PATENTABILITY, R&D DIRECTION, AND CUMULATIVE INNOVATION

• Author: Yongmin Chen,Shiyuan Pan,Tianle Zhang
• DOI: http://doi.org/10.1111/iere.12326
• Date: 01 November 2018
• Published date: 01 November 2018

INTERNATIONAL ECONOMIC REVIEW

Vol. 59, No. 4, November 2018 DOI: 10.1111/iere.12326

PATENTABILITY, R&D DIRECTION, AND CUMULATIVE INNOVATION∗

BYYONGMIN CHEN,SHIYUAN PAN,AND TIANLE ZHANG1

University of Colorado, U.S.A.; Zhejiang University,China; Lingnan University, Hong Kong

We present a model where firms conduct R&D in both a safe and a risky direction. As patentability standards

rise, an innovation in the risky direction is less likely to receive a patent, which decreases the static incentive for

new entrants to conduct risky R&D but can increase their dynamic incentive. These, together with a strategic

substitution and a market structure effect, result in an inverted-U shape in the risky direction but a U shape

in the safe direction for the relationship between R&D intensity and patentability standards. R&D is biased

toward (against) the risky direction under lower (higher) standards.

1. INTRODUCTION

A central issue in the economics of innovation is how patent policy may affect innovative

activities. The recent literature has examined this issue in the context of cumulative innova-

tion, where discoveries build on each other, under a standard assumption that firms pursue

innovations along a single research direction. In many industries, however, firms can conduct

R&D in multiple directions to achieve a specific goal, as, for example, the development of a

next-generation color copier in the early 1990s by Fuji Xerox, of a new mobile system by Eric-

sson in the mid-1990s, and of an X Terminal workstation by Hewlett Packard in the late 1980s

(Birkinshaw and Lingblad, 2001).2The purpose of this article is to inquire how patent policy,

specifically patentability standards, may affect the rate and direction of cumulative innovation

in an industry where firms can conduct R&D in multiple directions.

We consider a situation where there are two research directions, Aand B, for a sequence

of innovations (or new products) that deliver higher product qualities over time. The quality

improvement of an innovation in direction Bmay range stochastically from low to high, whereas

that in direction Ais at some intermediate level. Hence, if an innovation is patentable only

when its quality improvement (or innovation size) is sufficiently large, as, for instance, implied

by the requirement of a minimum inventive step, there will be a range of quality thresholds, or

patentability standards (S), under which innovations in direction Aare always patentable but

an innovation in direction Bmay not be. We will focus on patentability standards in the interior

of such a range and call Athe safe direction and Bthe risky direction.3

∗Manuscript received August 2016; revised October 2017.

1We thank two referees, an editor, and participants of several conferences for helpful comments. Shiyuan Pan grate-

fully acknowledges financial support from the Key Project of the National Social Science Foundation of China (No.

15AJY003) and the Project of the Scientific Research Foundation for the Returned Overseas Chinese Scholars of the

Ministry of Education of China. Tianle Zhang gratefully acknowledges the Hong Kong Research Grant Council for re-

search support under Grant #23501215. Please address correspondence to: Shiyuan Pan, School of Economics, Zhejiang

University, 38 Zheda Road, Hangzhou, 310027, China. Phone: 86-0571-87952252. E-mail: shiyuanpan@zju.edu.cn.

2In particular, facing the possible emergence of a third-generation mobile system with high bandwidth, Ericsson in

the mid-1990s funded research teams to separately develop two different standards: a more radical new technology

called WCDMA and a new standard called EDGE upgrading the existing technologies.

3That is, we consider patentability standards that are high enough to exceed the left tail of innovations in the risky

direction but not so high as to make a safe innovation not patentable. We rule out by assumption situations where

patentability standards are so high that a safe innovation is not patentable but a risky one can be (in which case the

problem would be reduced to analyzing innovations only in the risky direction).

1969

C

(2018) by the Economics Department of the University of Pennsylvania and the Osaka University Institute of Social

and Economic Research Association

1970 CHEN,PAN,AND ZHANG

If innovation is a one-time activity that ends with the successful introduction of a new product,

a (marginally) higher patentability standard would discourage R&D in the risky direction by

making it harder to obtain a patent and the rents associated with it through this direction, which

we shall call the threshold effect, whereas it would have no impact on R&D in the safe direction,

provided that there are no (dis)economies of scope in R&D and that the return to a successful

patentable discovery in one direction is not diminished by that in the other. In this static setting,

a higher Sreduces industry R&D through the threshold effect, and it also allocates relatively

more resources in the safe direction than in the risky direction, which can reduce the expected

size of innovation if a successful innovation through the risky direction has a higher expected

quality improvement than that through the safe direction.

The issue is more complex if innovations are cumulative, as we assume in this article. Specif-

ically, we consider the following model that builds on and extends Hunt (2004) by having two

research directions: Suppose that n+1 firms have entered an industry. At any time, one of

them is the leader and the other nfirms are challengers. The challengers are in a patent race

to develop a new product that improves upon the current leaders. When a challenger succeeds

in developing a patentable innovation, it becomes the new leader to replace the current one,

who then joins the rank of challengers, and this process repeats itself indefinitely. In this dy-

namic setting, a marginal increase in the patentability standard will increase the value of being

a leader because it will take longer before the leader is replaced by a successful challenger. This

incumbency-prolonging effect can potentially increase the incentive for R&D in both innova-

tion directions, even though the threshold effect from a higher Swill still have a negative impact

on the incentive for R&D in the risky direction.4

Moreover, the changes in the R&D incentives in the two different directions will interact

with each other, giving rise to a dynamic strategic substitution effect between the two directions:

When the R&D intensity in one direction becomes higher (or lower), it exerts an opposite

force on the R&D intensity in the other direction. In particular, an increase of R&D in one

direction induces the next innovation discovery to come sooner, which lowers the profit from

incumbency and thus reduces the incentive for R&D in the other direction. This turns out to

be the crucial force that leads to new effects of patentability standards under multiple research

directions.

Finally, as we shall assume, a firm needs to incur a fixed cost to enter the market to conduct

R&D and innovate. Therefore, patentability standards, by impacting the expected return to

R&D in each direction, also affect the number of entrants in the free entry equilibrium. Our

analysis will examine how this market-structure effect interacts with the other forces in the

model.

We find that as patentability standards rise, R&D intensity in the risky direction first rises

and then falls, exhibiting an inverted-U shape, whereas R&D intensity in the safe direction is

U shaped, initially decreasing and then increasing. Thus, the incumbency-prolonging effect is

the dominating force in the risky direction when Sis low, but it is dominated by the negative

threshold effect when Sis high. More surprising is that, despite the positive impact from the

incumbency-prolonging effect, increases in Sinitially lower R&D in the safe direction due to

the strategic substitution effect.

We also find that as Sincreases, the industry rate of innovation initially goes up and eventually

falls down, reaching its maximum at some intermediate level. The market-structure effect plays

a balancing role: There will be more firms when the expected return from R&D investment is

higher, which moderates the effects of patentability standards on R&D intensities both for each

firm and the industry.

4It has been found in the literature that innovation and competition may have a nonmonotonic relationship because,

while more intense competition may lower rents from a single innovation, it could increase innovation incentives

under cumulative innovation due to the dynamic effect and the desire to “escape competition” (Aghion et al., 2005).

The incumbency-prolonging effect is also due to the dynamic effect, but it works through the channel of patentability

standards.