Patent Thickets and the Market for Ideas: Evidence from Settlement of Patent Disputes

1
Patent Thickets and the Market for Ideas
Evidence from Settlement of Patent Disputes

• Alberto Galasso
• University of Toronto
• Mark Schankerman
• London School of Economics and University of
  Arizona
• Kauffman Summer Legal Institute (Dana Point,
  California)
• 10-13 July 2008

2
Motivation

• The market for ideas (patent licensing and sale
  of patents) is an important determinant of the
  rate of diffusion of new technologies. Also it is
  central to innovation incentives for small
  entrepreneurial firms, where patents are the key
  asset.
• We study the market for ideas through a new
  lens the settlement of patent infringement
  disputes.
• An effective market for ideas requires that
  rapid settlement of disputes. Delays in the
  settlement process create
• higher transaction costs
• slower diffusion of patented technology

3
Motivation (2)

• We focus on a key aspect of the patent
  environment fragmentation of patent rights
  (patent thickets').
• We investigate how fragmentation of patent
  rights affects the duration of patent disputes
  and, thereby, the speed of transactions in the
  market for ideas.

How does fragmentation affect the length of
settlement negotiations?
vs
4
Fragmentation and Negotiation Delay

• Two opposite views among legal scholars
• Mainstream Heller and Eisenberg (1998),
  Eisenberg (2001) greater fragmentation of IP
  rights creates higher transaction costs, longer
  delay and higher risk of bargaining failures.
• A dissenting voice Lichtman (2006) greater
  fragmentation of IP rights lowers the value at
  stake in each negotiation, and thus mutes the
  incentive of each claimant (patentee) to bargain
  aggressively. This speeds up settlement.

5
Intellectual Property

• As in Lerner and Tirole (2004), we focus on a
  firm with an innovation needing technology
  embodied in n patents
• symmetrical in importance
• each owned by a different patentee
• Users gross surplus from using n patents is V.
  Surplus from using mltn patents is

with
Perfect complements ?0 Perfect substitutes
?n/m
6
Intellectual Property (2)

• Value at stake in the nth negotiation
• Greater fragmentation is represented by larger
  n
• Greater complementarity is represented by larger
  ?

7
Litigation
Litigation between a patentee and an alleged
infringer (both risk neutral). Infringer has
private information she can estimate the
likelihood of the patentee prevailing at trial.
This probability has a (uniform) distribution.
We analyze a simple two-period settlement
bargaining game. In period 0 the patentee makes
a take it or leave it settlement offer to the
infringer. If rejected, a trial takes place in
period 1. If a trial takes place, both parties
sustain legal costs. Damages awarded are equal
to the value at stake z (unjust enrichment
doctrine generalizes for lost profit/reasonable
royalty doctrine)
8
Court of Appeal Federal Circuit

• We model the CAFC as reducing the uncertainty on
  trial outcome and damages. We introduce a simple
  form of pro-patent bias
• a fraction of courts always awards full
  damages
• the remaining fraction assesses correctly the
  probability of infringement
• The pro-patent bias of CAFC is not what matters
  here. Any bias that reduces uncertainty is
  equivalent for this argument.

9
Testable Predictions
H1 Settlement negotiations will be shorter for
infringers requiring access to fragmented patent
rights H2 Settlement negotiations will last
longer when patents have fewer substitutes
H3 Settlement negotiations will be shorter
for patents litigated after CAFC H4
Fragmentation of patent rights will reduce delay
by less after the introduction of CAFC (when
appellate uncertainty is reduced)
10
Data

• Patent litigation essentially all patents
  litigated from 1975-2000 (details in Lanjouw and
  Schankerman, 2001, 2004)
• NBER patent dataset USPTO for patent number,
  citations and assignees of litigated patents
• Focus on infringement cases. Manually matched
  infringer names with assignee names in NBER
  dataset to back-out patenting activity of
  infringers
• Final dataset 4489 cases with info on settlement
  duration, patent litigated, patentee and
  infringer

11
Data (2)

• Dependent Variable Dispute duration -- number
  of months between case filing date and case
  termination date.

12
Data (3)

• Fragmentation
• we take infringers patents in the technological
  sub-category of the litigated patent with
  application in a 5-year window from the suit
• we identify the share of citations in each
  n-class
• we construct a weighted average of class
  concentration (share of top-4 patentees)

13
Additional Controls

• CAFC dummy variable for cases litigated after
  1982
• Complementarity citations within a class/total
  cites received
• Patent Value number (self and non-self)
  citations received
• Tech-Field Dummies Drugs, Other Health,
  Chemical, Electronics (excluding computers),
  Mechanical, Computers, Biotechnology and
  Miscellaneous
• Full set of District Court Dummies
• Patent Age

14
Findings

• A standard deviation increase in fragmentation
  reduces dispute duration by
• almost 4 months before CAFC
• 15 days after CAFC
• On average, CAFC reduced disputed duration by 7.8
  months
• The impact of CAFC is twice as large for district
  courts that had the most uncertainty about patent
  validity before CAFC was introduced (this
  confirms that the CAFC effect is not just picking
  up some common third factor)

15
Aggregate Delay

• Fragmentation reduces per dispute settlement
  time. If negotiations with different licensors
  are concurrent, expected total delay also
  declines. If they are sequential, total delay can
  rise or fall with fragmentation, as there are now
  more licensors.
• This helps reconciles the two opposite views on
  patent thickets in legal literature Heller and
  Eisenberg, and Lichtman.
• Calculations show that total delay falls with
  fragmentation in some fields but not others,
  before CAFC. But not after CAFC.

16
Conclusion

• Fragmentation of IP rights reduces per-dispute
  settlement delay
• CAFC had large effect in reducing settlement
  delay. But it also reduced the beneficial effect
  of fragmentation.
• Theoretically the impact of fragmentation on
  total negotiation time is ambiguous. On our
  evidence, the anti-commons view of Heller and
  Eisenberg (2001) may be overly pessimistic.
• The impact of fragmentation on total delay
  depends critically on the timing/structure of
  negotiations. Need for legal rules and market
  intermediaries to reduce frictions. Also badly
  need case study evidence, including across
  technology fields.