shapiro (1985) and others, who show that voluntary licensing has an ambiguous effect on industry-wide innovation incentives, relative to no licensing. As in that paper, aggregate innovation incentives will rise if the bargaining strength of the licensor (here: the follower) is high, specifically if \(1-\sigma >1/2\).
See, e.g., Manzini and Mariotti (2003), Arguedas (2005) and Segerson and Miceli (1998). A notable difference in the bargaining context analysed in those papers is that all firms lose out as a result of environmental regulation. In the compulsory licensing context, the non-innovator actually benefits when compulsory licensing is imposed.
Note that the FRAND requirement is therefore equivalent to the requirement that \(P^{\textit{FRAND}}\le \overline{{P}}^{L}\) [see (4)]. That is, the FRAND price must be below the willingness of the non-innovator to pay for the licence.
Since the focus of the paper is the effect of compulsory licensing relative to a voluntary licensing benchmark, we exclude for brevity the no-licensing regime from these comparisons.
The derivation of this expression parallels the derivation of present discounted profits in Appendix 2.
See (1), which shows that the competitive threats are weighted by the reciprocal of r. This assumption of low r is, therefore, also implicit in the work of Ulph and Ulph (1998, 2001), for example, who similarly rely on the competitive threat as the principal determinant of firms’ innovation decisions.
This follows by inspection of the residual terms in (15), which may also be rewritten as \(X^{V}\left[ {{\textit{CS}}\left( {G+2g} \right) -{\textit{CS}}\left( {G+g} \right) } \right] -\left( {X^{V}+Y^{V}-2X^{C}} \right) \left[ {{\textit{CS}}\left( {G+2g} \right) -{\textit{CS}}\left( G \right) } \right] \). The more general problem with considering an environment in which r is high is that, in that case, it is no longer true that the competitive threats dominate firms’ innovation decisions. Instead, these will be based on their profit incentives (see footnote 26 above), moving this case outside the scope of the present framework. We therefore leave a full characterisation of the case where r is high as the subject of future work.
This follows because \(\Delta _{\textit{TW}} =\left( {g/18} \right) \left( {8\varepsilon -3g-14G} \right) \).
It is well known that tournament models provide incentives for socially excessive investment in R&D (see, e.g., Beath et al. 1995). It should be noted in this regard that the nature of our total welfare result is robust to the inclusion of spillovers (by which innovation incentives can be made arbitrarily small) and, as such, it is not driven by this average over-investment issue. See Appendix 4 for details.
With reference to (23), in order for there to be a possibility that the leader will refuse to license in this action-reaction context, it must therefore be true that \(g<G\).
In this sense, the refusal by the leader to license voluntarily can also be seen as an attempt to limit the extent of action-reaction in the market.
The results from the foreclosure standard carry over unchanged to the action-reaction setting.
This will of course tend to favour compulsory licensing over voluntary licensing in the total welfare comparison.
Implicitly, given that our model features only one licensee, we are focusing here on the “fair and reasonable” portion of the FRAND requirement. The extent to which innovators should be allowed to price discriminate between licensees under a FRAND regime is discussed in Swanson and Baumol (2005) and Mariniello (2011), for example. The incentives for owners of essential IP to charge excessive royalties and the associated impact on downstream competition is further discussed in Rey and Salant (2012) and Layne-Farrar and Schmidt (2010).
See O’Donoghue and Padilla (2013, pp. 689–711) for a discussion of FRAND pricing in the context of recent cases.
This argument implicitly assumes that aggregate innovation incentives were not socially excessive to begin with. Moreover, the claim made by the European Commission in Microsoft that compulsory licensing may actually increase innovation incentives (e.g. Komninos and Czapracka 2010) cannot hold in our framework. In other words, the trade-off between competition and innovation incentives is always real in our setting and our results in favour of compulsory licensing come about despite this trade-off.
See O’Donoghue and Padilla (2013, p. 684) for more detail on the variety in types of SSOs. Two recent cases with relevance to the SSO context are Microsoft Corp. v. Motorola Inc. in the US, which established that Motorola had violated its [F]RAND obligations with respect to its standard-essential patents, and Rambus in the EU, in which it was found that the firm had engaged in “patent ambush” by concealing its ownership of IP that was subsequently to be deemed essential as part of a new standard. Microsoft Corp. v. Motorola Inc., No. C10-1823JLR, 2013 WL 2111217 (WD Wash Apr 25, 2013); Rambus v. Commission, Case COMP/38.636 (2009).
See, e.g., Sidak (2013, pp. 949–950) for relevant commitments in the policies of the European Telecommunications Standards Institute (ETSI) and the Institute of Electrical and Electronic Engineers (IEEE) SSOs.
See, e.g., Chiao et al. (2007). Steady licensing revenue streams are one particular benefit of SSO membership.
In fact, most of the contributions to the debate on FRAND licensing have to-date been set in the context of SSOs. One specific method of determining FRAND licence fees in the SSO setting is based on the Georgia-Pacific factors, a framework put forward by the US District Court for the Southern District of New York in Georgia-Pacific Corp. v. United States Plywood Corp. to determine appropriate damages in patent infringement cases. This damages calculation is itself based on an assessment of a “reasonable royalty”, defined as the hypothetical voluntary licence price that would have been agreed between parties ex ante (see point 15 of the Georgia-Pacific factors). While this, at least conceptually, gets around the hold-up problem in SSOs, it is not helpful in the setting of our model absent standards, where an ex ante mutually agreeable voluntary licence price does not exist when the leader innovates. Georgia-Pacific Corp. v. United States Plywood Corp., 318 F.Supp. 1116, 1120 (SDNY 1970). See also Sidak (2013), Layne Farrar et al. (2007) and literature cited therein for a discussion of FRAND licensing in the context of SSOs.
It is worth noting that there may be other antitrust concerns that go along with the formation of standards, such as the potential for collusion among members of the standard, which we do not go into here. See, e.g., Schmalensee (2009). Furthermore, our framework does not cover non-vertically integrated firms and, since only one firm can innovate, it does not extend to competing technologies, which may raise further interesting questions in the SSO context.
Available at: https://www.wto.org/english/tratop_e/trips_e/t_agm0_e.htm.
Of course, there are other specific factors relating to the pharmaceutical industry and trade-related aspects of compulsory licensing that also need to be taken into account. See, e.g., Stavropoulou and Valletti (2015) for further discussion of these points.
In what follows, we will focus on the leader’s decision problem with no loss in generality—equivalent expressions to those presented for the leader also hold for the follower.
This model therefore also nests the voluntary licensing regime of Sect. 3.1 (when \(\theta ^{L}=\theta ^{F}=0)\) and the compulsory licensing regime of Sect. 3.3 (when \(\theta ^{L}=1\) and \(\theta ^{F}=0)\). Note that, in principle, it is also possible to allow \(\theta ^{F}\) to increase from zero under a compulsory licensing policy. This would shift the reservation prices, and therefore also the actual price at which the follower licenses towards \(P^{F(FRAND)}\). There are then two possibilities. To the extent that \(P^{F(FRAND)}\) is low, this would reduce the voluntary licence price \(P^{F}\), and therefore reinforce the negative incentive effect that is already present in the model. To the extent that \(P^{F(FRAND)}\) is very high (that is, close to the incremental value \(\pi _L \left( {G+g,g} \right) -\pi _L \left( {G,g} \right) )\), however, an increase in \(\theta ^{F}\) could actually increase the voluntary licence fee \(P^{F}\) and therefore provide an offsetting incentive effect relative to the voluntary licensing scenario. This possibility notwithstanding, since we typically think of compulsory licensing as a tool to combat the refusal by dominant firms to license, rather than one by which follower firms can extract higher prices for their licences, we stick with the model where \(\theta ^{F}\) remains constant at zero. This is, moreover, consistent with a setting in which compulsory licensing only affects dominant firms.
Clearly, setting \(s=0\) returns the corresponding expression from the no-spillovers case analysed in the main body of the paper, see (20).
References
Acemoglu, D., & Akcigit, U. (2012). Intellectual property rights policy, competition and innovation. Journal of the European Economic Association, 10(1), 1–42.
Aghion, P., Bloom, N., Blundell, R., Griffith, R., & Howitt, P. (2005). Competition and Innovation: An inverted-U relationship. Quarterly Journal of Economics, 120(2), 701–728.
Amir, R., Encaoua, D., & Lefouili, Y. (2014). Optimal licensing of uncertain patents in the shadow of litigation. Games and Economic Behavior, 88, 320–338.
Arguedas, C. (2005). Bargaining in environmental regulation revisited. Journal of Environmental Economics and Management, 50(2), 422–433.
Beath, J., Katsoulacos, Y., & Ulph, D. (1987). Sequential product innovation and industry evolution. Economic Journal, 97, 32–43. Conference 1987.
Beath, J., Katsoulacos, Y., & Ulph, D. (1989a). The game-theoretic analysis of innovation: A survey. Bulletin of Economic Research, 41(3), 163–184.
Beath, J., Katsoulacos, Y., & Ulph, D. (1989). Strategic R & D policy. Economic Journal, 99, 74–83. Conference 1989.
Beath, J., Katsoulacos, Y., & Ulph, D. (1995). Game-theoretic models of innovation. In P. Stoneman (Ed.), Handbook of the economics of innovations and technological change (pp. 132–181). Oxford: Basil Blackwell.
Bond, E., & Saggi, K. (2012). Compulsory licensing, price controls, and access to patented foreign products. Vanderbilt University Department of Economics Working Papers. No. 12-00006.
Budd, C., Harris, C., & Vickers, J. (1993). A model of the evolution of Duopoly: Does the asymmetry between firms tend to increase or decrease? Review of Economic Studies, 60(3), 543–573.
Carlton, D. W., & Gertner, R. H. (2002). Intellectual property, antitrust and strategic behavior. NBER Working Paper. No. 8976.
Carlton, D. W., & Shampine, A. L. (2013). An economic interpretation of FRAND. Journal of Competition Law and Economics, 9(3), 531–552.
Chen, Y. (2014). Refusals to deal, intellectual property rights, and antitrust. Journal of Law, Economics, and Organization, 30(3), 533–577.
Chiao, B., Lerner, J., & Tirole, J. (2007). The rules of standard-setting organizations: An empirical analysis. RAND Journal of Economics, 38(4), 905–930.
Coco, R. (2008). Antitrust liability for refusal to license intellectual property: A comparative analysis and the international setting. Marquette Intellectual Property Law Review, 12(1), 1–47.
DOJ-FTC (2007). Antitrust enforcement and intellectual property rights: Promoting innovation and competition. Retrieved 25 July 2015 from www.usdoj.gov/atr/public/hearings/ip/222655.pdf.
European Commission. (2007). Commission decision of 24 May 2004 relating to a proceeding pursuant to article 82 of the EC treaty and article 54 of the EEA agreement against Microsoft Corporation (Case COMP/C-3/37.792—Microsoft). Official Journal of the European Union, L32, 23–28.
Feldman, J. (2009). Compulsory licenses: The dangers behind the current practice. The Journal of International Business and Law, 8(1), 137–167.
Février, P., & Linnemer, L. (2004). Idiosyncratic shocks in an asymmetric Cournot oligopoly. International Journal of Industrial Organization, 22(6), 835–848.
Gilbert, R. J., & Newberry, D. M. G. (1982). Preemptive patenting and the persistence of monopoly. American Economic Review, 72(3), 514–526.
Gilbert, R. J., & Shapiro, C. (1996). An economic analysis of unilateral refusals to license intellectual property. Proceedings of the National Academy of Sciences USA, 93(23), 12749–12755.
Grossman, G. M., & Shapiro, C. (1987). Dynamic R and D competition. Economic Journal, 97(386), 372–387.
Häckner, J. (2000). A note on price and quantity competition in differentiated oligopolies. Journal of Economic Theory, 93(2), 233–239.
Harris, C., & Vickers, J. (1985). Patent races and the persistence of monopoly. Journal of Industrial Economics, 33(4), 461–481.
Iskhakov, F., Rust, J., & Schjerning, B. (2013). The dynamics of bertrand price competition with cost-reducing investments. Univ. of Copenhagen Dept. of Economics Discussion Paper. No. 13–05.
Kamien, M. I., & Tauman, Y. (1986). Fees versus royalties and the private value of a patent. Quarterly Journal of Economics, 101(3), 471–491.
Kamien, M. I., Oren, S., & Tauman, Y. (1992). Optimal licensing of cost-reducing innovation. Journal of Mathematical Economics, 21(5), 483–508.
Katsoulacos, Y. (2009). Optimal legal standards for refusals to license IP: A welfare-based analysis. Journal of Competition Law and Economics, 5(2), 269–295.
Katsoulacos, Y., & Ulph, D. (2009). On optimal legal standards for competition policy: A general welfare-based analysis. Journal of Industrial Economics, 57(3), 410–437.
Katsoulacos, Y., & Ulph, D. (2013). Antitrust penalties and the implications of empirical evidence on cartel overcharges. Economic Journal, 123(572), F558–F581.
Katz, M. L., & Shapiro, C. (1985). On the licensing of innovations. RAND Journal of Economics, 16(4), 504–520.
Komninos, A. P., & Czapracka, K. A. (2010). IP rights in the EU Microsoft Saga. In I. Kokkoris & F. Etro (Eds.), Competition law and the enforcement of article 102 (pp. 224–241). Oxford: University Press.
Kühn, K.-U., & Van Reenen, J. (2008). Interoperability and market foreclosure in the European Microsoft case. Centre for Economic Performance Special Paper, No. 20.
Layne Farrar, A., Padilla, J., & Schmalensee, R. (2007). Pricing patents for licensing in standard setting organizations: Making sense of FRAND commitments. Antitrust Law Journal, 74(3), 671–706.
Layne-Farrar, A., & Schmidt, K. M. (2010). Licensing complementary patents: Patent trolls, market structure, and excessive royalties. Berkeley Technology Law Journal, 25(2), 1121–1144.
Linnemer, L. (2003). Backward integration by a dominant firm. Journal of Economics and Management Strategy, 12(2), 231–259.
Manzini, P., & Mariotti, M. (2003). A bargaining model of voluntary environmental agreements. Journal of Public Economics, 87(12), 2725–2736.
Mariniello, M. (2011). Fair, reasonable and non-discriminatory (FRAND) terms: A challenge for competition authorities. Journal of Competition Law and Economics, 7(3), 523–541.
Novshek, W. (1985). On the existence of Cournot equilibrium. Review of Economic Studies, 52(1), 85–98.
O’Donoghue, R., & Padilla, J. (2013). The law and economics of article 102 TFEU. Oxford: Hart Publishing.
Pate, R. H. (2003). Antitrust and intellectual property. Arlington: American Intellectual Property Law Association.
Reinganum, J. F. (1983). Uncertain innovation and the persistence of monopoly. American Economic Review, 73(4), 741–748.
Reinganum, J. F. (1985). Innovation and industry evolution. Quarterly Journal of Economics, 100(1), 81–99.
Rey, P., & Salant, D. (2012). Abuse of dominance and licensing of intellectual property. International Journal of Industrial Organization, 30(6), 518–527.
Salant, S. W., & Shaffer, G. (1999). Unequal treatment of identical agents in Cournot equilibrium. American Economic Review, 89(3), 585–604.
Schmalensee, R. (2009). Standard-setting, innovation specialists and competition policy. Journal of Industrial Economics, 57(3), 526–552.
Segal, I., & Whinston, M. D. (2007). Antitrust in innovative industries. American Economic Review, 97(5), 1703–1730.
Segerson, K., & Miceli, T. J. (1998). Voluntary environmental agreements: Good or bad news for environmental protection? Journal of Environmental Economics and Management, 36(2), 109–130.
Sen, D., & Tauman, Y. (2007). General licensing schemes for a cost-reducing innovation. Games and Economic Behavior, 59(1), 163–186.
Sidak, J. G. (2013). The meaning of FRAND, part I: Royalties. Journal of Competition Law and Economics, 9(4), 931–1055.
Stavropoulou, C., & Valletti, T. (2015). Compulsory licensing and access to drugs. European Journal of Health Economics, 16(1), 83–94.
Swanson, D. G., & Baumol, W. J. (2005). Reasonable and nondiscriminatory (RAND) royalties, standards selection, and control of market power. Antitrust Law Journal, 73(1), 1–58.
Ulph, A., & Ulph, D. (1998). Bargaining and innovation, labour markets. European Economic Review, 43(35), 931–939.
Ulph, A., & Ulph, D. (2001). Strategic innovation with complete and incomplete labour market contracts. Scandinavian Journal of Economics, 103(2), 265–282.
Vickers, J. (1986). The evolution of market structure when there is a sequence of innovations. Journal of Industrial Economics, 35(1), 1–12.
Vickers, J. (2010). Competition policy and property rights. Economic Journal, 120(544), 375–392.
Acknowledgments
I am grateful to the Editor, Michael Crew, and two referees for constructive feedback during the review process. I would also like to thank David Ulph, Yannis Katsoulacos, Marco Mariotti, Vincenzo Denicolò, Yassine Lefouili, as well as participants of the 2013 CRESSE, EARIE and German Economic Association conferences, and the St Andrews Workshop on Competition, Innovation and Competition Policy for helpful comments and discussion. Financial support from the UK Economic and Social Research Council (grant number ES/J500136/1) and the Royal Economic Society (grant number RESJFS2014) is gratefully acknowledged. Any errors and omissions are my own.
Author information
Authors and Affiliations
Corresponding author
Appendix
Appendix
1.1 Appendix 1: Collected proofs
Proof of Lemma 1
The sum of marginal costs (equivalently, cost gaps) is the same, regardless of which firm innovates. Therefore, on the basis of Salant and Shaffer (1999), since the variance of marginal costs is greater when the leader wins, we know that industry profits are also greater in that case: \(\Sigma \left( {G+g,0} \right) >\Sigma \left( {G,g} \right) \).
Proof of Lemma 3
Lemma 2 confirms the equality between hazard rates under compulsory licensing, while (12) confirms the rankings in the voluntary licensing regime. We can also see that compulsory licensing cannot lead to an increase in the hazard rate chosen by the follower, since, by (13),
The inequality is strict if we have \(\phi <1\).
Proof of Lemma 4
We can write the rate-adjusted hazard rates under compulsory licensing as
The follower will remain the predicted winner of the race if and only if \(Y^{C}>X^{C}\), which is to say
Given (22), (27) and (28), this is always satisfied.
Proof of Lemma 5
Compulsory licensing reduces the hazard rate of the leader and (weakly) reduces the hazard rate of the follower. By (28)
Also,
since \(\phi \le 1\).
1.2 Appendix 2: Firms’ innovation behaviour in tournament R&D model
This appendix presents a more formal derivation of firms’ R&D investment behaviour. It should be noted that our description of the single-stage, tournament model of R&D competition follows standard treatments in the literature (e.g. Grossman and Shapiro 1987; Beath et al. 1989a, b).
In this setting, firms decide at each instant of time, over the infinite horizon of the model, how much to invest in R&D. Following the literature, it is assumed that a given firm’s probability of innovating, conditional on no firm having innovated up to that point (that is, its hazard rate), depends solely on its current flow rate of R&D expenditure (there is no learning by doing). Therefore, if a given firm spends a constant amount on R&D over time, its hazard rate will also be constant over time. Formally, this implies that if the leader, say, spends a constant flow amount on R&D, the probability that it will discover by date t is equal to \(F\left( {t;x} \right) =1-e^{-xt}\), yielding a hazard rate \(H\left( {t;x} \right) =\left[ {\partial F\left( {t;x} \right) /\partial t} \right] /\left[ {1-F\left( {t;x} \right) } \right] =x\), as required, and where x is a constant that is determined by the amount that the leader invests in R&D. (Similarly for the follower, \(F\left( {t;y} \right) =1-e^{-yt}\).) In particular, as noted in Sect. 2.1, both firms are assumed to face the same innovation technology, which implies an R&D cost that is quadratic in the chosen hazard rate (so the leader’s flow cost of achieving hazard rate x is \(c\left( x \right) =x^{2}\), and likewise for the follower). Note finally that, since the decision problem faced by the firms is stationary, firms will indeed select a constant hazard rate over time.
In these circumstances, assuming that the firms can borrow and lend at the constant risk-free rate of interest r, the present discounted payoff to the leader from choosing hazard rate x when the follower chooses hazard rate y, denoted by \(V\left( {x,y} \right) \), is implicitly defined byFootnote 55
Here \(\pi _L^{win} \) and \(\pi _L^{lose} \) represent the flow profits earned by the leader in the product market, conditional on having won and lost the innovation race, respectively, and \(\pi _L^0 \) is the leader’s status quo profit level. Equation (33) can be rearranged to yield
To derive the leader’s reaction function, the first-order condition with respect to x may be easily solved for. This shows that, as the follower’s chosen hazard rate becomes arbitrarily large, \(y\rightarrow \infty \) (so that the follower is almost certain to innovate), the leader’s choice of hazard rate is determined by the competitive threat, defined as \(\bar{{x}}=\frac{\pi _L^{win} -\pi _L^{lose} }{2r}\). If, on the other hand, the follower is certain not to innovate, \(y=0\), the leader selects its hazard rate on the basis of the profit incentive, defined as \(x_0 =\sqrt{\pi _L^{win} -\pi _L^0 +r^{2}}-r\).
It is important to note that, when r is low (consistent with our framework), the competitive threat is orders of magnitude larger than the profit incentive, and therefore dominates firms’ R&D investment decisions (in the sense that the firms’ reaction functions intersect at a point that is well-approximated by their competitive threats. See, e.g., Beath et al. 1989a). This is the reason for which much of the applied literature has focused on the competitive threat as the key determinant of firms’ innovation decisions (Ulph and Ulph 1998, 2001). As discussed in Sect. 2.1, in order to render the model solvable, we follow this approach by taking \(x=\bar{{x}}\) for the leader and likewise for the follower (see (1)).
1.3 Appendix 3: Voluntary licensing and the ‘regulatory threat’
We wish to show that, when firms anticipate the possibility of a compulsory licensing remedy in case no voluntary agreement is reached, the licensing conditions (5) and (8) do not change. As described in Sect. 3.2, let \(\theta ^{j},\, 0\le \theta ^{j}\le 1\), be the common probability with which firms anticipate a compulsory licence being imposed on the innovating firm \(j=L,F\) if a voluntary deal is refused. Moreover, let \(P^{j({\textit{FRAND}})}\) now denote the FRAND licence price that applies in case firm \(j=L,F\) innovates successfully and is forced to license by compulsory licence.
Then, in case the leader innovates and licenses, the cost gaps are \(\left( {g_L^{LV} ,g_F^{LV} } \right) =\left( {G+g,g} \right) \), while, if the leader innovates and does not license, the cost gaps would be \(\left( {g_L^{LN} ,g_F^{LN} } \right) =\left( {G+g,0} \right) \) with probability \(\left( {1-\theta ^{L}} \right) \), and \(\left( {g_L^{LN} ,g_F^{LN} } \right) =\left( {G+g,g} \right) \) with probability \(\theta ^{L}\). It follows that the minimum price that the leader would be willing to accept for the licence, and the maximum that the follower would be willing to pay, can be written as
and
respectively. That is, the reservation prices are now a weighted average of those that apply under voluntary licensing [see (3) and (4)] and the appropriate FRAND fee. Therefore, the leader will still license if and only if \(\Sigma \left( {G+g,g} \right) >\Sigma \left( {G+g,0} \right) \).
Similar derivations for the case when the follower innovates shows that the reservation prices in that case can be written as
and
Therefore, if the follower innovates, licensing will again take place if and only if \(\Sigma \left( {G+g,g} \right) >\Sigma \left( {G,g} \right) \).
Since, in our general set-up of Sect. 2.3, it must hold that \(\Sigma \left( {G+g,0} \right) >\Sigma \left( {G,g} \right) \) (Salant and Shaffer 1999), the same intuition concerning the incentives of the leader to refuse to license carry over to this setting. Given (10), the leader has no (strict) incentive to license for all \(0\le \theta ^{L}\le 1\). It follows that the threat of compulsory licensing alone cannot resolve the refusal to license problem.
Note that, when \(\theta ^{L}=1\), the leader’s decision to license has no perceived effect on industry profits, because firms anticipate a compulsory licence with certainty in case no voluntary deal is reached. Therefore, the licensing conditions break down, and firms are indifferent between licensing and not licensing. We may assume that the leader still does not license in that case. Alternatively, note that \(\theta ^{L}=1\) implies that \(\underline{P}^{L}=\overline{{P}}^{L}=P^{L({\textit{FRAND}})}\), so that the only voluntary deal that may be reached is the one which exactly replicates the compulsory licence. Therefore, from an analytical point of view, we can still talk of compulsory licensing with no loss in generality.Footnote 56
1.4 Appendix 4: Spillovers
We wish to demonstrate that our total welfare result is robust to the inclusion of spillovers, a standard method in the context of tournament models of R&D to combat the excess investment problem. Suppose, to that end, that a fraction \(s,\, 0<s<1\), of the technological progress engendered in any innovation spills over to the non-innovating firm. So, in the absence of licensing, if the leader innovates, the cost gaps would be \(\left( {g_L^{LN} ,g_F^{LN} } \right) =\left( {G+g,sg} \right) \), while if the follower innovates they would be \(\left( {g_L^{FN} ,g_F^{FN} } \right) =\left( {G+sg,g} \right) \). Therefore, we will have persistent dominance if and only if
This always holds (Salant and Shaffer 1999), so that we still have persistent dominance in our baseline scenario.
Now, given the assumption of fixed fee licensing, it is straightforward to see that, if the leader innovates, licensing will take place if and only if
while, if the follower innovates, we will have voluntary licensing if and only if
Given (34), it is still true that, if any firm were to refuse to license, it would be the leader, as in Sect. 3.1.
What we now require in order for total welfare to be higher under compulsory licensing rather than voluntary licensing when r is low (abstracting from cost savings) is that
This is equivalent to the requirement thatFootnote 57
Therefore, the total welfare effect is still positive when the industry is sufficiently un-competitive (that is, when \(\varepsilon \) is sufficiently high)—thus, the nature of Proposition 2 does not change. Nonetheless, the precise threshold level of “un-competitiveness” above which the total welfare effect turns positive is clearly increasing in the magnitude of the spillover parameter s. This is consistent with the idea that R&D investment is more likely to be excessive from society’s point of view when s is low.
Rights and permissions
About this article
Cite this article
Seifert, J. Welfare effects of compulsory licensing. J Regul Econ 48, 317–350 (2015). https://doi.org/10.1007/s11149-015-9288-9
Published:
Issue Date:
DOI: https://doi.org/10.1007/s11149-015-9288-9