Sequential innovation, patent policy, and the dynamics of the replacement effect

• DOI: http://doi.org/10.1111/1756-2171.12287
• Author: Álvaro Parra
• Published date: 01 September 2019
• Date: 01 September 2019

RAND Journal of Economics

Vol.50, No. 3, Fall 2019

pp. 568–590

Sequential innovation, patent policy, and the

dynamics of the replacement effect

´

Alvaro Par ra∗

I study how patent policy—characterized by patent length and forward protection—affects Re-

search and Development (R&D) dynamics, leadership persistence, and market structure. Firms’

R&D investments increase as the patent’s expiration date approaches. Through forward protec-

tion, followers internalize the leader’s replacement effect. In protective systems, this internaliza-

tion is substantial, reversing Arrow’straditional result: followers invest less than leaders at every

moment of the patent’s life. I study the policy that maximizes innovative activity.Overly protective

policies decrease innovation pace through two mechanisms: delaying firms’ investments toward

the end of the patent’s life and decreasing the number of firms performing R&D.

1. Introduction

Consider the incentives that a technology leader faces when deciding whether to improve

upon its currently patented technology. When new technologies cannibalize rents from existing

products, the cannibalization reduces the leader’s incentives to invest in replacing its patented

technology (i.e., Arrow’s replacement effect). The replacement effect is nonstationary, patents

lose value when the patent’s expiration date approaches. A leader’s incentives to invest in R&D,

therefore, increase as the patent term runs out. Time mitigates the replacement effect.

Firms that are behind in the technology race (or followers) are also affected by the leader’s

replacement effect. Stronger patent protection against future innovations increases the followers’

probability of infringing on existing patents. When an infringement occurs, license fees equal

to the leader’s profit-loss (damages) must be paid in order to commercialize the new innovation.

Through these fees, followers internalize the cost of replacing the leader, discouraging them

∗University of British Columbia; alvaro.parra@sauder.ubc.ca.

I thank twoanonymous referees and the Editor, Kathryn Spier, for their excellent comments and suggestions. I am indebted

to Michael Whinston and Wojciech Olszewski for their advice and support. I also thank Nancy Gallini, Joshua Gans,

Jorge Lemus, Igor Letina, Fernando Luco, Guillermo Marshall, Alessandro Pavan, JavieraPumarino, Bruno Str ulovici,

Ralph Winter, and seminar participants at Bates and White, CEMFI, ESEM 2013 (Gothenburg), FTC Microeconomics

Conference 2013 (Washington DC), IESE, IIOC 2015 (Boston), LAMES 2013 (Mexico DF), Mannheim University,

Northwestern University, TOI 2013 (Vi˜

na del Mar), University of California at Davis, University of British Columbia,

and WashingtonUniversity. This research was supported by Northwestern’s Center of Economic Theory and by the Social

Sciences and Humanities Research Council of Canada. A previous versionof this ar ticle wastitled “Sequential Innovation

and Patent Policy.”All remaining errors are my own.

C2019 The Authors. The RAND Journal of Economics published by Wiley Periodicals, Inc. on behalf of The RAND

Corporation.

This is an open access article under the terms of the Creative Commons Attribution-NonCommercial-NoDerivs License,

which permits use and distribution in any medium, providedthe original work is properly cited, the use is non-commercial

and no modifications or adaptations are made. 568

PARRA / 569

from investing in R&D. The extent of this internalization can be substantial. Under sufficiently

high protection against future innovations,Ar row’sresult reverses: followers haveless incentive to

invest than leaders at everymoment of the patent’s life. As the patent’s expiration date approaches,

the profit-loss of a replaced leader fades away and the expectedlicense fees paid by an infringing

follower decrease. Consequently, the followers’incentives to improve upon existing technologies

are also nonstationary. Both followers and leaders have greater incentive to investin R&D toward

the end of a patent’s life.

Patents of different length and strength against future innovation thus induce different

innovation patterns among technology leaders and followers; patent policy plays a crucial role in

determining the magnitude and timing of R&D investments as well as the degree of leadership

persistence that exists in the market. This article studies how patent policy—through its dynamic

impact on the replacement effect—shapes firms’ R&D incentives and market structure. Using

these results, I study optimal patent design in the context of a quality-ladder model (Grossman

and Helpman, 1991; Aghion and Howitt, 1992; Aghion et al., 2001).

Innovationscome from a technology leader trying to prolong its lead or from followers aiming

to become the new leader. A patent is represented by a two-dimensional policy determining

how long a leader will be able to exclude others from using its current technology—patent

length—and how enforceable its patent will be against future innovations—forward protection.

FollowingLemley and Shapiro (2005) and Farrell and Shapiro (2008), I treat forward protection as

probabilistic, capturing both the uncertainty that exists when a replaced leader tries to enforce its

patent against a new innovationand the possible leniency of courts toward new innovators. When

a follower developsa new innovation, the replaced leader files an infringement lawsuit against the

innovating follower. The patent authority—for example, a US federal court—may decide, with

certain probability, to uphold the claim or to declare it invalid.1Inthe for mer case, a compulsory

license fee, equal to the damages caused by the commercialization of the new innovation, must

be paid by the infringing firm before the firm can commercialize the new invention and obtain

economic profits.

This article contributes to the literature on leadership persistence. In the context of winner-

takes-all models, Arrow (1962) shows that, when an innovating leader cannibalizes part of its

existing rents, leaders have lower incentives to innovate than followers. In contrast, when an

innovating follower shares the market with the existing leader, Gilbert and Newbery (1982) show

that incumbents have an incentive to preempt followers, persisting as leaders. Using a stochastic

innovationmodel, Reinganum (1983) shows that the preemptive effect in Gilbert and Newbery can

dominate but only under nondrastic innovation.Building on these ar ticles, I connect patent policy

to leadership persistence. In particular, using a stochastic model that can accommodate drastic

innovations, I show that the infringement of existing patents introduces market-sharing effects

into the model. Unlike Gilbert and Newbery,where the reversal occurs due to preemptive motives,

followers invest less than leaders due to the internalization of the (leader’s) replacement effect

through license fees. Strong forward protection increases expected license fees, internalizing the

profit-loss of the leader, discouraging followers from investing in R&D.2

The value of possessing a patent, the number of competing firms, the extent of the replace-

ment effect, and firms’ investment decisions are endogenously determined by patent policy. In

contrast with the previous literature—discussed further in the next section—the finiteness of

patent protection induces nonstationary investments that are increasing throughout the patent’s

life. Although patents are necessary to incentivize innovation, longer protection intensifies the

replacement effect, which induces technology leaders to delay their investments toward the end

1Allison and Lemley (1998) find that in 46% of the litigated cases, the suing patent is found invalid. Price-

waterhouseCoopers (2018) documents that 37% of infringement claims were successful in US federal courts between

1998–2017, and that this success rate varies across sectors.

2Other factors that affect leadership persistence are: productivity differentials (Segerstrom and Zolnierek, 1999;

Segerstrom, 2007), leader’sability to precommit to its R&D (Etro, 2004), and the technology gap among firms (Harris

and Vickers, 1987; Denicol`

o and Zanchettin, 2012).

C2019 The Authors. The RAND Journal of Economics published by Wiley Periodicals, Inc. on behalf of The RAND Corporation.