An economic model of patent exhaustion
Published date | 01 October 2020 |
Author | Ted Sichelman,Olena Ivus,Edwin L.‐C. Lai |
DOI | http://doi.org/10.1111/jems.12393 |
Date | 01 October 2020 |
J Econ Manage Strat. 2020;29:816–833.wileyonlinelibrary.com/journal/jems816
|
© 2020 Wiley Periodicals LLC
Received: 5 November 2019
|
Revised: 24 June 2020
|
Accepted: 28 June 2020
DOI: 10.1111/jems.12393
ORIGINAL ARTICLE
An economic model of patent exhaustion
Olena Ivus
1
|Edwin L.‐C. Lai
2
|Ted Sichelman
3
1
Smith School of Business, Queen's
University, Kingston, Ontario, Canada
2
Department of Economics, Hong Kong
University of Science and Technology,
Kowloon, Hong Kong
3
School of Law, University of San Diego,
San Diego, California
Correspondence
Olena Ivus, Smith School of Business,
Queen's University, 143 Union Street,
Kingston, ON K7L 3N6, Canada.
Email: olena.ivus@queensu.ca
Abstract
The doctrine of patent exhaustion implies that the authorized sale of patented
goods “exhausts”the patent rights in the goods sold and precludes additional
license fees from downstream buyers. Courts have considered absolute ex-
haustion, in which the patent owner forfeits all rights upon an authorized
sale, and presumptive exhaustion, in which the patent owner may opt‐out of
exhaustion via contract. This paper offers the first economic model of do-
mestic patent exhaustion that incorporates transaction costs in licensing
downstream buyers and considers how the shift from absolute to presumptive
exhaustion affects social welfare. We show that when transaction costs are
high, the patent owner has no incentive to individually license downstream
users, and absolute and presumptive exhaustion regimes are equivalent. But
when transaction costs are at the intermediate level, the patent owner en-
gages in mixed licensing, individually licensing high‐valuation buyers and
uniformly licensing low‐valuation buyers. Presumptive exhaustion is socially
optimalwhensocialbenefitsfrombuyer‐specific pricing outweigh social
costs from transaction cost frictions in individualized licensing, which
requires sufficiently low transaction costs.
1|INTRODUCTION
Within the United States, the judge‐made doctrine of “patent exhaustion”(sometimes termed the “first‐sale”doctrine)
implies that authorized sales of patented goods—for example, by a manufacturer directly authorized by the patent
owner—“exhausts”any rights of the upstream patent owner to seek payment from downstream buyers. However, in
some instances, upstream patent owners attempt to contractually restrict downstream buyers—for example, prohibiting
resale of the good—to preserve their right to collect additional license fees. Whether and under what circumstances
such downstream contractual restrictions can overcome the doctrine of patent exhaustion has been the subject of many
conflicting, and often vague, judicial decisions over the past century. The legal literature has qualitatively examined the
arguments for and against absolute patent exhaustion, but the qualitative literature provides no clear framework to
resolve the issue of whether absolute exhaustion increases or decreases overall welfare. In the economics literature,
several scholars have examined the exhaustion doctrine in the international context but surprisingly, very few have
advanced a formal model of domestic patent exhaustion.
This paper offers the first economic model of domestic patent exhaustion that incorporates transaction costs in
buyer‐specific patent licensing, and also allows for two types of licensing of buyers by the patent owner—individualized
licensing (buyer‐specific) directly from the patent owner and uniform licensing via the manufacturer. In this context, we
examine how a shift in domestic patent policy from absolute to presumptive exhaustion (which presumes exhaustion
but allows the patent owner to opt‐out via contract) affects social welfare.
1
Our model is motivated by a transactional structure common in the high‐technology industries addressed by the
U.S. Supreme Court in the patent exhaustion case of Quanta Computer, Inc. v. LG Electronics, Inc. (2008), which we
discuss further below. In our stylized model of this transactional structure, there exists one patented component of a
complex product. In order for a manufacturer to produce the good, it must enter into a license with the patent owner.
We assume the patent provides market power to the patent owner, which in turn selects one of many competitive
manufacturers to produce and sell the good. The patent owner thus licenses the component to only one manufacturer,
making the manufacturer a monopolist in the market for the good.
2
Buyers differ in their willingness to pay for the
good. Note that “buyers”are often themselves downstream manufacturers of their own complex product.
3
The price a
buyer is charged depends on the regime of patent exhaustion.
Under absolute exhaustion (Regime AE), when the manufacturer sells the good to the buyer, the patent owner loses
all rights to proceed against the buyer for patent infringement. Under presumptive exhaustion (Regime PE), the patent
owner may opt out of exhaustion by imposing contractual restrictions on the buyer, in which case the buyer must enter
into a license with the patent owner or risk patent infringement. We allow for two types of licensing of buyers in
Regime PE—individualized and uniform—and assume that a license from the patent owner (of either type) provides
the buyer with a right to use the good but not to resell it.
4
Precluding resale in Regime PE implies arbitrage will not
occur, and the patent owner and manufacturer can price discriminate between buyers in the first degree. Regime AE, by
contrast, prevents contractual restrictions on buyers and as such, allows resale; consequently, pricing is uniform.
5
Importantly, we allow for a positive transaction cost in individualized licensing, which increases the unit cost of
producing the final good for those who are licensed individually. This transaction cost is critical in determining the
social welfare implications of patent exhaustion. It includes the time and effort involved in negotiating license fees and
executing separate licensing agreements with individual buyers.
6
The size of the transaction cost affects the patent owner's optimal licensing scheme in Regime PE. When the
transaction cost is positive but not excessively high, the patent owner will engage in mixed licensing, individually
licensing high‐valuation buyers and uniformly licensing low‐valuation buyers. Mixed licensing features a tradeoff
between social benefits in the form of price discrimination and social costs generated by transaction cost frictions. Two
insights, not recognized in the previous literature, arise. First, at intermediate levels of transaction costs, the patent
owner does not fully internalize the social costs of licensing in its single‐minded, private goal of rent maximization.
Thus, when allowed to do so in Regime PE, the patent owner will succumb to the classic economic problem of rent
dissipation, willing to burn up essentially all of a buyer's individual surplus to gain a penny of producer surplus.
Secondly, at sufficiently low levels of transaction costs, the patent owner's private incentive to maximize rents may lead
to net social gains. Specifically, the ability of the patent owner to price discriminate implies that more buyers may gain
access to the patented good than in Regime AE. In other words, while Regime PE burns up surplus in the transaction
costs when the patent owner engages in individualized licensing, these costs may be worth the candle if enough
additional producer surplus is generated and enough additional buyers gain access to the good. Our model shows that
when transaction cost frictions are sufficiently low, presumptive exhaustion is optimal.
We briefly discuss extensions to our model in Section 6and more formally discuss how to relax some of the
assumptions in a Supporting Information Online Appendix.
7
In particular, we note that dynamic gains in promoting ex
ante investment in product quality may outweigh any loss of static efficiency in Regime PE when the transaction cost is
high. Regime PE is also more likely to dominate (in static terms) when licensing is via heterogeneous intermediaries
(e.g., wholesalers, resellers), with each intermediary negotiating the license fees on behalf of a large number of buyers.
But when there are information frictions, which act as an increase in transaction costs—such as information asym-
metry in assessing the buyer's demand and search costs—Regime AE is more likely to dominate. Our results also hold
qualitatively when we allow for a positive transaction cost in manufacturer licensing or substitutes for the patented
component, consider goods comprising multiple patented components, or allow buyers to resell used patented products
on a secondary market.
In the economics literature, Maskus and Chen (2004), Valletti (2006), Valletti and Szymanski (2006), Grossman and
Lai (2008), and Saggi (2013,2014) are valuable contributions that offer models of patent exhaustion. But these models
are limited to parallel trade among countries and draw a distinction between the regimes of national and international
exhaustion, under both of which the exhaustion is absolute. This is in contrast to recent legal decisions and case
comments, which focused on the distinction between absolute and presumptive domestic exhaustion. Furthermore, the
effects of international exhaustion cannot be easily adapted to the domestic context.
8
Only one paper, Layne‐Farrar, Llobet, and Padilla (2014), formally models domestic patent exhaustion. In that
model, a patent owner can grant a license to an upstream manufacturer and, if absolute patent exhaustion does not
IVUS ET AL.
|
817
To continue reading
Request your trial